portfolio risk and dependence modeling: application of factor copula models 1 the international journal of banking and finance, volume 9 (number 3) 2012: pages 1-14 portfolio risk and dependence modeling: application of factor and copula models arsalan azamighaimasi wuhan university of technology, china ____________________________________________________________ abstract we consider portfolio credit risk modeling with a focus on two approaches, the factor model, and the copula model. while other models have received greater scrutiny, both factor and cupola models have received little attention although these are appropriate for rating-based portfolio risk analysis. we review the two models with emphasis on the joint default probability. the copula function describes the dependence structure of a multivariate random variable. in this paper, it is used as a practical to simulation of generate portfolio with different copula, we only use gaussian and t–copula case. and we generate portfolio default distributions and study the sensitivity of commonly used risk measures with respect to the approach in modeling the dependence structure of the portfolio. key words: gaussian copula, factor model, copula model jel classification: c15, c38 _____________________________________________ 1. introduction there is a need to understand components of portfolio risk and their interaction. the basel committee for banking supervision in its basel proposed (bis, 2001) to develop an appropriate framework for a global financial regulation system. several portfolio credit risk models developed in the industry have been made public since then. examples are: creditmetrics (gupton et al., 1997), creditrisk+ (credit suisse financial products, 1997) and credit portfolio view (wilson 1997a; 1997b). others systems remain proprietary, such as kmv’s portfolio manager (kealhofer, 1996). although the models appear quite different on the surface, recent http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#basle01 2 theoretical work has shown an underlying mathematical equivalence among them (gordy, 2000; and koyluoglu and hickman, 1998). the credit portfolio models to obtain portfolio loss distributions, which are statistical models, can be classified as based on credit rating systems; see crouhy et al. (2001) for exact description and discussion of the various models. frey and mcneil (2001) study the mathematical properties of the models and consider the modeling of dependent defaults in large credit portfolios using latent variable models and mixture models. crouhy et al. (2000) compared and reviewed models on benchmark portfolio using credit migration approach, the structural approach, the actuarial approach, and mckinsey approach. however, few studies have attempted to investigate aspects of portfolio risk based on rating-based credit risk models. gordy (2000) offered a comparative anatomy of two especially influential benchmarks for credit risk models, the risk metrics group's credit metrics and credit suisse financial product’s . kiesel et al. (1999) employ a mark-to-market model and stress the importance of stochastic changes in credit spreads associated with market values, an aspect also highlighted in hirtle et al. (2001). the aim of this paper is to contribute to the understanding of the performance of ratingbased credit portfolio models, long ignored in the field. we apply a default-mode model to assess the effect of changing dependence structure within the portfolio. first, in the ensuing section, we discuss about the copula model as one of the dependency approaches within the portfolio. second, we describe a factor model by focusing on the effects of default dependence model within the portfolio. finally, in the penultimate section, we simulated types of copula model with different degree of freedom within the portfolio. 2. copula modelling an overview of basic copula uses in structural systems and models is provided in this section. copulas provide a natural way to study and measure dependence between random variables. suppose we have specified a portfolio of obligors, with default times . the variable international journal of banking and finance, vol. 9, iss. 3 [2012], art. 1 http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#croughy-galai-mark01 http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#frey-mcneil01 http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#cy-gi-mk99 http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#kl-pn-tr99 http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#hirtle01 http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#hirtle01 3 of default of obligor , at time t, is donated as . the probability space is . this space has filtration : (1) for the joint default probability at time t, evaluated at time 0, as (2) the survival property as (3) we take for granted the copula definition as a joint distribution function with uniform margins, which implies that and take for granted the fundamental sklar’s theorem, in terms of a copula and the marginal distribution functions : (4) the joint survival probability with survival copula, and the marginal survival functions : (5) factor copula is, (6) in the credit risk case, since the variables are default time, the copula represents default dependence. it is donated as , (7) (8) 4 according to merton model (1974) if default of firm occurs, the values of asset or values of shares cross from barrier line of outstanding debt at debt maturity. default is occurred when the firm’s asset value falls to the liability one, , the time of default is: (9) the default probability at time is, (10) the marginal default probability can be easily computed to be (11) then, (12) and is the instantaneous return on assets, which equates the riskless rate under the risk neutral measure. the joint default probability of assets is (13) where, is the distribution function of a standard normal vector with correlation matrix r. the marginal default probabilities is follows ) (14) to study the effect of different copula on default correlation, we use the following examples of copula (further details on these copula can be found in embrechts et al., 2001). (i) gaussian copula: (15) where, denotes the joint distribution function of the variety normal with linear correlation matrix ,and the inverse of the distribution function of the univariate standard normal. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 1 http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#es-lg-ml01 5 (ii) a student copula: (16) where is the standardized multivariate student’s distribution, with correlation matrix and degrees of freedom, while is the inverse of the corresponding margin. gumbel copula: (17) where . this class of copula is a sub-class of the class of archimedean copula. according to the table[1], joint default probabilities of two obligors are represented through three types of obligors with individual default probabilities corresponding to rating classes.as you will see that and gumbel copula have higher joint default probabilities than the gaussian copula. the joint default probabilities of two obligors are represented through three types of obligors with individual default probabilities corresponding to rating classes. table 1: copula and default probability copula default probability class a class b class c 6.89 3.38 52.45 46.55 7.88 71.03 134.80 15.35 97.96 gumbel 57.20 14.84 144.56 gumbel 270.60 41.84 283.67 6 3. factor modelling another popular approach to default modeling allows us to switch to the so called product copula. the reduction technique, which is widely adopted for the evaluation of losses in highdimensional portfolios, with hundreds of obligors (see for instance laurent and gregory (2003)), is the standard approach of (linear) factorization, or transformation into a bernoulli factor model. in the typical portfolio analysis the vector is embedded in a factor model, which allows for easy analysis of correlation, the typical measure of dependence. we assume that the underlying variables are driven by a vector of common factors. (18) where is dimensional normal vector, and is independent normally distributed random variables. here is obligor to factor , i.e. the so-called factor loading and is volatility of the risk contribution. the default indicators of the obligor are independent bernoulli variables, with probability: (19) where is cut-off point for default obligor . the individual default probabilities are, (20) and the joint default probability is, (21) if we denote by the correlation of the underlying latent variables and by the default correlation of obligors and , then we obtain the default correlation formula (22) international journal of banking and finance, vol. 9, iss. 3 [2012], art. 1 7 under assumption above, we obtain the joint default probability, (23) where is bivariate normal density with correlation coefficient . 4. simulation results of copula model here, we want to generate portfolios with given marginal and the above copula. we only use gauss and copula case . we looking for random sample generation for this mean we obtain the generation of an -variety normal with liner correlation matrix ,to take realizations from a gaussian copula we simply have to transform the marginal: • set • to generate random varieties from the –copula we assume the random vector x act the stochastic process (24) with where z and y are independent, and then x is distributed with mean and covariance matrix we assume , while the stochastic process is still valid the parameters has to change for . we will have algorithm (this is algorithm in embrechts et al. (2001)): • set • set • . http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#es-lg-ml01 8 we can replace the with in order to have multivariate distribution with –copula and normal marginal, to obtain the –copula . figure 1 shows three simulation results with 1000, 500, and 50 observations from a multivariate normal distribution. as you see the represents tree types of observations from a multivariate normal distribution with mean vector mu and covariance matrix. the figure 2 shows to computes a scatterplot of a normal sample and in a second plot the contour ellipses for mu =# (3, 2) and sigma = # (1,-1.5) ~# (-1.5, 4) with different observations. figure1: simulation results from samples of 1,000, 500 and 50 observations a=1000 b=500 c=50 figure 2: scatter plots of normal sample and second plot of the contour ellipses a=1000 b=500 c=50 international journal of banking and finance, vol. 9, iss. 3 [2012], art. 1 9 further analyses of the same data are plotted in figure 2. these are scatterplots of a normal sample. in an adjacent plot next to each sample, we present a second plot as the contour ellipses for mu =# (3, 2) and sigma = # (1,-1.5) ~# (-1.5, 4) with different size. 4.1 portfolio for our first simulation exercise, we assume that the underlying variables are normally distributed within a single factor framework, i.e. and in formula as follow: (25) they are constant and are chosen so that the correlation for the underlying latent variables is (kiesel et al., 1999. note that we use three rating classes, named a, b, and c with default probabilities 0.005, 0.05, and 0.15 roughly corresponding to default probabilities from standard rating classes (ong, 1999). to generate different degrees of tail correlation, we link the individual assets together using a gaussian, a and a -copula. the information in table 2, 3 and 4 represent the effect tail-dependence has on the high quintiles of highly-rated portfolios at different quintiles: table 2 is for 99 percentile. table2: effect of normal copula with default probability set at 0.005 portfolio copula mean variance a=1000 normal 0.115 0.13391 1 2 a=500 normal 0.106 0.119 1 1 a=50 normal 0.18 0.19143 1 2 b=1000 normal 0.99 1.8277 4 6 b=500 normal 1.038 1.8442 4 6 b=50 normal 1.18 2.3955 4 6 c=1000 normal 3.029 7.0953 8 11 c=500 normal 2.998 6.9078 8 11 c= 50 normal 3.1 7.3163 9 10 http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#kl-pn-tr99 http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode40.html#ong99 10 the copula is more than three-times larger than the corresponding quintile for the gaussian copula. the same effect can be observed for lower rated portfolios although not quite with a similar magnitude. table3: effect of with default probability 0.05 portfolio copula mean variance a=1000 0.101 0.26907 1 2 a=500 0.098 0.15671 1 2 a=50 0.14 0.36776 1 4 b=1000 0.963 2.38 4 6 b=500 0.994 2.1984 4 6 b=50 1.06 2.9147 4 9 c=1000 3.008 7.9799 9 11 c=500 3.05 7.9474 9 12 c=50 3.42 8.9016 9 11 we assume the second factor, i.e. in (4), for a sub-portfolio of 100 obligors increasing the correlation of the latent variables within the sub-portfolio to 0.5 table4: effect of with default probability 0.15 portfolio copula mean variance a=1000 0.088 0.39665 0 2 a=500 0.084 0.24543 0 2 a=50 0.22 2.42 0 11 b=1000 0.924 3.1454 5 9 b=500 1 3.0261 7 5 b=50 1.02 3.5302 4 11 c=1000 2.997 9.5860 10 12 c=500 3.028 9.0213 9 13 c=50 3.34 9.2086 9 12 international journal of banking and finance, vol. 9, iss. 3 [2012], art. 1 11 table 5: the effect of correlation cluster with default probability 0.005 portfolio copula first subportfolio second subportfolio mean variance a=1000 normal 100 150 1.237 6.8447 5 13 a=500 normal 50 75 0.6 1.6433 2 7 a=50 normal 20 30 0.24 0.47184 1 4 b=1000 normal 100 150 12.723 204.41 41 71 b=500 normal 50 75 6.198 47.951 20 33 b=50 normal 20 30 2.58 7.3506 10 11 c=1000 normal 100 150 37.972 871.43 96 132 c=500 normal 50 75 18.832 200.1 49 63 c=50 normal 20 30 7.74 30.36 20 23 table 6: the effect of correlation cluster with default probability 0.05 portfolio copula first subportfolio second subportfolio mean variance a=1000 100 150 1.451 27.335 7 28 a=500 50 75 0.644 6.7668 3 11 a=50 20 30 0.2 0.32653 1 3 b=1000 100 150 11.76 299.29 52 83 b=500 50 75 6.28 85.605 24 44 b=50 20 30 2.32 11.365 10 17 c=1000 100 150 38.24 1104.7 105 148 c=500 50 75 18.638 263.7 52 75 c=50 20 30 7.5 31.235 17 24 12 table 7: the effect of correlation cluster with default probability 0.15 portfolio copula first subportfolio second subportfolio mean variance a=1000 100 150 1.635 70.278 7 42 a=500 50 75 0.682 14.554 3 21 a=50 20 30 0.36 2.1943 1 10 b=1000 100 150 13.385 592.25 65 128 b=500 50 75 6.266 132.82 28 61 b=50 20 30 2.26 16.074 13 18 c=1000 100 150 38.465 1395 117 157 c=500 50 75 18.676 331.96 56 80 c=50 20 30 7.56 41.109 23 27 for this reasaning we want to shows the effects of increased correlation within parts of the portfolio; we change the factor loading within parts of our portfolio. these results are shown in tables 7, 9 and 10. as expected, the results in tables 5, 6, 7 show increase in the quantiles due to the increased correlation within the portfolio. however, comparing the three tables we will see that the sensitivity of the portfolio loss quantiles is higher with regard to the underlying copula than to the correlation within the portfolio. 5. conclusions to investigate the riskiness of credit-risky portfolios is one of the big challenging in financial mathematics. an important thing for a model of credit-risky portfolios is the dependence structure of the underlying obligors. we studied two approaches, a factor structure, and the direct specification of a copula. we generated portfolio default distributions and studied the sensitivity international journal of banking and finance, vol. 9, iss. 3 [2012], art. 1 http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xfghtmlnode39.html#cortab 13 of commonly used risk measures with respect to the approaches in modeling the dependence structure of the portfolio using as a rating-based approach using cupola mathematics. the simulation results indicate that the degree of tail dependence of the underlying copula plays a major role. that is identified as a credit risk. the copula modeling links the underlying variables together, which is of crucial importance especially for portfolios of highly-rated obligors. author information: arsalan azamighaimasi is a faculty member in the department of management, wuhan university of technology, china. he may be contacted at e-mail: arsalan.azami2011@gmail.com. references bis (2001), "overview of the new basel capital accord," basel committee on banking supervision. caouette, j., altman, e. and narayanan, p., (1998). "managing credit risk, the next great financial challenge," wiley frontiers in finance,vol. wiley frontiers in finance, wiley & sons, inc, new york. m. carey, (2000). "dimensions of credit risk and their relationship to economic capital requirements.," federal reserve board. m. gordy, (1997). "a comparative anatomy of credit risk models," journal of banking and finance, vol. 24, pp. 119-149, 2000j. morgan, "creditmetrics," jp morgan, new york. crouhy, m., galai, d., and mark, r., (2001). risk management, mcgraw hill. crouhy, m., galai, d. and mark, r, (2000). "a comparative analysis of current credit risk models," ournal of banking and finance, vols. 24(1-2), pp. 59-117. embrechts, p., lindskog, f. and mcneil, a, (2001). "modelling dependence with copulas and applications to risk management," eth zürich. frey, r. and mcneil, a, (2001). "modelling dependent defaults," eth zürich. p. jorion, (2000). value at risk, new york. : mcgraw-hill. kiesel, r., perraudin, w. and taylor, a, (1999). "the structure of credit risk," birkbeck college. koyluoglu, h., and hickmann, a., (1998). "a generalized framework for credit portfolio models," wyman & company, mailto:arsalan.azami2011@gmail.com 14 luciano, e., (2008)."copula-based default dependence modelling: where do we stand?," in nickell, p., perraudin, w. and varotto, s, "ratings-versus equity-based credit risk models: an empirical investigation," unpublished bank of england mimeo. m. ong, (1999). internal credit risk models. capital allocation and performance measurement, london: risk books. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 1 a note on 'what drives share prices in the middle east?' the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 113 a note on ‘what drives share prices in the middle east?’ panos priftakis and m. ishaq bhatti la trobe university, australia abstract there are several hypotheses suggesting that some properties of oil prices make it interesting to focus on the predictive ability of oil prices for stock returns. this paper reviews some models recently used in the literature and selects the most suitable one for measuring the relationships and/or linkages of oil prices to the stock markets of the selected five oil producing countries in the middle east. in particular, the paper uses two methodologies to test for the presence of a cointegrating relationship between the two variables and an unobserved-components model to find a relationship between the two variables. the results rejects convincingly that there is no linkage between the prices of oil and the stock market prices in these oil-based economies. keywords: oil prices, stock market, econometrics, unobserved-components model, cointegration jel classification: g15 1. introduction oil, as a major source of inputs for a variety of petrochemical products, plays an important role throughout most, if not all, economies around the world (happe, 1984, p. 3). the price of oil had more of an influence on the world economy as compared to the price of any commodity over the past 30 years (patel, 2003). at least it appears to be so from the volume of reports dedicated to linking oil prices to economic cycles. although there has been little done relating the relationship between oil price and the stock market index, it is yet uncertain, in fact unproven, if oil prices contribute to the movements of the stock markets in an important way. there are numerous studies on the effect of oil price changes on the economy. the main objective of this paper is to measure the relationship between oil price and various stock markets. to carry out this research, five oil dependent countries, kuwait, oman, united arab emirates (uae), qatar and saudi arabia were selected because of the significance oil plays in their gross domestic outputs. the monthly data were analysed during the period between 01:2000 and 05:2004. observing the stock markets, the gulf cooperation council (gcc) countries have brought the highest returns worldwide over the period between 2001 and 2005 ijbf 114 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 (see woertz, 2005). stock markets in countries such as the saudi arabia and the uae have had increased more than fivefold in this period (woertz, 2005). fadlallah (2005) found that the saudi tadawal index and the uae general index outperformed the nasdaq between january 2002 and september 2006. considering the importance of oil revenue in the middle eastern countries, and these extraordinary performances of the stock markets in the oil dependent countries, it was logical to presume that there might be a relationship between the price of oil and stock market price. this paper only examines the relationship between the stock prices of the five oil dependent countries and the oil price. the paper uses two methodologies to test for the presence of a cointegrating relationship between the two variables, and then undertakes an unobserved-components model to find a relationship between the two variables. using these tests and different methodologies, the paper has attempted to draw better conclusions regarding the effect, if any, oil prices may have on the stock markets in the five oil dependent countries. the rest of the paper is organized as follows. section 2 presents the relevant literature and methods used by other researchers. section 3 describes the models development for this study. section 4 provides the results and a discussion. the paper ends with a conclusion in section 5. 2. relevant literature and methods there have been several studies which have examined the relationship between stock prices and the oil price. these studies have examined the relationships through the use of various econometric methods ranging from the augmented dickey fuller test to a number of ols applications. the uses of different methods are examined in bley (2002), harris (1995), granger (1969) and sims (1972). bley (2002), with the use of the daily historic prices on gcc country market indices, tested for the presence of unit roots using the adf test. harris (1995) on the other hand estimated an error correction model to investigate whether each of the exchanges was contributing to price discovery. we consider the use of the adf test and the error correction model were undertaken by these authors. the model which is of great importance to the research is the unobservedcomponents model, which is built with the addition of explanatory variables as in diebold (1989). this model allows us to observe other unspecified factors, which have not been tested. the analysis of various papers, allowed us to understand existing results achieved by other researchers. the findings proved to be mixed. mauldin (2003) for example, found that a rise in oil prices suggested a fall in the stock market prices and a drop in oil prices is associated with a rise in stock prices. on the other hand, driespong et al. (2003), after undertaking various tests, concluded that the initial market reaction to the oil price change depended on whether the country was a net energy importer or exporter. they found that an oil price increase had a significantly positive impact on the norwegian economy. similarly, countries such as venezuela and canada, which are net energy exporters, all reacted positively to the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 115 oil price increases (driespong et al., 2003 and bhatti et al. 2006; chapter 7). these research indicated that there was usually a relationship between oil price and the stock market price. the comparison of various articles from different journals reveals some common factors. there has been little done to establish the relationship between oil price and the stock market index, and therefore not too many alternative approaches have been tried. several researchers used the data to apply differnt multiple regression models, but did not attempt anything new or different. the main difference therefore with this paper, is that we use a new method, the unobserved-components model, not used to date to address this issue. this we consider to be a likely contribution to the future studies on this topic area. the use multi-country data may have also account for the results we observed on the relationship between the oil price and the stock market index. 3. the model the augmented dickey fuller test is used for all stock markets and the oil price variables. the method includes the testing for the existence of a unit root in the univariate time series. when conducting the adf test, if the null hypothesis was rejected, it implied stationarity of the process meaning that any of the variables tested was stationary. in the situation where the null hypothesis was not rejected, a test for a second root was conducted by testing the null hypothesis h0: yt ~ i (2), which can be done by testing h1: yt ~ i (1) (vogelvang, 2005; p. 287). when the variable in this test was rejected, we assumed that the variable was non-stationary. to conclude the testing of the data, what was then included was long run models and error correction models (ecm), which allowed us to determine whether cointegration exists among the variables in the short run and the long run models tested. when the conditions for cointegration were not satisfied, we assumed that there was spurious regression, and that the two variables were not related. what this then caused was that the residuals were not stationary, e t ~ i (1), causing them to have a unit root. the final model that was completed for all five countries was an extended timeseries model by the addition of extra explanatory variables, called an unobservedcomponents model. this meant that the dependent variable, which was the various stock markets, was determined by its various components, while also including the oil price variable as the explanatory variables (moosa, 2006). the model was displayed as follows y t = μ t + ø t + γ t + ∑βx t + ε t (1) where x t was the vector of explanatory variables and b a vector of unknown coefficients. in this case, there was only one coefficient. the explanatory variable for all five countries was the oil price, x t . equation (1.1), contains the dependent 116 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 variable and explains that it depends on the components trend, cycle, seasonal component, random components and also the oil prices, x t (moosa, 2006). the achievement of the coefficient values for the variables was found using a program called stamp (structural time series analyser, modeller and predictor) enabling us to complete the process of fitting the unobserved-components models (harvey, 1989; p. 15). the possible outcomes of the equation were as follows: if the explanatory variables were only partially successful in explaining the dependent variable, so if any of the level, slope, cycle or seasonal components were significant, then there was a role for the components. this demonstrated that there may be an effect of other economic indicators which are not included (moosa, 2006). in completing the model, if the outcome was that the explanatory variables (trend, cycle, seasonal, oil prices) were successful in explaining the dependent variable, and the oil prices were also found to be significant, the conclusion was that oil prices play an important role in affecting the stock market prices, though there will still be other unspecified variables still unspecified that are contributing an important role. the final possible outcome was the situation where the oil price is found solely to explain the dependent variable. this occurred when the other explanatory variables were insignificant, thus leading to the conclusion that oil prices alone will explain the movements of the stock prices (moosa, 2006; p. 11). this outcome allowed us to conclude that oil prices are of great importance in the movements of the stock prices. 4. results and discussion the results achieved for all the five oil dependent countries were found to be very similar, apart from some small differences. the adf unit root test was firstly conducted to determine the order of integration. the conclusion from the test was that all the five stock markets and the oil prices were integrated of order one i (1), displayed in table 1. the conclusion is that all variables were found to be nonstationary. the integration of order one i (1) allowed for various cointegration tests to be conducted on the relationships between each country’s two variables in the model. examining the long-run relationships between the stock market prices and the oil prices provided mixed results. the residuals of each equation, displayed in appendix 1, table 1, showed that cointegration existed in the long run for the relationships between the uae stock market and the oil price and for the saudi arabian stock market and the oil price. the remaining country stock markets, which included oman, kuwait and qatar, were not cointegrated with oil prices. thus three of the five markets exhibited behavior that led us to conclude that there is relationship between oil prices and stock market prices: only for two countries there is a relationship. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 117 table 1.1: adf tests for stationarity an error-correction model was then tested to examine as to whether any shortrun relationships existed: the results are in appendix table 2. the conclusion is that the uae stock market index was cointegrated in the short-run model, leading to the suggestion that oil prices may have been important in the short run changes to market prices. for others, oman, kuwait, qatar and saudi arabia, the conditions for cointegration were not satisfied, allowing for spurious regression amongst the oil prices and the respective stock markets.1 the implementation of an unobserved-components model provided more detailed results on the relationship of the oil price and the stock market. the first three countries analysed were kuwait (table 2), oman (table 3) and qatar (table 1 the findings of the granger causality test (appendix 1, table 1.3) found that oil prices do not “granger cause” the stock market, while on the other hand, causality in the opposite direction was statistically supported. this was not taken into consideration, since in the real world we would not expect the stock market to cause the oil price movements. no. of adf adf first conclusion country variables lags levels difference at the 5% level uae stock market index wti oil prices 12 -2.04 -1.79 -5.27 -5.63 1 (1) 1 (1) kuwait stock market index wti oil prices 12 -2.67 -1.79 -5.64 -5.63 1 (1) 1 (1) oman stock market index wti oil prices 12 -0.94 -1.79 -6.69 -5.63 1 (1) 1 (1) qatar stock market index wti oil prices 12 -2.88 -1.79 -6.88 -5.63 1 (1) 1 (1) saudi arabia stock market index wti oil prices 12 0.20 -1.79 -5.43 -5.63 1 (1) 1 (1) * the results include both a constant and a trend. in level, intercept and no intercept or trend, where all were not rejected. once completed in first difference, the constant and a trend were immediately rejected. 118 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 4). the coefficients of the unobserved-components models for the three countries provided evidence that there was no apparent significant relationship between oil prices and stock prices. for each model it was found that the level, slope or the seasonal components of the models were significant, providing an indication that for all three countries there were other factors that were affecting the stock market index values. table 2: kuwait the analysis of the explanatory variable which was the oil price for all three countries led us to conclude that in all cases it was insignificant, putting forward our major conclusion that oil prices did not play an important role in the movement of the stock prices in the five middle eastern countries. the outcome that there was no cointegration between the variables of these countries also emphasized the outcome obtained. this did not mean that oil prices had absolutely no affect, as mentioned earlier. rather discussions about oil prices in the countries may have changed the minds of investors before there were any movements in the oil prices. factors such as these would not have appeared in the results, since the changes may have already occurred. the completion of an unobserved-component models on the uae (table 5) and saudi arabia (table 6) provided similar results, though with different conclusions. the coefficients of the unobserved-components models for the two countries again provided evidence that there was no apparent significant relationship between oil prices and stock prices, whereas the oil price variable in both cases appear to be the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 119 insignificant, allowing us to assume, that oil price did not play an important role in the movement of the stock prices in the other two middle eastern countries as well. table 3: oman table 4: qatar 120 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 table 5: uae table 6: saudi arabia the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 121 the results from the cointegration suggested that the uae stock market is cointegrated in the short-run model, and that cointegration existed in the long-run relationships for both the uae and saudi arabia meant that the two models may be inaccurate. in other words, a relationship may exist between oil price and the stock prices for these two countries. 5. conclusion this paper provided a decent test of a relationship between oil prices and the stock market prices. the findings from the unobserved-components model enabled us to reach a more valid conclusion that there appears to be no relationship between the two variables in the five markets. thus, this paper makes a useful contribution to the literature, since it proves that maybe the effect of oil prices is not the most important influence on the movements of the stock market indices in the five oil dependent countries, something not considered in the literature. author statement: panos priftakis is the submitting author. he is a postgraduate student at the la trobe university. m. ishaq bhatti is an associate professor in the same institution. the authors express their thanks to imad moosa, anonymous referees, and the editor of the journal for their constructive comments which helped improve the final version of this article. references bhatti, m. i., al-shanfari, h., and hossain, m. z. (2006), econometrics analysis of model selection and moel testing, ashgate, england, uk. bley, j, chen, h, k. (2002). gulf cooperation council (gcc) stock markets: the dawn of a new era, american university of sharjah, uae. driespong, g, jacobsen, b, maat, b., (2003). striking oil: another puzzle? rotterdam school of management, working paper. diebold, f. x., (1989). structural time series analysis and modelling package: a review, software review, journal of applied econometrics 4: 195-204. fadlallah, t., (2005). the great arabian bubble, red alert, nomura, bahrain. freeman, j., (1983). granger causality and the time series analysis of political relationships, american journal of political science 27 (2): 327-358. fuller, w., (1976). introduction to statistical time series, wiley, new york, usa. granger, c.w.j., (1969). investigating causal relations by econometric models and cross-spectral methods, econometrica 37: 424-438. hanna, d., (2004). the uae’s economy continues to fly high, scb economic update. happe, l, n., (1984). south korea in the wake of the oil price rise, garland publishing, inc. new york and london, p 3. harris, f, mclnish, t, shoesmith, g, wood, r., (1995). cointegration, error correction, and price discovery on informationally linked security markets, journal of finance and quantitative analysis 30 (4): 563-579. 122 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 harvey, a., (1989). forecasting, structural time series models and the kalman filter, redwood press limited, great britain. hong, h, stein, j., (1999). differences of opinion, rational arbitrage and market crashes, nber working papers, national bureau of economic research. jones, m, and kaul, g., (1996). oil and the stock markets, the journal of finance 51(2): 463-491. lesage, j., (1990). a comparison of the forecasting ability of error correction models and var models, the review of economics and statistics 72 (4). macdonald, g, a., (2000). critical values for unit root and cointegration test statistics the use of response surface equations, economics department, curtin university, perth, wa. 2000. mauldin, j., (2003). oil, stock prices and the dollar, editorial, usa. moosa, i., (2006). structural time series modelling, applications in economics and finance, the icfai university press, india, 2006. patel, k., (2003). the “crude” economics of oil, crsp research, usa. sims, c, a., (1972). money, income and causality, the american economic review 62: 540-552 woertz, e., (2005).gcc stock markets at risk, gulf research center, dubai, united arab emirates. vogelvang, b., (2005). econometrics, theory and applications with eviews, pearson education limited, u.k. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 123 appendix 1: table 1: adf tests of the residuals * there is a trend and intercept ** there is only an intercept ***no intercept or trend the df statistic that will be used will be -3.50, the table of b.9 in fuller (1976) based on fuller was used. table 2: error correction model variables conclusion at in the no. of adf the 5% level country equation lags levels against df stat residuals uae stock market index and wti oil prices 12 -3.83* the null hypothesis of no cointegration is rejected residuals kuwait stock market index and wti oil prices 12 -2.19*** the null hypothesis of no cointegration is not rejected residuals oman stock market index and wti oil prices 12 -2.87*** the null hypothesis of no cointegration is rejected residuals qatar stock market index and wti oil prices 12 -3.31* the null hypothesis of no cointegration is rejected residuals saudi arabia stock market index and wti oil prices 12 -3.70* the null hypothesis of no cointegration is rejected 124 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 113-124 table 3: granger causality test international journal of banking and finance 3-1-2008 a note on 'what drives share prices in the middle east?' panos priftakis m. ishaq bhatti recommended citation day-of-the-week effect and volatility in stock returns: evidence from east asian financial markets the international journal of banking and finance 3-4 (special issue): 2005-2006: 153-164 153 ijbf day-of-the-week effect and volatility in stock returns: evidence from east asian financial markets chiaku chukwuogor-ndu eastern connecticut state university, united states of america abstract the presence of the day-of-the-week effect has been documented in finance literature. this paper investigates the presence of the day-of-the-week effect and return volatility in ten east-asian financial markets in the post asian financial crisis period, after 1998. a set of parametric and non-parametric tests is used to test the equality of mean returns and standard deviations of returns. the results indicate the presence of the day-of-the-week effect and insignificant daily returns volatility in most markets. some of these results reinforce some previously documented evidence and others are at variance with published results for the same markets. this effect, unlike in devloped markets, is still persistent. jel classification: g14 & g15 key words: returns, volatility, standard deviation, anomalies, day-of-the-week effect, kurtosis, skewness. 1. introduction and anomaly literature the objective of this paper is to examine the day-of-the-week stock return patterns and determine the volatility of returns of ten east asian stock markets to provide empirical evidence in the post-1997 asian financial crisis period. the presence of anomalies in returns of common stocks reported for developed markets has intrigued finance researchers since the 1980s. there is need to examine this phenomenon in asia’s emerging capital markets (especially after the crisis), since all countries in asia have introduced pro-market reforms and strengthened the institutional structures of the market pricing processes. the results reported for most developed markets have led to further challenges to the appropriateness of the efficient market hypotheses (emh). 1 the international journal of banking and finance, vol. 5.(number 1): 2007: 153-164154 the day-of-the-week effect in share returns in the us has been well documented. the mean return on mondays is negative, and is significantly lower than the mean return on any other day of the week. the mean returns are significantly higher on fridays relative to other days (see fama (1965), gibbons et al. (1981), lakonishok et al. (1982), keim et al. (1984), and mehdian and perry (2001)). jaffee et al. (1985a; b) found a negative monday effect in canada and the u.k., but a negative tuesday effect in japan and australia. condoyanni et al. (1987) confirms these findings on the japanese and australian markets. kato (1990) also found that the tuesday return is negative and wednesday and saturday returns are strongly positive in japan. jaff et al. (1989) wrote a significant paper providing international evidence. colorful phrases have crept into the literature. bad-friday effect, which refers to a decline of the market on fridays, usually precedes monday with increased stock selling pressure. turning now to asian markets, the-day-of-the-week effect on stock has also been evidenced in several asian stock markets before the asian financial crisis, 1997. these markets were characterized as having very high returns, high volatility in returns, and high illiquidity. east asian markets in the post-crisis years still remains of great interest because of high economic growth rates that lead to high market returns. wong et al. (1986) reported evidence of day-of-the-week and seasonal effect in the singapore market. the pattern was similar to that of the us market. one difference noted is that, in japan and australia, mean return is negative on mondays and the highest positive return occurs on fridays. the mean daily return in january is higher than that for other months, although the end-of-year effect is not significant at all. aggarwal et al. (1989) noted day-of-the-week effect in hong kong, korea, taiwan, japan, and singapore, while ho’s (1990) paper finds strong, seasonality effect – an evidence against the emh in ten asia-pacific markets, further confirming the day-of-the-week effect in singapore, malaysia, hong kong, and thailand. chen, kwok, and rui (2001) reported dayof-the-week effect in china, showing negative tuesdays after 1995 and highlighting that this anomaly disappears once non-normality and spill-over from other countries is taken into account. the rest of the paper is organized into four more sections. the following section is an introduction to the ten markets, selected for this study. in section 3 the reader may find a description of the data, variables, and the measurement methods. the results are presented in section 4 and the paper is concluded as section 5. 2. overview of the markets included most of the documented studies on the day-of-the week-effect on returns and returns volatility in eafm were done before july 1997, the time of the asian financial crisis. the choice of post-crisis period is appropriate for two important 155the international journal of banking and finance, vol. 5.(number 1): 2007: 153-164 reasons. the crisis was perpetrated by the financial fragility of south korean and thai economies, which first led to the sudden collapse of their two currencies in the first two weeks of july 1997. thereafter, the contagion of currency decline spread to other neighboring trading partner countries, resulting in both political and economic changes of far-reaching magnitude. the south korean won depreciated by 88.40 percent, and the thai baht depreciated by 64 percent. with that magnitude of currency collapses in the two and other countries, governments had to call the imf for restructuring their economies with massive bailout loans. imf initially approved us $35 billion financial support for reform programs in indonesia, south korea, and thailand. later, the amount ballooned to us $77 billion of additional financing (see imf (1999)). the gross domestic product growth rates of asian countries dropped from very close to double-digit, to negative territory. singapore, south korea, thailand, and taiwan registered negative growth rates in 1998. after the recovery brought on through imf bailout, reforms, and restructuring, several asian countries achieved very high gdp growth rates (see figure 1). figure 1:economic growth rates in the ten selected markets: 1990-2002 source: this figure is constructed from data from imf international financial statistics, 2003, cd-rom, at the university of connecticut, u.s. as at june 2004. the international journal of banking and finance, vol. 5.(number 1): 2007: 153-164156 the post-crisis recovery phase is from 1999, whereas the charts show the pre-crisis growth rates as well. except taiwan, growth rates were around 9 or more percent per year in the years before 1996. but when the crisis occurred in 1998, growth rates declined suddenly to a region below the 10 percent level, and in three cases, below 0. even though figure 1 shows that indonesia achieved a very high gdp growth rate in 1998, (probably due to the gross inflow of funds from multilateral and bilateral sources), the situation in all the other east asian countries was generally very poor. in addition to the severe effects, the crisis put pressure on emerging markets outside the region contributing to a virulent contagion and volatility increases in international financial markets. the impact was more pronounced on the asian economies because of the traditional trade links that exists among them. prior to the crisis, these financial markets were characterized as having yet completed prudential market regulation that would be required to secure stability. following the severe impact of the crisis, and under the tutelage of the imf, and other international bodies, necessary reforms were put in place, and these reforms not only improved the economic performance, but also contributed to the capital markets improved liquidity and prudential management. if that is true, it is a worthwhile effort to motivate this study to see if the post-crisis behavior is different from that documented for an earlier era. 3. data and methodology to do this, the daily closing prices of ten east asian stock market indices were identified for scrutiny. parametric and non-parametric tests were used to test the equality of mean returns across the-day-of-the-week and the equality of the standard deviations across the-day-of-the week. we use the daily closing values market indices from january 1998 to october 31, 2003. the period covered 5 years and ten months. the financial market indices included are: • china’s shanghai composite (ssec) • india’s bse 30 (bsesn) • indonesia’s jakarta composite (jk) • japan’s nikkei 225 (n225) • malaysia’s klse composite (klse) • philippines’ pse composite (psi) • singapore’s straits times (sti) • south korea’s seoul composite (ksii) • thailand’s set (seti) for thailand • taiwan’s taiwan weighted (twii) 157the international journal of banking and finance, vol. 5.(number 1): 2007: 153-164 the daily stock returns for the east asian stock indices are calculated as follows: ln (p t /p t-1 ) * 100 (1) where p t : the stock market index value at date t and t-1. the ratio multiplied by 100 gives the cent returns for each day for each market. except for the returns on monday , we excluded any returns that are preceded by a holiday. this exclusion was done to avoid the speculation that observed day-of-the-week-effect could be partially due to these non-trading days. the returns are then calculated for each day of the week. a similar procedure is applied to compute the sample standard deviations of each day’s observations in each market. we used the standard deviations in the returns to measure volatility. then, we tested the hypothesis of equal variance as substantiated by non-parametric tests. to test the hypothesis of an equal mean return across all days of the week, we applied the non-parametric kruskal-wallis statistics: where: k: the number of observations in each of the samples, n j : the number of values in the jth sample, n : ∑n j = total number of values, r j : the sum of ranks in the sample when n values are ranked together. this statistic is approximately chi-square distributed with degrees of freedom equal to k-1. we employ the levene’s (1960) test to check the results on equality of variance. in measuring the variation within a class, levene’s test uses the average of the absolute deviations instead of the mean square of deviations. this avoidance of squaring makes the test criterion much less sensitive to nonnormal distributions (see snedecor et al. (1976)): where with rij is the return for week i for the weekday j for j =1, 2,…., j =5 if the last trading day of the week is a friday (it is the case in all countries) m = is the mean of the sample d = denotes the difference being tested. these tests are robust enough to yield reliable results on the comparisons of mean and standard deviations, without the results being biased due to the nonnormality of distributions. the international journal of banking and finance, vol. 5.(number 1): 2007: 153-164158 4. empirical results the results are grouped in two sub-sections. first to be presented are the results for the day-of-the-week mean returns in sub-section below. a. day-of-the-week effect table 1:daily mean and standard deviations of returns of 10 markets, 1998-2003 table 1 provides a summary of statistics on the mean returns for each day of the week and the standard deviations across the days of the week for the ten east asian financial markets. similar to the us case and the case of results relating to the pre-asian financial crisis asian case, seven of the ten east asian markets 159the international journal of banking and finance, vol. 5.(number 1): 2007: 153-164 show negative mean monday returns. the markets include china, indonesia, malaysia, japan, singapore, south korea, and thailand. the returns of indonesia, malaysia, and thailand are marginally significant at the 10% acceptance level. philippines, thailand, and taiwan show negative tuesday returns. in precrisis studies on day-of-the-week returns, taiwan showed positive returns on tuesday. consistent with past results, the stock market returns of philippines and taiwan are significantly negative at the 10% acceptance level. indonesia, japan, and the philippines show negative wednesday returns. china, japan, south korea, and taiwan show negative thursday returns. in pre-crisis studies, thursday returns in all asian financial markets were positive (see ho (1990)). all the east asian stock markets except those in japan and india have positive friday returns. but only a few were significantly positive: south korea and thailand at 10% acceptance level and malaysia at 5% acceptance level. table 2 summarizes the days in which the highest and lowest return or standard deviations occur. five of the stock markets record the lowest return on monday: china, malaysia, singapore, thailand, and indonesia. lowest returns also occur on tuesday for two of the stock markets, namely in the philippines and in taiwan. table 2: maximum, minimum and standard deviations of returns, 1998-2003 the international journal of banking and finance, vol. 5.(number 1): 2007: 153-164160 this is an interesting finding because in previous studies, the lowest return did not occur on tuesday for the philippines and taiwan. in fact, it is documented that except for new zealand, philippines, and taiwan, all asia-pacific stock markets had their lowest return on tuesday (ho, 1990: this previous study contained ten asian pacific stock markets and excluding indonesia, china, and india). in this study, japan and south korea recorded their lowest returns on thursday. china, malaysia, south korea, and taiwan show their highest returns on wednesday. highest returns also occurred on friday for singapore and thailand. india and philippines recorded their highest returns on monday. japan and indonesia recorded their highest return on tuesday and thursday, respectively. even though there is mixed results with respect to which day the highest return occurs, in general, high returns mostly occur on friday and wednesday. previous studies document a highest return on the last trading day of the week (ho (1990)). as the skewness and kurtosis statistics of the returns for each day are generally well above zero and 3 respectively, normality of the returns is rejected. this is confirmed by the results of the w-test for normality shown below in table 3. for all the stock markets, the null hypothesis that the sample is normally distributed is rejected, which justifies our use of the no-parametric tests for further analysis. to substantiate the evidence of the day-of-the-week effects shown in table 1, the kruskal-wallis test is carried out to test the null hypothesis of equality of mean returns across the days of the week. this would correct for the non-nortable 3:kruskal-wallis, w-test for normality and levene test, 1998-2003 note: ** indicates significant at 5% level. 161the international journal of banking and finance, vol. 5.(number 1): 2007: 153-164 mality problem invalidating the earlier results. as shown in table 3, the values of the _2-statistics are significant at 5% level for the markets except those in thailand, pakistan, and malaysia. these results support the existence of the day-of-the-week effect in stock returns in china, india, japan, philippines, singapore, south korea, and taiwan. the evidence leads to the conclusion that there is no day-of-the-week effect in thailand and malaysia. this result is similar to the previous findings of wong et al. (1992). hence, it may be suggested that, although the days on which the effects occur have changed over the recent period after the crisis, day-of-theweek effect is absent for two countries over the preand post-crisis period. for others, there is still the day-of-the-week effect significantly present in 8 markets. b. volatility of returns in general, there seems to be less volatility in the daily returns in these east asian markets since the asian financial crisis. the countries with the highest standard deviation (sd) are those of south korea (standard deviation of 1.2372), malaysia (1.0848), and taiwan (0.8971). these were all emerging economies severely affected by the asian financial crisis. table 1 examined earlier showed the standard deviation of returns. the highest sd of 0.8883 percent is in taiwan on tuesday, and the lowest standard deviation of 0.7263 occurs on thursday. the days on which the maximum and minimum occurs differ from previously reported findings. for example (ho and cheung, 1994) documented that the highest standard deviation was for taiwan occurring on monday and the lowest occurring on friday. in addition, there seems to be a more even distribution across the days of the week regarding the day in which the lowest standard deviations occur. japan, philippines, singapore, and thailand experience their lowest volatility on tuesday. indonesia and india experienced the lowest volatility on wednesday, while china and south korea experienced their lowest volatility on friday. china, japan, malaysia, philippines, singapore, south korea, and india all had the highest volatility on monday. this evidence is more consistent with past documentation regarding the fact that most of the asian countries show highest volatility on monday (see ho et al. (1994)). indonesia experienced the highest standard deviation on friday; taiwan on wednesday; and thailand on tuesday. in testing the equality of variance, we used the levene’s test in preference to the more popular bartlett’s test because the non-normality of the distribution had been established for the ten markets included in this study. it has been found that bartlett’s test tends to reject the hypothesis of homoskedasticity more often in the case of non-normal or lepokurtosis distributions. the results of the levene’s test show that in none of the ten countries can the homoskedasticity hypothesis be rejected. we therefore conclude that both the developed and emerging markets the international journal of banking and finance, vol. 5.(number 1): 2007: 153-164162 in the post-crisis period have a low degree of variation across the day-of-theweek returns and this variation is not significant at any level. this observation is at variance with the pre-crisis findings. the earlier findings had indicated that emerging markets have a higher degree of variation in returns across the day-ofthe-week than the developed markets (ho et al. (1994)). in this study, according to the levene’s statistics, japan’s nikkei 225 (the most developed market in the test sample) has the highest variation of 3.491. other emerging markets that ranked high in terms of variation include india’s bse 50, china’s ssec, and singapore’s strait times index with variations of 4.509, 3.451 and 3.893, respectively. 5. conclusions not knowing if the reforms and restructuring that occurred during the postasian financial crisis period has posed a question as to whether the conclusions regarding day-of-the-week effect and volatility patterns in the pre-1998 period still holds, the study on this important issue confirms the existence of the day-ofthe effect week effect in the ten asian stock markets. the days in which the negative or positive returns or even the minimum and maximum returns had occurred seem to have shifted to other days in some countries. nevertheless, there is evidence of the day-of-the-week still lingering in the post-crisis period.it has not been traded away. it further confirms that the variation in stock returns is not significant for almost all the asian countries in this study. however, the results of the levene’s test for homoskedasticity across the days of the week show evidence of lower degree of variation in standard deviation across day-ofthe-week for all the asian stock markets, which is an important finding. so far, this seems to make sense in that the reforms would be expected to reduce the volatility somewhat. prior to the asian financial crisis, the asian markets were characterized by high returns and volatility. after the crisis, there seems to be a trend toward less volatility and subdued day-of-the-week effect in few of the ten markets tested. author statement: chiaku chukwuogor-ndu is a staff of eastern connections state university. as the editors are unable to contact chiaku in time for the production of this issue, we are unable to provide information on research interests. 163the international journal of banking and finance, vol. 5.(number 1): 2007: 153-164 references aggarwal, r. and p. rivoli (1989). seasonal day-of-the week effects in four emerging stock markets, the financial review, 24: 541-50. chen, g., kwok, c.c.y. and o.m. rui (2001). the day-of-the-week regularity in the stock markets of china. journal of multinational financial management 11 (2): 139-163. condoyanni, l., j. o’hanion and c. w. r. ward (1987). day of the week effect on stock returns; international evidence, journal of business finance and accounting 14: 159-74. fama, e. f. (1965) the behavior of stock market prices, journal of business, 38: 34-105. french, k., (1980) stock returns and the week-end effect, journal of financial economics 8: 55-70. gibbons, r.s., and p. hess (1981). day of the week effects and asset returns, journal of business 54: 579-96. ho, y. k., 1990, stock return seasonalities in asia pacific markets, journal of international financial management and accounting, 2, 47-77. ho, r. y. and y. l. cheung, 1994, seasonal pattern in volatility in asia stock markets, applied financial economics, 4, 61-67. international monetary fund (1999). the imf’s response to the asian crisis, available from http://www.imf.org/external/np/exr/facts/asia.htm#2. jaff, f., westerfield, r.l., and c. ma (1989). a twist on the monday effect in stock prices: evidence from the u.s. and foreign stock markets, journal of banking and finance 13: 641-50. kato, k. (1990). weekly patterns in japanese stock returns, management science, 36: 1031-43. keim, b.d. and and r. f. stambaugh, 1984, a further investigation of the weekend effect in stock returns, the journal of finance, 39, 819-840. lakonishok, j., and s. smidt, 1988, are seasonal anomalies real? a ninety-year perspective, review of financial studies, 1, 403-25. levene, h. (1960) robust tests for equality of variances in contribution to probability and statistics, (ed) 1, olkin: stanford university press, palo alto. mehdian, s. and m. perry (2001). the reversal of the monday effect: new evidence from us equity markets, journal of business finance and accounting 28: 1043-1066. rogalski, r. (1984). new findings regarding day of the week returns over trading and non-trading period, the journal of finance 39: 1603-14. snedecor, g. w., and w.g. cochran (1976) statistical methods . ames: iowa state university press. 164 smirlock, m., and starks, l. (1986) day of the week and intraday effects in stock returns, journal of financial economics 17: 197-210. wong, k.a., and h. d. ho (1986). the weekend effect on stock returns in singapore, hong kong journal of business management 4: 31-50. wong, k.a., t.k. hui and c.y. chan (1992). day-of the week effects: evidence from developing stock markets, applied financial economics 2: 49-56. international journal of banking and finance 1-1-2007 day-of-the-week effect and volatility in stock returns: evidence from east asian financial markets chiaku chukwuogor-ndu recommended citation ijbf7-marina.indd the international journal of banking and finance, vol. 7. number 1: 2010: 161-170 161 a note: debt selling and their impact on islamic bank value1 abdul ghafar ismail and karmila hanim kamil universiti kebangsaan malaysia _______________________________________________________________ abstract this paper shows how a risk management mechanism through selling debt can affect the value of islamic banks. islamic banks are able to maximize their value from the sale of murabahah on housing debt in order to manage their risk arising from fl uctuations in interest rates. a tractable theoretical model is developed to maximize the islamic banks’ values from the sale of housing debt fi nancing in order to hedge against fl uctuations in interest rates. our fi ndings showed that islamic banks could improve their earnings and rectify the problem in aligning their assets and liabilities through the benefi ts of debt selling. a rise in the market interest rates leads to an increase in the base fi nancing rate and the mark-up rate in islamic banks, since market interest rates serve as benchmarks in determining profi ts or mark-ups. if the islamic banks engage in debt selling to decrease their risk exposure, their earnings or value may be amplifi ed since they have the opportunity to undertake other positive npv projects from the payoffs on the murabahah debt selling. keywords: bank risk management; financing; investment jel classifi cation: g21, g31, g32. _______________________________________________________________ 1. introduction risk management has become a challenge in islamic banking because the growth in these banks may not occur without taking risk beyond the norm of conventional banking. thus, taking additional risk is unavoidable for the islamic banks to progress. nevertheless, taking excessive risks will hurt investment and may deter future growth. therefore, consideration of effective risk management is important for the bank’s fi nancial stability, the soundness of its business, its charter, and its value and profi tability. 1 note from editors: this paper was one of the three best papers selected by a review panel of three professors at the symposium held on november 26-28, 2008, in melbourne, australia. the symposium was funded by the australian research council grant 20072009/10 for research on islamic banking and finance. ijbf ht tp :// ijb f.u um .e du .m y 162 the international journal of banking and finance, vol. 7. number 1: 2010: 161-170 effi ciently engaging in risk management activities, as demonstrated by mason (1995), could increase the value of the banks by reducing the chance of occurrence of costly lower tail-end in earnings and cash fl ows. the costly lower tail outcomes are normally associated with the likelihood of default of borrowers and the consequent costs of fi nancial distress for a fi rm. this is particularly true since these costs are higher for banks as fi nancial institutions compared to nonbank fi rms. banks have the central role of creditworthiness in the provision of fi nancial services and they assume various kinds of fi nancial risks. hence, the potential loss of their earnings and value could lead to high costs of default and unforeseen external fi nancing need. these are particularly costly to the banks especially during emergency situations. in recent years, risk management has become a bank’s central activity. the steps related to properly managing risk for ultimate savers and for the banks themselves, as proposed by scholten and wensveen (2003), create economic value both for the bank as well as for its clients. accordingly, it is through the management of risk that the banks contribute to the economic welfare of the society wherein they operate. banks undertake risk management activities as an effort to lessen the chances of extreme fl uctuations in their fi nancial conditions. these efforts subsequently minimize the probability of becoming insolvent. a number of studies had documented the effect of various kinds of fi nancial risks on the sensitivity of banks earnings and profi tability ( hanweck and ryu, 2005; saunders and schumacher, 2000; zarruk and madura, 1992; angbazo, 1997; and wong, 1997). for instance, hanweck and ryu (2005) show that quarterly changes in the net interest margins (one of the main elements of bank’s earnings and value) are sensitive to credit and interest rate risks of a commercial banks in different bank groups. the bank groups are defi ned in terms of their specializations and asset size. their empirical tests showed that the greater the proportion of net short term assets and non-maturing liabilities of deposits that a bank holds, the more positive the effect of an increase in short term interest rate on the bank’s net interest margins. another line of studies also supported that fi nancial risks could lead to an adverse impact on the common stock returns of banks and accordingly would affect banks’ value (flannery and james, 1984; scott and peterson, 1986; kwan, 1991; choi, elyasiani and kopecky et al., 1992; elyasiani and mansur, 1998; benink and wolff, 2000; jianping and zheng wang, 2000). for example, considering three different interest rate variables, flennery and james (1984) suggested that commercial bank stock returns are very sensitive to the changes in interest rate. from the above studies, it should be clear that banks are exposed to interest rate risk. this paper considers one of the several mechanisms that are suggested to manage the risk factor in the islamic banks, i.e., selling murabahah on house fi nancing to a third party in order to hedge against the movement of the market interest rates that, in turn, affects the repayments of the sum lent. ht tp :// ijb f.u um .e du .m y a note: debt selling and their impact on islamic bank value 161-170 163 building on the risk management literature, this paper theoretically examines the sale of debt as a hedging strategy and its impact on the value of the islamic banks.2 the remaining parts of this paper are presented in three sections. section 2 provides a review of previous studies on risk management. section 3 presents a banking model that maximizes profi t from selling debt. the last section provides the conclusions. 2. related literature risk management in banking institutions has become more sophisticated along with the growing levels of fi nancial risk undertaken by the banks. banks conduct risk management in a variety of forms. among the major sets of actions are elimination or avoidance of risks, transfer of risks, and absorption/management of risks (allen and santomero, 1998). generally banks can eliminate or avoid certain risks (i.e., systematic risk) that are not consistent with their fi nancial characteristics by selling instruments in the spot market, hedging risk using derivatives instruments, and diversifying portfolios to minimize risk. islamic banks, for example, allow the selling of debt created by installment sales (murabahah) and securitization of debt/asset in a way that is permitted by the shari’ah (chapra and khan, 2000). these mechanisms as applied by the islamic banks are part of the risk management that helps the banks to guard risk against market uncertainties. cebenoyan and strahan (2004) empirically test on how credit risk management in all domestic commercial banks in the us is done through active loan purchase and sales activity; and how these affect the bank’s capital structure, lending, and profi t. they showed that banks take advantage of the risk reducing benefi ts of loan sales by taking more profi table but higher risk actions along with greater fi nancial leverage. there are also some types of risks that can be mitigated through the technique of risk transfer. interest rate risk, for example, can be transferred by such products as swaps or other derivatives. brewer, jackson and moser (2001) showed that banking organizations that use derivatives experience more business lending than banks that do not use derivatives. banks that use derivatives to manage interest rate risk also hold lower levels of costly capital than other banks. the islamic banks are also exposed to the interest rate movement that can be reduced by hedging via islamic profi t rate swap. 2 the market risk in murabahah on house fi nancing exists due to the uncertain movement of market interest rate that is used as a benchmark in setting profi t margin or mark-up. the islamic banks opt to manage against this risk by selling the debt for a lump sum plus fees. thus, they create an opportunity to economize on the use of cash and manage capital structure. ht tp :// ijb f.u um .e du .m y 164 the international journal of banking and finance, vol. 7. number 1: 2010: 161-170 some risks, however, should be absorbed or managed at the bank level due to their unique characteristics, such as, illiquid structure, competitive advantage, or moral hazard reasons. in this case, banks are assumed to diversify and hold suffi cient amount of capital in order to assure that risk is always kept at a relatively low level. whether risk management at the bank level can maximize or enhance value of the bank will depend on the bank’s competitive advantage. basically, risk management consists primarily of reducing earnings volatility and avoiding large losses. the modern literature on corporate fi nance has offered a number of suggestions as to why a fi rm may be interested in reducing the volatility of its value through some hedging policy. hedging can affect the payoff of a risk-neutral fi rm in that some market imperfections make the fi rm’s payoff a concave function of some state contingent variable. the rationales for the concavity of the payoff function might be related to the fi rm’s tax schedule (smith and stulz, 1985; graham and smith, 1999), to the costs of fi nancial distress (smith and stulz, 1985; stulz, 1996), to agency costs to asymmetric information problems (demarzo and duffi e, 1995), to underinvestment problems (froot scherfstein and stein, 1993), or to a combination of some of these factors (leland, 1998). some important deviations from the perfect capital markets in the modigliani-miller setting that have been identifi ed. these induce fi rms to care about risk management.3 the plausibility for the above explanations on various market imperfections varies, especially when the benefi ts of hedging and costs of risk management are taken into account. it is clear that the cost of fi nancial distress is viewed as the most compelling reason for risk management by fi rms. volatility of earnings may lead to costly bankruptcy which fi rms always try to avoid. thus for that reason, some studies offer signifi cant insight into why the banks themselves may choose low risk strategies (marcus, 1984). there is an extensive literature dating back to warner (1977) on costs of bankruptcy and then to weiss (1990) that cover evidence of fi nancial distress. as for banks, they face the same types of bankruptcy costs as other fi rms and have an incentive to manage their risks (systematic and nonsystematic) prudently so that the probability of bankruptcy is minimized. diamond (1991) showed that banking institutions have an incentive to manage risk because bankruptcy costs can be reduced and at the same time charter values of the bank can be preserved when the bank is hedged against large losses. therefore, mastering the stability of banks’ profi t is a critical credibility issue, because, if the current capital structure of a bank is not strong enough, it can have fi nancial distress costs and some additional external funding will be required. such funding can prove costly and dangerous due to the discrete transaction costs to obtain external fi nancing, agency costs driven by asymmetric information, and the high cost of potential future bankruptcy. 3 modigliani and miller (1958) suggested that in a theoretically perfect market fi rms should not waste their resources in managing risks because shareholders can do it more effi ciently by holding a well diversifi ed portfolio. ht tp :// ijb f.u um .e du .m y a note: debt selling and their impact on islamic bank value 161-170 165 in a series of papers by froot et al. (1993; 1998), they propose a model which endogenizes distress costs. external fi nancing is more costly than internally generated funds due to capital market imperfections. in other words, the volatility of profi tability may lead to reduction in the fi rm’s value. if fi rms face constraints on their internal fi nancing, they incur either higher costs of raising external funds or higher costs of forgone investment opportunity. the more diffi cult it is for a bank to raise external funds, the more risk averse it behaves. thus, risk reduction is always desirable for the risk-averse bank in the froot scharfstein and stein (1998) setting. this would suggest that the optimal risk management strategy for a bank in order to maximize its value is to fully hedge all tradable risks as long as they can be hedged at little or no cost in the capital markets. in sum, assuming that value maximization is the ultimate objective function of the islamic or conventional banks, risk management suggests maximization of bank value through reduction of costly capital while pursuing suffi cient internal funds, if available, and taking advantage of attractive investments. 3. the model we considered a banking model that has three time periods, 0, 1, and 2. in the fi rst two periods, the islamic bank decides on its capital structure, and then makes investments and debt sale decisions. the last period is required to end the model by showing that the bank achieves a sound objective function in value maximization from its hedging strategy in managing their risk exposures. a. period 0: capital structure and investment decisions the sources of the islamic banks’ capital can come from subordinated long-term debt and equity. hence at t=0 an islamic bank decides to lend/sell a portfolio of securities (securities held-for-trading) g t, to investors under repurchase agreement. random payoff from the portfolio will be repaid to the bondholders at t=2. for simplicity, a standard assumption is made wherein the borrowing and lending rates in the capital markets are zero. the islamic bank also decides on how much equity capital, e, to hold at t=0. more specifi cally, the bank’s decision at t=0 is how much of its equity capital to invest in risky asset, f. this can be interpreted as extending house fi nancing using murabahah ("cost-plus fi nancing" or "mark-up fi nancing") as a mode of fi nancing. the proceeds is to be invested in riskless securities, gs (for instance, malaysian government securities (mgs) for liquidity purposes to guard against unexpected withdrawals by depositors or draw downs by borrowers (saidenberg and strahan, 1999). the bank also has to decide how much to hold as capital reserves. it is assumed that reserves are perfectly liquid and have a zero return. ht tp :// ijb f.u um .e du .m y 166 the international journal of banking and finance, vol. 7. number 1: 2010: 161-170 both capital and investment decisions made by the bank at t=0 are assumed to involve direct costs. each bank is subject to the following standard quadratic cost functions model as presented in kopecky and van hoose (2004). the bank’s cost functions are modelled with a specifi c quadratic form to allow the marginal costs to increase in response to the quantities of gs, d, f, and e. government securities cost : c g = (g/2) gs2 (1) deposit cost : c d = (d/2) d2 (2) financing cost : c f = (f/2) f2 (3) equity cost : c e = (e/2) e2 (4) where g, d, f, and e are the positive cost parameters associated with c j = bank resource costs. gs = government securities, d = deposits transactions, f = bank fi nancing, and e = equity. with regards to minimum capital that needs to be hold, we assume that islamic banks are required to put aside an amount of equity equal to a fraction of bank fi nancing and fi xed percentage against deposit transactions. b. period 1: selling decisions the islamic bank uses murabahah in a sale agreement, whereby the bank discloses the true cost of the house and then adds a mark-up (m) to sell it at an agreed price to the customer.4 however, the mark-up is determined by adding the benchmark, i.e. base fi nancing rate plus the spread. if we set the contract at t=0, the islamic bank will receive a constant stream of income through the duration of contract. therefore at t=1, when the market base fi nancing rate will change, the islamic bank is exposed to interest rate risk. to hedge against the movement of market interest rate, the islamic bank chooses to sell a fraction or all of its sale receivable (murabahah housing contract) to a third party at t=1. the number of murabahah housing contract is given by h. a position of h will generate a random payoff equal to h(m-r), where the mark-up rate, m, is known at t=0 and r represents the future spot base fi nancing rate, which is realized at t=1. if we sell the contract today, during the higher rate of base fi nancing, the islamic bank has a competitive advantage and may charge a higher mark-up that leads to more generation of income. the payoff from the selling position can then be reinvested by the islamic bank on other lucrative projects or provide higher return to depositors at t=2. 4murabahah is an acceptable form of credit sale under shari’ah. similar in structure to a rent to own arrangement, the intermediary retains ownership of the property until the loan is paid in full. ht tp :// ijb f.u um .e du .m y a note: debt selling and their impact on islamic bank value 161-170 167 c. period 2: cash flow realization at t=2, the payoffs from the riskless assets, debt selling positions, reinvestment positions, and the amount of capital raised at t=0 by the islamic bank are realized. it is assumed that the islamic bank maximizes profi ts: max = g t + r e e + r g gs + r f f+ h (m-r) – r d d – c j (5) subject to a balance sheet constraint: f + gs + f + d = d + e (6) where, = a fraction of bank fi nancing that required to hold as capital reserve,  = fi xed percentage against deposit transaction, r g = the government securities rate, r l = the fi nancing rate, r d = the deposit transaction rate, and r e = required return on equity. by using current-valued hamiltonian, h = (g t + r e e + r g gs + r f f+ h (m-r) – r d d – c j + (f + gs + f + d – de) (7) or h = ( ) + (k) (8) from equation (7), we can have the following optimality conditions: it follows that (15) (16) (17) (18) (19) / gs = rg – ggs + (10) / e = re – ee (11) / f = rf – ff + + (12) / d = – rd – dd+ (13) / h = m – r (14) h/ gt = m – r h/ gs = m-r/ rg – ggs + h/ e = m-r/ re – ee h/ f = m-r/ rf – ff + + h/ d = m-r/– rd – dd+ / gt = 1 (9) ht tp :// ijb f.u um .e du .m y 168 the international journal of banking and finance, vol. 7. number 1: 2010: 161-170 equation (15) – (19) expresses the rate of change in h with respect to gt, gs, e, f, and d accordingly. the increase or decrease in gt, gs, e, f and d depends on the value of m and r. during the high rate of interest, m will also rise along with r. therefore, gt, gs, e, f will also tend to increase. it can be interpreted that the islamic bank could make a decision to restructure its investment during the high rate of interest in order to reduce its risk through debt selling; and switch the benefi ts or payoff from the risk management position to other potential investments in gt, gs, e, and f. if the value of m is less than r, equation (19) shows that the rate of change in d will increase. this implies that during the low and unattractive movement in the market rate of interest, the islamic bank prefers to hold or not to sell its murabahah housing debt to the third party and continues receiving constant stream of income from the house buyer. the result is an increase in the deposit transactions of the islamic bank. consistent with cebenoyan and strahan (2004), we also found it optimal for the islamic bank to reduce its risk of movement in interest rate through a debt selling process in accordance with the shari’ah principal. by using debt selling, the islamic bank can increase its value in venturing into other lucrative investments. 4. conclusion we have shown that the islamic banks are able to maximize their value from the sale of murabahah or debt on housing in the process of managing their risk that arises from fl uctuation in interest rates. by constructing a model that considers banks’ capital structure, investments, and the decision to sell debt that is open to exposure due to interest rates fl uctuations, we have shown that the payoff from the debt selling decision enables the bank to generate additional income. the value of the islamic banks could further be enhanced from the risk reducing benefi ts of risk management, if they adopt more profi table projects or activities. future research may use an expanded framework of the islamic bank models that include debt-based and equity-based (i.e. mudharabah and musharakah) fi nancing contract together in the bank’s maximization model. this would enable us to analyse the cash fl ow realization of the islamic banks when implementing the appropriate hedging strategy to manage risk associated with both types of fi nancing. author information: submitting author karmila hanim kamil, universiti kebangsaan malaysia, bangi, 43000 selangor, malaysia; fax: +603-8921 5789, e-mail: agibab@ukm.my abdul ghafar ismail is professor of financial econometrics at the same university. ht tp :// ijb f.u um .e du .m y a note: debt selling and their impact on islamic bank value 161-170 169 references allen, f., and santomero, a.m. (1998). the theory of fi nancial intermediation. journal of banking and finance 21, 1461-1485. angbazo, l. (1997). commercial bank net interest margins, default risk, interestrate risk, and off-balance sheet banking. journal of banking and finance, 21(1), 55–87. benink, h. and wolff, c. (2000). survey data and the interest rate sensitivity of us banks stock returns. economic notes, 29(2), 201-213. brewer, e iii, jackson, w.e and moser, j.t. (2001). the value of using interest rate derivatives to manage risk at u.s. banking organizations. 3q economic perspectives, federal reserve bank of chicago, 49-66. cebenoyan, a.s. and strahan, p.e. (2004). risk management, capital structure and lending at banks. journal of banking and finance, 28, 19-43. chapra, m.u and tariqullah khan, (2000). regulation and supervision of islamic banks jeddah. islamic research and training institute, islamic development bank. choi, j., elyasiani, e. and kopecky, k. (1992). the sensitivity of bank stock returns to market, interest and exchange rate risks. journal of banking and finance, 16, 983-1004. demarzo, p., and duffi e, d.,(1995). corporate incentives for hedging and hedge accounting. review of financial studies, 8, 743-771. diamond, d., (1991). monitoring and reputation: the choice between bank loans and directly placed debt. journal of political economy, 99, 689-721. elyasiani, e. and mansur, i. (1998). sensitivity of bank stock returns distribution to changes in the level and volatility of interest rate: a garch-m model. journal of banking and finance, 22, 535-563. flannery, m. and james, c. (1984). the effect of interest rate changes on the common stock returns of fi nancial institutions. journal of finance, 39, 1141-1153. froot, k. a., scharfstein, d.s. and stein, j.c., (1993). risk management: coordinating corporate investment and fi nancing policies. journal of finance, 48, 1629-1658. froot, k. a., scharfstein, d.s. and stein, j.c., (1998). risk management: capital budgeting and capital structure policy for fi nancial institutions: an integrated approach. journal of financial economics, 47, 55-82. graham, j.r. and c.w. smith, jr., (1999). tax incentives to hedge. journal of finance, 54, 2241-2262. hanweck, g. and ryu, l., (2005). the sensitivity of bank net interest margins and profi tability to credit, interest rate, and term-structure shocks across bank product specializations. working papers, school of management, george mason university. jianping, m. and wang, z. (2000). excess risk premia of asian banks. international review of finance, 1(2), 143-159. ht tp :// ijb f.u um .e du .m y 170 the international journal of banking and finance, vol. 7. number 1: 2010: 161-170 kopecky, k.j., vanhoose, d. (2004). a model of the monetary sector with and without binding capital requirements. journal of banking and finance, 28, 633-646. kwan, s. (1991). re-examination of interest rate sensitivity of commercial bank stock returns using a random coeffi cient model. journal of financial services research, 5, 61-76. leland, hayne e., (1998). agency costs, risk management, and capital structure, journal of finance, 53(4), 1213-1243. mason, s.p. (1995). the allocation of risk. working paper 95 – 060. harvard business school. marcus, alan j., (1984). deregulation and bank financial policy. journal of banking and finance, 8, 557-65. modigliani, f. and miller, m. h. (1958). the cost of capital, corporate fi nance and the theory of investment. american economic review, 48, 261-97. saidenberg, m.r., strahan, p.e., (1999). are banks important for fi nancing large businesses? current issues in economics and finance, 5(12), 1-6. saunders, a. and schumacher, l. (2000). the determinants of bank interest rate margins: an international study. journal of international money and finance, 19(6), 813–832. scholten, b. and wensveen, d. (2003). the theory of fi nancial intermediation: an essay on what it does (not) explain. the european money and finance forum. suerf vienna. scott, w. and peterson, r. (1986). interest rate risk and equity values of hedged and unhedged fi nancial intermediaries. journal of financial research, 9, 325-329. smith, jr., c.w. and stulz, r.m., (1985). the determinants of fi rms’ hedging policies. journal of financial and quantitative analysis, 20, 391-405. stulz, r.m., (1984). optimal hedging policies. journal of financial and quantitative analysis, 19, 127-140. stulz, r.m., (1996). rethinking risk management. journal of applied corporate finance, 9, 8-24. warner, j., (1977). bankruptcy costs: some evidence. journal of finance, 32 337-347. weiss, l., (1990). bankruptcy resolution: direct costs and violation of priority of claims. journal of finance and economics, 27, 285-314. wong, k.p. (1997). on the determinants of bank interest margins under credit and interest rate risks. journal of banking and finance, 21(2) 251–271. zarruk, e. and madura, j. (1992). optimal bank interest margin under capital regulation and deposit insurance. journal of financial and quantitative analysis, 27(1) 143–149. ht tp :// ijb f.u um .e du .m y the relationship between risk propensity, risk perception and risk-taking behaviour in an emerging market 134 the international journal of banking and finance, volume 10 (number 1), 2013: pages 134-146 the relationship between risk propensity, risk perception and risk-taking behaviour in an emerging market fazelina sahul hamid a , gary john rangel b , fauziah m. taib a , and ramayah thurasamy a a universiti sains malaysia and b monash university malaysia ____________________________________________________________________ abstract this paper reports evidence to support a relationship between risk propensity, risk perception, and risk-taking behaviour of investors in an emerging market. primary data were gathered using a validated structured questionnaire, which was self-administered by respondents: there were 162 investors from 8 stockbroking companies. a multiple regression was used to test the direct and indirect effects of the identified behavioural characteristics on investment decision. risk propensity was found to be positively related to risk-taking behaviour whereas risk perception was negatively related to risk-taking behaviour. it was further found that risk perception partially mediates the effect of propensity to take risk. this suggests that the perceptual framing of a situational context in the investors’ thought processes reduces but it does not totally overwhelm the innate personality traits with respect to either the investor’s risk-seeking or risk-averseness. the tendency to engage in risky behaviour is more psychological in nature. the implications of the research are further explored. keywords: risk propensity, risk perception, risk-taking behaviour, empirical study jel classification: d81, g02. ____________________________________________________________________ 1. introduction malaysia’s benchmark kuala lumpur composite index (klci) reached an all-time high of 1,872.52 points in december 2013. the previous high coincided with the super bull-run of 1993 when the klci value was 1,275.30. however, the circumstances now and then seem to be different. retail investors who made up 60 per cent to 70 per cent of the daily trading volume during the super bull-run have not returned in droves this time around. in fact, retail investors have continually shunned the stock market since the losses retail investors took during the 1997 asian financial crisis as well as during the 2008 subprime crisis. the nations’s prime minister declared in march 2010 that the state-controlled employees provident fund (epf) accounted for 50 per cent of the daily trading volume in the equity and bond markets with most individual investors shifting to mutual funds after the malaysian economy slid into a recession in 1998 (chan, 2010). retail investors are still wary of trading on the market because they are not convinced that malaysia’s economic recovery is real and sustaining. moreover, their appetite for risk seems to have abated (chieh, 2010). therefore, the objective of this paper is to understand the current risk-taking behaviour of the malaysian retail investors. it seeks to understand whether risk aversion is indeed a significant contributing factor to the lack of participation in the local stock market. by examining the risktaking behaviour of retail investors, it would help in identifying the salient factors that influence investors to participate less in the malaysian stock market. we use a model developed for exploring risk-taking behaviour in the corporate context and apply it to individual investors’ trading behaviour. it is posited that risk perception and risk propensity have a significant influence on risk-taking behaviour. we also posit that retail investors’ risk perception may have a mediating effect on risk propensity (i.e., the willingness to take risks). the rest of the paper is organised in the following manner. the next section provides a discussion of risk-taking behaviour and the factors that could relate to it. section 3 describes the relationship, develops the hypotheses and the methodology to test the relationship. section 4 is a discussion of the data and summary of empirical results. the conclusion in section 5 completes the paper. 2. literature review although risk-taking behaviour is a widely researched topic in finance, a search of empirical literature show that consumer decision-making research in the context of financial products is surprisingly scarce (byrne, 2005). past research on risky decision-making has focused on individual risk-taking behaviour in an organisational context especially in management decisionmaking. there is a dearth of decision-making studies in the context of the securities market in general and none exists from an emerging economy in particular. as such, this paper seeks to address a gap in literature. traditional theories, some of which have won nobel accolades, have classified financial risk as something quantifiable, so it is measured by the volatility of returns and individual trade-off between risk and return (diacon, 2004). this is the essence of the expected utility model proposed by von neumann & morgenstern (1944). the axioms of the utility theory argue that investors are (1) completely rational, (2) able to deal with complex choices, (3) risk-averse, and international journal of banking and finance, vol. 10, iss. 1 [2013], art. 7 (4) wealth maximising agents. this assumes that investors select a portfolio that maximises their returns while minimising their risks. the utility theory is also the central tenet of the efficient market hypothesis with its concept of investor rationality. ibrahim & lim (1995), however, found this relationship to be unstable, as has also been shown by recent studies. most retail investors speculate in a bullish market but revert toward fundamental analysis in a bearish market. however, an updated study by lai, low, & lai (2001) found malaysian retail investors to exhibit stable rational behaviour. these contradictory findings lead us to believe there may be behavioural explanations with regard to retail investors’ stock market participation: this is exactly what the momentum theory explains. proponents of behavioural finance argue investors may not be rational at all times. kahneman & tversky (1979) proposed the prospect theory as an alternative to the expected utility theory. it is one of the most widely used theories of individual decision-making that looks at the cognitive limitation of the decision-makers. specifically, they argue that individuals will be risk-averse in a gain situation and risk-seeking in a loss situation. in reviewing the prospect theory, sitkin & pablo (1992) found contradictions where past success led to the willingness to engage in risky behaviour and they proposed an alternative model of the determinants of risky behaviour. they suggested that risky behaviour is determined by two individual factors, namely risk propensity and risk perception. they further postulated that the relationship between risk propensity and risk behaviour is mediated by risk perception described in figure 1 below. figure 1: proposed research model risk perception is defined as an individual’s assessment of the inherent risk in a given situational problem (sitkin & wiengart, 1995). this assessment is based on one’s probabilistic estimation of the degree of uncertainty, controllability, and confidence in a problematic situation. risk perception is likely to be affected by cognitive biases that arise out of ways of thinking risk propensity risk perception risk behaviour known as heuristics (diacon, 2004). this acts as a shortcut to enable processing and simplification of information. the disposition effect postulated by the prospect theory is consistent with a negative relationship between risk perception and risk-taking behaviour. positive gain situations lead to conservative, risk-averse behaviour whereas situations labelled as negative elicit risk-seeking behaviour. some studies in the literature have reported the opposite results. one example is the threat-rigidity hypothesis postulated by staw, sandelands, & dutton (1981). they found that, under threatening situations, individuals tend to rely heavily on prior expectations, have increased attention to dominant cues, and are inclined to emit well-learned responses which are all consistent with conservative, risk-averse behaviour. in positively labelled contexts, others have also reached opposite conclusions concerning the prospect theory. march & shapira (1987) and thaler & johnson (1990) suggest that, when conditions are perceived positively, individuals tend to focus on opportunities inherent in those situations and thus will be inclined to behave in a risk-seeking manner as a result of their past successes. the argument here is not for an absence of a relationship between risk perception and risk behaviour but the sign of the relationship. figure 2: juxtaposing extant theoretical models and predictions of risk behaviour situational characteristics (objective or perceived) positive negative risk averse risk propensity risk seeking source: adapted from sitkin & pablo (1992), p. 27. prospect theory – threat rigidity conservation of prior gains (staw, sandelands, & (kahneman & tversky,1979) dutton, 1981) loss prevention bias hypervigilance (jackson & dutton, 1988) (janis & mann, 1977) prediction: low risk prediction: low risk behaviour behaviour cell 1 cell 2 attention to prospect theory – opportunities going for broke (march & shapira, 1987) kahneman & tversky 1979; singh, 1986) prediction: high prediction: high risk behaviour risk behaviour cell 3 cell 4 international journal of banking and finance, vol. 10, iss. 1 [2013], art. 7 one way to reconcile these apparent contradictory findings, instead of accepting one and rejecting the other, is the identification of previously latent variables which may help explain the contradictory results. there are two possible scenarios that such a relationship could occur. first, risk perception-to-risk-behaviour relationship is moderated by a hidden variable that modifies the relationship between perception of risk and risky behaviour. second, risk perceptions may be correlated with some previously unexamined variable that is actually driving the relationship. sitkin & pablo (1992) made the case that missing variable is risk propensity. in order to reconcile the mixed results, we illustrate them in figure 2 by forming a matrix between risk propensity (risk-averse behaviour versus risk-seeking behaviour) with risk perceptions (positive situation versus negative situation). quadrants 1 and 4 are consistent with the prospect theory where risk-averse investors are prone to exhibit low-risk behaviour in circumstances of low perceived risk (quadrant 1) and vice versa (quadrant 4). quadrants 2 and 3 are consistent with its competing alternatives. nevertheless, it can be argued that the settings in which the studies that fall in these two quadrants would determine the results obtained. specifically, studies in quadrant 2 is meant to examine more bureaucratic, risk-averse contexts (janis & mann, 1977; staw, sandelands, & dutton, 1981) and studies in quadrant 3 focuses on more entrepreneurial situations (march & shapira, 1987). prior literature on risk propensity concentrates on two main aspects. the first relates to individual investors’ differences that influence risk-taking behaviour. the second aspect relates to the situational influence. as a result, there are also contradictory opinions on risk propensity (huff, keil, kappelman, & prybutok, 1997). one school of thought would argue that risk propensity is a trait that is stable over time (gerrans, faff, & hartnett, 2012) while another school of thought is that it is a trait that changes through a learning process. prior research on the influence of personality on risk-taking behaviour argues that risk propensity is determined by individual characteristics. for example, personality traits such as impulsivity, sensation-seeking, and low self-control generally exhibit risk-taking behaviour (mishra & lalumière, 2011). prior studies have also shown that personality constructs, especially risk-taking, becomes a consistent and emergent factor in decision-making within the positive domain whereas in negative domains, personality constructs associated with risk becomes diminished and are less of a factor in decision-making (skeel, neudecker, pilarski, & pytlak, 2007; garvey, 2010). this confirms the findings of masters (1989) who found through the choice dilemmas questionnaire (cdq) that risk-taking investors display more risk-prone cdq scores. the review of past research as summarised in figure 2 does suggest an interaction effect between risk propensity and risk behaviour. the main effect of risk propensity on risk behaviour might well be strengthened as the level of perceived risk rises. investors, who are risk-averse, will exhibit such behaviour if their perceived situational risk rises. conversely, investors who are risk-seeking are more prone in exhibiting such behaviour the higher the perceived risk (sitkin & wiengart, 1995). a. the relationship the description by maccrimmon & wehrung (1986) of risk propensity as a “willingness to take risks” clearly indicates that this willingness will eventually affect actual risk-taking. individual tendency to take or avoid risks has been shown to affect risk-taking (brockhaus, 1980). thus, this study proposes that: h1: risk propensity will be positively related to risk-taking behaviour. h2: the effect of risk propensity on risk behaviour will be mediated by risk perception individuals who perceive higher levels of risk in a problematic situation relate risks with negative outcomes and thus will make less risky decisions. individuals, who perceive lower levels of risk, will relate it with positive outcomes thus making more risky decisions (sitkin & wiengart, 1995). in line with this, it is hypothesised that: h3: risk perception will be negatively related to risk-taking behaviour. 3. methodology the unit of analysis for this paper is the individual investors as they are major players in the stock market. the operational definitions of the variables used are summarised in table 1. the questionnaire used was adopted from sitkin & pablo (1992) with modification to suit the context of this study. the questionnaire was administered to individual investors who frequented the public galleries of eight stockbroking companies. the questionnaire focuses on a typical investment scenario faced by investors. two sets were developed to depict a situation in a positive scenario and a negative scenario respectively. one hundred questionnaires for each scenario were distributed. only individuals who have had experience in investing in the stock market were interviewed. a total of 177 completed questionnaires were returned (response rate = 88.5%) and international journal of banking and finance, vol. 10, iss. 1 [2013], art. 7 out of these, 15 were removed due to incompleteness or inconsistencies in the responses. of the remaining 162 questionnaires, 89 and 73 were positive and negative scenario versions respectively. table 1: operational definitions of variables variables operational definition prior studies risk behaviour degree of uncertainty associated with the chosen decision outcome (certain outcome vs. probable outcome) kahneman & tversky (1979) sitkin & pablo (1992) sitkin & weingart (1995) pablo (1997) risk propensity individuals’ current period tendency toward risk-taking maccrimmon & wehrung (1986) sitkin & pablo (1992) sitkin & weingart (1995) pablo (1997) wong (2005) risk perception individuals’ assessment of the degree of risk inherent in a decision situation sitkin & pablo (1992) sitkin & weingart (1995) 4. results table 2: personal profile of the respondents profile description total percentage gender male female 93 67 58 42 age below 20 years old 21 – 30 years old 31 – 40 years old 41 – 50 years old above 50 years old 2 19 79 42 20 1 12 49 26 12 race malay chinese indian others 56 92 14 0 35 57 9 0 education level spm or lower certificate diploma bachelor’s degree master’s or higher 28 55 40 31 8 17 34 25 19 5 average investment amount less than rm10,000 rm10,001 – 50,000 more than rm50,000 94 45 23 58 28 14 the profile of the respondents is presented in table 2. the frequency analysis shows that 58.1 per cent of those sampled were male. almost half of the investors belong to 31-40 years age group. approximately 60 per cent of the cohort was chinese. about half of the investors had ten years of education with a certificate or had lower educational qualifications. about 58 per cent of the investors invested less than rm10,000 (us$3,400) in the stock market. a. goodness of measures we establish the goodness of fit measures. two of the tests are validity and reliability. although there are various measures of validity and reliability, we have opted for construct validity and the inter-item consistency measures for this paper. factor analysis is a data reduction technique that purports to reduce a number of items into a manageable number of factors (see table 3). it basically tests the factor validity of the measures. we used a factor analysis with a varimax rotation to validate the 10 items used to measure risk perception and risk propensity. the factor analysis yielded a 2-factor solution explaining 76 per cent of the variation with a significant test value using bartlett’s test of sphericity ( 2 = 501.145, p < 0.01). the kaiser meyer olkin’s measure of sampling adequacy was 0.70, which can be described as moderate based on harris and woodward’s (1974) classification. thus, we can conclude that the factors are distinct and valid. table 3: component analysis for mediating variables components items 1 2 risk perception 1 0.880 -0.103 risk perception 3 0.874 -0.024 risk perception 5 0.603 0.074 risk perception 7 0.927 -0.118 risk propensity 1 0.026 0.940 risk propensity 2 -0.097 0.935 eigenvalue variance (75.998%) 2.771 46.183 1.789 29.815 note: risk perception 2, 4, and 6, and risk propensity 3 were dropped due to low loadings. next, the reliability of the measures was assessed by using the cronbach's alpha coefficient. reliability refers to the degree of consistency, as kerlinger (1986) puts it; if a scale possesses a high reliability, the scale is homogeneous. according to nunnally (1978), alpha values equal to or greater than 0.70 are considered to be a sufficient. the alpha values ranged from 0.84 to 0.87, international journal of banking and finance, vol. 10, iss. 1 [2013], art. 7 and were much higher than cut-off of 0.7. thus, we conclude that the measures have sufficient reliability. conclusion made based on the results from the survey is acceptable. table 4 depicts the summary of the reliability analysis. table 4: summary of reliability analysis variables items items deleted cronbach’s alpha risk preference 3 0 0.85 risk propensity 2 1 0.87 risk perception 7 3 0.84 b. hypotheses testing a regression analysis was conducted in order to test two hypotheses. table 5 is a summary of results. as hypothesised, risk propensity was found to be positively related ( = 0.296, p < 0.01) to risk-taking behaviour whereas risk perception was negatively related ( = -0.242, p < 0.01) to risk-taking behaviour. thus, both h1 and h3 are supported. further to that, we also tested the mediation effect of risk perception on the risk propensity and risk-taking behaviour. table 5: results of the regression analysis unstandardised beta standardised beta t value p value risk propensity risk perception 0.484 -0.384 0.296 -0.242 3.799 -3.110 0.000 0.002 f value r 2 adjusted r 2 13.354** 0.16 0.148 **p < 0.01 a mediator specifies how (or the mechanism by which) a given effect occurs (james & brett, 1984; baron & kenny, 1986). baron and kenny (1986, pp. 1173, 1178)> it may be described as the following: the generative mechanism through which the focal independent variable is able to influence the dependent variable of interest . . . (and) mediation . . . is best done in the case of a strong relation between the predictor and criterion variable. according to mckinnon et al. (1995), mediation is generally present when: 1. the independent variable (iv) significantly affects the mediator, 2. the iv significantly affects the dependent variable (dv) in the absence of the mediator, 3. the mediator has a significant unique effect on the dv, and 4. the effect of the iv on the dv shrinks upon the addition of the mediator to the model. baron & kenny (1986) formulated the steps and conditions to ascertain whether full or partial mediating effects are present in a model. the beta value for risk propensity was 0.320 (p < 0.01) when directly regressed against risk-taking behaviour. when risk perception is included as shown in table 5 below, the beta value for risk propensity reduces to 0.296 and is still significant, which is consistent with partial mediation effect. we can thus conclude that h2 is supported, albeit partially. 5. conclusion this study considered whether investors’ risk-behaviour in an emerging market (malaysia) is influenced by two variables; risk perception and risk propensity. an additional measure included is whether risk perception mediates the effect of risk propensity on risk-behaviour. the results suggest both risk propensity and risk perception do influence an investor’s risky behaviour. however, the effect of risk propensity on risky behaviour is partially mediated by risk perception. this is in line with findings of sitkin & weingart (1995). investors with tendencies to engage in risky investment have a behavioural motivation to project actual risky behaviour. the results also indicate that risk perception is significant and is negatively related to risk behaviour. the findings are consistent with those of kahneman & tversky (1979) which suggest that an individual’s risk choices will be context dependent. however, the situational domain (positive or negative) the individual investor is only partially influences the relationship between personality traits which are an inherent determinant of risky behaviour. in other words, this suggests that the framing of a situational context in the investor’s thought processes reduces but not totally overwhelms the innate personality traits (either risk-seeking or risk-averse). the evidence presented is indicative that personality preferences emerge in the face of both positive and negative situations. these results reinforce a conclusion that providing more education to individual investors does not necessarily mean that they will be more risk-averse or risk-seeking. the tendency to engage in these behaviours is more psychological in nature. the resultant lack of participation of retail investors may be due to the over extension of risky investing in the stock market based on the bad experiences over the years. it is therefore international journal of banking and finance, vol. 10, iss. 1 [2013], art. 7 imperative for regulatory authorities to strengthen the information dissemination infrastructure so retail investors would be better informed when making decisions on their investment and allow them to re-align their risk tendencies. the standardisation of financial reporting should be one of the key initiatives undertaken. for example, the u.s. securities and exchange commission sponsors edgar, a website which provides detailed information on public securities and companies that issue them (shiller, 2008). with real-time and simplified financial reporting provided to the malaysian securities and exchange commission (sec), retail investors would be able to obtain vital and easy-to-use information in deciding which stocks to invest that best suit their risk profile. in conclusion, better measures could be implemented to mitigate the information asymmetry experienced by retail investors. finally, further research on the stability of risk-taking personality traits could be carried out. this requires a longitudinal study. future research could also look at the determinants of risk propensity, namely risk preference, inertia, and previous outcome history. author information: fazelina sahul hamid is an economics lecturer, universiti sains malaysia (usm), email: fazelina@usm.my. gary john rangel is a lecturer of finance at the school of business, monash university malaysia, email: gary.rangel@monash.edu. fauziah m. taib is a professor and dean, usm, email: mfauziah@usm.my. ramayah thurasamy is professor at usm and dy dean, email: ramayah@usm.my. the authors acknowledge with thanks the editorial review suggestions and comments as well as the useful editing of the paper by the journal’s coeditors. the authors alone are responsible for any errors. references baron, r. m., & kenny, d. a. (1986). the moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. journal of personality and social psychology, 51(6), 1173-1182. brockhaus, s. r. h. (1980). risk taking propensity of entrepreneurs. academy of management journal, 23(3), 509-520. doi: 10.2307/255515. byrne, k. (2005). how do consumers evaluate risk in financial products? journal of financial services marketing, 10(1), 21-36. chan, t. h. (2010). malaysia lures individual investors to market shunned since asian crisis. bloomberg. retrieved september 28, 2012, from http://www.bloomberg.com/news/201006-09/malaysia-seeks-to-lure-individual-stock-traders-decade-after-asian-crisis.html chieh, y. h. (2010). ordinary malaysians shun stock market amid stalling recovery. the malaysian insider. retrieved september 28, 2012, from mailto:fazelina@usm.my mailto:gary.rangel@monash.edu mailto:mfauziah@usm.my mailto:ramayah@usm.my http://www.bloomberg.com/news/2010-06-09/malaysia-seeks-to-lure-individual-stock-traders-decade-after-asian-crisis.html http://www.bloomberg.com/news/2010-06-09/malaysia-seeks-to-lure-individual-stock-traders-decade-after-asian-crisis.html http://www.themalaysianinsider.com/business/article/ordinary-malaysians-shun-stockmarket-amid-stalling-recovery/ diacon, s. (2004). investment risk perceptions: do consumers and advisers agree? the international journal of bank marketing, 22(3), 180-198. dutton, j. e., & jackson, s. e. (1987). categorizing strategic issues: links to organizational action. academy of management review, 12(1), 76-90. doi: 10.5465/amr.1987.4306483 garvey, j. (2010). an investigation into risk propensity in bull and bear markets. journal of risk research, 13(6), 789-804. doi: 10.1080/13669870903560283 gerrans, p., faff, r., & hartnett, n. (2012). individual financial risk tolerance and the global financial crisis. doi:http://dx.doi.org/10.2139/ssrn.1990811 harris, d. r., & woodward, j. a. (1974). kaiser's little jiffy, mark iv for a small computer using basic fortran iv. educational and psychological measurement, 34(1), 119-120. doi: 10.1177/001316447403400116 huff, r. a., keil, m., kappelman, l., & prybutok, v. (1997). validation of the sitkin-weingart business risk propensity scale. management research news, 20(12), 39-48. ibrahim, m. i., & lim, c. f. (1995). profile of individual investors in the klang area. capital markets review, 3, 1-15. jackson, s. e., & dutton, j. e. (1988). discerning threats and opportunities. administrative science quarterly, 33(3), 370-387. james, l. r., & brett, j. m. (1984). mediators, moderators, and tests for mediation. journal of applied psychology, 69(2), 307-321. janis, i. l., & mann, l. (1977). decision making: a psychological analysis of conflict, choice, and commitment. new york: free press. kahneman, d., & tversky, a. (1979). prospect theory: an analysis of decision under risk. econometrica, 47(2), 263-291. kerlinger, f. n. (1986). foundations of behavioral research (3rd ed.). new york: holt, reinhart and winston. lai, m.-m., low, k. l. t., & lai, m.-l. (2001). are malaysian investors rational? journal of psychology and financial markets, 2(4), 210-215. doi: 10.1207/s15327760jpfm0204_5 maccrimmon, k. r., & wehrung, d. a. (1984). the risk in-basket. the journal of business, 57(3), 367-387. mackinnon, d. p., warsi, g., & dwyer, j. h. (1995). a simulation study of mediated effect measures. multivariate behavioral research, 30(1), 41-62. doi: 10.1207/s15327906mbr3001_3 march, j. g., & shapira, z. (1987). managerial perspectives on risk and risk taking. management science, 33(11), 1404-1418. masters, r. (1989). study examines investors' risk-taking propensities. journal of financial planning, 2(3), 151. mishra, s., & lalumière, m. l. (2011). individual differences in risk-propensity: associations between personality and behavioral measures of risk. personality and individual differences, 50(6), 869-873. doi: 10.1016/j.paid.2010.11.037 nunnally, j. c. (1978). psychometric theory (2nd ed.). new york: mcgraw-hill. pablo, a. l. (1997). reconciling predictions of decision making under risk: insights from a reconceptualized model of risk behaviour. journal of managerial psychology, 12(1), 420. shiller, r. j. (2008). the subprime solution: how today's global financial crisis happened, and what to do about it. princeton: princeton university press. singh, j. v. (1986). performance, slack, and risk taking in organizational decision making. academy of management journal, 29(3), 562-585. doi: 10.2307/256224 international journal of banking and finance, vol. 10, iss. 1 [2013], art. 7 http://www.themalaysianinsider.com/business/article/ordinary-malaysians-shun-stock-market-amid-stalling-recovery/ http://www.themalaysianinsider.com/business/article/ordinary-malaysians-shun-stock-market-amid-stalling-recovery/ sitkin, s. b., & pablo, a. l. (1992). reconceptualizing the determinants of risk behavior. academy of management review, 17(1), 9-38. sitkin, s. b., & wiengart, l. r. (1995). determinants of risky decision-making behavior: a test of the mediating role of risk perceptions and propensity. academy of management journal, 38(6), 1573-1592. skeel, r. l., neudecker, j., pilarski, c., & pytlak, k. (2007). the utility of personality variables and behaviorally-based measures in the prediction of risk-taking behavior. personality and individual differences, 43(1), 203-214. doi: 10.1016/j.paid.2006.11.025 staw, b. m., sandelands, l. e., & dutton, j. e. (1981). threat-rigidity effects in organizational behavior: a multilevel analysis. administrative science quarterly, 26, 501-524. thaler, r. h., & johnson, e. j. (1990). gambling with house money and trying to break even: the effects of prior outcomes on risky choice. management science, 36(6), 643-660. von neumann, j., & morgenstern, o. (1944). theory of games and economic behavior (first ed.). princeton: princeton university press. wong, k. f. e. (2005). the role of risk in making decisions under escalation situations. applied psychology, 54(4), 584-607. doi: 10.1111/j.1464-0597.2005.00236.x. optimal inflationary and reserve requirement policies: a study of an economy with an informal sector the international journal of banking and finance, vol. 8, number 1, march 2011: 21-34 21 optimal inflationary and reserve requirement policies: a study of an economy with an informal sector hamid e. ali the american university in cairo ________________________________________________________________ abstract governments in developing economies often resort to taxing bank money balances through imposition of high reserve requirements and also by relying on seigniorage to fi nance their defi cits. in the context of those practices, this research attempts to answer the following questions. first, why do developing economies with an informal sector resort to infl ationary measures to fi nance their activities? second, how does a government induce an agent to choose the formal economy? as to the fi rst question on the trade-off between infl ation and reserve requirements, it is shown that of maximum infl ation and minimum reserve requirements will increase the steady-state utility of an optimizing agent. regarding the second question, the agents prefer the informal economy if policy relies on a maximum reserve requirement. keywords: informal market, financial development, financial repression, seigniorage jel classifi cation: j33, j61, h2 ________________________________________________________________ 1. introduction1 this paper addresses the unintended consequences of the regulations on bank reserve and infl ation fi nancing on the informal and formal sectors of an economy. such regulations are often observed in several developing countries. as a matter of public policy, the terms “informal sector” and “informal economy” 1 i am greatly indebted to professor scott freeman and professor bruce smith for their academic tutelage and intellectual nourishment. i express my deepest gratitude for their many constructive comments and for fi ne-tuning my thoughts with extensive discussions. my thanks also go to my colleague, dilip k. ghosh, to the editors of this journal and to the anonymous referees for their critical suggestions that have resulted in this revised version of the paper. this author is solely responsible for any errors that may remain. ijbf ali: financial suppression under inflation/seigniorage policy 22 the international journal of banking and finance, vol. 8. number 1, march 2011: 21-34 are synonymous, and they refer to a marginally unregulated economy that sometimes not only subordinates itself to the formal economy, but also provides income and a safety net for the economic agent (ilo, 2003; ilo, 2005; castells & portes, 1989). these informal economic arrangements are a rational response by micro-entrepreneurs to defeat over-regulation by the government (de soto, 2000). moreover, greater shares of national wealth in developing countries are uncounted; the average size of the informal sector i n developing economies is 41 percent of gross national in come (gni) (schneider, 2002). effectively, it can be noted that there is an increasing interdependence between the formal and the informal sectors, because the policy designed to target the formal sector has broader implications for the social welfare of the economic agents in the informal sector. in addition, governments in developing economies with informal sectors often resort to taxing money balances through imposition of high reserve requirements on bank deposits; they also rely on seigniorage to fi nance their defi cits.2 so, to fi nance its expenditures, the government uses either bank reserves or seigniorage. however, a higher reserve requirement implies crowding-out of private capital while by using seigniorage the government creates money from thin air, which drives infl ationary pressure. in such policy trade-offs, there is a continuous tension between the macroeconomic stability programmes advised by th e imf and the domestic needs of the developing economies to meet their social and economic obligations. consequently, countries are trapped in cycles of severe reserve requirements and anti-infl ationary measures that have unintended consequences for economic growth and development. therefore, the conventional prescriptions derived from neo-liberal policies have failed to address the existence of the informal economy and the ways in which it complicates a government’s fi scal and monetary options and further represses the fi nancial sector. the fi rst attempt to address this fi nancial repression was made by mckinnon and shaw (1973), who defi ned fi nancial repression as the set of regulatory restrictions that limited the capacity of the fi nancial intermediaries to achieve their full potential. however, fi nancial development is important in promoting economic growth (stiglitz & uy, 1996; levine, 1997; beck, levine & loayza, 2000). for example, based on the fi ndings from surveys of formal and informal fi nancial institutions in ghana, malawi, nigeria, and tanzania, nissanke and aryeetey (2006) stated that the continuous poor performance of fi nancial systems could be partly explained by the high degree of fi nancial market fragmentation. on the basis of the evidence in a recent study by lu and yao (2009), it has been argued that the rise of privileged and unprivileged sectors in the economy can be regarded as a consequence of fi nancial repression. any government action has implications for the welfare of the agent when the government fi nances its defi cits through money creation and reserve 2 defi cits are part of government expenditures, since the government will pay for debt services. using the terms “defi cits” or “expenditures” will not alter the underlying results. with that caveat or clarifi cation, the term “defi cits” instead of “expenditures” is used consistently throughout the paper. also, for simplicity, we allow the government to fi nance defi cits through money creation, holding taxes constant. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 2 optimal infl ationary and reserve requirement policies: a study of an economy with an informal sector: 21-34 23 requirements. steel, aryeetey, hetige and nissanke (1997) have contended that informal fi nancial institutions are an important vehicle for mobilizing household savings, and this study has recommended that informal fi nance be better integrated into fi nancial development strategies. freeman (1987), and freeman and haslag (1993), have used an overlapping generation framework in a fi nancially-repressed environment, which provides an important baseline for subsequent theoretical research on optimal reserve requirements. this paper extends freeman’s and haslag’s models into the fi nancially repressive dual economy model to answer the following questions: first, why do developing economies with informal sectors resort to infl ationary measures to fi nance their activities? secondly, how does a government induce the agent to choose the formal economy? in our case, the model is applied to an environment in which homogenous agents hold different assets in two markets—formal and informal—allowing the government to fi nance defi cits through a reserve requirement and money creation or seigniorage. consistent with the fi ndings of freeman and haslag (1993), we demonstrate that for the fi rst question, on the trade-off between infl ation and reserve requirements, the optimal policy is maximum infl ation and minimum reserve requirements that will increase the steady-state utility of an optimizing agent. regarding the second question, the agents prefer the informal economy if the policy relies on a maximum reserve requirement. therefore, the government is not only optimizing the infl ationary policy but also inducing the agent to favour the forma l economy by minimizing the reserve requirement. the agent uses the storage technology in the informal sector because the optimizing agent predicts that the government is inclined to regulate and suppress fi nancial intermediaries. the best response of the agent in the informal sector is to hide some goods, because the agents do care about the weighted average rate of return on the assets. in a dual economy with formal and informal sectors with an expanding nominal stock of money, the reserve requirement not only serves as a tax on deposits, but also keeps the agent from using the most productive capital to match the return from the informal sector. the reserve requirements will not induce growth in the fi nancial sector. instead, it does the opposite and promotes the informal sector. the heavier the degree of fi nancial regulation, the bigger the scope of the informal economy, and the less seigniorage the government can collect. we use a relatively simple model expounded in section 2 to arrive at the conclusions. section 3 describes the maximization of welfare, and examines policy analysis. section 4 discusses the steady states under different policy regimes, and section 5 is the conclusion of this paper. 2. the model this model is an overlapping-generation model following eastman op cit. we provide an appendix that includes further details on the model derivation. there are three types of assets: money, capital, and informal goods. assets are held ali: financial suppression under inflation/seigniorage policy 24 the international journal of banking and finance, vol. 8. number 1, march 2011: 21-34 either because they are required (money balances) or because they have a high rate of return (capital). we describe the evolution of the economy from time t = 1 onwards. the economy is populated by agents who each live for two periods. in each period t ,1 tn two-period-living agents are born. we assume that 1 tt nnn for each period ,t where n is population growth with a positive constant and implying that total endowment of the economy grows at this rate. agents maximize the twice–continuously differentiable additive utility function ),()( 21 cvcu  where ic denotes the agent’s consumption in i period of life of the economy’s sole consumption goods. the function ),()( 21 cvcu  is strictly concave and strictly increasing in each argument. let u' denote the derivative of ),()( 21 cvcu  with respect to ic . the marginal rate of substitution approaches 0 as approaches infi nity, and approaches infi nity as approaches 0. an agent born in period 1t is endowed with y units of consumption goods when young, and with nothing when old. 2.1 asset allocation decisions there is storage technology in the informal sector. an agent stores th2 of his savings in the informal sector at time t and returns )h(f t2 units of goods at 1t . the minimum requirement to invest in capital technology is yk  . this assumption forces the agent to use intermediation; if the rate of return on capital is x and by assumption )( nx  , the return on capital is higher than the money rate of return. if the economy is growing, then money supply should grow and the rate of return of intermediated capital x should grow even faster as well (champ & freeman, 1994). we assume that the current old generation holds the fi at money, which is an intrinsically worthless piece of paper. the monetary authority can determine the growth of the money supply: 1 tt zmm z is the rate of growth of the fi at money. the government uses seigniorage to fi nance its expenditures. the government requires that a fi xed fraction  of all deposits at fi nancial intermediaries must be held as a reserve of fi at money. )1(  is a fraction of the agent’s investment in the intermediated capital market with rate of return x . ' ' v u 2 1 c c 2 1 c c international journal of banking and finance, vol. 8, iss. 1 [2011], art. 2 optimal infl ationary and reserve requirement policies: a study of an economy with an informal sector: 21-34 25 2.2 behaviour of young agents taking the price level sequence tp , as given, we defi ne tt hh 21 , as the value of agents (fi at money and intermediated capital) and stored goods in the informal sector respectively. young agents at each date t choose tt hh 21 , to maximize: )()( 21 tt cvcu  subject to yhhc ttt  211 (1) )( 212 ttt hfhc   (2) the stock of fi at money available in this intermediated economy is ttt phnm 1 (3) where gh 1t is required reserve and n t gh 1t is aggregate reserve. the price sequence evolves over time to keep up with infl ation. given that the intermediated stock of fi at money is nominal, it was multiplied by the price sequence to obtain the real value of money (champ and freeman 1994). m is the nominal stock of fi at money. 2.3 competitive equilibrium the agent’s problem can be transformed as follows; the young agent allocates his endowment, and diversifi es the allocation of assets in formal and informal markets, because the agent cares about the weighted rate of returns. the entire endowment is allocated for current consumption when young, and saving when old, in the informal and formal markets. let tt hh 21 , denote the total savings of a young agent at time t in the formal and informal sectors, so that the young agent chooses  yehit ,0, to maximize: ))(()( 2121 tttt hfhvhhyu   (4) the fi rst-order condition: (5) (6) ))(()(max 2121 , 21 tttt hh hfhvhhyu )((' )(' 21 21 tt tt hfhv hhyu ali: financial suppression under inflation/seigniorage policy 26 the international journal of banking and finance, vol. 8. number 1, march 2011: 21-34 (7) in the competitive equilibrium, the agent equalizes the marginal rate of substitution to the rate of return in both economies. given ),(1 th any policy that lowers the rate of return has a negative implication for the welfare of the agent, which requires further characterization to the welfare property of our model. 2.4 characterization of stationary equilibrium the market-clearing condition is the real rate of return on fi at money using equation (3): (8) under the assumption ( .xn  ), if money is not dominated in the rate of return, then the agent has no reason to use the storage technology in the informal sector. moreover, no agent or intermediary will choose to hold more fi at money than legally required. the rate of return on saving is the weighted average of the rates of return on money and capital in the formal economy: (9) we can deduce from this equation the following rules:3 i) formal sector, if z = o p = (1-) x ii) informal sector, if z = o  p = )(' 2thf results (i) and (ii) imply that if z = o  )(' 2thf = (1-) x. this simply means that if z = 0, the rates of return in the formal and informal sectors are the same. if the fi at money is not growing, then the government is running a balanced budget, and the agent is indifferent and will allocate its endowment to either market. however, the government does create money as a device to alter the welfare and asset allocation between the formal and informal markets. also from equations (7), and (9) we defi ne the rate of return in the informal sector: x z n hf t )1()(' 2   (10) finally, the government’s budget constraints require that thz g 1 1 1        (11) )(' )((' )(' 2 21 21 t tt tt hf hfhv hhyu z n hn m hn m p p tt t tt t t t 1 1 11 x z n )1( international journal of banking and finance, vol. 8, iss. 1 [2011], art. 2 optimal infl ationary and reserve requirement policies: a study of an economy with an informal sector: 21-34 27 the deduction from the government budget constraints has greater public policy implications as the result of corner solutions.4 for equation (11), the deduction, if z = 0  g = ∞ is unrealistic, so we make further assumptions to maximize the welfare of the future generation of the current young by setting  z 1 to optimize in equation (12) as a counter-reaction of the agent to the government’s action. the government is always faced with trade-offs between infl ation and reserve requirements, as a matter of public policy. our response to our research question—which policy maximizes the welfare of the economic agent?—is that excessive infl ationary policy is preferred to extreme reserve requirements. it is true that the extreme cases remain as extreme, and they rarely occur; the bottom line is that infl ationary policy is the second-best public policy. 3. maximizing the welfare of the future generation the future generation chooses  , 1  z to maximize its welfare. (12) (13) the economy on the right side of the laffer curve, conducting monetary policy through reserve requirements to gain seigniorage, will lower the rate of return in money. given that the government fi nances its expenditures through money creation, equation (13) shows the steady-state utility of the agent under an infl ationary policy regime. since we defi ne  z 1 , for simplicity, we need to defi ne     to determine the sign of equation (13). after cancelling the terms out in further equations, we get equation (15): (14) (15) ))(h(f)(hv)(h)(hyumax 2121 , g)(h)1( 1 n 0 )(1 ' 1 1 h nxh vnx 12 11 )1(' h nhvhn ali: financial suppression under inflation/seigniorage policy 28 the international journal of banking and finance, vol. 8. number 1, march 2011: 21-34 using  from equation (13) in equation (15), we obtain equation (16): (16) for the government to fi nance its expenditures, the optimal repression is to set  as low as possible, which is to say, zero, i.e., 0 1   z . given that the level of current government expenditure is constant, the optimal policy is to set z , which will relax the constraint on reserve requirements and create more opportunity for the agent to obtain the most productive capital in the formal sector. the optimal seigniorage should be obtained by the policy that minimizes the reserve requirements and maximizes infl ation, which is consistent with the fi ndings of freeman and haslag. 3.1 policy to induce agents into the formal sector the agent’s criteria to choose either sector depend on the rate of return  and )(' 2thf , so if )(' 2thf , the agent will prefer the formal sector to the informal economy. if )(' 2thf , the agent will prefer the storage technology in the informal economy to saving in the formal economy. conducting the policy experimentation as shown in table 1, we consider the extreme case of conducting monetary policy and its implication for the rate of return in both sectors. the government monetary policy will change the composition of the assets held by the agent in both markets through the rate of return. using the rate of return in the informal economy in equation (10) we obtain equation (17):      xn xhf t )(' 2 (17) solving for government expenditure in terms , and setting either maximum reserves or minimum reserves, we obtain the comparative results in table 1 below: )1( 1h g  (18) 4. the steady state under policy regimes from table 1 we conclude that for the government to induce agents to hold assets in the formal sector, the optimal policy is a combination of infi nite infl ation and minimum reserve requirements. however if the government conducts its policy 0 )()1( )( 1 1 2 1 h nxh xnh international journal of banking and finance, vol. 8, iss. 1 [2011], art. 2 optimal infl ationary and reserve requirement policies: a study of an economy with an informal sector: 21-34 29 towards maximum reserve requirements, such a policy promotes the informal sector, because the agent will favour storing his/her goods rather than allocating one’s portfolio in the formal economy with a lower rate of return. table 1: comparative static under extreme policy regimes rate of return in the formal sector rate of return in the informal sector policy of infi nite infl ation policy of maximum reserve requirement most governments of developing economies do not rely heavily on reserve requirements, in order to avoid further growth of the informal sector. promoting the size of the informal economy is not in the interest of the government; this could be one of the reasons that developing economies use the infl ationary policy. we might say that the imf prescription for holding maximum reserve requirements is not really helpful to the economies that rely on seigniorage as their main source of revenue. this could be one of the sources of the resistance and tension between the global institutional demand and what country-level policymakers need. 5. conclusions in economies with developed fi nancial intermediation, where the size of the informal sector is small to be negligible, the agents prefer the informal economy if the policy relies on a maximum reserve requirement. the government that minimizes the reserve requirements is not only optimizing the infl ationary policy but also inducing the agent to favour the formal economy. in searching for pareto improvement, using the open-market operations remains a reasonable mechanism designed to offset the transfers of wealth between generations when the government is paying a return on reserves. open-market operations will reward deposit in the formal economy and mitigate the need to use the informal economy unless the rate of return is equalized between the two sectors. this issue is left for future research. author information: hamid ali is an assistant professor in the school of global affairs and public policy at the american university cairo, egypt. he may be contacted at: p.o. box 74, new cairo 11835, egypt. e-mail: hali@aucegypt. edu. 1h gx x 1h ng n 1 2 )(' h gx xhf 1 2 )(' h ng nhf ali: financial suppression under inflation/seigniorage policy 30 the international journal of banking and finance, vol. 8. number 1, march 2011: 21-34 appendix )()( 21 tt cvcu  subject to yhhc ttt  211 (1) )( 212 ttt hfhc   (2) the stock of fi at money available in this economy is ttt phnm 1 (3) 2.3 competitive equilibrium the agent’s problem can be transformed as follows. let tt hh 21 , denote the total savings of a young agent at time t in the formal and informal sectors, so that the young agent chooses  yehit ,0, to maximize: ))(()( 2121 tttt hfhvhhyu   (4) the fi rst-order condition: ))(()(max 2121 , 21 tttt hh hfhvhhyu   (5)      )((' )(' 21 21 tt tt hfhv hhyu (6) )(' )((' )(' 2 21 21 t tt tt hf hfhv hhyu     (7) 2.4 characterization of stationary equilibrium the market clearing condition is the real rate of return on fi at money from (3): z n hn m hn m p p tt t tt t t t                1 1 11 (8) x z n )1(   (9) international journal of banking and finance, vol. 8, iss. 1 [2011], art. 2 optimal infl ationary and reserve requirement policies: a study of an economy with an informal sector: 21-34 31 also from equations (6) and (7) we defi ne the rate of return in the informal sector: x z n hf t )1()(' 2   (10) finally, the government’s budget constraints require that thz g 1 1 1        (11) 3. maximizing the welfare of future generations (12)                                1 121 ''..'.' hvv hh u h u  0)1()1(')(' 11 2 2                         h hv h hf (13) using the envelope theorem we cancel out the following terms: '..' 11 v hh u                  =0 (14) '.)('.' 22 2 v h hf h u                =0 (15) from equation (9) we get 0       0         xnx z n    (16) h(f)(hv)(h)(hyumax 2121 , g)(h)1())( 1 ali: financial suppression under inflation/seigniorage policy 32 the international journal of banking and finance, vol. 8. number 1, march 2011: 21-34 (17) (18) since the government fi nances its expenditures through money creation, equation (19) shows the steady-state utility of the agent under this policy regime.                                1 121 ''..'.' hvv hh u h u  +                         1 1 2 2 )1((')(' h hv h hf (19) using the envelope theorem the following terms in (20) and (21) will cancel out. 0'..' 11            v hh u        (20) 0'.)('.' 22 2            v h hf h u      (21) since we defi ne  z 1 , for simplicity we need to defi ne     to determine the sign of equation (19). after cancelling out the terms in equations (20) and (21) we get equation (26):     n   (22)               12 11 )1(' h nhvhn  (23) 0 l if 0 0)()1()1(' 111 h nxhvhnx l 0 )(1 ' 1 1 h nxh vnx international journal of banking and finance, vol. 8, iss. 1 [2011], art. 2 optimal infl ationary and reserve requirement policies: a study of an economy with an informal sector: 21-34 33 using  from equation (21) in equation (26) we obtain equation (24): 0 )()1( )( 1 1 2 1                   h nxh xnh (24) using the rate of return in the informal economy in equation (11) we obtain equation (25):      xn xhf t )(' 2 (25) references auernheimer, l. (1974). the honest government’s guide to the revenue from creation of money. journal of political economy, 82, 595-606. beck, t., levine, r. & loayza, n. (2000). finance and the source of growth. journal of financial economics, 58(1-2), 261-300. castells, m. & portes, a. (1989). world underneath: the origins, dynamics, and effects of the informal economy. baltimore: john hopkins university press. champ, b. & freeman, s. (1994). modeling monetary economies. new york: john wiley & sons. de soto, h. (2000). the mystery of capital: why capitalism triumphs in the west and fails everywhere else. london: bantam. freeman, s. (1987). reserve requirement and optimal seigniorage. journal of monetary economics, 19, 307-314. freeman, s. (1996). on the optimality of interest-bearing reserves in economies of overlapping generations. economic theory, 7, 557-565. freeman, s. & haslag, j. (1993). on the optimality of interest-bearing reserves in economies of overlapping generations. research paper 9328, federal reserve bank of dallas. international labor offi ce (ilo). (2005). report to the government on employment injury insurance in cambodia: legislation, fi nancing and administration 91 sources. international labor offi ce (ilo). (2003). conclusions on decent work and informal economy. report of the director general. international labor offi ce (ilo). (2002). decent work and the informal economy. report vi, 6th item on the agenda. international labor conference (ilc), 90th session, geneva. levine, r. (1997). financial development and economic growth: views and agenda. journal of economics literature, 35(june), 688-726. ali: financial suppression under inflation/seigniorage policy 34 the international journal of banking and finance, vol. 8. number 1, march 2011: 21-34 lu, susan & yao, y. (2009). the effectiveness of law, fi nancial development, and economic growth in an economy of fi nancial repression. world development, 37(4), 763-777. mckinnon, ronald i. (1973). money and capital in economic development. washington, dc: brookings institution. nissanke, m. & aryeetey, e. (2006). institutional analysis of fi nancial market fragmentation in sub-saharan africa. wider research paper no. 2006/87. helsinki: world institute for development economics research. samuelson, paul a. (1958). an exact consumption loan model with or without the social contrivance of money. journal of political economy, 66, 467482. schneider, f. (2002). size and measurement of the informal economy in 110 countries around the world. working paper. rapid response unit, world bank. shaw & edward. (1973). financial deepening in economic development. new york: oxford university press. stiglitz, j. & uy, m. (1996). financial market, public policy and the east asian miracle. the world bank research observer, 11(2), 249-276. steel, w., aryeetey, e., hetige, h. & nissanke, m. (1997). informal fi nancial markets under liberalization in four african countries. world development, 25(5), 817-830. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 2 performance of china-owned banks in hong kong 72 the international journal of banking and finance, volume 9 (number 3) 2012: pages 72-87 performance of china-owned banks in hong kong xiaoxi zhang and kevin daly university of western sydney, australia ____________________________________________________________ abstract this paper reports results on the performance of mainland china-owned banks operating in hong kong and compares them to hong kong (sar) owned banks and foreign owned banks. in general, the test model performs well under diagnostic tests on variables such as net interest margin, non-interest expense, impaired loans ratio, equity multiplier and ownership structures. profitability, as measured by return on assets and return on equity for chinese owned banks increased over the period 2004-2010. chinese owned banks recorded increased performance in terms of net interest margin and equity multiplier but decreased with respect to non-interest expense and impaired loans ratio. banks having a license also appears to be a major contributor to banks profitability across hksar. compared to hong kong based foreign banks and local hong kong banks, we found that in general the mainland china banks tend to perform poorly across a number of key banking performance indicators. key words: bank performance; china’s financial outward investment; chinese banks in hong kong jel classification: g21, f21, f23 _____________________________________________ 1. introduction this paper constructs a model of the profitability of china owned banks operating in hong kong (hk). compared to other industries, china’s outward investment across the financial sector represents a relatively small proportion of its total dollar investment abroad. however outward financial investment has being increasing gradually especially since 2006. to give an idea of this growth, shown in a figure at later page of this paper, we examine china’s total outward financial investment over 20062009. from the figure, it is clear that the growth in outward financial investments is almost entirely explained by financial investment to hk. 73 discovering the factors driving the outward investment in the area of finance should be as important as that of china’s outward investment in manufacturing and resources sectors. the objective of this research is therefore to analyse the performance of china’s banks with a particular focus on hk. by using firm level data across a range of performance indicators, this research enables researchers to discover the key determinants of china’s outward investment in the banking sector. chinese owned banks comprise approximately twenty five percent of all foreign owned banking assets operating in hong kong. the presence of chinese owned banks operating in hksar allows us to apply firm-specific data in particular parent-specific characteristics of chinese banks performance in hong kong. in order to investigate this issue fully, we compared chinese owned banks with two other groups’ namely local owned banks i.e. hksar and owned foreign banks. the rest of the paper is structured as follows: section one provides an overview of foreign banks (including china’s owned banks) operating in hksar; section two reviews previous studies conducted into measuring banks performance in hk including hypotheses testing; section three provides a description of the data sources and methodology; section five discusses the empirical results; the final section offers some concluding comments and provides some suggestion for chinese bank expansion overseas in the future. figure 1: chinese outward investment in financial industry stocks, total and in hong kong, 2006-2009, $100 million 156.1 167.2 366.9 459.9 137.3 311.826 400.0364 0 100 200 300 400 500 2006 2007 2008 2009 total outward financial investment outward financial investment to hong kong source: statistical bulletin of china’s outward foreign direct investment, 2006, 2007, 2008, 2009. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 6 74 2. foreign banks in hong kong hong kong is one of the world’s largest offshore banking centers, with approximately sixty of world’s largest banks operating within the sar of hong kong: see table 1. table 1: selected list of licensed banks incorporated in hong kong name type country/region bank of china (hong kong) limited china bank of east asia limited hong kong china construction bank (asia) corporation limited china chiyu banking corporation limited china chong hing bank limited hong kong citibank (hong kong) limited united states citic ka wah bank limited china dah sing bank limited hong kong dbs bank (hong kong) limited singapore fubon bank (hong kong) limited taiwan hang seng bank limited hong kong hongkong and shanghai banking corporation limited united kingdom industrial and commercial bank of china (asia) limited china mevas bank limited hong kong nanyang commercial bank limited china public bank (hong kong) limited malaysia shanghai commercial bank limited hong kong standard chartered bank (hong kong) limited united kingdom tai sang bank limited hong kong tai yau bank limited hong kong wing hang bank limited hong kong wing lung bank limited china source: the financial services sector in hong kong, hong kong monthly digest of statistics. traditionally hong kong has been acknowledged globally as having a strong, independent and stable banking system, in addition the hk dollar has over time been recognized as a stable managed currency, the hk economic model is viewed by foreigners as somewhat more liberal than that of mainland china, in addition hk administers a lower taxation regime relative to most developed economies, taken together these characteristics make hk an attractive financial investment destination for multinational banks.. hong kong as an offshore banking center 75 operates a three-tier banking system that includes licensed banks; 1 restricted license banks (rlbs) 2 and deposit-taking companies (dtcs). 3 this three-tier banking system allows hong kong authorities to keep domestic and offshore banking sectors separate from each other. foreign banks which do not qualify for a full bank license usually get rlb or dtc license status which allows them to conduct wholesale and investment banking business on the asian dollar market. in 2005 hong kong was host to 133 incorporated licensed banks, 39 restricted license banks and 35 deposit-taking companies in addition of the 207 authorized institutions 182 were foreign owned. 3. hypotheses and tests this research employs firm-level panel data to examine several performance specific hypotheses. as acknowledge relatively few studies employ industry-level data to test other countries’ financial institutions performance in a third country. the data considered in this study employs chinese, foreign and hong kong owned banks in hong kong from 2004 to 2010, separately. this paper takes into account institutional features such as expenses management, capital liquidity and firm’s ownership. profitability measures the profitability of banks operating across borders can be measured by several variables. this study will consider the following dimensions; return on assets (roa) and return on equities (roe): the roa is used to measure earning ability which is net income after taxes for the year relative to total average assets. roa is an internal performance measure of shareholder value, and it is by far the most popular measure of performance because: (1) it proposes a direct assessment of the financial return of a shareholder’s investment; (2) it is easily available for analysts, only relying upon public information; and (3) it allows for comparison between different companies or different sectors of the economy. the return on equity (roe) is used to measure bank’s profitability which is net income after taxes divided by total equity. however, roa is biased upwards for banks that earn significant profits from off-balance sheet operations 1 they can accept deposits of any size and maturity. they are also allowed to provide current and savings accounts to their clients. 2 they are allowed to accept deposits of hk$500,000 and above. 3 they are licensed to accept deposits of hk$100,000 and above with the minimum maturity of 3 months. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 6 76 such as derivative activities, as these activities generate revenue and expenses but are not recorded as assets (rhoades, 1998). therefore, we employ roe as an alternative measure of profitability. hypotheses in the context, of factors within the control of management the immediate factors, which would have an impact on bank profitability, would be those factors, which affect a bank’s net interest income (guru, et al, 2000). to this extent, the net interest margin could be expected to have a positive influence on bank’s profitability. hypothesis 1: net interest margin has a positive impact on profitability in hong kong generally, bank’s efficiency regards cost control would be reflected in profit. a federal reserve bank of new york study found that enhanced expenses such as salaries and wages, pensions, and other (non-interest) expenses resulted in relatively slower growth in profitability (federal reserve bank of new york 1973, pp. 70-71). hypothesis 2: non-interest expense has a negative impact on profitability in hong kong licensed banks are expected to have some positive value; otherwise no institution would be willing to bear the regulatory costs associated with gaining a licence. previous research has indicated that foreign banks expressed a strong preference for banking licence status upon entering into a foreign market, davis & lewis (1982). thus a bank licence may be considered as a firm-specific advantage. in our model a bank licence is treated as an exogenous variable, as the chinese banks applied for a bank licence, but the decision as to their success or failure was outside their control. hypothesis 3: the possession of a bank license has a positive impact on profitability in hong kong. the ratio of impaired loans to gross loans (npl) is included to capture the effect of credit risk on bank profitability. evidence suggests that impaired loans may increase after periods of increased lending often when senior bank managers under competitive pressure to satisfy shortterm profit targets imposed by owners employ less rigorous lending standards (salas and saurina, 2002; berger and udell, 2006; ruckes, 2004). petersen & rajan (1994) posited that banks concerned for short-run reputation will ease credit standards to boost assets growth and profitability in order to maintain comparability with peer banks. furthermore, disaster myopia 77 may occur when banks find it difficult to acknowledge and assess the possibility of a major shock to their loan books (guttentag and herring, 1986; herring, 1999). consequently, decisions are based on recent information, which is assumed to continue into the future. this myopia may be accompanied by overconfidence (where bank managers believe in their investing skills), which lead them to undertake excessive lending and risk leading to increases in impaired loans in the longer term. overall, we expect a negative relationship between npl and profitability. hypothesis 4: impaired loans ratio has a negative impact on profitability in hong kong the equity multiplier (defined as assets divided by equity) is the reciprocal of the capital-toasset ratio. it provides a gauge of bank’s leverage (debt-to-asset ratio), or the dollar amount of assets pyramided on the bank's base of equity capital. a higher em indicates that the bank has borrowed more funds to convert into asset with the share capital. the higher value of em indicates greater risk for a bank. samad and hassan (2000) found that the higher returns might have been due to higher risky investments by the bank. this is supported by the increased debt equity and equity multiplier ratio. hypothesis 5: equity multiplier has a positive impact on profitability in hong kong. 4. data and methodology this research considers the period from 2004 to 2010, the data was provided by bankscope, the hong kong monetary authority and the chinese financial yearbooks. we cannot obtain earlier data because most data in these databases started from 2004. the general consensus from the literature regards modelling bank profitability favours a linear analysis approach. to this extent, short (1979) and bourke (1989) considered several functional forms and concluded that the linear model produced results as good as any other functional form. in support of this, molyneux et al, (1994) and guru, et al, (2000) had also considered a linear model in their studies on bank profitability. thus, in this study as well we have considered a linear model to analyse pooled cross-section time series data. our modelling begins with running four regressions simultaneously, using return on assets (roa) and return on equity (roe) as dependent variables we itemised our independent variables as net interest margin (margin), non-interest expense (nie) and the ratio of impaired loans to gross loans (npl). net interest margin is a measure of the difference between the interest international journal of banking and finance, vol. 9, iss. 3 [2012], art. 6 78 income generated by banks and the amount of interest paid out to their lenders (for example, deposits), relative to the amount of their (interest-earning) assets. based on the definition, we expect there is a positive relationship between profitability and net interest margin if the ratio is bigger the roa and roe should be larger. theoretically, higher non-interest expense (including employee salaries and benefits, equipment and property leases, taxes, loan loss provisions and professional service fees) will reduce bank profitability which is showed by roa and roe in our study. impaired loans ration represents banks’ loan quality, we expect it has a negative effect on roa and roe. our initial empirical model considered seven of china’s largest owned banks operating in hong kong (table 2). since licensed banks in hong kong can accept deposits of any size and maturity we also separated our banks into licensed/non-licensed groups to find if having a license can be an advantage to bank performance. next, we compared chinese owned banks with two other groups namely: hong kong locally owned banks and foreign (other countries except china) owned banks. finally we ran a regression employing our enlarged sample of banks which included the top 30 banks (nine chinese owned banks, nine local owned banks and eleven foreign owned banks) in hong kong (table 3). in this regression we added one more independent variable s called equity multiplier (em)4 to carry out regression. table 2: top 7 china’s owned banks in hong kong bank name licensed/u nlicensed chinese shareholder name latest total assets (mil hkd) citic bank international limited licensed china citic bank corporation limited 148,209 wing lung bank ltd licensed china merchants bank co ltd 137,075 china construction bank (asia) corporation limited licensed china construction bank corporation 102,728 industrial and commercial bank of china (asia) limited icbc (asia) licensed industrial & commercial bank of china (the) icbc 266,939 boc hong kong (holdings) ltd unlicensed bank of china limited 1,661,040 bank of china (hong kong) limited licensed 1,612,194 nanyang commercial bank ltd licensed 192,175 source: bankscope database, financial services sector in hong kong, and hong kong monthly digest of statistics. 4 equity multiplier = total equity / total assets. 79 table 3: top 30 banks in hong kong according to total assets ranking bank name country/region latest total assets (mil hkd) 1 hong kong and shanghai banking corporation limited gb 5,039,918 2 bank of china (hong kong) limited cn 1,612,194 3 hang seng bank ltd. gb 916,911 4 standard chartered bank (hong kong) limited gb 758,555 5 bank of east asia ltd hk 534,193 6 industrial and commercial bank of china (asia) limited cn 266,939 7 dbs bank (hong kong) limited sg 247,423 8 nanyang commercial bank ltd cn 192,175 9 wing hang bank ltd hk 159,297 10 citic bank international limited cn 148,209 11 wing lung bank ltd cn 137,075 12 dah sing bank, ltd hk 130,509 13 citibank (hong kong) limited us 119,299 14 shanghai commercial bank ltd tw 118,642 15 china construction bank (asia) corporation limited cn 102,728 16 chong hing bank limited hk 74,289 17 fubon bank (hong kong) limited tw 61,780 18 hong kong mortgage corporation limited hk 55033 19 chiyu banking corporation ltd. cn 41,067 20 public bank (hong kong) limited bm 36,848 21 china everbright limited hk 33,942 22 sun hung kai & co. limited hk 20,151 23 boci securities limited cn 11,502 24 daiwa capital markets investments asia hk 1,481 25 china construction bank (asia) finance limited cn 6,175 26 china merchants securities (hk) co.ltd cn 5,011 27 kookmin bank hong kong limited kr 608 28 standard bank asia limited za 457 29 sun hung kai investment services limited hk 3,340 30 orix asia limited jp 409 source: bankscope database. the basic regression model takes the form with parameters (γ, α, βs) and error terms (µ, ε): bank performance measure = γ + )(1 marginβ + )(2 nieβ + )(3 nplβ + µ (1) bank performance measure = α + )(1 marginβ + )(2 nieβ + )(3 nplβ + )(4 aeβ + ε (2) the variables specified as following with time subscript not shown: international journal of banking and finance, vol. 9, iss. 3 [2012], art. 6 80 dependent variables roa : profits after tax / total average assets, roe : profits after tax / total average equity, independent variables margin : (interest income interest payment) / interest-earning assets, nie : non-interest expense/ total average assets, npl : impaired loans (npls) / gross loans, and em : total assets / total equity. in this study, all regression equations were estimated using pooled time series crosssection (tscs) data in order to use the information in the data set. 4. empirical results table 4 presents the regression results for the major seven chinese owned banks operating in hong kong from 2004 to 2010. rows (1-7) provide an indication of the effect of net interest margin on roa and roe for each bank. for example, in row one (citic) banks results from the regression of net interest margin on roa and roe are provided. these results indicate that roa and roe are positively related to margin but only significant for wing lung bank (wl) and nanyang commercial bank (ny). interestingly row seven (nie) indicates the relationship between non-interest expense and roa and roe. the result here indicates that in almost one hundred percent of cases, wl’s nie reduces its return on assets while non interest expenses are also highly significant and negatively related to the bank’s return on equity. however, for other banks this effect is insignificant and positive (except nanyang commercial bank which is showed a negative relation). the last seven rows indicate that negative and significant relationship appears for impaired loans and roa in the case of wl, and negative and significant in the case of citic bank. when we examine the results from table 5 we conclude the following: overall the results for margin indicate that net interest margin has a significant and positive relationship with roa and roe for two banks. our results suggest that licensed banks are more profitable than unlicensed banks. for example, when net interest margin increases by 1%, licensed banks can 81 table 4: top 7 chinese owned banks performance in hong kong, 2004-2010 dependent variable roa roe independent variables margin citic 0.789763 (0.5026) 11.6889 (0.3398) wl 2.155831 (0.0024)*** 14.8658 (0.0323)** ccb 0.40550 (0.2150) 2.06029 (0.5369) boch 0.21412 (0.7858) 2.52278 (0.7569) boc 0.89406 (0.5431) 10.7575 (0.4798) ny 0.96444 (0.0161)** 7.58278 (0.0601)* icbc 0.51584 (0.4993) 7.74798 (0.3292) nie citic 1.04061 (0.3904) 13.1931 (0.2943) wl -3.36211 (0.0024)*** -23.6044 (0.0093)** ccb 0.17781 (0.8058) 1.29047 (0.8630) boch 0.87243 (0.5165) 12.9179 (0.3553) boc 3.24912 (0.3304) 43.5914 (0.2100) ny -0.71437 (0.3586) -5.80317 (0.4691) icbc 0.07584 (0.9750) 1.96536 (0.9374) npl citic -0.36246 (0.3502) -5.72297 (0.0591)* wl -1.33174 (0.0442)** 12.9012 (0.1700) ccb -0.09124 (0.9661) 0.52723 (0.9810) boch 0.00320 (0.9987) -6.61851 (0.7499) boc 1.35090 (0.3912) 8.84538 (0.5852) ny 0.24533 (0.5532) 2.53301 (0.5537) icbc 0.18541 (0.8682) 2.55833 (0.8248) international journal of banking and finance, vol. 9, iss. 3 [2012], art. 6 82 adjusted r-squared 0.61150 0.59241 notes: *, ** and *** denote significance at 0.10, 0.05 and 0.01 levels, respectively. increase (roa) by 0.53305 per cent and (roe) by 15.0740 per cent while unlicensed banks returns are 0.21412 per cent (roa) and 2.3734 per cent (roe) respectively. we also found a positive relationship between nie and profitability (roa and roe), but the results are only significant for roe in the case of licensed banks. for npl and roa, the results are significant and positive for unlicensed banks (0.9989). related to roa, an increase of 1 per cent in npl will cause roe to decline by 2.0610 per cent for licensed banks and 1.001196% for unlicensed banks. table 5: licensed/unlicensed china-owned bank performance in hong kong, 2004-2010 roa roe margin licensed 0.53305 (0.0978)* 15.0740 (0.0800)* unlicensed 0.21412 (0.0819)* 2.37346 (0.0922)* nie licensed 0.21598 (0.8362) 1.70692 (0.0254)** unlicensed 0.87243 (0.5934) 1.23015 (0.8413) npl licensed 0.20962 (0.0799)* -2.06108 (0.0355)** unlicensed 0.9989 (0.0032)*** -1.00119 (0.0524)* adjusted r-squared 0.56629 0.70264 notes: *, ** and *** denote significance at 0.10, 0.05 and 0.01 levels, respectively. the statistics in table 6 indicates the results for hk’s top 15 banks over 2004-2010. in terms of the relationship between net interest margin and profitability the results appear to be mixed across chinese owned banks. margin on bank of china (hong kong) and citic bank is associated with a negative sign but insignificant relationship for roa and roe while for the other two bank groups this effect is positive. regards nie and profitability we found the results ambiguous. for example, the highly significant and negative relationship appears for nie and roa in the case of wing lung bank (-3.3572) and bank of east asia (-1.7503) while for hong kong and shanghai banking corporation this relationship is significantly positive (1.242263). regards the results for nie and roe, we only found evidence for 5 significant results including 2 for chinese owned banks (boc and wlu), 2 for hong kong local banks (boea and wla) 83 table 6: top 15 bank performance in hong kong, 2004-2010 dependent variable roa roe independent variables margin china boc -0.894061 (0.4037) -10.75756 (0.3466) icbc 0.515843 (0.3535) 7.747985 (0.1933) ny 0.964440 (0.0009)*** 7.582789 (0.0117)** citic -0.789763 (0.3571) -11.68895 (0.2031) wlu 2.152847 (0.0000)*** 14.83528 (0.0041)*** ccb 0.405504 (0.0880)* 2.060298 (0.4108) hong kong boea 1.583841 (0.0697)* 16.67085 (0.0733)* wla 2.287652 (0.0218)** 30.64399 (0.0045)*** ds 0.639456 (0.4257) 7.308102 (0.3936) foreign hksh 0.310197 (0.6563) 10.16891 (0.1751) hs 1.612816 (0.1606) 21.82092 (0.0769)* sc 0.945729 (0.0435)** 8.742491 (0.0785)* dbs 0.720075 (0.1323) 6.835456 (0.1793) citi 4.551043 (0.1253) 4.234877 (0.0001)*** scb 0.648829 (0.3288) 4.828662 (0.4945) nie china boc 3.249124 (0.1809) 1.707784 (0.0941)* icbc 0.075847 (0.9657) -1.965366 (0.9168) ny -0.714374 (0.2070) -5.803179 (0.3348) citic 1.040611 (0.2379) 13.19317 (0.1618) wlu -3.357284 (0.0000)*** -23.55725 (0.0034)*** ccb 0.177811 (0.7358) 1.290475 (0.8184) hong kong boea -1.750362 (0.0427)** -16.82504 (0.1856) wla -2.862145 -39.68585 12 international journal of banking and finance, vol. 9, iss. 3 [2012], art. 6 http://epublications.bond.edu.au/ijbf/vol9/iss3/6 84 (0.0826)* (0.0255)** ds 0.008683 (0.9950) 0.039009 (0.9979) foreign hksh 1.242263 (0.1993)* 5.530305 (0.7241) hs -0.623514 (0.0998)* -6.886503 (0.7928) sc -0.352015 (0.0514)* 1.316591 (0.8132) dbs -0.044082 (0.0642)* 1.235044 (0.9062) citi -8.029151 (0.0615)* 4.942662 (0.0000)*** scb 1.172952 (0.5099) 7.375564 (0.6972) npl china boc -1.350908 (0.0387)** 8.845383 (0.4674) icbc 0.185415 (0.8199) 2.558332 (0.7683) ny 0.245330 (0.4154) 2.533016 (0.4304) citic -0.362462 (0.0992)* -5.722973 (0.0599)* wlu -1.334513 (0.0438)** -12.92733 (0.0661)* ccb -0.091242 (0.9535) 0.527238 (0.9748) hong kong boea 0.467813 (0.4459) 2.258221 (0.7295) wla -0.358646 (0.0611)* -4.018856 (0.5828) ds -0.611951 (0.0822)* -6.645590 (0.0769)* foreign hksh -0.738506 (0.0183)** -6.559722 (0.4998) hs -1.071641 (0.0454)** -12.38169 (0.1154) sc -0.287842 (0.0551)* -1.326421 (0.7729) dbs -0.200391 (0.6298) -1.771156 (0.6894) citi -2.071532 (0.0826)* -62.87232 (0.0000)*** scb -0.454543 (0.1850) -2.935596 (0.4194) r-squared 0.811509 0.910842 notes: *, ** and *** denote significance at 0.10, 0.05 and 0.01 levels, respectively. 85 and 1 for a foreign owned bank (citic). in conclusion for these 5 banks we find the return on equity to increase along with non-interest expense for bank of china and citi bank, therefore boc and citi are less efficient than the other three banks due to higher roe used and higher nie. for example when we examine the results of citi and wla banks we find that a rise in non-interest expense by 1 per cent, causes roe to increase by 4.9426 per cent, while for wla roe will increase by 39.6858 per cent if nie decreases 1 per cent. the last fifteen rows suggest that a negative relationship exists between npl and profitability. table 7: top 30 bank performance in hong kong by groups, 2004-2010 dependent variable roa roe independent variables margin chinese owned banks 0.75107 (0.0676)* 5.51634 (0.3565) hong kong local banks -0.16319 (0.8510) 0.61473 (0.8603) foreign owned banks 1.81734 (0.0765)* 1.94773 (0.9255) nie chinese owned banks -0.00866 (0.9904) -0.6861 (0.8134) hong kong local banks 1.16432 (0.4573) -11.1892 (0.0964)* foreign owned banks -0.59615 (0.5559) -3.35882 (0.0436)** npl chinese owned banks -0.05875 (0.0920)* -1.17442 (0.7654) hong kong local banks 0.15767 (0.8955) -15.1238 (0.0104)** foreign owned banks -0.09840 (0.9421) -2.13943 (0.6958) em chinese owned banks -0.00191 (0.9954) 0.39581 (0.0674)* hong kong local banks 0.02214 (0.8511) 3.04913 (0.0001)*** foreign owned banks -0.08582 (0.9390) 1.86858 (0.0801)* adjusted r-squared 0.5693 0.9013 notes: *, ** and *** denote significance at 0.10, 0.05 and 0.01 levels, respectively. table 7 examines the relationship between the top banks (30) performance indicators in hk against the profitability measures roa and roe. rows (1-3) indicate that foreign owned banks will be more profitable on assets than chinese owned banks if they charge the same international journal of banking and finance, vol. 9, iss. 3 [2012], art. 6 86 percentage of net interest margin. unfortunately, we cannot provide any indication of the effect of margin on roe in terms of insignificant results. based on the results for nie indicate that non-interest expense has a negative and significant relationship with roe in local owned banks (-11.1892) and foreign owned banks (-3.3588) overall only one bank group here showed a significant result. regards the results for bank equity multiplier and roe we find a positive and significant relationship between em and roe in all cases for chinese owned banks, and for hong kong owned banks and foreign owned banks. 5. conclusions in this study, we constructed a model of the profitability of chinese owned banks in hong kong. generally, the model proposed performs well for net interest margin, non-interest expense, and impaired loans ratio and equity multiplier. the purpose of this study was to evaluate how banking profitability in hong kong has been affected by these measures. based on four regressions with annual data for the period from 2003-2010 our empirical results provide some interesting outcomes. the results indicate that for the top seven chinese owned banks, wing lung bank has superior recorded profits in terms of returns on assets and return on equity compared to other chinese owned banks. it is interesting to note that licensed chinese owned banks will make more profits than unlicensed chinese owned banks where both have similar net interest margin, management efficiency (non-interest expense) and asset’s risk (impaired loans ratio). comparing the top fifteen banks across our groupings (chinese,hksar and foreign), we found that in general chinese banks tend to perform poorly in terms of margin while profitability is ranked average in terms of non-interest expense and impaired loans. when we applied the equity multiplier indicator, the results support claims that chinese owned banks is the least profitable amongst foreign owned and local owned banks. the choices for china’s banks future are therefore to continue expansion via merger and acquisition with overseas banks in hong kong. this is happening to a degree with european banks at present. the apparent advantages in terms of transferring technical and managerial efficiency gains to china’s banks involved in m&a with foreign banks appears to be the major advantage of such activities. in addition foreign banks operating in hong kong have added to their overall performance in their foreign operations in the fastest growing banking markets globally. 87 author information: xiaoxi zhang is a doctoral student, business school, university of western sydney, australia: email: xiaoxi.zhang@uws.edu.au; phone: 61 04 2540 8180 with post address as building 11, campbelltown, locked bag 1797, penrith, nsw, australia 2751. kevin daly is an associate professor at the same university: email: k.daly@uws.edu.au; phone: 612 4620 3546. references berger, a n, and udell g f., (2006). a more complete conceptual framework for sme finance. journal of banking & finance, , 30(11): 2945-2966. bourke, p., (1989). concentration and other determinants of bank profitability in europe, north america and australia. journal of banking & finance, 1989, 13(1): 65-79. davis, k., and lewis, m., (1982). foreign banks and the financial system, australian financial system inquiry. commissioned studies and selected papers. part 1: macroeconomic policy: internal policy. australian government publishing service, canberra. guru, b k, vaithilingam, s., ismail n., (2000). electronic banking in malaysia: a note on evolution of services and consumer reactions. journal of internet banking and commerce, 5(1): 234-256. guttentag, j, and herring r., (1986). disclosure policy and international banking. journal of banking & finance, 10(1): 75-97. herring, r. j., (199). credit risk and financial instability. oxford review of economic policy, 15(3): 63-79. molyneux, p., lloyd-williams d. m., and thornton, j., (1994). competitive conditions in european banking. journal of banking & finance, 18(3): 445-459. petersen, m. a., rajan, r. g., (1994). the effect of credit market competition on lending relationships. national bureau of economic research. rhoades, s., (1998). the efficiency effects of bank mergers: an overview of case studies in nine mergers. journal of banking and finance, 22 (3), 273–291. ruckes, m., (2004). bank competition and credit standards. review of financial studies, 17(4): 1073-1102. salas, v., saurina, j., (2002). credit risk in two institutional regimes: spanish commercial and savings banks. journal of financial services research, 22(3): 203-224. samad, a., and hassan, m. k., (2000). the performance of malaysian islamic bank during 1984-1997: an exploratory study. thoughts on economics, 10(1-2): 7-26. short, b. k., (1978). the relation between commercial bank profit rates and banking concentration in canada, western europe, and japan. journal of banking & finance, 3(3): 209-219. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 6 mailto:xiaoxi.zhang@uws.edu.au mailto:k.daly@uws.edu.au how to cite this article: kamaruddin, m. i. h., & auzair, s. m. (2020). accountability in malaysian islamic social enterprises (ises): stakeholder versus management perspectives. international journal of banking and finance, 15(2), 47-64. https://doi.org/10.32890/ijbf2020.15.2.3 accountability in malaysian islamic social enterprises (ises): stakeholder versus management perspectives 1muhammad iqmal hisham kamaruddin faculty of economics and muamalat, universiti sains islam malaysia, malaysia 2sofiah md auzair faculty of economics and management, universiti kebangsaan malaysia, malaysia 1corresponding author: m.iqmalhisham@gmail.com; 2sofiah@ukm.edu.my a r t i c l e i n f o article history: received 3 november 2019 revised form 18 january 2020 accepted 31 january 2020 published 31 july 2020 jel code: e 16, l 14, m 41, o 53, p 31, z 12 keywords: accountability, islamic accountability, islamic social enterprise (ise), stakeholder perspective, management perspective. a b s t r a c t this study compares stakeholder and management perspectives on accountability practices in malaysian islamic social enterprises (ises). two sets of questionnaires were used in the survey for this study. the first set was answered randomly by 100 ise stakeholders, and the second set was answered by the management from 102 malaysian ises. the findings showed that both ise stakeholders and ise management have mixed agreement for all six proposed accountability dimensions. based on this result, malaysian ises need to improve their accountability practices, particularly with regard to input, output and procedures. the accountability measurement proposed in this study could also be used by other ises as indicators to evaluate their accountability practices. the international journal of banking and finance, vol. 15. number 2, 2020: 47-64 47 48 the international journal of banking and finance, vol. 15, no 2, 2020 : 47-64 1. introduction the conflict between stakeholders and management of an organisation has long been a topic of debate. based on agency, stewardship and stakeholder theories, this issue is derived primarily from the accountability relationship between the stakeholders and management where there is continuous demand and supply of information founded on the responsibilities of the stakeholders (principal) assigned to the management (agent). the management is responsible for their mandate and accountable for all their actions and activities. this accountability relationship emerged with the agency theory. previous studies on accountability found that the relationship between the principal and agent were opposed to each other in determining the direction of the organisation. later, due to development and the growing population covered by the organisation, scholars shifted focus to the stakeholder theory in order to recognise other interest groups or multiple relationships, especially in nonprofit organisations (ebrahim, 2003). these relationships became more complex with the introduction of the concepts of social enterprise (se) and islamic social enterprise (ise). in this case, a combination of social and economic objectives has increased the number of se/ise stakeholders due to increased participation from various interest groups. this change comes with additional tension because of the different interests of the various groups of stakeholders either among themselves or with the management (ebrahim, battilana & mair, 2014). despite constructive studies and discussions on accountability in se (ebrahim et al., 2014; izaguirre, 2015; sarman, zainon, atan, bakar, yoke, ahmad & shaari, 2015), there are limited studies and discussions on accountability in the ise context. studies on ises are more focused on development concepts (hati & idris, 2014; mohiuddin, 2017; muhamed, kamaruddin & nasrudin 2018), sustainability (abdul kadir & mhd sarif, 2015); governance (muhamed, ramli, shukor & kamaruddin, 2016), financial management (ramli, muhamed & kamaruddin, 2016) and classification (kamaruddin & auzair, 2018). therefore, this has prompted this study to examine ise accountability from both the stakeholder and management perspectives. 2. literature review to date, there is no specific registration, organisational classification, legal standard nor specific authority for se and ise in malaysia. therefore, any nonprofit organisation in malaysia adopting the se concept, which is a combination of both social and economic activities in an organisation, is recognised as se. in addition, there is no specific authorised database or sources of se in malaysia (zainon et al., 2014). furthermore, there is no authority responsible for se and ise registration and monitoring in malaysia. as a result, there have been limited studies conducted on se and ise in malaysia. accountability in malaysian islamic social enterprises (ises): stakeholders vs management perspectives: 47-64 49 for instance, sarman et al. (2015) used the social enterprise alliance (sea) database developed by public initiatives which is still limited to only about 30 ses in malaysia registered in this database. this database is incomplete, and there are no specific characteristics for identifying ses. meanwhile, kamaruddin and auzair (2018) classified ise in malaysia from their economic sectors whether in public, private or non-profit based on their legal establishment. ise could adopt a similar definition of se, but it has to be distinguished by additional islamic practices. islamic organisations like ise should have islamic values such as worship or obedience to allah and operate based on islamic teachings or shariah principles as stated by ahmad (1988). therefore, it can be concluded that ise is an entity which is driven by both social and business objectives according to islamic principles and values (kamaruddin & auzair, 2019b). as se and ise are driven primarily by social objectives, most researchers agreed that se and ise are classified as part of the non-profit sector (defourny & nyssens, 2008). in malaysia, non-profit organisations include societies, associations, foundations and companies limited by guarantee (clbg). each of these non-profit organisations is bound to specific legislations such as the trustee (incorporation) act 1952, companies act 2016, societies act 1966 or states enactment. therefore, it is believed that any non-profit organisation in malaysia that aligns with the objectives of ise can be recognised as ise. meanwhile, defining ise accountability indicators is complex and challenging. as ises involve large groups of stakeholders compared to other nonprofit organisations, any decision and action performed by an ise affects these groups of stakeholders (hyndman & mcmahon, 2011). besides, as an islamic organisation, ise accountability must also encompass islamic accountability dimensions in addition to existing conventional accountability dimensions. this includes islamic social objectives, islamic economic objectives and other islamic principles and values (kamaruddin & auzair, 2019a; 2018). ebrahim et al. (2014) stressed that the problem of multiple accountabilities faced by ises is how to align the interests of various stakeholders, and whose interests to prioritise when those interests conflict. besides, conflicts between social and economic performance indicators in ise due to a combination of social and economic objectives add tension, especially concerning accountability (cheah, amran, & yahya, 2019). moreover, ises tend to have poor financial management practices due to a lack of financial knowledge. this limits their ability to secure financing and expand their economic activities (hynes, 2009). ises are responsible not only for the social needs of various groups of stakeholders, but are required to balance this response with economic needs. moreover, ises normally focus on upward accountability (such as funders and government) and ignore full accountability with downward accountability (such as beneficiaries) (o’dwyer & unerman, 2008). where different groups of stakeholders’ interests come into conflict, there is a need for ises to bring these interests into an optimal balance to ensure fairness and equity among the 50 the international journal of banking and finance, vol. 15, no 2, 2020 : 47-64 stakeholders, as suggested in the ethical model of stakeholder theory (dhanani & connolly, 2012). this ‘balance’ action is needed based on the belief that an organisation has accountabilities and responsibilities to all those whose life experiences may be affected by the organisation’s decisions (cooper & owen, 2007). it is a difficult task as stakeholders often have goals which are sometimes contrary to ise goals (wronka-pospiech, 2016), whereas accountability from the perspective of stakeholder theory is ensuring responsibility to all stakeholders (yuesti, novitasari, & rustiarini, 2016). therefore, in an attempt to identify suitable ise accountability indicators, accountability dimensions must be viewed to concern all possible ise stakeholders and management interests. in this case, candler and dumont (2010) proposed three accountability dimensions which consist of accountability for input, output and procedures as the best measurement to indicate, multiple accountability. under accountability for input, four indicators are suggested which include fund collection, income generation, volunteer resources and brand name (reputation capital). meanwhile, under accountability for output dimension, several indicators are proposed by focusing on goods and services delivered by ise, the impact of these goods and services on stakeholders and also its capabilities to achieve both social and economic objectives. whereas, under accountability for the procedural dimension, the accountability indicators consist of: laws and regulations, mission and vision, ethical and legitimacy aspects. however, depending on these accountability dimensions alone is insufficient without additional islamic accountability dimensions. muhamed et al. (2018) stated that ises differ from social enterprises (se) where ises must have islamic objectives apart from conventional social and economic objectives. this is because islamic objectives known as maqasid shariah, is a vital foundation for every islamic organisation. in addition, ises should embed islamic practices in their daily operations. islamic values such as worshipping or obedience to allah and operating based on islamic teachings, known as shariah principles, as stated by ahmad (1988), are supposedly embedded within ises. in this case, another three accountability dimensions derived from the islamic perspective are proposed as part of the ise accountability indicators. these include accountability for islamic social objectives, islamic economic objectives and other islamic principles and values (kamaruddin & auzair, 2019a, 2019b). based on the above discussion, this study examines ise accountability practices from the stakeholder and management perspectives in order to gain a better understanding for malaysian ises so as to discharge their accountability. specifically, this study attempts to answer the following research questions: 1. do malaysian ises comply with accountability practices from the stakeholders’ perspective? 2. what are the most significant accepted accountability practices from the ise management’s perspective? accountability in malaysian islamic social enterprises (ises): stakeholders vs management perspectives: 47-64 51 3. does stakeholder perception of the accountability practices of ises differ significantly from the perceived importance of ise management? 3. methodology this study utilised a survey to collect the required data. two questionnaires; one for ise stakeholders and another for ise management were developed by focusing on accountability in malaysian ises. based on the literature on accountability for ises, this study proposed six accountability dimensions to be tested in order to examine accountability practices in malaysian ises, which were: (i) accountability for input; (ii) accountability for output; (iii) accountability for procedures; (iv) accountability for islamic social objectives; (v) accountability for islamic economic objectives; and (vi) accountability for other islamic principles and values (candler & dumont, 2010; kamaruddin & auzair, 2019a; 2019b; muhamed et al., 2018). there was a total of 25 items for these six accountability dimensions. each item was tested using a five-point likert-scale where respondents were asked to indicate the level of agreement for each accountability item with ‘1’ for an item that is highly disagreeable and ‘5’ for an item that is highly agreeable. for ise stakeholders, an open survey via google form was selected as a medium of data collection and all respondents involved with ises regardless of their type of involvement were encouraged to answer the questions. using a 95 percent confidence level and a 10 percent margin of error in calculating the sample size (sekaran & bougie, 2016), 96 respondents were required as a minimum sample size from the total malaysian muslim population in malaysia which was 19.42 million (department of statistics malaysia [dosm], 2016). in this case, a total of 100 respondents were received. meanwhile, for the ise management group, companies limited by guarantee (clbg) with additional criteria such as adopting islamic objectives and involvement with islamic charity funds were identified as samples for this study. as of 31 december 2018, there are 2,040 registered clbgs in malaysia. after filtering, some 301 clbgs consisting of 73 clbgs with the name ‘berhad’ and 228 clbgs without the name ‘berhad’ were identified as a suitable sample size for this study. for this cohort, a mail questionnaire approach was chosen by posting the questionnaire to the selected 301 malaysian ises. 105 questionnaires (34.88%) were returned and after cleaning the data, only 102 questionnaires were deemed useable for analysis. for data analysis, this study employed descriptive statistics and independent t-test analysis using the statistical package for social sciences (spss) software in identifying ise accountability practices from the perspectives of ise stakeholders and ise management. detailed information on both ise stakeholders and ise management respondents are shown in tables 1 and 2. 52 the international journal of banking and finance, vol. 15, no 2, 2020 : 47-64 table 1. information profile of ise stakeholders details n = 100 demographic: north region (perlis, kedah, penang & perak) south region (negeri sembilan, melaka & johor) east region (kelantan, terengganu & pahang) central region (selangor, kuala lumpur & putrajaya) sabah & sarawak 15 36 9 36 4 gender: male female 49 51 age: below 20 years old 21 – 40 years old 41 – 60 years old 61 years old and above 3 91 5 1 highest degree of education: mce/spm diploma/a level degree master/phd 4 9 48 39 profession: government employee private employee self-employed housewife student 17 26 10 5 41 income: less than rm1,000 rm1,000 – rm2,000 rm2,001 – rm4,000 rm4,001 – rm6,000 rm6,001 – rm8,000 more than rm8,000 30 21 27 8 7 7 involvement with malaysian ises: regulator/government (upward accountability) donor/contributor (upward accountability) activist/volunteer/member (inward accountability) staff/adviser (inward accountability) partner/ally (downward accountability) beneficiary/client (downward accountability) media/public follower (horizontal accountability) 7 40 44 9 3 15 42 accountability in malaysian islamic social enterprises (ises): stakeholders vs management perspectives: 47-64 53 table 2. information profile of ise management details n = 102 core business: education & training service manufacturing & trading health investment & asset others 52 22 14 7 5 2 years of operation: < 5 years 5 – 10 years 11 – 15 years 16 – 20 years > 20 years 38 28 12 7 17 nature of activity: social and economic activities conducted separately within an entity social and economic activities conducted as separate entities social and economic activities are conducted together/embedded within an entity 26 22 54 financial resource nature: fully funded partially funded self-sustained 42 33 27 social activity level: district state region national international 17 14 10 31 29 economic activity level: district state region national international 21 17 12 30 21 54 the international journal of banking and finance, vol. 15, no 2, 2020 : 47-64 4. analysis of results this study proposed six accountability dimensions: (i) accountability for input; (ii) accountability for output; (iii) accountability for procedures; (iv) accountability for islamic social objectives; (v) accountability for islamic economic objectives; and (vi) accountability for other islamic principles and values (candler & dumont, 2010; kamaruddin & auzair, 2019a, 2019b). ise stakeholders and ise management were required to state their level of agreement on accountability measurements based on these six accountability dimensions. 4.1 accountability for input accountability for input comprised four indicators as suggested by candler and dumont (2010) which are: (i) fund collection; (ii) income generation; (iii) volunteer resources; and (iv) brand name. based on table 3, 72.95 percent of ise stakeholders and 86.4 percent of ise management agreed that malaysian ises complied with accountability practices for input. among these four indicators, the most significant indicator agreed by ise stakeholders was volunteer resources (75.4%), followed closely by fund collection (74.2%), brand name (72.2%) and income generation (70%). on the other hand, the most significant indicator agreed by ise management was fund collection (88.4%), followed closely by income generation (87.8%), brand name (87.6%) and volunteer resources (81.8%). this result was probably due to ise stakeholders prioritising their workforce impact rather than monetary impact, whereas ise management was mostly concerned with the use of financial resources as an accountability issue. regardless of the differences in priorities, all four indicators were accepted by the majority of ise stakeholders and ise management as accountability practices for input in malaysian ises. table 3. accountability for input indicator ise stakeholder perspective (n=100) ise management perspective (n=102) mean % mean % fund collection 3.71 74.2 4.42 88.4 income generation 3.50 70.0 4.39 87.8 volunteer resources 3.77 75.4 4.09 81.8 brand name (reputational capital) 3.61 72.2 4.38 87.6 overall mean 3.65 72.95 4.32 86.4 accountability in malaysian islamic social enterprises (ises): stakeholders vs management perspectives: 47-64 55 4.2 accountability for output accountability for output comprises three indicators according to candler and dumont (2010), which are goods and services, social capital and policy impact. these indicators were tested separately for the social and economic activities conducted on malaysian ises, which comprised a total of six indicators as follows: (i) goods and services on social activities; (ii) goods and services on economic activities; (iii) impact of goods and services on social activities toward stakeholders; (iv) impact of goods and services on economic activities toward stakeholders; (v) achieving social objectives from goods and services delivered; and (vi) achieving economic objectives from goods and services delivered. based on table 4, 73.03 percent of ise stakeholders and 86.97 percent of ise management agreed that malaysian ises complied with accountability practices for output. table 4. accountability for output indicator ise stakeholder perspective (n=100) ise management perspective (n=102) mean % mean % goods and services on social activities 3.70 74.0 4.42 88.4 goods and services on economic activities 3.67 73.4 4.36 87.2 impact of goods and services on social activities toward stakeholders 3.71 74.2 4.32 86.4 impact of goods and services on economic activities toward stakeholders 3.65 73.0 4.3 86.0 achieving social objectives from goods and services delivered 3.63 72.6 4.36 87.2 achieving economic objectives from goods and services delivered 3.55 71.0 4.33 86.6 overall mean 3.65 73.03 4.35 86.97 among these six indicators, the most significant indicator agreed by ise stakeholders was the impact of goods and services on social activities (74.2%), followed closely by goods and services on social activities (74%), goods and services on economic activities (73.4%), impact of goods and services on 56 the international journal of banking and finance, vol. 15, no 2, 2020 : 47-64 economic activities toward stakeholders (73%), achieving social objectives from goods and services delivered (72.6%) and achieving economic objectives from goods and services delivered (71%). the most significant indicator agreed by the ise management was goods and services on social activities (88.4%), followed by goods and services on economic activities and achieving social objectives from goods and services delivered (87.2%), achieving economic objectives from goods and services delivered (86.6%), impact of goods and services on social activities toward stakeholders (86.4%) and the impact of goods and services on economic activities toward stakeholders (86%). these results showed that both ise stakeholders and ise management were more focused on social activities output as compared to economic activities. therefore, it can be concluded that all six indicators were accepted by the majority of ise stakeholders and ise management as accountability practices for output in malaysian ises. 4.3 accountability for procedures accountability for procedures comprises four indicators as suggested by candler and dumont (2010) which are: (i) laws and regulations; (ii) mission and vision; (iii) ethical aspect; and (iv) legitimacy aspect. based on table 5, 81.85 percent of ise stakeholders and 87.85 percent of ise management agreed that malaysian ises complied with accountability practices for procedures. table 5. accountability for procedures indicator ise stakeholder perspective (n=100) ise management perspective (n=102) mean % mean % laws and regulations 4.13 82.6 4.41 88.2 mission and vision 4.17 83.4 4.42 88.4 ethical aspect 4.23 84.6 4.44 88.8 legitimacy aspect 3.84 76.8 4.3 86.0 overall mean 4.09 81.85 4.39 87.85 among these four indicators, the most significant indicator agreed by ise stakeholders was ethical aspect (84.6%), followed closely by mission and vision (83.4%), laws and regulations (82.6%) and legitimacy aspect (76.8%). similarly, the most significant indicator in accountability practices for procedures agreed accountability in malaysian islamic social enterprises (ises): stakeholders vs management perspectives: 47-64 57 by ise management was ethical aspect (88.8%), followed closely by mission and vision (88.4%), laws and regulations (88.2%) and legitimacy aspect (86%). this result indicated that ethics was the top priority under accountability practices for procedures by both ise stakeholders and ise management. therefore, it can be concluded that all four indicators were accepted by the majority of ise stakeholders and ise management as accountability practices for procedures in malaysian ises. 4.4 accountability for islamic social objectives accountability for islamic social objectives comprises four indicators based on the review of literature mentioned, which are: (i) ta’awanu alal birri wattaqwa (islamic cooperation); (ii) amar ma’ruf nahi munkar (commanding the good and forbidding the evil); (iii) fastabiqul khairat (competitive in doing good deeds); and (iv) maslahah ummah (public interest) (kamaruddin & auzair, 2019a, 2019b). based on table 6, 85.45 percent of ise stakeholders and 83.95 percent of ise management agreed that malaysian ises complied with accountability practices for islamic social objectives. table 6. accountability for islamic social objectives indicator ise stakeholder perspective (n=100) ise management perspective (n=102) mean % mean % ta’awanu alal birri wattaqwa (islamic cooperation) 4.27 85.4 4.45 89.0 amar ma’ruf nahi munkar (commanding the good and forbidding the evil) 4.25 85.0 4.36 87.2 fastabiqul khairat (competitive in doing good deeds) 4.27 85.4 3.92 78.4 maslahah ummah (public interest) 4.30 86.0 4.06 81.2 overall mean 4.273 85.45 4.2 83.95 among these four indicators, the most significant indicator agreed by ise stakeholders was maslahah ummah with a score of 86 percent. this was 58 the international journal of banking and finance, vol. 15, no 2, 2020 : 47-64 followed closely by both ta’awanu alal birri wattaqwa and fastabiqul khairat with 85.4 percent. amar ma’ruf nahi munkar scored the lowest percentage in agreement as an islamic social accountability dimension with 85 percent. on the other hand, the most significant indicator agreed by ise management was ta’awanu alal birri wattaqwa (89%), followed closely by amar ma’ruf nahi munkar (87.2%), maslahah ummah (81.2%) and fastabiqul khairat (78.4%). it was predicted that ise stakeholders would demand maslahah ummah as it entailed the most significant impact for stakeholders. whereas, ise management viewed cooperation with its stakeholders as the most important to ensure that they would be able to achieve the targeted social objectives. therefore, it can be concluded that all four indicators were accepted by the majority of ise stakeholders and ise management as accountability practices for islamic social objectives in malaysian ises. 4.5 accountability for islamic economic objectives accountability for islamic economic objectives comprises two indicators, namely: (i) involvement with islamic funds (international shari’ah research academy for islamic finance [isra], 2012); and (ii) avoidance of prohibited economic activities in islam such as riba’ (interest), gharar (uncertainty), and maysir (gambling) (kamaruddin & auzair, 2019a, 2019b). based on table 7, 85.2 percent of ise stakeholders and 87.2 percent of ise management agreed that malaysian ises complied with accountability practices for islamic economic objectives. table 7. accountability for islamic economic objectives indicator ise stakeholder perspective (n=100) ise management perspective (n=102) mean % mean % involvement with islamic funds 4.19 83.8 4.26 85.2 avoidance of prohibited economic activities in islam 4.33 86.6 4.46 89.2 overall mean 4.26 85.2 4.36 87.2 between the two indicators, the most significant indicator agreed by ise stakeholders was avoidance of prohibited economic activities in islam (86.6%) accountability in malaysian islamic social enterprises (ises): stakeholders vs management perspectives: 47-64 59 followed by involvement with islamic funds (83.8%). similarly, the most significant indicator agreed by ise management was avoidance of prohibited economic activities in islam (89.2%) followed by involvement with islamic funds (85.2%). therefore, it can be concluded that both indicators were accepted by the majority of ise stakeholders and ise management as accountability practices for islamic economic objectives in malaysian ises. 4.6 accountability for other islamic principles and values accountability for other islamic principles and values comprises five indicators, as follows: (i) ‘adala (social justice); (ii) ihsan (benevolence); (iii) amanah (trust); (iv) ikhlas (sincerity); and (v) rahmah (compassion) (muhamed et al., 2016; kamaruddin & auzair, 2019a, 2019b). based on table 8, 85.8 percent of ise stakeholders and 85.68 percent of ise management agreed that malaysian ises complied with accountability practices for other islamic principles and values. table 8. accountability for other islamic principles and values indicator ise stakeholder perspective (n=100) ise management perspective (n=102) mean % mean % ‘adala (social justice) 4.26 85.2 4.28 85.6 ihsan (benevolence) 4.31 86.2 4.24 84.8 amanah (trust) 4.32 86.4 4.31 86.2 ikhlas (sincerity) 4.35 87.0 4.34 86.8 rahmah (compassion) 4.21 84.2 4.25 85.0 overall mean 4.29 85.8 4.29 85.68 among these five indicators, the most significant indicator agreed by ise stakeholders was ikhlas with 87 percent, which was the highest score among all indicators suggested in this study. this was followed closely by amanah (86.4%), ihsan (86.2%), ‘adala (85.2%) and rahmah (84.2%). similarly, the priorities in accountability for procedures by ise management were also ikhlas (86.8%), followed closely by amanah (86.2%). despite this, the priorities which came after 60 the international journal of banking and finance, vol. 15, no 2, 2020 : 47-64 was ‘adala (85.6%), rahmah (85%) and ihsan (84.8%). the results indicated that ise stakeholders and ise management seek ikhlas and amanah as the top most important islamic principles and values that need to be realised by malaysian ises. therefore, it can be concluded that all five indicators were accepted by the majority of ise stakeholders and ise management as accountability practices for other islamic principles and values in malaysian ises. 4.7 t-test analysis last but not least, an independent t-test analysis was conducted in order to compare agreement between ise stakeholders and ise management on accountability practices of malaysian ises. the findings are presented in table 9. table 9. accountability for other islamic principles and values accountability dimension t-test for equality of means t df sig mean difference accountability for input -6.841 200 0.000* -0.67113 accountability for output -7.082 200 0.000* -0.69964 accountability for procedures -2.645 200 0.009* -0.30211 accountability for islamic social objectives 0.596 200 0.552 0.07397 accountability for islamic economic objectives -0.841 200 0.401 -0.10275 accountability for other islamic principles and values 0.031 200 0.975 0.00373 * p-value = < 0.05, significant value table 9 shows that accountability for input (p = 0.000), accountability for output (p = 0.000) and accountability for procedures (p = 0.009) are significantly different in practice between ise stakeholders and ise management perspective. in contrast, accountability for islamic social objectives, islamic economic objectives and other islamic principles and values recorded no significant difference in practice between ise stakeholders and ise management. the accountability for other islamic principles and values was the most agreed upon indicator with 97.5 percent for both ise stakeholders and ise management, that malaysian ises observed this accountability practice. this was followed by accountability for islamic social objectives and accountability for islamic economic objectives with 55.2 percent and 40.1 percent for both ise stakeholders accountability in malaysian islamic social enterprises (ises): stakeholders vs management perspectives: 47-64 61 and ise management. therefore, it can be concluded that there are differences in priorities for accountability and islamic accountability practices in malaysian ises for both ise stakeholders and ise management. 5. conclusion this study has provided significant information on the accountability practices of malaysian ises from the perspectives of ise stakeholders and ise management. this study tested six accountability dimensions, which were: (i) accountability for input; (ii) accountability for output; (iii) accountability for procedures; (iv) accountability for islamic social objectives; (v) accountability for islamic economic objectives; and (vi) accountability for other islamic principles and values. all indicators that measured accountability dimensions in malaysian ises were agreed by the majority of ise stakeholders and ise management. among these six accountability dimensions, the most significant for ise stakeholders was accountability for other islamic principles and values (85.8%), followed closely by accountability for islamic social objectives (85.45%), accountability for islamic economic objectives (85.2%), accountability for procedures (81.85%), accountability for output (73.03%) and accountability for input (72.95%). on the other hand, the most significant accountability dimension for ise management was accountability for procedures (87.85%), followed closely by accountability for islamic social objectives (87.2%), accountability for output (86.97%), accountability for input (86.4%), accountability for other islamic principles and values (85.68%) and accountability for islamic economic objectives (83.95%). in addition, the t-test results showed differences in the levels of agreement between accountability and islamic accountability practices in malaysian ises from the perspective of ise stakeholders and ise management. there was a significant difference in the level of agreement on accountability for input and output, and accountability for procedures by malaysian ises, whereas accountability for islamic social objectives, accountability for islamic economic objectives and accountability for other islamic principles and values did not differ significantly between both perspectives. the empirical findings from this study can be used by malaysian ises to enhance their accountability practices in order to gain their stakeholders’ trust, especially for sustainability concerns. the difference in priorities for accountability indicators also provides a basis for which malaysian ises can understand and manage its numerous and diverse groups of stakeholders and thus achieve more efficient value creation for its stakeholders in the future. the following are some limitations of this study and suggestions for future research. as this study selected clbgs as ise management to serve as 62 the international journal of banking and finance, vol. 15, no 2, 2020 : 47-64 respondents, other types of organisations such as societies, associations and foundations which may have different views on accountability could not be considered. therefore, other types of ises could be explored in future for their views on ise accountability practices. apart from this, the sample size of ise stakeholders employed was relatively small; a larger sample for each stakeholder group (upward, downward, inward and horizontal accountability) could produce different views on ise accountability practices. besides, comparisons between these stakeholder groups could also reveal different interests across stakeholder groups. furthermore, due to differences in agreement between ise stakeholders and ise management on accountability practices for input, output and procedures, future research could be conducted to investigate the reasons for such differences in views. last but not least, factors that enhance ise accountability practices could be investigated to improve the performance of ises in the future. 6. acknowledgements the authors would like to thank universiti kebangsaan malaysia for the financial assistance from the research grant, governing the islamic social enterprise towards accountability (gup-2017-101). references abdul kadir, m. a. b., & mhd sarif, s. (2015). social enterprise sustainability: an exploratory case of selected private islamic schools in malaysia. international academic research journal of social science, 1(2), 255263. ahmad, s. f. (1988). the ethical responsibility of business: islamic principles and implications. journal of objective studies, 3(1), 23-43. candler, g., & dumont, g. (2010). a non-profit accountability framework. canadian public administration, 53(2), 259-279. cheah, j., amran, a., & yahya, s. (2019). external oriented resources and social enterprises’ performance: the dominant mediating roles of formal business planning. journal of cleaner production, 236, 117693. cooper, s. m., & owen, d. l. (2007). corporate social reporting and stakeholder accountability: the missing link. accounting, organizations and society, 32(7), 649-667. defourny, j., & nyssens, m. (2008). social enterprise in europe: recent trends and developments. social enterprise journal, 4(3), 202-228. accountability in malaysian islamic social enterprises (ises): stakeholders vs management perspectives: 47-64 63 dhanani, a., & connolly, c. (2012). discharging not-for-profit accountability: uk charities and public discourse. accounting, auditing & accountability journal, 25(7), 1140-1169. dosm. (2016). current population estimates, malaysia, 2014–2016. the office of chief statistician malaysia. department of statistics (dos) malaysia. retrived from https://www.dosm.gov.my ebrahim, a. (2003). making sense of accountability: conceptual perspectives for northern and southern nonprofits. nonprofit management and leadership, 14(2), 191-212. ebrahim, a., battilana, j., & mair, j. (2014). the governance of social enterprises: mission drift and accountability challenges in hybrid organizations. research in organizational behavior, 34, 81–100. hati, s. r. h., & idris, a. (2014). antecedents of customers’ intention to support islamic social enterprises in indonesia. asia pacific journal of marketing and logistics, 26(5), 707–737. hyndman, n., & mcmahon, d. (2011). the hand of government in shaping accounting and reporting in the uk charity sector. public money & management, 31(3), 167–174. hynes, b. (2009). growing the social enterprise–issues and challenges. social enterprise journal, 5(2), 114–125. isra. (2012). islamic financial system: principles & operations. kuala lumpur: international shari’ah research academy for islamic finance (isra). izaguirre, g. a. g. (2015). social enterprises: examining accountability for social and financial performance (unpublished master’s dissertation). queensland university of technology, australia. kamaruddin, m. i. h., & auzair, s. m. (2018). classification of islamic social enterprise in malaysia based on economic sectors. management and accounting review, 17(2), 17-42. kamaruddin, m. i. h., & auzair, s. m. (2019a). conceptualizing islamic social enterprise (ise) from islamic perspective. international journal of management, accounting and economics, 6(4), 368–381. kamaruddin, m. i. h., & auzair, s. m. (2019b). integrated islamic financial accountability model for islamic social enterprise (ise). the journal of muamalat and islamic finance research, 16(1), 17–36. mohiuddin, m. f. (2017). islamic social enterprises in bangladesh: conceptual and institutional challenges. cogent business & management, 4(1), 1305674. muhamed, n. a., ramli, n. m., shukor, s. a., & kamaruddin, m. i. h. (2016). governing the islamic social enterprise (ise). in n. a. muhamed (ed.), critical readings in islamic social finance. vol. 2, yti lecture series (pp. 205-227). negeri sembilan: usim press. 64 the international journal of banking and finance, vol. 15, no 2, 2020 : 47-64 muhamed, n. a., kamaruddin, m. i. h., & nasrudin, n. s. m. (2018). positioning islamic social enterprise (ise). journal of emerging economies and islamic research, 6(3), 28–38. o’dwyer, b., & unerman, j. (2008). the paradox of greater ngo accountability: a case study of amnesty ireland. accounting, organizations and society, 33(7), 801–824. ramli, n. m., muhamed, n. a., & kamaruddin, m. i. h. (2016). the financial management perspectives of islamic social enterprise. in nurul aini muhamed & m. obaidullah (eds.), critical readings in islamic social finance. vol. 2, yti lecture series (pp. 229-254). negeri sembilan: usim press. sarman, s. r., zainon, s., atan, r, bakar, z. a., yoke, s. k., ahmad, s. a., & shaari, n. h. m. (2015). the web-based accountability practices in social enterprises: validating the stakeholder theory. procedia economics and finance, 31, 243–250. sekaran, u., & bougie, r. j. (2016). research methods for business: a skill building approach (7th ed.). john wiley & sons. wronka-pospiech, m. (2016). the identification of skills and competencies for effective management in social enterprises. a managerial perspective. management, 20(1), 40–57. yuesti, a., novitasari, l. g., & rustiarini, n. w. (2016). accountability of non-government organization from the perspective of stakeholder theory. international journal of accounting and taxation, 4(2), 98–119. zainon, s., ahmad, s. a., atan, r., wah, y. b., bakar, z. a., & sarman, s. r. (2014). legitimacy and sustainability of social enterprise: governance and accountability. procedia-social and behavioral sciences, 145, 152157. ijbf7-marina.indd the international journal of banking and finance, vol. 7. number 1: 2010: 35-50 35 do exchange rates affect the stock performance of australian banks? jing chi, david tripe and martin young massey university, new zealand _____________________________________________________ abstract it is expected that banks with signifi cant foreign business should be impacted by relative changes in the currency values of the foreign countries where they do business. using data from january 1997 to march 2007, this study explored this relationship for the four major australian banks. contrary to expectations, no signifi cant relationships between australian bank stock returns and foreign exchange rates were found, raising questions as to the effi ciency of stock markets in recognising banks’ foreign exchange exposures arising from their offshore assets and business. keywords: exchange rates, offshore assets, stock returns, australia, bank international expansion. jel classifi cation: g12, g21. _____________________________________________________ 1. introduction a number of theories has been proposed as to why banks should expand internationally, although all of these, ultimately, relate to banks’ ability to earn profi ts from doing so. major theories to explain international expansion, many of which also apply to multinational fi rms more generally, include the industrial organization theory, the comparative advantage theory, the international investment theory, portfolio theory, the internalisation theory and the eclectic theory. the industrial organisation theory has a number of different strands, including banks following their customers into foreign markets, higher concentration in the home country market providing higher profi ts to support expansion, greater strength or importance of the home country currency and a desire to secure (retail) deposits in the host country. the comparative advantage theory proposes that banks expand from countries with a comparative advantage in the supply of banking services. international investment theory is based on ijbf ht tp :// ijb f.u um .e du .m y 36 the international journal of banking and finance, vol. 7. number 1: 2010: 35-50 the idea that banks expand internationally in order to exploit or avoid market externalities, while portfolio theory argues that international expansion is a risk diversifi cation decision. internalisation theory proposes that banking fi rms expand internationally to get around imperfections in, for example, fi nancial market information, and generate an internal advantage by retaining fi rmspecifi c advantages for their own benefi t. the eclectic theory combines a number of strands, largely from the internalisation theories: these are ownership-specifi c advantages, internalisation specifi c advantages and location-specifi c variables.1 within this range of options, relatively little attention has been given to portfolio theory, which could be used to argue that banks ought to expand internationally to reduce the risks of investing solely in their home markets. the risks that might be reduced include not only the economic conditions applying in particular countries, which have direct effects on bank performance, but also the values of currencies of the countries in which multinational banks do business. with the equity of larger banks now being traded on international markets, a concentration of assets in only one currency would expose shareholders to risks in respect of that single currency. this might be less of an issue for banks whose home markets are larger zones such as the united states or the eurozone, but it will be more important for banks from smaller (and more volatile) currency areas such as australia. grosse and goldberg (1991) found that banks from countries considered risky were more likely to have a foreign offi ce, but an alternative perspective has been that countries can diversify their loan portfolios internationally without an extensive network of foreign offi ces (although a bank’s lending portfolio might then be limited to low-margin, wholesale business). other studies have considered exchange rates as a factor in banks’ foreign investment decisions, but the focus of these has often been on the issue of whether an investment is likely to generate a foreign exchange gain or a loss (e.g. hultman and mcgee, 1989; moshirian and pham, 1999; moshirian, 2001). relatively little attention has been given to looking at internationally-active banks’ assets as a portfolio, with sub-portfolios in different countries exposed to different risks and returns. this paper attempts to fi ll some of the gap in the exploration of the portfolio theory of the international expansion of banks by looking at the four major australian banks, each of which has signifi cant operations outside australia. the specifi c focus is on the impact of changes in the exchange rate of the australian dollar relative to the currencies of the main countries into which these major australian banks have expanded their operations. the effect was measured in terms of the effect on stock returns for those australian banks; the sample period was from 1 january 1997 to 31 march 2007 and the methodology employed is the capital market method. surprisingly, we fi nd no signifi cant foreign exchange impact on the four major australian banks’ stock returns, or 1this summary of the theories and their strengths and weaknesses is very brief. for a more extensive review, please refer to cho (1985), williams (1997), and tripe and matthews (2003). ht tp :// ijb f.u um .e du .m y do exchange rates affect the stock performance of australian banks?: 35-50 37 on the stock returns of the fi ve australian regional banks whose focus is on the domestic market only. questions are then raised as to the effi ciency of stock markets in recognising banks’ foreign exchange exposures arising from their overseas assets and business. the rest of the paper is structured as follows. the next section provides some detail on the international operations of the major australian banks and their history. the third section outlines the key hypothesis that will be explored in this study, while section 4 introduces the data that are used for the analysis. section 5 reports and reviews the results and discusses their implications, while section 6 concludes and identifi es issues for further research. 2. the major australian banks and their international expansion the australian banking market is dominated by four major banks, which together control around 66% of the market by assets (as at january 2007). these are the australia and new zealand banking group limited (anz), commonwealth bank of australia (cba), national australia bank (nab), and the westpac banking corporation (westpac). each of these banks is listed on the australian securities exchange (asx), where they are among the largest listed companies, and each of them has both retail and other banking business outside australia.2 these banks are also signifi cant in international terms, ranked at 59, 60, 38 and 71 respectively by tier one capital in the banker magazine’s 2007 listing. the cba is the youngest of these banks, originally established as a government-owned institution in 1911, but fully privatised since 1996. like all the major australian banks, it has offi ces in the major fi nancial centres of london, new york, tokyo, hong kong and singapore. it expanded into new zealand with the acquisition of 75% of asb bank in 1989, with the remaining 25% acquired in 2000. its other signifi cant international expansion has been into fiji, where it acquired a business with the acquisition of colonial limited (an australian bancassurance group) in 2000. there are some small operations in other countries, but the new zealand business is the largest outside australia. the trend in all banks’ international activities is shown in figure 1. westpac is the longest established of the australian banks, having been originally established as the bank of new south wales in 1817. it has had business in new zealand since the bank of new south wales’ acquisition of the new zealand business of oriental bank in 1861, with the new zealand business increasing in importance following the acquisition of trust bank new zealand in 1996. westpac embarked on an ambitious plan to try and turn itself in to a major multinational bank in the late 1980s, but the somewhat disappointing ensuing 2the following summary of australian banks and their international operations is relatively brief. for a more extensive discussion, please see tripe and matthews (2003) and wilkins (2003). ht tp :// ijb f.u um .e du .m y 38 the international journal of banking and finance, vol. 7. number 1: 2010: 35-50 performance and lending losses in its home market forced the abandonment of this strategy.3 its international activities now comprise the standard network of offi ces in major fi nancial centres and business in new zealand and a number of smaller countries in the pacifi c. figure 1. percentage of assets outside australia major australian banks nab was originally among the more australian focused of the major banks, but in 1987 it purchased a network of retail banks in scotland and ireland. this was followed by further expansion by acquisition into the united kingdom, new zealand and the united states, although in more recent years it has sold both its united states business and that in ireland. in addition to its network of offi ces in international fi nancial centres, it thus now has signifi cant business in england and scotland (trading as the clydesdale and yorkshire banks) and new zealand (where it trades as bank of new zealand). some arguments have been put forward to suggest that a major benefi t of the nab’s expansion into the uk market was to relieve the bank from the consequences of its previous concentration of exposures in the australian banking market (robinson, 1990), consistent with the portfolio theory. for the period from 1988 to 2000, the correlation coeffi cient between the returns on assets for the australian and uk businesses was 0.22, which does lend support to that portfolio effect argument. anz is now the major australian bank with the greatest proportion of its business in new zealand. all of its major constituent banks, the union bank, 0% 10% 20% 30% 40% 50% 60% 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 anz cba nab westpac 3 for a more extensive discussion of these events, refer to carew (1997). ht tp :// ijb f.u um .e du .m y do exchange rates affect the stock performance of australian banks?: 35-50 39 the bank of australasia and the english scottish & australian bank (es&a), started life as british overseas banks, with both the union bank and the bank of australasia having operated in new zealand. the bank sought to increase its international involvement still further in 1984 with the acquisition of grindlays, although this was not wholly successful, and after successive scaling back of the grindlays network, the remains of the bank were sold in 2000. in 2003, the anz increased the proportion of its business in new zealand through the acquisition of the national bank of new zealand. its international network now comprises the standard network of offi ces in major fi nancial centres, its new zealand business (comprising nearly 25% of assets as at 30 september 2006), and in the smaller countries of the pacifi c and through a number of offi ces in east and south-east asia. the position of the preceding big four banks contrasts with that of the other australian banks listed on the asx. these other fi ve so-called regional banks, adelaide bank (adb), bank of queensland (boq), bendigo bank (ben), st george bank (sgb) and suncorp metway (sun), are much smaller and do not, in general, have any non-australian business. we would thus not expect these banks to show the same sensitivity to exchange rates for the australian dollar as the big four, and they have been included in this research in the expectation that there should be different effects for these banks.4 against this background, the next two sections of the paper describe the methodology used to explore the signifi cance of any portfolio effects in the international expansion of the australian banks, and look at the data used to undertake this analysis. 3. methodology hypotheses there are in fact three ways in which banks can be exposed to changes in exchange rates. it is common to recognise the short run exposure that arises from foreign exchange trading, and the medium term exposure arising from mismatches between the currencies in which a bank’s assets and liabilities are denominated. there is also, however, a longer-term exposure, which might be related to a translation exposure that arises from banks’ investments in the banking business in other countries where different currencies are used. in the case of the major australian banks, exposures from foreign exchange trading and from asset and liability mismatches are small. although the banks raise signifi cant portions of their funding in international markets, the vast bulk of this is hedged into australian dollars (staff, 2000; 2002). value-at-risk from 4macquarie bank is also listed on the australian securities exchange, but its business is primarily in investment banking, and bears only limited resemblance to the banks included in this study. ht tp :// ijb f.u um .e du .m y 40 the international journal of banking and finance, vol. 7. number 1: 2010: 35-50 trading exposures is also very small (staff, 2000). the reserve bank of australia further notes that the main foreign exchange exposure of banks arises on the asset side from their equity investment in offshore operations (staff, 2002). becker et al. (2005) showed that banks continued to have relatively low levels of foreign currency exposure, other than in respect of their offshore business. the general principle in this research was to perceive banks as comprising a portfolio of assets. our proposition is that the market performance of banks should respond to changes in the value of their underlying assets making up these portfolios. if these underlying assets are held in different currencies, changes in the values of these currencies ought to impact on the market value of the banks, except insofar as the net foreign currency position relative to those assets is hedged into australian dollars. such hedging would indicate that the bank did not want the relative currency exposures. while the banks we considered primarily in this study do not state precisely their hedging policy in relation to the value of assets held offshore, the common practice is to hedge offshore profi ts but not the offshore assets themselves. there are thus alternative hypotheses that may apply to the different scenarios. it may be that banks do not hedge their currency exposures, and that markets are effi cient, in which case a strengthening of that foreign currency relative to the australian dollar would result in an increase in the bank’s relative share price (refl ected in positive stock returns). alternatively, if banks hedge their foreign currency positions, we would expect no relationship to be evident. the third possibility is that share markets fail to adequately refl ect information about the currencies to which the banks are exposed: this is not the expected outcome, but it is noted that relatively little attention is given to the effects of exchange rate changes in the discussion of bank performance.5 in our study, we use stock returns to evaluate the four major australian banks, with fi ve regional banks included to provide a comparison (and for which different results might be expected). we hypothesize that three major exchange rates, namely us dollar (usd) to australian dollar (aud), sterling pound (gbp) to australian dollar and new zealand dollar (nzd) to australian dollar, would have an impact on the bank’s stock performance. the higher the exchange rates, the stronger the australian dollar is and the weaker the foreign currencies are. the weak foreign currencies will reduce the value of the overseas assets of the australian banks. other things being equal and if the banks do not hedge their foreign exchange exposure, we expect banks’ stock returns to be impacted negatively when exchange rates increase, due to the high proportion of banks’ foreign assets. in our study, we also consider the impact from the market index returns and interest rate and real gdp differentials between two countries on the bank stock returns. market index returns are considered directly, but interest rate and gdp differentials are considered through their expected impact on the exchange rate. 5an exception to this is the case of westpac, where exchange rate differences relative to the new zealand dollar in particular are identifi ed as impacting on reported results. ht tp :// ijb f.u um .e du .m y do exchange rates affect the stock performance of australian banks?: 35-50 41 4. data in this study, we focused on the sample of the four major australian banks (anz, nab, cba and wbc) and use quarterly data from 1 january 1997 to 31 march 2007. we also studied fi ve regional australia banks as a comparison with the four major banks, since the fi ve regional banks have little overseas assets in comparison with the major four. we start our sample period in january 1997 to avoid possible problems caused by two signifi cant events that took place in 1996, namely westpac’s acquisition of a major new zealand asset, trust bank new zealand, and the completion of privatisation of cba. the stock returns of nine banks, the market index returns (asx200), foreign exchange rates between the united states dollar, pound sterling, new zealand dollar and australian dollar, three month inter-bank rates of the us, britain, new zealand and australia, and real gdp of these four countries are collected from the datastream. the summary statistics for our data are provided in table 1. table 1 statistical summary of variables in the study mean median sd min max adb 0.037 0.027 0.074 -0.083 0.194 boq 0.036 0.040 0.078 -0.132 0.169 sgb 0.048 0.038 0.068 -0.056 0.177 ben 0.041 0.038 0.108 -0.207 0.254 sun 0.052 0.071 0.084 -0.152 0.235 anz 0.045 0.045 0.085 -0.140 0.282 nab 0.035 0.047 0.080 -0.219 0.241 cba 0.047 0.054 0.076 -0.160 0.185 wbc 0.041 0.046 0.079 -0.138 0.248 asx200 0.032 0.033 0.052 -0.090 0.109 aud/nzd 0.858 0.864 0.045 0.770 0.938 aud/gbp 2.537 2.527 0.200 2.076 2.882 aud/usd 1.547 1.532 0.232 1.280 1.984 inter-bank (nz-a) 0.998 1.010 1.070 -0.620 3.960 inter-bank (us-a) -1.380 -1.069 1.795 -4.820 0.910 inter-bank (uk-a) -0.110 -0.305 1.308 -1.794 2.611 real gdp (nz/a) 0.120 0.120 0.007 0.108 0.132 real gdp (us/a) 19.137 19.010 2.931 15.510 24.766 real gdp (uk/a) 3.340 3.307 0.267 2.824 3.801 note: the variables include the stock returns of four major australian banks (anz, nab, cba and westpac) and fi ve australian regional banks (adelaide bank, bank of queensland, bendigo bank, st george bank, and suncorp-metway); the asx200 market index returns; three exchange rates (aud/usd, aud/gbp, and aud/nzd); the difference between the three month inter-bank rates of new zealand, us, uk and australia; and the ratio of the real gdp of new zealand, us, uk and australia. since we use gdp fi gures, the frequency of our data is quarterly. the sample period is from 1 jan 1997 to 31 march 2007. ht tp :// ijb f.u um .e du .m y 42 the international journal of banking and finance, vol. 7. number 1: 2010: 35-50 4. empirical results and their implications we use the capital market approach to estimate the sensitivity of stock returns to the change of foreign exchange rates while controlling for market movements. this approach is shown in equation 1: r t =  0 +  m r m,t +  x x t +  t (1) where r t is the stock return adjusted for dividend and rights issues for time t; r m,t is the market index return for time t; x t is the exchange rate factor for time t which comes from equation 2;   the intercept;  m is the market exposure;  x is the foreign exchange exposure; and  t is the error term for time t. previous efforts to fi nd signifi cant exchange rate exposure have not been very successful, although most research so far has used a trade-weighted exchange rate index. since in this study, we are interested in how overseas assets and business affect the stock performance of the four major australian banks, we choose key currency pairs that are most relevant to our study and use them individually in the equation to avoid any offsetting exposure among foreign exchange rates. we also include all three exchange rate factors in the same equation to check how the exchange rate factors impact jointly on banks’ stock performance. besides using key currency pairs, another adjustment to the traditional approach, following martin and mauer (2005), is that the exchange rate factors are constructed to be orthogonal to the interest rate and real economic activity differentials. the orthogonal exchange rate factor is generated from x j,t , the residual or the unexplained part of the exchange rate which is not captured by macroeconomic variables. the model to estimate x j,t is shown in equation 2: xrt j,t =  0 +  1 int j,t +  2 gdp j,t + x j,t (2) where xrt j,t is the value of the australian dollar in terms of country j currency at time t; int j,t is the difference in 90-day interest rates of country j and australia at time t; gdp j,t is the ratio of the real economic activity level in country j to australia at time t; x j,t is residual exchange rate factor for country j currency at time t; and    2 are regression coeffi cients for country j. the exchange rate factor used in equation 1 is the residual exchange rate factor, x j,t, defi ned in equation 2. recent studies have also modifi ed the traditional capital market model to allow an asymmetric sensitivity of returns to exchange rate risk. miller and reuer (1998) and koutmos and martin (2003) argued that stock returns likely respond differently to currency appreciation and depreciation. koutmos and martin (2003) found that asymmetric exposure is common in the fi nancial sector. since our sample fi rms are banks, we also seek to test an asymmetric version of the traditional capital market model, which ht tp :// ijb f.u um .e du .m y do exchange rates affect the stock performance of australian banks?: 35-50 43 includes a dummy variable to capture the potential asymmetries. when x t > 0, then d t = 1. the signifi cance of  d,x shows the asymmetric exposure: r t =  0 +  m r m,t + ( d,x d t ) x t +  t (3) the correlation matrix among independent variables in equations 1-3 is shown in table 2.1 to table 2.3. table 2.1 correlations matrix of independent variables in the study from 1 january 1997 to 31 march 2007 nz-a us-a uk-a nz/a us/a uk/a inter bank (nz-a) 1.000 inter bank (us-a) 0.340 1.000 inter bank (uk-a) 0.496 0.751 1.000 real gdp (nz/a) 0.266* -0.494 -0.252 1.000 real gdp (us/a) -0.147 0.197* 0.264 -0.649 1.000 real gdp (uk/a) -0.173 0.064 0.200* -0.572 0.907 1.000 this table shows the correlation matrix among all independent variables in equation 2. since there are three regressions involved in equation 2, only the three numbers with stars are relevant to test whether there is any multicollinearity in each of the three regressions. the results show that there is no multicollinearity in each of the three regressions using equation 2. nz, us, uk and a stand for new zealand, the us, the uk and australia. table 2.2 correlations matrix of independent variables in the study from 1 january 1997 to 31 march 2007 asx200 renz reuk reus asx200 1.000 renz -0.052 1.000 reuk -0.290 0.433 1.000 reus -0.144 0.336 0.604 1.000 table 2.2 shows the correlation matrix among all independent variables in equation 1. when running equation 1, we have four regressions for each bank. the fi rst three regressions include the independent variables of asx200 index returns and one of the residuals showing the foreign exchange factor. the fourth regression includes the independent variables of asx200 index returns and all three foreign exchange residuals. renz, reuk and reus stand for residuals of nzd, gbp and usd. for individual foreign exchange factor regressions, the correlations are low enough to avoid multicollinearity. however, since the correlation between sterling pound residual and us dollar residual is 0.6, we need to analyse the regression consisting all three foreign exchange factors with caution. ht tp :// ijb f.u um .e du .m y 44 the international journal of banking and finance, vol. 7. number 1: 2010: 35-50 table 2.3 correlations matrix of independent variables in the study from 1 january 1997 to 31 march 2007 renz reuk reus renz*d reuk*d reus*d asx200 renz 1.000 reuk 0.433 1.000 reus 0.336 0.604 1.000 renz*d 0.858 0.308 0.254 1.000 reuk*d 0.139 0.816 0.422 0.096 1.000 reus*d 0.141 0.404 0.753 0.170 0.413 1.000 asx200 -0.052 -0.290 -0.144 -0.165 -0.412 -0.288 1.000 table 2.3 shows the correlation matrix among all independent variables in equation 3. d stands for the dummy variable showing the asymmetry effect of foreign exchange movement. due to the very high correlations among the independent variables, we cannot use equation 3 to test the asymmetric foreign exchange exposure on bank stock returns. the results of the regressions using equation 1, corrected for heteroskedasticity whenever necessary, are presented in table 3. before running equation 2, we used the dickey and fuller (1979) unit root test to check the stationarity of the data and found the data series of foreign exchange, interest and real gdp differentials to be non-stationary, as were the residuals of the regressions. we therefore took a fi rst difference of all variables and re-ran equation 2 as a result of which the residuals become stationary. the residuals from equation 2 were then used in equation 1 to investigate the foreign exchange effects on the bank stock returns. we ran equation 1 for all nine banks. for each bank, there were four regressions, three regressions with single foreign exchange factors, and one with all three factors together. this table shows the results on equation 1: r t  m r m,t   x x t   t where rt is the stock return adjusted for dividend and right issues for time t of nine banks; r m,t the market index return (asx200) for time t; x t the exchange rate factor for time t which are the residuals from equation 2. renz, reuk and reus stand for residuals of nzd, gbp and usd. we run the equation 1 for all nine banks. for each bank, there were four regressions, including three regressions with three foreign exchange factors respectively and one with all three factors together. the numbers in the parentheses are t-statistics. *, **, and *** denote statistical signifi cance at the 10%, 5%, and 1% level, respectively. full results are reported in table 3, and we noted that the market index has a signifi cant positive impact on the bank returns, except for bank of queensland6. 6the bank of queensland has been subject to some activity by large investors, which would be likely to account for the differences in its share price performance relative to the wider market. ht tp :// ijb f.u um .e du .m y do exchange rates affect the stock performance of australian banks?: 35-50 45 however, in all individual foreign exchange factor regressions, the coeffi cients of the foreign exchange factors are insignifi cant, with no difference between the major four banks and fi ve regional banks. when putting all three foreign exchange factors together in the same regression, the new zealand dollar factor table 3 regression results on foreign exchange exposure of australian banks constant asx200 renz reuk reus adjusted r-square durbin-watson stat anz 0.017 0.875 -0.601 0.253 2.648 (1.257) (3.853)*** (-0.409) 0.020 0.802 -0.968 0.276 2.581 (1.432) (3.437)*** (-1.153) 0.017 0.873 -0.303 0.251 2.609 (1.257) (3.802)*** (-0.217) 0.020 0.795 0.066 -1.340 0.980 0.241 2.629 (1.412) (3.318)*** (0.040) (-1.190) (0.557) nab 0.013 0.699 1.664 0.184 2.202 (1.025) (3.412)*** (0.942) 0.012 0.741 0.687 0.169 2.333 (0.841) (3.152)*** (0.811) 0.013 0.704 0.743 0.161 2.364 (0.933) (3.078)*** (0.533) 0.012 0.724 1.422 0.339 -0.037 0.140 2.239 (0.865) (3.013)*** (0.858) (0.300) (-0.021) cba 0.023 0.775 1.006 0.247 2.399 (1.857)* (3.807)*** (0.764) 0.022 0.803 0.452 0.242 2.460 (1.756)* (3.769)*** (0.589) 0.024 0.747 -0.828 0.244 2.370 (1.920)* (3.630)*** (-0.660) 0.022 0.802 1.019 0.941 -2.067 0.244 2.344 (1.758)* (3.751)*** (0.691) (0.938) (-1.317) wbc 0.010 0.977 1.861 0.398 2.312 (0.875) (5.125)*** (1.510) 0.012 0.900 -0.768 0.380 2.290 (1.059) (4.457)*** (-1.057) 0.012 0.930 -1.350 0.383 2.208 (0.990) (4.773)*** (-1.136) 0.014 0.868 3.136 -1.106 -1.267 0.441 2.287 (1.204) (4.509)*** (2.361)** (-1.223) (-0.897) (continued) ht tp :// ijb f.u um .e du .m y 46 the international journal of banking and finance, vol. 7. number 1: 2010: 35-50 constant asx200 renz reuk reus adjusted r-square durbinwatson stat adb 0.019 0.588 1.986 0.166 2.020 (1.483) (2.824)*** (1.476) 0.020 0.554 -0.221 0.118 1.946 (2.076)** (2.483)** (-0.299) 0.019 0.582 0.435 0.119 1.984 (1.984)* (2.714)*** (0.399) 0.021 0.518 2.592 -1.142 0.727 0.148 1.966 (1.622) (2.350)** (1.706)* (-1.104) (0.449) boq 0.026 0.315 1.496 0.014 2.017 (1.829)* (1.320) (0.971) 0.026 0.307 0.060 -0.011 2.110 (1.800)* (1.221) (0.067) 0.027 0.284 -0.761 -0.004 2.148 (1.874)* (1.172) (-0.514) 0.027 0.292 1.933 0.103 -1.474 -0.020 2.049 (1.824)* (1.148) (1.104) (0.086) (-0.791) sgb 0.027 0.656 1.130 0.221 2.018 (2.447)** (3.537)*** (0.942) 0.027 0.657 0.122 0.203 1.979 (2.387)** (3.355)*** (0.173) 0.027 0.664 0.719 0.211 2.002 (2.402)** (3.523)*** (0.625) 0.028 0.638 1.208 -0.474 0.798 0.185 2.026 (2.411)** (3.206)*** (0.881) (-0.508) (0.547) ben 0.019 0.699 0.680 0.066 2.035 (0.988) (2.171)** (0.327) 0.016 0.786 1.147 0.086 2.102 (0.844) (2.364)** (0.959) 0.019 0.700 0.303 0.064 2.098 (0.982) (2.154)** (0.153) 0.016 0.794 -0.110 1.626 -1.245 0.041 2.123 (0.810) (2.323)** (-0.047) (1.013) (-0.496) sun 0.027 0.769 1.884 0.211 2.028 (1.987)* (3.349)*** (1.268) 0.027 0.770 0.201 0.178 2.086 (1.921)* (3.148)*** (0.228) 0.027 0.783 1.225 0.192 2.170 (1.928)* (3.339)*** (0.856) 0.029 0.738 2.012 -0.808 1.374 0.182 2.076 (1.996)* (3.011)*** (1.190) (-0.702) (0.764) table 3 regression results on foreign exchange exposure of australian banks ht tp :// ijb f.u um .e du .m y do exchange rates affect the stock performance of australian banks?: 35-50 47 has some impact on adelaide bank and westpac. the p-values for the two coeffi cients are 9.69% and 2.39% respectively.7 moreover, because we have run a total of 36 regressions, chance is likely to cause some to appear to be signifi cant, even if there were no valid relationships in reality.8 the adjusted r2 statistics for the regressions reported in table 3 are, except for westpac, all below 0.3, and in westpac’s case the highest adjusted r2 reported was 0.441. a signifi cant proportion of the variability in banks’ stock returns is thus not explained, and may thus be attributable to interest rate levels, bank-performance specifi c factors, etc. because there is a corpus of literature (samuelson, 1945; flannery and james, 1984) that has looked at the impact of interest rates on banks’ share prices, we tried including interest rate returns as an additional explanatory variable in our regression analyses (equation 1). interest rate factors were not in general found to be signifi cant, and there was no clear improvement in the explanatory power of our new regressions. as a further test for the robustness of our fi ndings, we ran regressions looking at all four major banks together as a panel, fi rstly looking at the currency effects individually, and then together. we also ran these models with interest rate returns as an additional explanatory variable. once again no signifi cant relationships were found. a comment on an earlier version of this paper suggested that we should be looking at the raw exchange rate changes as well as the residual exchange rate factors. as a further check on our results, we therefore undertook regressions using only the raw exchange rates, but once again failed to fi nd any signifi cant results for anything other than the market index. we also sought to use equation 3 to test the asymmetric exposure of the three currencies on the banks’ stock returns. however, when we explored the correlations among the independent variables, we found that each foreign exchange residual and the product of the residual and the dummy variable were highly correlated (table 2.3). we were therefore not able to use equation 3 to further our study. 5. conclusions the fi ndings of our research are surprising. there is no evidence that the market reacts to the changing value of bank assets held outside of australia for the four major australian banks. the basic theoretical proposition that we sought to explore is the portfolio theory of the international expansion of banks, 7 we were concerned at the possibility of multicollinearity, in view of the relatively high correlations between the gbp and usd exchange rate factors in particular, but values of the variance infl ation factors were low, suggesting that this was not a problem in practice. 8on the other hand, until 2003, westpac was the bank with the largest proportion of its assets in new zealand, although its fi gures were not much larger than for anz. adelaide bank has not at any stage had any new zealand business. ht tp :// ijb f.u um .e du .m y 48 the international journal of banking and finance, vol. 7. number 1: 2010: 35-50 which suggests that banks expand internationally to diversify their risks, and to reduce their relative exposure to their home country markets. it also entails a view of banks as a portfolio of businesses operating in different countries, and that banks’ market values should refl ect the international composition of their portfolios, with share values rising (and positive stock returns) as the home country’s currency weakened against those of the foreign countries in which the bank invested. in presenting our research hypothesis, however, we identifi ed a number of potential outcomes, not dependent on the portfolio theory of the international expansion of banks. we suggested that banks might not hedge their currency exposures, and that markets are effi cient, in which case a strengthening of that currency relative to the australian dollar would result in an increase in the bank’s relative share price (refl ected in positive stock returns). alternatively, if banks hedged their foreign currency positions, we would expect no relationship to be evident. the third possibility is that share markets fail to adequately refl ect information about the currencies to which the banks are exposed through their asset portfolios.9 review of the banks’ reported risk management policies and the commentary provided by the reserve bank of australia both suggest that banks do not hedge their structural foreign exchange exposures arising from their international business operations. this would therefore suggest that markets are not effi ciently pricing the shares of australian banks. another possible explanation is that investors are not primarily focused on asset values, but rather on profi t streams. insofar as banks are hedging current profi ts (as they suggest in their policies for management of foreign exchange risk), investors may perceive banks as protected against exchange rate changes, and share prices would thus not be sensitive to these changes. alternatively, exchange rate changes may be perceived as relatively short term, with the currency likely to revert to a mean in due course. this might apply particularly in the case of the new zealand dollar which is a commoditybased currency similar to the australian dollar. another possibility would be that investors might desire the currency exposure consequent upon investing in the major banks, meaning that they would be largely unfazed by exchange rate changes. it may be noted that the period from 1997 to 2007 (as covered by this research) was a positive one for the australian economy in general, and for the banks as well. banks have also enjoyed positive outcomes in new zealand. different results might be found if there were to be some less benign economic experience. it would also be interesting to look at banks from other countries which have undertaken signifi cant international expansion. martin and mauer (2005) have applied an approach similar to this to look at a group of united 9the question then arises as to whether this might also be the case for other classes of fi rms with net foreign exchange exposures. ht tp :// ijb f.u um .e du .m y do exchange rates affect the stock performance of australian banks?: 35-50 49 states banks, but it would also be possible (and meaningful) to look at banks from canada, the united kingdom, spain and a number of other countries whose banks have expanded into different currency areas. author information: submitting author, professor martin young, department of economics and finance, massey university, private bag 11-222, palmerston north, new zealand, phone: +(64-6)-3505799 ext 2482, fax: +(64-6)-3505651, email: m.young@massey.ac.nz. references becker, c., debelle, and g., fabbro, d. (2005, december). australia’s foreign currency exposure and hedging practices. reserve bank of australia bulletin, 1-8. carew, e. (1997). westpac: the bank that broke the bank. sydney: doubleday. cho, k.r. (1985). multinational banks: their identities and determinants. ann arbor: umi research press. dickey, d.a., and fuller, w.a. (1979). distribution of the estimators for autoregressive time series with a unit root. journal of the american statistical association, 74 427-431. flannery, m.j., and james, c.m. (1984). the effect of interest rate changes on the common stock returns of fi nancial institutions. journal of finance, 39,(4) 1141-1153. goldberg, l.g., and kabir, r. (2002). the stock market performance of the central banks of belgium and japan. journal of economics and business, 54, 137-152. grosse, r., and goldberg, l.g. (1991). foreign bank activity in the united states: an analysis by country of origin. journal of banking and finance, 15, 1093-1112. hultman, c.w., and mcgee, l.r. (1989). factors affecting the foreign banking presence in the u.s. journal of banking and finance, 13, 383-396. koutmos, g., and martin, a.d. (2003). asymmetric exchange rate exposure: theory and evidence. journal of international money and finance, 22, 365-383. martin, a.d., and mauer, l.j. (2005). a note on common methods used to estimate foreign exchange exposure. journal of international financial markets, institutions and money 15, 125-140. miller, k.d., and reuer, j.j. (1998). asymmetric corporate exposures to foreign exchange rate changes. strategic management journal, 19, 1183-1191. moshirian, f., and pham, t. (1999). cost of capital and australia’s banking investment abroad. applied financial economics, 9, 295-303. moshirian, f. (2001). international investment in fi nancial services. journal of banking and finance, 25, 317-337. ht tp :// ijb f.u um .e du .m y 50 the international journal of banking and finance, vol. 7. number 1: 2010: 35-50 robinson, d. (1990, august). aussie charmers nab the profi ts. euromoney, 4445. samuelson, p.a. (1945, march). the effect of interest rate increases on the banking system. american economic review, 16-27. staff. (2000, august). foreign exchange exposures of australian banks. reserve bank of australia bulletin, 43-49. staff. (2002, august). australia’s foreign currency exposure and hedging practices. reserve bank of australia bulletin, 56-60. tripe, d. and matthews, c., (2003). the international expansion of australian banks. in lőnnborg, m; olsson, m.; rafferty, m. & nalson, i (eds.), money and fi nance in transition (pp. 155-180). huddinge, sweden: sődertőrns hőgskola. wilkins, s. (2003). the international diversifi cation of australian banks. in lőnnborg, m: olsson, m.; rafferty, m. & nalson, i (eds.), money and fi nance in transition (pp. 181-196). huddinge, sweden: sődertőrns hőgskola. williams, b. (1997). positive theories of multinational banking: eclectic theory versus internalisation theory. journal of economic surveys, 11(1), 71100. ht tp :// ijb f.u um .e du .m y leverage, maturities of debt and stock performance 74 the international journal of banking and finance, volume 10 (number 1), 2013: pages 74-93 leverage, maturities of debt and stock performance tristan nguyen and alexander schüßler whl graduate school of business and economics, germany and hhl leipzig graduate school of management, germany ___________________________________________________________ abstract we add to the prior literature that test the influence of total leverage on stock returns by focusing on an extended ratio, namely, ‘total debt to (total capital + long term debt)’, td/(tc+ltd)’, the ratio henceforth. further, and in contrast with others, we account for different maturities of debt. the link between this ratio and stock returns for periods of one to sixty months are considered for germany, the uk and the us. we control for beta and form quintiles based on the ratio to compute mean returns. our findings indicate a robust negative relation between the ratio and returns for germany and the uk. in these two markets, the lowest ratioquintile performs better than the highest ratio-quintile for all the periods studied. interestingly, the results for the united states are less clear. due to a number of known factors, market efficiency might be higher in the us than in the other two markets. keywords: stock returns, leverage, debt maturity, long term debt, short term debt jel-classification: g11, g12, g17 _______________________________________________________________ 1. introduction the effect on stock returns by different accounting-based ratios has been studied via numerous empirical tests. for example, the relation between price-book ratio and stock returns is found to be negative (debondt and thaler, 1987). dividend yield, in contrast, has a positive link with stock returns (keppler, 1991). in contrast and though tested several times, the empirical relation between leverage and stock returns is still unsettled. in a perfect market, the relation between returns and leverage should be positive. modigliani and miller (1958) confirm this prediction for a restricted sample of utilities. several other studies also analyze the link between total debt and stock returns. however, findings are inconclusive. some studies find a positive relation (bhandari, 1988; dhaliwal, heitzman and zhen, 2006), but others find a negative relation between leverage and stock returns. (arditti, 1967; korteweg, 2004; george and hwang, 2009) given the mixed results, we want to shed light upon this issue. in comparison to other studies, we also distinguish between short-term and long-term debt. this is an important point because short-term debt and long-term debt have fundamentally different characters and are employed for divergent goals. that affects cost of equity. kose (2011) studies the relation between ‘short-term debt/total capital’ and equity returns as well as the relation between ‘long-term debt/total capital’ and equity returns. he finds that stock returns increase with ‘short-term debt/total capital’; however, the relation between ‘long-term debt/total capital’ and stock returns is negative, albeit insignificant. we use a different approach. contrary to kose (2011), we distinguish between different maturities of debt within one ratio, namely ‘total debt to (total capital + long term debt)’ [td/(tc+ltd)]. in comparison to ‘total debt/total equity’, ‘total debt/total capital’ or other conventional debt ratios, our ratio has the advantage that it treats short term debt and long term debt differently. therefore, it accounts for the different character of debt, depending on maturity. we run regressions for stock markets in germany, the united kingdom, and the united states. these countries have the largest exchanges by value of share trading in the western world (world federation of exchanges, 2012). consequently, these markets are particularly important for investors and firms. moreover, we form quintiles depending on td/(tc+ltd) and calculate average abnormal returns for those quintiles for each market. in the process, we demonstrate what investment performance to achieve by choosing stocks based on td/(tc+ltd). the holding periods that are chosen range from one month to five years. therefore, results are highly relevant for short-term traders as well as long-term investors. in comparison to other indicators like price-earnings ratio or price-book ratio, leverage is often ignored in the financial community. therefore, in addition to academia, findings are also interesting for practitioners. the structure of the paper is as follows. in the next section we give an overview of the relevant literature. section 3 describes the methodology of the study. section 4 shows the results and section 5 interprets the findings. finally, in section 6 we conclude the paper and highlight major results. 2 international journal of banking and finance, vol. 10, iss. 1 [2013], art. 5 2. literature review in an influential work, modigliani and miller (1958) show that in a perfect capital market, risk should increase with leverage. as a result, they expect cost of equity and stock returns to increase with leverage, which is known as proposition 2. modigliani and miller confirm their theory using a constricted sample of utilities. considering corporate level taxes, they suggest that the deduction of interest expenses leads to a tax benefit for levered firms. this tax benefit raises the company value (modigliani and miller, 1963). later, miller (1977) accounts for the influence of personal taxes. according to miller, the relative taxation of debt and equity income increases the equity risk premium from leverage. after the work of modigliani and miller however, empirical studies show inconclusive results. korteweg (2004) and muradoglu and sivaprasad (2012a) conduct more comprehensive tests and find that stock returns and leverage are negatively related. these results contradict one of the basic principles in traditional financial theory (kose, 2011). korteweg (2004) uses a time series approach and conducts a study based on capital structure changes. he supposes that investors are overly optimistic about the prospects of highly levered companies and overbuy those stocks. muradoglu and sivaprasad (2012a) show that investing in firms with low leverage earns abnormal returns of about 4.4 percent per annum. they find that abnormal returns are not timevarying and exist in most industries. caskey, hughes and liu (2012) fragment leverage into a target component and an excess component and show that the negative relationship is driven by the excess component. they attribute their finding to investors´ failure to respond immediately to news about the likeliness of distress as well as future asset growth which is included in the excess component of leverage. muradoglu and sivaprasad (2012b) discover that while abnormal returns decline in firm-leverage, they increase in industry-leverage, meaning that industry returns rise as the average leverage ratio within an industry rises. in contrast to some of the prior studies, bhandari (1988) finds a positive association between equity returns and leverage on the firm level, confirming the results of modigliani and miller for a more extensive sample. he controls for beta and size and uses inflation adjusted returns. in his analysis, he includes financial firms and does not account for different risk classes. penman, richardson and tuna (2007) disaggregate the leverage component of the book to market ratio relating to financial risk from the leverage component relating to operational risk. they find that stock returns decrease in book leverage but increase in market leverage. dhaliwal, heitzman and zhen (2006) illustrate that the effect of leverage on cost of equity is mitigated by corporate taxes. kose (2011) accounts for different maturities of debt; he divides total debt into short-term debt and long-term debt. kose finds that stocks with higher short-term debt earn higher returns and stocks with higher long-term debt earn lower returns. however, the relationship between longterm debt and stock returns is not significant. in asset pricing models, leverage is mostly ignored due to mixed results. the conflicting evidence of the relation between leverage and equity returns motivates us to shed light on this issue. in practice, knowing the empirical relation between returns and leverage is highly relevant for investors and managers because of its importance for increasing investment performance and lowering cost of capital. we focus on the ratio td/(tc+ltd) because it takes account of different maturities of debt. due to mixed results of former studies, we start with an open relation between td/(tc+ltd) and stock returns. 3. data and methodology we obtain both accounting and stock market data from datastream and use the time series from 1996 to the first quarter of 2012. this time-span ensures that we have a reasonable number of observations. we conduct our study for three countries: germany, the united kingdom and the united states. for each country we use 100 companies. for the united states and the united kingdom we take the companies of the s&p100 and the ftse100, respectively. the s&p100 is an index which incorporates 100 us-american large cap stocks. the ftse100 is the most prominent stock market index in the united kingdom for large companies. for germany, there is no equivalent index which contains exactly 100 stocks; therefore our german sample is made up of companies from different stock market indices. we take the companies of the dax 30, the companies of the mdax 50 and the 20 largest companies of the sdax to get 100 companies for germany as well. for all countries, we exclude financial companies because they have higher leverage. financial companies are defined as banks, insurance companies and investment firms. moreover, we eliminate companies which alter their fiscal year end date within our research period and companies for which we do not obtain matching stock prices or fundamental ratios for the time span. after eliminating these companies, our resultant sample for the german market consists of 67 companies and 1070 observations. for the united kingdom we have 60 companies and 944 observations and for the united states we have a resultant sample of 78 companies and 1244 observations. using the above data, we conduct a panel analysis. our regressions are for holding periods of one, three, six, twelve, 36 and 60 months. the dependent variable is abnormal return (ar). we employ consecutive closing prices adjusted for dividends splits and right-issues and calculate international journal of banking and finance, vol. 10, iss. 1 [2013], art. 5 returns from the first trading day of june. hence, we ensure to have a time interval of at least five months between fiscal year end and the date from which returns are computed. thereby, we avoid the look-ahead bias. even though, abnormal returns refer to holding periods of up to 60 months, they are calculated on a monthly basis. we calculate abnormal returns with the market model. the market model has the advantage that it lowers the variance of abnormal returns. this raises the ability to detect the outcome (campbell, lo and mackinley, 1997; muradoglu and sivaprasad, 2012a). the abnormal return is defined as: ari=ri –e(ri) (1) where ri is the return and e(ri) is the expected return. the subscript i refers to the holding period in months. stock returns are calculated as nominal returns in local currencies. thus, returns for the united states are in us-dollar, returns for the united kingdom in pound sterling and returns for germany is in euro to ensure that performance is measured from the perspective of a local investor. according to the market model, we expect the return of a share to be the return of the benchmark index. the indices are described in the prior section. the mixed german benchmark index has the following weightings: the weight of the dax 30 is 0.3, the weight of the mdax 50 is 0.5 and the weight of the sdax is 0.2. for the holding period of i months (i = 1, …, 60), we run the following regressions for each country: *i td ar a b c beta tc ltd ε= + + + + (2) where, ari is the monthly abnormal stock return for a holding period of i months (i = 1, ..., 60), beginning at the first trading day in june of month 0 and ending at the first trading day of month i. ltdtc td + stands for ‘total debt to (total capital + long term debt)’. total capital is the sum of equity, short term debt and long term debt. total debt is the sum of short term debt and long term debt. a is a constant and b is the regression coefficient, ε is the error term. to calculate the leverage ratio td/(tc+ltd), we use book values instead of market values because we want to concentrate on financial decisions rather than the influence of market fluctuations. graham and harvey (2001) show that executives focus on book values when determining the capital structure. using market values is, furthermore, problematic since market values might correlate with exogenous variables (kose, 2011). we employ td/(tc+ltd) because this is a more sophisticated leverage ratio than ´debt to equity´ or ´debt to total capital´ which are used in most of the cited studies. the ratio we employ treats long term debt and short term debt differently. long term debt and short term debt fundamentally differ and are used for different purposes (kose, 2011). additionally, in each regression we control for beta. beta is calculated as the price movement of the share divided by the price movement of the corresponding index over a five year time span. furthermore, we use a portfolio approach based on td/(tc+ltd) to realize direct investment implications of leverage. once a year, we form five equal sized portfolios depending on td/(tc+ltd) for every holding period from one to 60 months and compute average monthly abnormal returns for each quintile. moreover, we calculate average cumulative abnormal return for every holding period. in each of the quintiles, included stocks are equally weighted. we do this sorting for every country and use t-tests to detect differences from zero. in our portfolio approach, we link accounting data of fiscal year t-1 to abnormal stock returns which are calculated from the first trading day of june in year t. 4. findings in this section, we show the result of our empirical study for the three markets (germany, united kingdom and the united states). the regression is an ols. after regression coefficients for independent variables, corresponding standard errors are reported in parentheses. * indicates statistical significance at the 10%-level, ** at the 5%-level and *** at the 1%-level. a. empirical results for germany table 1 shows the results of the cross sectional panel regression for one, three and six month holding periods, which are estimated with equation (2). the dependent variable is monthly abnormal return for respective holding periods. the model includes the independent variables td/(tc+ltd) and beta. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 5 contrary to the traditional financial theory, the coefficients for td/(tc+ltd) are negative for all three holding periods, although only the one month estimate is significant. it has a value of -0.005, implying that low td/(tc+ltd)-stocks outperform high td/(tc+ltd)-stocks for this time span. the coefficients for three and six months are not significantly different from zero. for the three month holding period, beta coefficient is significant. interestingly, however, it has an unexpected sign: the coefficient is negative which is not in line with capm. table 1: regression for germany, abnormal returns for 1-, 3and 6-month periods 1 month 3 months 6 months variable coefficient st.error coefficient st. error coefficient st. error intercept 1.689** (0.691) 0.96** (0.415) 0.186 (0.271) td/(tc+ltd) -0.005* (0.003) -0.002 (0.002) -0.001 (0.001) beta -0.981 (0.763) -0.969** (0.458) -0.238 (0.299) r2 0.006 0.007 0.002 f-statistics 2.544* 3.308** 0.854 observations 1070 1070 1070 table 2: regression for germany, abnormal returns for 12-, 36and 60-month periods 12 month 36 months 60 months variable coefficient st.error coefficient st. error coefficient st. error intercept 0.74*** (0.232) 0.336 (0.244) 0.527** (0.253) td/(tc+ltd) -0.002** (0.001) -0.001 (0.001) -0.002* (0.001) beta -0.019 (0.256) 0.028 (0.269) -0.089 (0.280) r2 0.005 0.003 0.004 f-statistics 2.192 1.271 1.671 observations 1070 1070 1070 table 2 shows the regressions for holding periods of twelve, 36 and 60 months. for twelve and 60 months, both td/(tc+ltd)-coefficients have a value of -0.002. they are significant at the 5%and 10%-level, respectively. regression coefficients for beta are not significant for reported holding periods. table 3: mean monthly abnormal returns for the german market ar1 ar3 ar6 ar12 ar36 ar60 p1 (lowest) 0.809 0.185 0.146 0.608 0.335 0.281 p2 0.967 0.074 -0.002 0.261 0.055 0.088 p3 0.854 -0.018 -0.102 0.104 -0.138 -0.070 p4 -0.640 0.205 -0.091 -0.262 0.110 0.061 p5 (highest) -0.580 -0.344 -0.158 -0.072 -0.119 -0.118 p1-p5 1.389 0.529 0.304 0.680 0.454 0.399 t-value 1.30 0.83 0.70 1.78* 1.00 0.79 table 3 shows mean abnormal returns for different portfolios. portfolios are sorted based on td/(tc+ltd). again, we focus on holding periods between one and 60 months and in table 3, we report monthly returns for those periods. the abnormal returns are averages across the whole time series. the reason why abnormal returns of different portfolios do not exactly add up to zero is that benchmark indices are not equally weighted. instead, major indices like the dax30, mdax50, s&p100 and ftse100 are weighted according to market cap and/or free float of included companies. in our sample stocks are equally weighted. this is common practice and leads to the most meaningful results. however, it explains why abnormal returns do not exactly add up to zero. a second reason is that we exclude financial companies. benchmark indices, of course, contain financial companies. portfolio 1 has the lowest mean td/(tc+ltd), portfolio 5 the highest. we report spreads between p1 and p5 as well as t-values to get to know if those spreads are significantly different from zero. results show that holding low td/(tc+ltd)-stocks is highly profitable. the lowest quintile performs better than the highest quintile for every holding period. considering all portfolios, those with lower leverage mostly perform better than portfolios with higher leverage, but this does not hold for all cases. for example, p4 has a better return than p3 for the three, six, 36 and 60 month holding period, although it has a higher td/(tc+ltd) than p3. according to ttests, differences between mean abnormal returns of p1 and p5 are not random for twelve and 60 month investment periods. spreads are significant at the 10%-level. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 5 table 4: mean cumulative abnormal returns for the german market car1 car3 car6 car12 car36 car60 p1 (lowest) 0.81 0.55 0.88 7.30 12.07 16.84 p2 0.97 0.22 -0.01 3.13 1.99 5.30 p3 0.85 -0.05 -0.61 1.25 -4.95 -4.18 p4 -0.64 0.61 -0.55 -3.15 3.96 3.64 p5 (highest) -0.58 -1.03 -0.95 -0.86 -4.28 -7.07 table 4 shows how abnormal returns add up along the time span. for longer investment horizons, the magnitude of the outperformance of p1 is astonishing: p1 has a cumulative abnormal return of 12.1 and 16.8 percent for holding periods of 36 and 60 months. p2 still has an abnormal return of 2.0 and 5.3 percent for these time periods. in comparison, for high td/(tc+ltd)-portfolios, it is the downside potential which seems to be relatively high: p5 has a negative cumulative abnormal return for all holding periods. figure 1: mean cumulative abnormal returns for the german market -10 -5 0 5 10 15 20 1 3 6 12 36 60 c um ul at iv e ab no rm al r et ur n in % investment period in months p1 p2 p3 p4 p5 figure 1 illustrates investment implications graphically. it shows cumulative abnormal returns along the time bar. on the x-axis, for clarity, distances between indicated values are identical. the profitability of p1 turns out to be quite persistent: p1 has the highest returns for holding periods from six to 60 months. particularly for longer investment horizons of twelve to 60 months, it outperforms all other portfolios by a large margin. p5, the portfolio with the highest td/(tc+ltd), is the worst performing group for three of six holding periods. especially for long term investors, the spread between p1 and p5 is sizable: for three years, there is a cumulative return difference of more than 16 percent; for five years, the spread between p1 and p5 even exceeds 23 percent. b. empirical results for the united kingdom uk-regression results (table 5) show that td/(tc+ltd) coefficients are negative for holding periods of three, six and twelve months after controlling for beta. for three and six months, coefficients of -0.004 and -0.003 are significantly different from zero at the 5%-level. table 5: regression for the uk, abnormal returns for 1-, 3and 6-month periods 1 month 3 months 6 months variable coefficient st.error coefficient st. error coefficient st. error intercept 2.847*** (0.714) 0.433 (0.362) 0.076 (0.258) td/(tc+ltd) -0.006 (0.004) -0.004** (0.002) -0.003** (0.001) beta -1.992*** (0.703) -0.245 (0.356) 0.154 (0.253) r2 0.011 0.006 0.005 f-statistics 5.099*** 2.740* 2.426* observations 944 944 944 table 6: regression for the uk, abnormal returns for 1-, 3and 6-month periods 12 month 36 months 60 months variable coefficient st.error coefficient st. error coefficient st. error intercept -0.026 (0.237) 0.103 (0.172) 0.411*** (0.139) td/(tc+ltd) -0.001 (0.001) -0.00008 (0.001) -0.00005 (0.001) beta 0.115 (0.233) 0.044 (0.169) -0.043 (0.136) r2 0.001 0.000 0.000 f-statistics 0.630 0.040 0.052 observations 944 944 944 international journal of banking and finance, vol. 10, iss. 1 [2013], art. 5 similar to the germany sample, the beta coefficient is negative for three and six month periods. for the one month holding period, negative beta has significant predictive power for returns. after being negative for three months as well, the beta coefficient turns positive for the twelve month period, albeit it is insignificant. f-statistics show that models are significant at the 1% and 10% level; therefore results should hold for the population, too. in contrast, f-statistics for longer holding periods indicate weaker overall models. td/(tc+ltd) coefficients are all negative but not significant. beta coefficient as well, which is positive for twelve and 36-months, but negative for 60-months periods, is not significantly different from zero. table 7: mean monthly abnormal returns for the uk also 12-, 36and 60-month periods ar1 ar3 ar6 ar12 ar36 ar60 p1 (lowest) 0.755 0.650 0.558 0.360 0.313 0.057 p2 -0.180 -0.033 0.070 0.083 -0.091 -0.072 p3 0.477 0.010 0.126 -0.116 0.009 0.083 p4 0.988 0.228 0.065 0.012 0.096 0.319 p5 (highest) -0.967 -0.801 -0.397 -0.226 -0.213 -0.294 p1-p5 1.722 1.450 0.955 0.586 0.526 0.352 t-value 1.89* 3.00*** 2.60** 1.63 2.05** 2.10** the uk-stocks are also sorted into td/(tc+ltd)-quintiles. table 7 shows that especially for large holding periods, the distribution of portfolio returns is not as clear as for the german market. for example, p2 performs relatively poorly for holding periods of 36 and 60 months (0.091% and -0.072% monthly abnormal return). in contrast, p4 which has a higher average td/(tc+ltd), performs better for these holding periods. (+0.096% and +0.319% respectively). however, on balance, like for germany, low td/(tc+ltd)-portfolios perform better than high td/(tc+ltd)-portfolios. according to ttests, differences between mean abnormal returns of p5 and p1 are significant for five of six studied holding periods. this fact is impressive and raises the question if significant spreads must be attributed to investors’ mistakes to overbuy high td/(tc+ltd)-stocks and oversell low td/(tc+ltd)-stocks. obviously buying low leverage stocks is successful for medium term periods: p1 has the highest return for three, six, twelve and 36 months. for the longest period, in contrast, p1 performs hardly above average with a cumulative abnormal return of 3.4%. two other portfolios, namely p4 and p3, are on average more successful in this time span. table 8: mean cumulative abnormal returns for the uk market for different periods car1 car3 car6 car12 car36 car60 p1 (lowest) 0.75 1.95 3.35 4.33 11.25 3.43 p2 -0.18 -0.10 0.42 1.00 -3.28 -4.34 p3 0.48 0.03 0.76 -1.39 0.33 4.99 p4 0.99 0.68 0.39 0.15 3.44 19.16 p5 (highest) -0.97 -2.40 -2.38 -2.71 -7.68 -17.66 figure 2: mean cumulative abnormal returns for the uk market for different periods -20 -15 -10 -5 0 5 10 15 20 25 1 3 6 12 36 60 c um ul at iv e ab no rm al r et ur n in % investment period in months p 1 p 2 p 3 p 4 figure 2 reports that very high leverage portfolios perform weak for all holding periods. actually p5 has the lowest returns for every studied period. interestingly, however, p2, a lowleverage portfolio, also performs relatively poor for longer periods of 36 and 60 months. in particular for the longest period, figure 2 demonstrates that the magnitude between the best and worst performing quintile is higher for the uk market than for the german market. c. empirical results for the united states as table 9 shows, the us results are mixed for short investment periods of one to six months. while we find a significant positive relation between td/(tc+ltd) and stock returns for one month, for three and six months the link is negative and the coefficients are insignificant. in line with rational models, beta coefficient is significantly positive for the one month period. but for longer horizons, it turns insignificant. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 5 table 9: regression for the usa, abnormal returns for 1-, 3-, and 6-month periods 1 month 3 months 6 months variable coefficient st.error coefficient st. error coefficient st. error intercept -3.357*** (0.856) 0.793** (0.366) -0.083 (0.279) td/(tc+ltd) 0.027** (0.013) -0.001 (0.006) -0.003 (0.004) beta 1.880*** (0.670) -0.284 (0.286) 0.021 (0.219) r2 0.011 0.001 0.000 f-statistics 6.729*** 0.511 0.287 observations 1244 1244 1244 considering longer periods (see table 10), the findings for the united states are also ambiguous. even though the td/(tc+ltd) coefficient for 36 months of -0.007 is significant at the 5%-level, other td/(tc+ltd) coefficients are not significant. in summary in comparison to the german and the uk markets, regression results for the us market are inconclusive and do not show a clear relation. table 10: regression for the usa, abnormal returns for 12-, 36and 60-month periods 12 month 36 months 60 months variable -0.014 (0.279) 0.452** (0.220) 0.058 (0.132) intercept 0.001 (0.004) -0.007** (0.003) -0.001 (0.002) td/(tc+ltd) 0.002 (0.218) -0.090 (0.172) 0.118 (0.103) beta 0.000 0.004 0.001 r2 0.016 2.387** 0.681 f-statistics 1244 1244 1244 observations -0.014 (0.279) 0.452** (0.220) 0.058 (0.132) mixed relations for the us are also reflected in the distribution of quintile-returns. apart from one and 36 months, average returns do not follow a clear pattern. for the one month holding period, higher leverage portfolios yield higher mean returns and lower leverage portfolios yield lower mean returns. however, for the 36-month period, we identify a pattern which is well followed in germany and the uk markets: lower td/(tc+ltd) quintiles have higher returns than higher td/(tc+ltd) quintiles. t-tests show that for the six and 36 months period, differences between mean abnormal returns of p1 and p5 are statistically significant at the 10%-level. table 11: mean monthly abnormal returns for the us market for different periods ar1 ar3 ar6 ar12 ar36 ar60 p1 (lowest) 0.197 0.954 0.487 0.533 0.583 0.240 p2 -1.793 -0.373 -0.259 -0.229 -0.085 -0.104 p3 -0.096 -0.502 -0.238 -0.328 -0.179 -0.116 p4 0.547 0.047 -0.432 -0.069 -0.251 -0.031 p5 (highest) 0.663 0.461 -0.144 0.224 -0.061 0.018 p1-p5 -0.467 0.492 0.631 0.308 0.644 0.221 t-value -0.57 0.95 1.64* 0.69 1.64* 0.91 cumulative average abnormal returns demonstrate that holding p1 for long periods proves profitable for the us market, too. the average cumulative abnormal return of p1 is 21.0 percent for 36 months and 14.4 percent for 60 months. table 12: mean cumulative abnormal returns for the us market for different periods car1 car3 car6 car12 car36 car60 p1 (lowest) 0.20 2.86 2.92 6.39 20.99 14.37 p2 -1.79 -1.12 -1.55 -2.75 -3.07 -6.26 p3 -0.10 -1.51 -1.43 -3.94 -6.45 -6.99 p4 0.55 0.14 -2.59 -0.82 -9.02 -1.84 p5 (highest) 0.66 1.38 -0.86 2.69 -2.19 1.09 international journal of banking and finance, vol. 10, iss. 1 [2013], art. 5 figure 3 is a summary of the performance of p1. moreover, it shows that for the 36 month holding period, apart from p1, all the other portfolios have negative returns. in addition to the three months holding period, for all other investment horizons, p1 is the most successful quintile and p5 the second in rank. figure 3 also shows that in comparison to the german and uk markets, p5 performs relatively well. figure 3: mean cumulative abnormal returns for the us market for different periods -15 -10 -5 0 5 10 15 20 25 1 3 6 12 36 60 c um ul at iv e ab no rm al r et ur n in % investment period in months p1 p2 p3 p4 p5 5. interpretation regarding the inconclusive results of prior research, the negative relation between td/(tc+ltd) and stock returns which holds at least for germany and the uk seems surprising at first sight. what might be the reasons? one possibility is that our findings for these two countries might be an anomaly: that lower td/(tc+ltd) firms indeed earn higher risk adjusted stock returns than higher td/(tc+ltd) firms. this implies that due to mispricing, low td/(tc+ltd) stocks are on average bargains and high td/(tc+ltd) stocks are pricey. in this case investors could be able to exploit mispricing by buying low td/(tc+ltd) stocks and selling (shortselling) high td/(tc+ltd) stocks. another explanation is that the betafactor is not a suitable substitute for risk? assuming that this is true, the following question comes to mind: does it make sense that higher returns of low td/(tc+ltd) stocks are a compensation for risk (which the betafactor does not measure)? before answering this question let´s demonstrate what td/(tc+ltd) actually captures. if a firm has no debt, td/(tc+ltd) is zero. if debt increases, td/(tc+ltd) increases, too. however, long term debt and short term are treated differently by the ratio. therefore, td/(tc+ltd) rises faster and obtains higher values with increasing short term debt (std) than with increasing long term debt (ltd). in addition, let´s focus on the relation between risk and leverage as well as between risk and debt maturity to answer the question if risk differences are a plausible explanation for our finding. kose (2011) addresses the relation between risk and leverage in a model that accounts for debt maturity. the model predicts that risk increases with leverage. however, it also suggests that risk decreases with debt maturity: the rationale is that companies opt for their optimal maturity of debt. companies which act in risky environments and have a relatively low credit quality have to bear particularly high issuance costs of long term debt. therefore according to the model these firms try to avoid long term debt or choose a relatively low fraction of long term debt. on the contrary, companies which operate in less risky environments and have higher credit quality opt for a higher fraction of long term debt. the reason is that issuance costs of long term debt are relatively low for those companies. as a result the model predicts that those firms with a high fraction of std are riskier than firms with a high fraction of ltd. for our ratio the implication of the model is that risk should clearly increase with rising td/(tc+ltd). for one thing that is the case because of the prediction that risk increases with leverage and for another thing because of the prediction that risk decreases with debt maturity. investment-based theories also suggest a negative relation between risk and debt maturity. for example, chen, novy-marx and zhang (2010) suggest that companies with higher investment are less risky and have lower expected returns than companies with lower investment. the reasoning is that low cost of equity lead to higher investment. carlson, fisher and giammarino (2004) also predict that investment is negatively related to risk and expected returns. it is plausible to infer that higher investment involves a higher fraction of long term debt. more importantly, kose (2011) demonstrates that this relation also holds empirically. accordingly, investmentbased models also suggest a positive relation between risk and td/(tc+ltd). considering these theories, our finding [the negative relationship between td/(tc+ltd) and stock returns for markets in germany and the united kingdom] is even more puzzling: in our view this result cannot be explained by rational behavior of investors. another question that arises is: why are results for the us market different from results for the german and uk market? why is the relation between td/(tc+ltd) and stock returns mixed for the united states while it is negative for germany and the united kingdom. one possible answer is that the us market is more efficient than the other two markets. although the german and the uk markets are not small by any measure, the us market is observed from all parts of the world and followed by many more investors than the other two. there is certainly a distinct international journal of banking and finance, vol. 10, iss. 1 [2013], art. 5 difference in terms of international investor participation: for example an average diversified german investor might well hold us-stocks. in contrast, it is rather untypical for a us-american investor to hold european or even german stocks. notice that the value of share trading at nyse and nasdaq is more than ten times the value of share trading at london stock exchange and more than 17 times the value of share trading at deutsche börse. in 2011, share trading at nyse and nasdaq was 30,751 billion usd in comparison to 2837 billion usd at london stock exchange and 1758 billion usd at deutsche börse (world federation of exchanges, 2012). the difference cannot be explained by gdp differences between countries. studies find that stock markets with higher value of share trading are more efficient than stock markets with lower value of share trading (butler and malaikah, 1992, or claessens, dasgupta and glen, 1995). it is plausible that higher competition among investors in the us market leads to a leveling out of investment mistakes. this would imply that in contrast to respective stocks from other two markets, low leverage s&p100-stocks get bought by rational investors before severe mispricing occurs. in the german and uk market, however, major mispricing might lead to substantial return differences between low td/(tc+ltd)and high td/(tc+ltd)-stocks. 6. conclusion this paper contributes to understanding the relation between leverage and stock returns. contrary to other studies, we choose the ratio of ‘total debt to (total capital + long term debt)’ [td/(tc+ltd)] and account for different maturities of debt. that is crucial because short-term debt and long-term debt are used for different purposes. we test the link between td/(tc+ltd) and equity returns for markets in germany, the united kingdom and the united states. interestingly, for germany and the united kingdom the relation between td/(tc+ltd) and stock returns is negative and often significant; whereas for the united states the results are mixed. our portfolio-based approach confirms this finding: low-td/(tc+ltd) portfolios perform considerably better than high-td/(tc+ltd) portfolios in the german and the uk markets while in the us market there is no clear pattern. under classical models, our findings for the german and the uk markets are not explicable by traditional finance paradigm: a risk-based explanation is not plausible. therefore, we attribute our findings to mispricing. however, one burning question arises: why do results for the usa differ from results of the other two countries? one possible explanation is that the us market is more efficient due to size and international investor participation. a limitation of our study is that samples are relatively small. more comprehensive samples may result in even more significant findings. future research may conduct tests with larger samples. furthermore, it would be interesting to perform a similar study in developing markets. according to our reasoning, a negative link between td/(tc+ltd) and stock returns would not be surprising due to lower efficiency of those markets. ___________________________________________________________ authors information: tristan nguyen, whl graduate school of business and economics, landsberger str. 257 b, 80687 munich, germany. tel: +49(0)7821-923865. e-mail: tristan.nguyen@whl-lahr.de. alexander schüßler, hhl leipzig graduate school of management, oetztalerstrasse 12, 81373 munich, germany. tel: +49(0)173-9521060. e-mail: alexander.schuessler@hhl.de references arditti, f. d., (1967). risk and return on equity. journal of finance, 22 (1): 19-36. bhandari, l. c., (1988). debt/equity ratio and expected common stock returns: empirical evidence. journal of finance, 43: 507-528. butler, k. c., and malaikah, s. j., (1992). efficiency and inefficiency in thinly traded stock markets: kuwait and saudi arabia. journal of banking and finance, 16 (2): 197-210. campbell j. y., lo a. w., and mackinlay a. c., (1997). the econometrics of financial markets (1st ed.). princeton: princeton university press: 149-178. carlson, m., fisher, a., and giammarino, r., (2004). corporate investment and asset price dynamics: implications for the cross-section of returns. journal of finance, 59: 2577–2603. caskey, j., hughes, j. s., and liu, j., (2012). leverage, excess leverage and future returns. review of accounting studies, 17: 443-471. chen, l., novy-marx, r., and zhang, l., (2010). an alternative three-factor model. working paper. available at ssrn: http://dx.doi.org/10.2139/ssrn.1418117. claessens, s., dasgupta, s., and glen, j., (1995). return behavior in emerging stock markets. the world bank economic review, 9 (1): 131-51. debondt, w., and thaler, r., (1987). further evidence on investor overreaction and stock market seasonality. journal of finance, 42: 557-581. dhaliwal, d. s., heitzman, s., and zhen l. o., (2006). taxes, leverage, and the cost of equity capital. journal of accounting research, 44 (4): 691–723. george t., and hwang c., (2009). leverage, financial distress and the cross section of stock returns. journal of financial economics, 96 (1): 56-79. gombola, m. j., and liu, f. j., (1993). dividend yields and stock returns: evidence of time variation between bull and bear markets. financial review, 28: 175-192. graham, j. r., and harvey, c. r., (2001). the theory and practice of corporate finance: evidence from the field. journal of financial economics, 60: 187-243. grinblatt, m., and han, b., (2004). prospect theory, mental accounting, and momentum. journal of financial economics, 78 (2): 311-339. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 5 mailto:tristan.nguyen@whl-lahr.de mailto:tristan.nguyen@whl-lahr.de mailto:alexander.schuessler@hhl.de mailto:alexander.schuessler@hhl.de keppler, a. m., (1991). the importance of dividend yields in country selection. journal of portfolio management, 17: 24-29. korteweg, a., (2004). financial leverage and expected stock returns: evidence from pure exchange offers. working paper. available at ssrn: http://dx.doi.org/10.2139/ssrn.597922 kose, e., (2011). dissecting the leverage effect on stock returns. working paper. available at:http://www.business.uconn.edu/finance/seminars/papers/dissecting%20the%20leverage %20effect%20on%20stock%20returns.pdf miller, m., (1977). debt and taxes. journal of finance, 32: 261–275. modigliani, f., and miller, m., (1958), the cost of capital, corporation finance, and the theory of investment. american economic review, 48 (3): 261–297. modigliani, f., and miller, m., (1963). corporate income taxes and the cost of capital: a correction. american economic review, 53: 433–443. muradoglu, y., and sivaprasad, s., (2012a). using firm level leverage as an investment strategy. journal of forecasting, 31: 260-279. muradoglu, y., and sivaprasad, s., (2012b). capital structure and abnormal returns. international business review, 21: 328-341. patel, p., yao, s., and barefoot, h., (2006). high yield, low payout, credit suisse quantitative equity research. penman, s. h., richardson, s. a., and tuna, i., (2007). the book-to-price effect in stock returns: accounting for leverage. journal of accounting research, 45 (2): 427-467. world federation of exchanges (2012). 2011 wfe market highlights, retrieved june 14, 2012 from http://www.world-exchanges.org/files/file/stats%20and%20charts/2011% 20wfe%20market%20highlights.pdf http://www.jstor.org/stable/1809766 the law of one price and arbitrage on china's dual-listings 58 the international journal of banking and finance, volume 9 (number 2) 2012: pages 58-76 the law of one price and arbitrage on china’s dual-listings in hong kong and new york liu, lixian and timofei bogomolov university of south australia ____________________________________________________________ abstract traditionally, arbitrage refers to simultaneously buying and selling the same financial assets by taking advantage of a price difference in two or more markets. however, the strict sense of arbitrage is hardly obtained after consideration the issues concerning transaction costs and time value of money. by using the identical assets such as chinese adrs and their underlying securities traded in different markets in hong kong in hk dollar and in new york in us dollar and by constructing a very simple arbitrage trading strategy, this study demonstrates that arbitrage profits are still available with monthly return ranging from 0.5 per cent to 3.8 per cent after considering transaction costs and non-overlap trading time issues. this is a new study to verify this behaviour of an emerging market’s adrs traded in two financial market locations, so adding evidence of inefficiency in trading of china-listed stocks in foreign locations. key words: arbitrage, transaction costs, adrs, financial markets, trading strategy jel classification: g10, g19, c18 _____________________________________________ 1. introduction theoretically, law of one price (lop) states that securities traded as cross-border transactions in different currencies – for example the securities located in china and traded in hong kong and in the us in the form of american depository receipts (adrs) should have identical prices given the ease with which arbitrage could be conducted across markets. therefore, it is claimed, all securities are close substitutes for investors in that investors would price these securities identically. the price disparity idea is readily confirmed for some of the dual-listed chinese securities that are traded in both hong kong and new york markets although malkiel 59 (2007) states that h-shares and n-shares (chinese shares listed and traded in hong kong and new york markets respectively) are reasonably priced relative to their growth rates. since chinese adrs (american depository receipts) can be freely converted to hshares, arbitrage should force the prices of these chinese securities in the new york stock exchange and stock exchange of hong kong to reach equilibrium. if the hong kong (hk) and new york (ny) markets are integrated in terms of the trading of cross-listed securities, the close price in hong kong and the open price in ny for those cross-listings should not exhibit any exceptional levels of volatility, volume or spreads regarding the trading activity (domowitz et al. 1998). this study investigates the price disparity between these dual-listed chinese securities and explores if there is arbitrage opportunities for traders meaning that the prices are identical. most of the dual-listed chinese securities traded in the hk and ny markets also have their shares traded in the mainland china market. since the short selling and free capital flows are restricted in the chinese mainland stock markets, does the arbitrage idea still hold is an unanswered policy issue. hence the examination on the arbitrage opportunities in this set up is of research interest so we focus on the dual-listed chinese securities traded in hk and ny markets in the form of adrs. the rest of the paper is organised into four sections. a brief review of literature is attempted to show that there is very little consensus that there is arbitrage-free trades in the cases of dual-listed chinese shares despite there being evidence of no arbitrage profits in trades of dual-listed shares in developed markets. in the following section, we provide a description of the data and research process to estimate share price returns adjusted for currency and for transaction costs in hong kong and new york of the china shares. the results are presented in section 4 while the paper is concluded in section 5. there is evidence of 0.5 per cent or more arbitrage profit opportunities in trades of 14 china shares. 2. review of arbitrage literature on dual listings there is no consensus in the literature regarding the price discrepancy and arbitrage of the crosslisted securities. the existing studies present different views on cross-listing and arbitrage. maldonado and saunders (1983), kato et al. (1991), park and tavakkol (1994), ding (2000), and lok and kalev (2006) found no significant price disparity between the cross-listed shares. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 4 60 so they suggest no obvious arbitrage opportunities exist. in contrast, wahab, lashgari and cohn (1992), kaul and mehrotra (2000) and suarez (2005) did find arbitrage opportunities. bailey et al. (1999) and miller and morey (1996) found price differences, but suggested that arbitrage is difficult. maldonado and saunders (1983) examine the one-year price behaviour of 37 us multinational corporation stocks traded simultaneously on the london and new york stock exchanges under the restricted and unrestricted periods of the british investment controls. their empirical results strongly supported lop for internationally traded stocks after adjustment for exchange rates and transactions costs. note that these markets do not frictions in terms of restriction in trades or even ownership. they also suggested that lop is not violated even under a one-sided government that imposes foreign exchange restriction on trade. wahab et al. (1992) studied arbitrage between adrs and their underlying stocks using an implicit pairs trading strategy with two portfolios: one is adr, the other is underlying home shares portfolio. they sell the winner (portfolio with the highest return over a period of two weeks) and buy the loser (portfolio with the lowest returns over the same two-week period). they found limited profits for their pairs-trading strategy when about four per cent transaction costs were considered and data limitations overcome. miller and morey (1996) examined one of the most heavily traded british stock (glaxowellcome) that is cross-listed in the us using intraday data to test arbitrage on european adrs: they also investigate the intraday price difference pattern between adrs and the underlying securities. the results showed that the price difference in these two markets was small and insignificant throughout their two-month sample period, suggesting the market is efficient with respect to arbitrage opportunities. ding (2000) examined the cross-listed stocks that were traded on the stock exchange of singapore and the kuala lumpur stock exchange (now called bursa malaysia) to see if the two markets are closely linked and the price parity is maintained from the same malaysian stock traded on both the kuala lumpur stock exchange and singapore’s clob international.1 the results showed that the two markets were well linked in terms of their returns and volatility. the 1 in the stock exchange of singapore and the kuala lumpur stock exchange, stocks were cross-listed and traded on both exchanges in their local currencies. however, on 1 january 1990, the cross-listing arrangement was abolished. to facilitate singaporean investors to trade in malaysian shares and other foreign shares and to keep the business in singapore, the stock exchange of singapore set up an over-the-counter market known as clob international. 61 markets of singapore and kuala lumpur also conformed to the law of one price reasonably. arbitrage opportunities appeared to exist when stock-broking houses trade for their own accounts, obviously with no transaction costs. meanwhile, arbitrage opportunities for other type of investors were significantly diminished due to their high transaction costs. conversely, some of the stock markets have different trading hours, suggesting that two closing prices must not be equal. a number of studies on lop had results that cannot reject the hypothesis of arbitrage-free pricing because the cross-listed stocks are indeed one security mainly using daily closing prices with non-overlapping trading time (maldonado and saunders 1983; kato et al. 1991; park and tavakkol, 1994). regarding the trading time issue, some of the studies focus on the overlap of trading activities and concluded with that arbitrage opportunities do exist. de jong et al. (2007) evaluated a simple trading rule involving a long position in the relatively underpriced part of the twin shares and shorting an equal dollar amount in the relatively overpriced part of the twin shares. they found that this rule produced an abnormal return of up to almost 10 per cent per annum adjusted for systematic risk, transaction costs, and margin requirements.2 suarez (2005) analysed the price disparity between the adrs and their corresponding french stocks and us companies cross-listed in france using high frequency intra-daily data sets as well as the intra-daily french franc/us dollar exchange rates. the sample included top 10 companies that are extensively traded stocks in both markets. this comparison was done on a 1.5 hours overlap of trading day, when both markets were open. all possible transaction costs were taken into account, and the quotes data rather than traded data were used. the result showed that there was a large deviation from lop, suggesting that an arbitrage trading profits could be made on these large disequilibria. 3 suarez also asserts that these markets were disintegrated and not fully efficient as measured by a lack of arbitrage opportunities. 2 a twin share is also called siamese twin, which involves two companies incorporated in different countries contractually agreeing to operate their businesses as if they were a single enterprise, while retaining their separate legal identity and existing stock exchange listings. from this sense, it is lightly different from cross-listing. a wellknown example is royal dutch/shell. the shares of the siamese twin parents represent claims on exactly the same underlying cash flow (froot and dobora 1999; bedi et al. 2003; de jong et al. 2007). 3 suarez’s paper established a point estimate for the no-arbitrage band of ±0.563 per cent, inside which a mispricing cannot be profitably arbitraged due to transaction costs and foreign exchange risk. about 2.1 per cent of the observations in their sample were outside this band and they were thus classified as arbitrage opportunities. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 4 62 alves and morey (2003) examining eleven most traded brazilian adrs listed on the new york stock exchange used intraday quote price data and the realis/dollar exchange rate, taken on an intraday basis during the period when the two stock markets overlap over a onemonth period in 2001. the transaction costs for different type of investors were also considered for the arbitrage possibilities. the results showed that there seems to be little arbitrage opportunities for the ordinary traders due to higher transaction costs. however, for the brazilian financial institution, the result showed that there was a profitable arbitrage opportunity available, which the researcher attributed to the special tax laws for financial institutions. kaul and mehrotra (2000) studied cross-listed canadian stocks that were traded in us markets and found that infrequent arbitrage opportunities did exist there, particularly with stock pairs that present a combination of relatively low spreads and low trading volume. 3. chinese adrs and arbitrage a split-share system has existed in china stock market since the re-establishment of stock market in the early 1990s, which allows chinese companies to issue different classes of shares to domestic and foreign investors. this is an attempt in the earlier days to force foreigners to bring foreign currency money to buy local shares. only these chinese firms with the best financial integrity and corporate governance are permitted to be dually listed on the stock exchange of hong kong as h-shares (named the red chips). in the united states, if listed and traded as adrs, it is an easy way for an american traders in the us to invest in a foreign company (mak et al. 2005). h-shares and their related shares traded in new york are essentially shares of mainland chinese companies registered in mainland china or hk and traded both on the stock exchange of hong kong and the new york stock exchange, respectively trading in hk dollar and us dollar. however, under the currency board arrangements, hk dollar is fixed to the us dollar. therefore the exchange rate risk in this case might be very limited. it can be observed that the prices of chinese shares listed in hk are different from their adrs listed in the us. several reasons might cause this price discrepancy. the time difference between trading hours in hk and the us might cause different valuation of the shares (hsu and wang 2008). market news and sentiments might be other factors that are responsible for the phenomenon of price discrepancy. the price differences between different markets make the arbitrage activity to take place (han 2004). 63 if there is no time issue, chinese adrs traded during hk market hours offer a live arbitrage opportunity as there is very little risk in such trading and the gap between the adrs and underlying stock is minimal (hsu and wang 2008). if there is a spread between the perceived values in the two markets, arbitrage can be expected. if the adr is trading at a value lower than what the underlying is in hong kong market, one can purchase the adr and expect to make money as its value converges, and vice versa. 4. examining arbitrage opportunities 4.1 data collection to overcome the time difference between hk and new york, and following lau and diltz’s (1994) methods, the closing price of each listing in hk and the opening price of each listing in new york are recorded for analysis of evaluation. for the price comparison, all the stock prices are adjusted for exchange rates with denominated hk dollar, which is calculated by multiplying the stock prices in one market with the country’s exchange rate, and the ratio of the adr is considered. all the data used for analysis, including the opening, closing price, foreign exchange rate are obtained from datastream. in this study, a total of 14 chinese adrs and their underlying securities are examined, because only these 14 dual-listed chinese securities satisfy the requirement to conduct the arbitrage activity, for example, no short selling restrictions. the data covers the period for each individual company from the same trading date (the listing date for chinese securities listed in hong kong and in new york it is the different for most of the cases) in the hk and new york markets through to 31 december, 2008. this covers the years from 1993 to 2008. therefore, the data period for each company may be different from each other. the day when shares traded in the hk market but not in the new york market was eliminated from the dataset. 4.2 transaction costs involved in arbitrage to examine arbitrage possibilities, transaction costs are important factors to influence the arbitrage decision. these transaction costs create a no-arbitrage band, therefore, the arbitrage opportunity must be sizeable enough to generate a profit that could cover and exceed the costs involved. not all the misprised securities have the profitable arbitrage opportunities. this study international journal of banking and finance, vol. 9, iss. 2 [2012], art. 4 64 examines whether arbitrage opportunities exist for china-based stocks traded in the new york market as adr and in the hk market when transaction costs are considered. the transaction costs structure faced by each category of investors is significantly different from each other. these differences can affect the magnitude of profits that they can reap from arbitraging. the transaction costs to an investor in different markets can largely consist of several elements, including brokerage fees, clearing fees, stamp duty, and foreign exchange cost. arbitrage between the chinese adrs and their underlying securities may present transaction costs arrived during the process of buying the issue, selling the issue at the existed adr market, or creating or cancelling the adrs. according to citibank, the cost of creating an adr is four us cents per adr, and the cost of cancelling an adr is five us cents per adr. these costs such as foreign exchange conversion fee, stamp duty can also be significant. table 1 presents a list of the related transaction costs when buying or selling the chinese adrs and their underlying securities. table 1: computation of transaction costs this table provides the related transaction costs when trading chinese adrs in hk market and their underlying shares in new york market based in 2009. when the transaction is initiated in hk adr conversion fee us$0.05 per adr stamp duty on stock transaction 0.10% foreign exchange rate fee 0.01% brokerage fee at ny market 0.10% ccass fee 0.01% custody fee free when the transaction is initiated in ny adr conversion fee us$0.05 per adr foreign exchange rate fee 0.01% brokerage fee at hk market 0.25% custody fee free in studying the arbitrage opportunities with transaction costs, this process focuses on one typical type of investors that are likely to be involved in cross-border arbitrage activities. this type of investors could be institutional investors such as stockbroking houses, banks and qualified foreign institutional investor. it is expected that institutional investors have much lower transactions costs than an ordinary trader. 65 the transaction cost in the arbitrage is defined as c, which is also the no-arbitrage band for the arbitrage activity when risk is assumed to exist. according to the transaction costs based in 2009 as presented in table 1, it is assumed that all of the transaction costs are 25 basis points for both markets of hk and new york: there is a five cents adr subscription and cancellation fee applying in new york. therefore, the estimated total transaction cost for one deal is assumed to be 0.5 per cent, and 1 per cent for opening and closing one transaction. however, transaction costs are not constant. stock exchanges change transaction costs from time to time. to overcome the changes in transaction costs in different periods, the arbitrage trading strategy in this study only focus on the period between 2002 and 2008, when there were no changes to the transaction fee. 4.3 trading strategy direct arbitrage between duallisted chinese securities traded in hk and new york is impossible as trading hours in new york and hk stock exchanges do not overlap. the time gap between two legs of the trade brings some degree of uncertainty into the trading, and the trade over the pairs of chinese adrs in new york and underlying shares in hk executed by the right trading signals of prices divergence and convergence can result in loss. however, an idea of applying arbitrage-style trading rules still looks very attractive as there are two freely traded assets over the same underlying stock and it is expected that any possible mispricing will be fixed over time. we take log prices of the stocks in hk and adrs in new york and create a spread process also as price premium as: (1) where are prices denominated in the same currency (hk dollars) in new york and hong kong; and represent a time difference between new york and hk prices, in this case, is the hk market closing time; t is the same day new york market opening time. if the difference between prices is greater than a chosen trigger level , we open position on the spread towards zero – initiate a sequence of trades over the pair of stocks in hk and new international journal of banking and finance, vol. 9, iss. 2 [2012], art. 4 66 york. when the spread hits zero in the first time we close position on the spread – initiate a sequence of trades to liquidate all open positions over the pair of stocks. so, if the spread is positive and greater than , we sell short adrs in new york market and commit ourselves to buy equal volume of shares in hk market regardless any possible price changes. when the spread gets negative for the first time we unwind position on the spread – buy adrs in new york and commit ourselves to sell shares in hk. the expected return from this trading strategy can be estimated as: (2) where: are shares and adrs prices denominated in the same currency in hk and us; – an error term due to the time difference between execution trades, which can be assumed independent and identically distributed with mean zero; is a trigger level to generate a trading signal to open position on the spread. in fact the expected return is greater than . we trade in discrete manner and value of the spread between shares and adrs is very unlikely to be equal to or zero at any point of time . trading signal to open or close position on the spread will be generated with some overshot. trading in the arbitrage style suggests that we will have a positive expected return in each trade, so we want to trade as often as possible to generate more profits. however we have to be sure that the expected profit from each trade is not less than the transaction costs. transaction costs are reported to reduce profitability dramatically under arbitrage and pairs trading strategies (bowen et al. 2010; do 2009). we estimate transaction costs for the round trip trading over the dual-listed securities in hk and new york as 1 per cent and choose two trigger levels : 0.01 and 0.02. the greater trigger level increases the expected return in each trade but reduces number of trades, which reduces the total profit from the trading strategy. 67 since there is a time issue, we consider four scenarios in this study which examines the most logical way to organise trading in two markets with not overlapping working hours: 1. analyse market and make trading decision on new york market open based on known hk close and new york open. trade according to trading signals to open/close position on spread at new york open and hk open next day; 2. analyse market and make trading decision on us market close based on known hk close and new york close. trade according to trading signals to open/close position on spread at new york close and hk open next day; 3. analyse market and make trading decision on hk market open based on known new york close of the previous day and hk open. trade according to trading signals to open/close position on spread at hk open and new york open same day; and 4. analyse market and make trading decision on hk market close based on known new york close of the previous day and hk close. trade according to trading signals to open/close position on spread at hk close and new york open same day. if the market in new york or hk is closed due to holidays, then we do not use those days in our analysis. so no trades can be done on those days even if there is a trading signal to open or close position on the spread. 4.4 excess return computation following an approach widely used in the literature on contrarian strategies and pairs trading, we trade the same amount of hk$1 in each short and long position and the total payoff of this strategy has the interpretation of excess return (gatev et al. 2006). daily returns are not calculated as the results could be misleading due to the time differences between the two markets. however, the total payoff after the closing of the last position would be computed as: (3) where: , are the opening prices on securities in hk and new york; , are the closing prices on securities in hk and new york; is a dummy variable equals to 1 if we have long international journal of banking and finance, vol. 9, iss. 2 [2012], art. 4 68 position on security 1 and short on security 2 and equals to -1 if we have short position on security 1 and long on security 2. the average holding time is less than three days and there are 14 companies traded simultaneously with average number of trades varying from 10 to 80 per month, so the resulting average total monthly payoff is then divided by the number of companies. the resulting number could be considered as a good estimation of monthly return. it is necessary to mention that the above method uses leverage 2:1, so the investor should be very careful comparing its results with the performance of the non-leveraged strategies as in a naive buy-and-hold strategy. we run tests for each trading scenario twice – without transaction costs and with transaction costs, which are estimated to be 1 per cent in total for the round trip. to account the transaction cost we reduce the total payoff of the each arbitrage trade by 1 cent. it’s a pretty rough estimation, trading on booming market attracts higher level of transaction costs while trading on falling market costs less. however, as we made reasonably conservative estimation about the size of transaction costs, we believe this simplified approach could provide an adequate representation of the effect of transaction costs on the performance of the proposed trading strategy. 4.5 empirical results table 2 is a summary of descriptive statistics on the daily price premium or discount series for all the 14 chinese dual-listed securities that are traded in both hong kong and new york. the table shows that chinese adrs take a relatively higher premium up to 63.3 per cent and discount down to -32.14 per cent during the examined period. these entire 14 price premium series exhibit a low dispersion level with modest standard deviation of one per cent to two per cent; that is, most of the price premium series are near their mean value. the excess kurtosis for all the price premium series are high, suggesting that there could be some extreme observations that lie at the tail of the distribution. 69 table 2: descriptive statistics of daily price premium/discount dual-listed stocks this table reports the descriptive statistics of daily price premium or discount for 14 chinese adrs and their underlying shares between 1st january, 1993 and 31st december, 2008. the price premium or discount is computed as: . mean st.dev. skewness kurtosis min max range observations aluminum -0.0010 0.0143 -0.2555 9.5458 -0.1013 0.0960 0.1973 1759 china eastern -0.0001 0.0270 -1.1243 13.3878 -0.3214 0.1634 0.4848 2796 china life 0.0008 0.0113 -0.2935 6.6334 -0.0653 0.0510 0.1163 1267 china mobile -0.0007 0.0118 -0.1113 12.9121 -0.1075 0.0772 0.1847 2813 china petro 0.0011 0.0118 0.1200 5.9273 -0.0555 0.0739 0.1294 2057 china telecome 0.0010 0.0104 0.4545 11.4357 -0.0593 0.0941 0.1534 1537 china southern 0.0016 0.0256 -1.1575 12.1432 -0.2335 0.1226 0.3561 2830 china unicom 0.0006 0.0128 0.1792 8.1976 -0.0731 0.0992 0.1723 2139 cnooc -0.0001 0.0105 0.5690 14.5006 -0.0633 0.1187 0.1821 1969 guangshen railway 0.0002 0.0220 -0.3079 5.8816 -0.1677 0.1146 0.2823 3170 hua neng 0.0020 0.0190 0.6819 9.3553 -0.0942 0.1347 0.2289 2755 petro china -0.0002 0.0108 0.3878 6.9271 -0.0466 0.0689 0.1155 2194 sinopec shanghai 0.0003 0.0213 -0.6574 8.3611 -0.1592 0.1142 0.2734 3855 yanzhou -0.0012 0.0251 -0.7137 12.7324 -0.2000 0.1883 0.3884 2516 the detailed analysis of each of the 14 dual-listings is shown in table 3. the summary there suggests that not all companies perform equally well. some companies generated small but consistent loss through the entire testing period from 2002 to 2008 while other companies (predominantly the company of china eastern with longer history of trading in new york and hk) demonstrated up to 10 per cent monthly return after transaction costs. more rigorous approach to the selection of the companies based on the past performance could help to exclude potential losers and improve strategy performance even further. after the examination of individual companies, as in tables 3 to 6, we provide average monthly returns of the trading strategy for each scenario with two different trigger levels. trading statistics related to individual trades are also reported. for the entire scenario, statistically significant excess returns from 2.71 per cent to 8.83 per cent before transaction costs are recorded. average monthly return after accounting for the transaction costs varies from 0.5 per cent to 3.8 per cent, which are statistically significant for all scenarios except for the scenario 3 for trigger level 0.01 (analysis and first trade on new york open). international journal of banking and finance, vol. 9, iss. 2 [2012], art. 4 70 table 3: individual company average monthly excess returns after transaction cost for trigger level 0.01 under the scenario 1 company mean st.dev. p-value aluminum 0.0122 0.0728 0.1324 china eastern 0.1121 0.1152 0.0000 china life 0.0115 0.0566 0.0691 china mobile 0.0037 0.0540 0.5384 china petro and chemical 0.0203 0.0898 0.0440 china southern 0.0928 0.0929 0.0000 china telcom 0.0172 0.0570 0.0079 china unicom 0.0493 0.1442 0.0027 cnooc 0.0142 0.0549 0.0221 guangshen railway 0.0740 0.0886 0.0000 huaneng power 0.0335 0.0738 0.0001 petro china 0.0025 0.0548 0.6823 sinopec shanghai 0.0526 0.0817 0.0000 yanzhou 0.0363 0.0871 0.0003 table 4: scenario 1, new york market open based on known hk close vs new york open, and hk open next day this table presents the summary statistics for the monthly average returns on the equally weighted portfolio that consists of 14 pairs of dual-listings between 2002 and 2008 with or without transaction costs. trigger level 0.01 0.02 0.01 0.02 monthly average return 0.0739 0.0412 0.038 0.0275 standard deviation 0.0482 0.0404 0.0406 0.0335 t-statistics 13.8903 9.2521 8.4724 7.4377 median 0.0679 0.0314 0.0332 0.0224 skewness 3.7445 4.7312 3.6857 4.9742 kurtosis 24.5748 34.1485 25.663 37.2954 sharpe ratio 1.5339 1.0217 0.9356 0.8214 m2 (modigliani rap) 0.1448 0.0965 0.0883 0.0775 min monthly return -0.0147 0.0016 -0.0476 -0.0174 max monthly return 0.3904 0.3321 0.309 0.2749 correlation with benchmark -0.3524 -0.3841 -0.2771 -0.3245 months with non-negative return 99% 100% 94% 93% individual average return per trade 0.0211 0.0309 0.0111 0.0209 average return of positive trade 0.0315 0.0429 0.0265 0.0363 average return of negative trade -0.0213 -0.0256 -0.0209 -0.0268 total number of trades 4163 1598 4163 1598 average holding time 3.1446 3.4524 3.1446 3.4524 trades with excess return > 0 80% 83% 67% 76% without transaction costs with transaction cost standard deviations range between 2.92 per cent and 4.82 per cent. further, an equalweighted portfolio of the underlying shares traded in hk and adrs traded in new york of all 71 14 companies are constructed, and used as a benchmark market. the equal-weighted portfolio has an average monthly return 0.88 per cent only and standard deviation 9.6 per cent. table 5: scenario 2, new york market close hk close vs new york close and hk open next day this table presents the summary statistics for the monthly average returns on the equally weighted portfolio that consists of 14 pairs of dual-listings between 2002 and 2008 with or without transaction costs. trigger level 0.01 0.02 0.01 0.02 monthly average return 0.0598 0.0404 0.01 0.0148 standard deviation 0.0398 0.0428 0.0386 0.036 t-statistics 13.6189 8.5556 2.3532 3.7062 median 0.0557 0.0318 0.0077 0.0115 skewness 2.4874 3.4982 1.0062 2.5769 kurtosis 12.9549 20.7876 8.8465 16.8963 sharpe ratio 1.504 0.9448 0.2599 0.4093 m2 (modigliani rap) 0.142 0.0892 0.0245 0.0386 min monthly return -0.0052 -0.031 -0.1001 -0.0638 max monthly return 0.2658 0.3059 0.178 0.2237 correlation with benchmark -0.1471 -0.2005 0.0407 0.0065 months with non-negative return 99% 99% 63% 71% individual average return per trade 0.0124 0.0162 0.0024 0.0062 average return of positive trade 0.021 0.0266 0.0181 0.0234 average return of negative trade -0.0134 -0.0166 -0.0144 -0.017 total number of trades 5800 3014 5800 3014 average holding time 2.8714 2.9695 2.8714 2.9695 trades with excess return > 0 75% 76% 52% 57% without transaction costs with transaction cost most scenarios have reasonably high sharpe ratio which exceed market benchmark sharpe ratio of 0.096. modigliani risk-adjusted performance (rap) as another measurement of risk is also used, which is derived from the sharpe ratio and shows risk-adjusted return as well. basically, the higher the sharpe ratios number the better is the result, which suggests that the asset with the higher sharpe ratio gives more return for the same risk. correlation analysis between the naive buy-and-hold strategy over the benchmark portfolio and an arbitrage-style trading strategy demonstrate that for all scenarios there is a low negative correlation with the benchmark or no correlation at all. that confirms a market neutrality of the proposed trading strategy. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 4 72 the most profitable scenarios are scenario 1 and scenario 4 with average monthly return of 3.8 per cent and 2.06 per cent based on the analysis of hk closing prices and trades executed with the shortest possible delay after hk closed. these results are consistent with the analysis of granger causality and demonstrate that most of the trading action happens during trading session in hk, i.e. home market is dominant in price discovery (lok and kalev 2006). table 6: scenario 3, hk market open new york close vst hk open and new york open this table presents the summary statistics for the monthly average returns on the equally weighted portfolio that consists of 14 pairs of dual-listings between 2002 and 2008 with or without transaction costs. trigger level 0.01 0.02 0.01 0.02 monthly average return 0.0468 0.0271 0.005 0.0103 standard deviation 0.0343 0.0305 0.0292 0.023 t-statistics 12.3442 8.0479 1.5351 4.0428 median 0.0438 0.0225 0.0052 0.0083 skewness 1.2339 2.5194 0.4956 1.741 kurtosis 7.2595 12.0712 7.4569 11.5996 sharpe ratio 1.3632 0.8887 0.1695 0.4465 m2 (modigliani rap) 0.1287 0.0839 0.016 0.0421 min monthly return -0.045 -0.0402 -0.0971 -0.0681 max monthly return 0.182 0.1764 0.1262 0.1132 correlation with benchmark -0.1988 -0.2128 -0.0556 -0.0943 months with non-negative return 98% 94% 62% 76% individual average return per trade 0.0113 0.016 0.0013 0.006 average return of positive trade 0.029 0.0367 0.0255 0.0321 average return of negative trade -0.0247 -0.0299 -0.0255 -0.0311 total number of trades 4870 1978 4870 1978 average holding time 2.9811 3.2209 2.9811 3.2209 trades with excess return > 0 67% 69% 53% 59% without transaction costs with transaction cost mispricing between the underlying shares and adrs during hk trading session probably will not be fixed on the trading in new york, so the investor has enough time to open positions on both markets and profit from that mispricing. 73 table 7: scenario 4, hk market close new york lose vs hk close and new york open this table presents the summary statistics for the monthly average returns on the equally weighted portfolio that consists of 14 pairs of dual-listings between 2002 and 2008 with or without transaction costs. trigger level 0.01 0.02 0.01 0.02 monthly average return 0.0883 0.0693 0.0206 0.0244 standard deviation 0.039 0.0383 0.0347 0.0295 t-statistics 20.5064 16.3597 5.3869 7.4962 median 0.0807 0.0595 0.018 0.0197 skewness 1.2104 1.8195 0.3928 1.132 kurtosis 7.7086 8.5765 6.9314 8.2275 sharpe ratio 2.2646 1.8066 0.5949 0.8278 m2 (modigliani rap) 0.2138 0.1706 0.0562 0.0782 min monthly return -0.0236 0.0081 -0.0901 -0.0499 max monthly return 0.2644 0.2532 0.1622 0.161 correlation with benchmark -0.3366 -0.3674 -0.2158 -0.2445 months with non-negative return 99% 100% 82% 90% individual average return per trade 0.0131 0.0155 0.0031 0.0055 average return of positive trade 0.0172 0.0197 0.0137 0.0157 average return of negative trade -0.0106 -0.0112 -0.0098 -0.0102 total number of trades 7847 5221 7847 5221 average holding time 2.3921 2.4465 2.3921 2.4465 trades with excess return > 0 85% 86% 55% 61% without transaction costs with transaction cost 5. conclusion arbitrage strategy is a well-known international finance topic relating to trades in the world’s biggest currency markets. the review of literature suggests that, in regard to the law of one price in the trading of share-like instruments, the results on the lop is very mixed. hence the motivation of this study to conduct a longitudinal study of 14 major china firms, listed and traded in hong kong and in new york (as adrs). the issue examined is whether there is arbitrage such that the trading of these 14 shares in three markets in different currencies are arbitrage-free, meaning that the prices are not significantly different. in practice, arbitrage which is technically riskless will still involve a certain level of risk due to uncertain future cash flows (dividends), trading risks, and so on (kleinbard et al. 1993). international journal of banking and finance, vol. 9, iss. 2 [2012], art. 4 74 under certain circumstances, most of the chinese adrs that are traded in the us markets do offer better gaps comparing their underlying asset prices traded in hong kong. hence, arbitrageurs could pocket a profit. however there is inherent risk when conducting this strategy. the execution of the arbitrage might take longer time than an arbitrageur expects. in an ideal world, trades placed to capture an arbitrage opportunity would be instantaneously executed. however, in practice, execution takes time, especially for adrs trading and in two nonoverlapping markets. in most cases, price variations between the time when an arbitrage opportunity is entered into and the time when the trade is actually executed might differ greatly. further, some of the adrs might not be very liquid. therefore, establishing the long position in the us market might not be possible. we therefore attempted to ascertain if price parity is maintained between the dual-listed chinese shares traded in hong kong and new york. studies have shown that these identical assets may be traded freely across international borders; arbitrage trading by investors should ensure such asset prices will not differ by more than the transaction costs involved. the two markets of hong kong and new york in terms of dual-listed shares appear to confirm lop reasonable well. however, price disparities do exist, and so do the arbitrage opportunities. it might be asked why there are so many profitable disequilibria in these two developed markets. the possible explanation is the non-overlapping trading time. however, when the price disparity becomes greater, the large price difference gap would be arbitraged away immediately. although the overnight risk makes the arbitrage difficult, the empirical results in this study show the monthly return ranging from 0.5 per cent to 3.8 per cent could be obtained for this simple strategy when transaction costs are considered. this evidence could be taken as possibly arising from data deficiency or method deficiency or that it is simply that these 14 stock prices are inefficiently priced across borders. either way, the law of one price appears to be violated in the trades of these otherwise very liquid and visible chinese shares, although in exchange rate studies, there is more evidence of arbitrage-free trades in major currencies. author information: the submitting author is dr bogomolov and he may be contacted at his email address timofei.bogomolov@unisa.edu.au. the co-author is lixian liu with the e-mail address of lixian.liu@unisa.edu.au. both are staff members at the university of south australia. mailto:timofei.bogomolov@unisa.edu.au mailto:lixian.liu@unisa.edu.au 75 references: alves, m. v. b., and matthew r. m., (2003). intraday arbitrage in emerging market american depositary receipts. unpublished working paper. amary, beatriz, and otavio ottoni, (2006). adr arbitrage opportunities for dummies. unpublished working paper. bailey, w., chung, p., and kang, j., (1999). foreign ownership restrictions and equity price premiums: what drives the demand for cross-border investments?, journal of financial and quantitative analysis, 34: 489-511. bowen, d., hutchinson, m. c.. and o'sullivan n., (2010). high frequency equity pairs trading: transaction costs, speed of execution and patterns in returns. journal of trading, forthcoming. cai, h., (2007). bonding, law enforcement and corporate governance in china. stanford journal of law, business & finance 13: 82-111. de jong, l. r., and a. van dijk, m., (2009). the risk and return of arbitrage in dual-listed c ompanies. review of finance 13: 495-520. ding, d. k., (2000). price parities of stocks listed on both the kuala lumpur stock exchange and singapore’s cloc international. advances in pacific basin financial market, 21-44. do, b, (2009). are pairs trading profits robust to trading costs?, unpublished working paper. domowitz, i., glen, j., and madhavan, a., (1998). international cross-listing and order flow migration: evidence from an emerging market. journal of finance 53: 2001-2027. gande, a., (1997). american depositary receipts: overview and literature survey. financial markets, institutions & instruments 6: 61-83. gatev, e., goetzmann, w.n., and rouwenhorst, k.g., (2006). pairs trading: performance of a relative-value arbitrage rule. review of financial studies, 19: 797-827. han, b., (2004). limits of arbitrage, sentiment and pricing kernel: evidences from index options. unpublished working paper. hsu, j., and wang, hsin-yi, (2008). why do price spreads between domestic shares and their adrs vary over time?. pacific economic review 13: 473-491. kato, k., linn, s., and schallheim, j. (1991). are there arbitrage opportunities in the market for american depository receipts? journal of international financial markets, institutions & money 73: 13-32. kaul, a., and mehrotra, v., (2000). in search of international integration: an examination of intra-daily north american trading of canadian dually listed stocks. working paper, university of alberta. kim, m.,. szakmary, a.c., and mathur, i., (2000), price transmission dynamics between adrs and their underlying foreign securities. journal of banking and finance 24: 1359-1382. kleinbard, e.d., clearly, g., and hamilton, s., (1993). risk and riskless positions in securities. taxes, 783-799. lau, s.t., diltz, j. d., and apilado, v.p., (1994). valuation effects on international stock exchange listings. journal of banking and finance 18: 743-755. lok, e., and kalev, p., (2006). the intraday price behaviour of australian and new zealand cross-listed stocks. international review of financial analysis 15: 377-397. mak, b.s.c., and ngai, a.m.s., (2005). market linkage for dual-listed chinese stocks. the chinese economy 38: 88-107. maldonado, r.,, and saunders, a., (1983). foreign exchange restrictions and the law of one price. financial management, 12: 19-23. malkiel, b.g., (2007). the efficiency of the chinese stock markets: some unfinished business on the road to economic transformation. ceps working paper no. 154. miller, d.p., and morey, m.r., (1996). the intraday pricing behaviour of international dually international journal of banking and finance, vol. 9, iss. 2 [2012], art. 4 76 listed securities. journal of international financial markets, institutions and money 6: 79 89. park j., and tavakkol, a., (1994). are adrs a dollar translation of their underlying securities? the case of japanese adrs. journal of international financial markets, institutions and money 4: 77-87. pontiff, j., (2006). costly arbitrage and the myth of idiosyncratic risk. journal of accounting and economics 42: 35-52. suarez, e. d., (2005). arbitrage opportunities in the depositary receipts market: myth or reality?. journal of international financial markets, institutions and money 15: 469-480. wahab, m., lashgari, m., and cohn, r., (1992). arbitrage opportunities in the american depository receipts market revisited. journal of international financial markets, institutions and money 2: 97-130. an analysis of bank efficiency in the middle east and north africa 27 the international journal of banking and finance, volume 9 (number 4) 2012: pages 28-47 an analysis of bank efficiency in the middle east and north africa saeid eisazadeh and zeinab shaeri bu ali sina university, iran ____________________________________________________________ abstract this paper reports institutional factor effects on bank efficiency in middle eastern and north african countries during a recent 14 years. the methods used are: stochastic frontier analyses and second-stage tobit regression to investigate the impact of institutional-cum-financial as well as bank-specific variables on efficiency. overall, the analysis shows that banks could save 20 percent of their total costs if they were operating efficiently. factors that affect production efficiency are: macroeconomic stability, financial development, the degree of market competition, legal rights and contract laws, better governance and political stability. differences in technology seem to be crucial in explaining efficiency differences. our findings point to the importance of policies that aim to build stronger institutions, promote more competition, and improve governance. policies should be aimed at giving banks incentives to improve their capitalization and liquidity. improvements in the legal system and in the regulatory and supervisory bodies would also help to reduce inefficiency, areas of immediate concerns for this vast region. finally, increased investments and upgrading of the stock markets in the region would help banks improve their performance through market-based investor actions. key words: bank efficiency, stochastic frontier analysis, competition jel classification: g21, o16 _____________________________________________ 1. introduction a large number of developed and developing countries have deregulated their banking systems over the past two decades: see ariff and can (2008). the primary objective of such reforms was to improve productivity, efficiency, and profitability of the banking systems and also to increase international competitiveness. in particular, developing countries, mostly following international monetary fund (imf) and/or world bank initiated programs, sought to improve performance and efficiency of financial sectors to enhance their overall economic performance. indeed, a strong and stable banking system has been advocated as being the cornerstone in many liberalization programs (saunders and sommariva, 1993). this is pertinent for the region studied. the banking industry may have benefitted considerably from advances in both nonfinancial and financial technologies. banks have used information processing to process deposit and loan customer information more efficiently and applied telecommunications technologies to transmit information and to process payments more quickly with less resources. banks have also used new financial technologies to provide new services, to evaluate risks more efficiently, and to unbundle and repackage risks in new ways. this paper uses bank level data to study the efficiency of the banking sectors in 19 mena (middle east and north africa) countries. we focus on the cost efficiency of the banks, and rely on the stochastic frontier analysis (sfa) to compute the efficiency score document the difference between bank's actual cost efficiency levels relative to an economy’s cost efficiency production frontier in the spirit of the studies that have focused on these newer research method away from accounting measures of efficiency. we find that, on average, banks would save about 20 percent of their total costs if they were operating on the frontier. the mena region is strategically located between asian economies and the western world. mena countries were colonized by the french or the british until the middle of the past century. major institutions including financial intermediaries were established on the line of western institutions. financial sectors these countries are generally still in the early phases of economic development. capital markets are weak or almost non-existent, and financial markets are dominated by bank-financed credit mechanisms. in this framework, banks are the main suppliers of credit to private and public investment projects and they also finance government deficits (turk-ariss, 2009). the major purpose of this paper is to find the common factors that could help explain the differences in efficiency among banks. the study focuses on the aggregate influence rather than bank or country-specific factors. therefore, the findings of this paper could be seen as an initial step toward understanding the production efficiency of the banking sectors in these countries. policy implications for specific countries, however, would certainly require more detailed follow-up country studies, which could be the next step in our research. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 3 the paper is organized as follows. section 2 reviews the literature on cost efficiency analysis. section 3 provides an overview of the banking sector in the mena countries. in section 4 are the results and discussion on the cost efficiency analyses while in section 5 the readers will see evidence on what determinants affect bank efficiency levels. the paper ends with a conclusion in section 6. 2. literature on production efficiency efficiency can be defined as the extent to which a decision-making unit (dmu) meaning a bank can increase its outputs without increasing its inputs; or reduce its inputs without reducing its outputs. efficiency is generally classified into three forms: scale efficiency; scope efficiency; and x-efficiency. x-efficiency measures whether banks are operating with an efficient mix of inputs, and has been the focus of recent bank efficiency studies. it represents the ability of management to control costs and use of resources to produce output. measuring the efficiency levels of individual banks is usually the first step. after all, understanding the determinants behind the differences among banks’ efficiency levels is more interesting, as is done in this paper. efficiency studies using frontier approaches did not start until sherman and gold (1985) initiated a study: prior to this, the fashionable method was malmquist’s data envelopment analysis. they applied the frontier approach to the banking industry by focusing on the operating efficiency of the branches of a saving bank. since then, frontier approaches have become popular as measures of banking efficiency. there have been extensive studies on bank efficiency of the us and european countries and most of them focused on conventional banking (berger and humphrey, 1997; goddard et al., 2001). only few efficiency studies on islamic banking can be found (elzahi saaid, 2002; hussein, 2003, kabir hassan, 2004; hamim mokhtar et al., 2006; bader, et al., 2007). however, studies dedicated to some south saharan countries remain limited (demirguc-kunt, et al, 2004 and chuling chen, 2009). mena countries have been also studied for some periods (olson and zoubi, 2008; chaffai and dietsch,2006; kobeissi, 2004;srairi,2009;sufian,2008). the relationship between efficiency and market structure is not so clear-cut in those studies since the studies focused on efficiency only. beck and hesse (2006) find that market structure in uganda played a limited role in determining bank efficiency, and structural impediments were more significant in lower spreads and margins charged. demirguc-kunt, et al. (2004) finds no robust association between bank concentration and interest rate margins. however, the paper by turk-ariss (2009) attempted to examine competition structures of region, of which tunisia is a part. based on revenue elasticity to input prices, and retaining a set of market and contestability indicators, the degree of competition measured for the test period of 7 years demonstrates that the region is characterized, for the most part, by a monopolistic competition. 3. overview of the banking sectors in mena countries the middle east and north african (mena) region is strategically located between asian economies and the western world. except for turkey, mena countries were colonized by the french or the british until the mid-last-century. the region is important for a number of reasons. it represents a bridge between europe and asia. in the current century, it is a fast growing region in terms of both population and wealth while its banking sector is relatively young with most banks being established since the 1970s. the region includes the rapidly expanding oil rich countries of the gulf cooperation council (gcc) as well as the arab countries of the near east and north africa. the world’s largest islamic banks are located in the mena region such as in bahrain and uae and its mix of conventional and islamic banks permits a comparison of efficiency and profitability by types of bank, a feat rarely done (see bader et al. op cit). financial sectors are generally still in the early phases of economic development. capital markets are weak or almost non-existent, and financial markets are dominated by bank-financed credit mechanisms. in this framework, banks are the main suppliers of credit to private and public investment projects while they also finance government deficits. while these features are common to many other emerging economies, banking sectors in this region are unique in three aspects. first, the recent oil price hike marks the beginning of a new era that was last witnessed a quarter of a century ago. investment opportunities in the united states using petrodollars coming from the middle east became more restricted after 9/11 event, so that oil surplus funds have to be channeled to productive uses elsewhere in the world. monetary authorities in these countries generally require banks to adopt international accounting standards as well as comply with international regulatory requirements such as basel ii and the anti-money laundering recommendations. still, a major concern is raised regarding the absorptive capacity of banks in the region to recycle oil surplus funds. this is pointed by policymakers with appropriately international journal of banking and finance, vol. 9, iss. 4 [2012], art. 3 designing policies for more efficient and stable banking systems. the banking systems in the region have traditionally been very highly concentrated markets, meaning low competition. in some countries where a large number of banks operate as in lebanon, turkey and the uae, he large banks have sought to consolidate their position domestically before expanding in the region. in other countries (jordan and kuwait) leading banks are strategically investing across borders in order to enhance their growth potential. ongoing consolidation of financial institutions within each country and regionally justly intensifies public policy debates on issues of concentration and competition in the banking industry. third, the governance structure of banks is evolving, following accession to the world trade organization (wto) and greater commitment to financial liberalization. traditionally, banking institutions were either mostly family-owned businesses managed by major shareholders who cater for their own personal interests, or dominated by state authorities, thus making it more difficult for new firms to compete in the industry. in the past two decades, however, ownership of large shares of the banking systems turned over from government to private control and from domestic to foreign control. such changes occurred as governments privatized many of their state-owned banks while providing reduced barriers to entry in line with the wto accession requirements. foreign investors bring in state-of-the-art technology, sophisticated risk management techniques and qualified human capital thus forcing domestic banks to undergo major structural reforms in order to compete on an equal platform with their peers. together, financial liberalization measures and incentives attract foreign banks accept better disclosure requirements which, together with better regulatory environment, ultimately should be a positive effect on growth performance (turk-ariss, 2008). it is long believed that advanced economies are characterized by a sophisticated financial systems, although, this myth may be put to rest after the 2008 world crisis. policy makers in developed and developing countries realized the importance of improving and restructuring their financial systems and mena region has taken a lesson to do so. mena countries have recognized the importance of financial sector reforms under the auspice of the international monetary fund (imf) so that their banks could become modern financial operators to help allocating investment, enhance productivity and effectively participate in economic growth. more recent investment in the financial sector has surged in these countries and the size of their economy has increased significantly since 2000. the gross domestic product (gdp) has increased from $ 709 billion in 2000 to $ 1,276 billion in 2006 (cherif et al., 2008). however, commercial banks are the dominant part in the mena financial system (alfayoumi and abuzayed, 2010). although, restructuring initiatives in the region are not as energetic as those taking place in eastern europe or in some parts of asia, several countries are witnessing new eras in privatization, bank regulation, and market-organizational structures. serious steps have been taken to improve bank efficiency and corporate governance (ben naceur and omran, 2008). a recent paper (cherif et al., 2008) demonstrates that the mena countries have relatively well developed financial systems compared with the european accession countries using as judged by three measures of broad money to gdp ratio, domestic credit in the banking sectors to gdp ratio, and the ratio of private credit to gdp. 4. efficiency analyses 4.1. methodology efforts to measure how efficiently a firm produces outputs with a given mix of inputs have led to the development of a number of efficiency concepts: scale efficiency, scope efficiency, economic efficiency, and x-efficiency. economic efficiency builds on scale and scope efficiency by incorporating prices and thereby allowing the firm to react to price changes thus potentially gaining market power. the concept of x-efficiency or managerial efficiency goes one-step further in the sense that it measures efficiency in implementing an existing production plan with given prices and technologies. there are several econometric (parametric) and linear programming (non-parametric) techniques used to measure efficiency: berger et al. (1993) and berger and humphrey (1997). the parametric approach has the advantage of allowing noise in the measurement of inefficiency. however, the approach needs to specify the functional from for production, cost or profit. the non-parametric approach is simple and easy to calculate since it does not require the specification of the functional form (coelli, 2004). to examine the efficiency of banks using frontier approaches, there are two models. parametric technique, such as stochastic frontier analysis (sfa), thick frontier approach (tfa) and distribution free approach (dfa), uses econometric tools and specifies the function form for the cost or profit function. on the contrary, the non-parametric approaches (such as dea) and free disposable hull analysis (fdha) do not make an assumption concerning the functional form of frontier and use a linear program to international journal of banking and finance, vol. 9, iss. 4 [2012], art. 3 calculate efficiency level. in the present study, we use the sfa, as developed by aigner et al. (1977), to estimate cost efficiency frontier. the main advantage of sfa over dea is that it allows us to distinguish between inefficiency and other stochastic shocks in the estimation of efficiency levels. in addition, by using this model, it would be easier to add control variables, such as country-level variables, in the equation of this model than in non-parametric techniques. hence, this approach allows us to compare efficiency between countries (srairi, 2010). applying sfa approach in this paper is appropriate to estimate the efficiency frontier. the sfa approach is one of the structural approaches to study efficiency. it is based on the economics of cost minimization or profit maximization by banks. thus it starts with a standard cost or profit function with factors of input, output, and their respective prices. it estimates the minimal cost or maximum profit based on these functions, and generates an efficiency frontier for the sample. the efficiency of each bank is then measured as the distance of its cost or profit to the frontier value (chen, 2009). a bank is labeled inefficient if it is behaving less optimally with respect to cost than the frontier value after taking out the random error. a bank’s total cost can be modeled as follows: ltc = f(w, y) + ln uc + ln vc (1) where lntc is the total cost variable, f denotes some functional form, y is the vector of output variables, w is the vector of prices of input variables, ln uc is the inefficiency factor that may raise cost above the bestpractice optimal cost and ln vc is the random error incorporated to capture the measurement error and luck, which may temporarily increase or decrease a bank s costs. basically, the cost function above describes the relationship between the cost variables with prices of input variables, quantities of output variables plus the inefficiency and random error. we estimate the following standard multi-product translog cost function: where, lntc= the natural logarithm of total costs; lny= the natural logarithm of output quantities; lnw= the natural logarithm of input prices; 𝐸𝑖= v+u in equation (iv.1); α, β, δ and ρ are coefficient to be estimated. 4.2. data and results we obtain bank level data from bankscope, which covers 90 percent of banks worldwide. the study used 266 panel data from the reports of 19 middle east and north africa countries from 1995 to 2008. all the variables used in the cost function are obtained from the balance sheet and income statement information in the bankscope database. the computer program frontier version 4.1, developed by coelli, is used to obtain the maximum likelihood estimates of the parameters of technical and cost efficiency (coelli, 1996 and coelli et al., 1998). the program can accommodate cross sectional and panel data; cost and production function; half-normal and truncated normal distributions; time-varying and invariant table 1: stochastic technical frontier ols parameter estimates t-ratio standard error coefficient parameter variables 4.697 1.218 5.723 intercept beta0 1.438 0.153 0.220 𝐿𝑛 𝑌 beta1 3.315 0.143 0.476 ln(w1 w2⁄ ) beta2 4.925 0.0107 0.0531 lnylny beta3 5.386 0.00497 0.0267 ln(w1 w2⁄ )ln(𝑊1 w2⁄ ) beta4 3.9130.00993 0.0388ln(w1 w2⁄ )lny beta5 0.133 sigma-squared 106.72log likelihood function notes: y= total earning assets (financing/loans, trading& investment securities and placement to other financial institution). w1=price of labor and capital, w2=price of deposits. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 3 table 2: stochastic cost frontier maximum likelihood parameter estimates t-ratio standard error coefficient parameter variables 2.971 1.1646 3.4604 intercept 3.483 0.1462 0.5094 𝐿𝑛 𝑌 0.6758 0.1178 0.07967 ln(w1 w2⁄ ) 4.907 0.01723 0.08458 lnylny 2.824 0.009726 0.02746 ln(w1 w2⁄ )ln(𝑊1 w2⁄ ) 0.48650.01717 0.008357ln(w1 w2⁄ )lny 4.540 0.03710 0.1684 sigma square 7.801 0.07765 0.6057 gamma 45.223log likelihood function table3: sfa cost efficiency (ce) estimates by banking system cost efficiency country 0.7679 algeria 0.8438 bahrain 0.7427 djibouti 0.8298 egypt 0.8475 emirate 0.7246 iran 0.8046 iraq 0.7629 israel 0.7752 jordan 0.8693 kuwait 0.8046 libya 0.6902 lebanon 0.7307 morocco 0.8034 oman 0.8739 qatar 0.8090 saudi 0.8883 syria 0.7568 tunisian 0.8379 yemen 0.80 mean efficiencies; and functional forms which have a dependent variable in logged or original units. these features of frontier 4.1 exhaustive, but provide an indication of its capabilities. we first estimate the model by pooled ordinary least squares (ols) regression (regression 1), and then conduct the frontier analysis by making the assumption that follows half-normal distribution and by mle regression (regression 2). we also estimate the model controlling for country specific effects (regressions 3). the estimation results for efficiency are reported in table 1, table 2 and table3. the results show that individual inefficiency can explain a large part of the variance we see in the production process of the banks. the estimation also shows consistency in terms of the efficiency levels and ranking of the banks. the overall efficiency levels are about 0.8, meaning 20 percent of total cost can be saved if banks were operating efficiently. 5. determinants of efficiency having obtained the individual bank’s cost efficiency; we next investigate if the efficiency levels can be explained by several different groups of country-specific or bank-specific factors. we will first determine the variables and then include them as explanatory variables in the following equation: where costeffi,t, is the bank level cost efficiency score from the sfa analysis, and mi,t includes the variables that could have potential impact on the cost efficiency levels of the banks. more specifically, we consider two groups of variables. the first group includes factors that are more specific to individual banks, and the second encompasses the external environment that banks operate in, such as macroeconomic conditions, financial depth, market structure, regulatory framework, and overall institutions. since we study cross bank-specific characteristics alone might not be enough to explain the difference in efficiency levels observed across the sample. in fact, there could be important country-specific factors that are omitted, but significantly correlated with both efficiency levels and the bank-specific characteristics we use. to disentangle the impact of bank-specific factors from that of environmental factors, we keep bank-specific variables in each regression. to avoid the possible multicollinearity between the different groups of variables, we also include each group of factors one at a time. 5.1. bank-specific factors the characteristics of a typical bank, such as size, ownership, organization forms, service quality and so on, can affect bank efficiency. in this research we are particularly interested in international journal of banking and finance, vol. 9, iss. 4 [2012], art. 3 specific factors: risk profile, business specialty, and service quality. these factors are interrelated. the amount of risk a bank takes on can change the efficiency results significantly. banking service quality may considerably change a bank’s efficiency. in this study the ratio of loan loss provisions to total loans (loss) and the loan-to-asset ratio (loan/asset), and the ratio of total other operating income to total asset (other income) are proxy for the risk level, the table 5: second stage regression results )8( )7( )6( )5( )4( )3( )2( )1( 0.76*** )0.015( 0.86*** )0.02( 0.81*** )0.007( 0.791*** )0.032( 0.73*** )0.019( 0.77*** )0.047( 0.803*** )0.012( 0.78*** )0.013( constant 0.038 )0.15( 0.1-*** )0.08( 0.075 )0.397( 1.324 )1.18( 0.097 )0.145( 0.020 )0.102( 0.093 )0.304( 0.090 )0.11( loss 0.073*** )0.015( 0.015 )0.017( 0.1-*** )0.023( 0.052 )0.089( 0.08*** )0.015( 0.08*** )0.015( 0.08-*** )0.015( 0.046*** )0.012( loan/asset 0.84-*** )0.22( 0.03 )0.14( 0.69 )0.54( 7.41-*** )1.62( 0.33 )0.29( 0.9-*** )0.32( 0.60-* )0.38( 0.65-*** )0.22( other income 0.0024*** )0.001( gdp per capita 0.001-*** )0.0006( inflation 0.01*** )0.003( political stability 0.036** )0.018( concentration 0.062*** )0.027( competition 0.01*** )0.009( financial depth 0.01*** )0.003( credit right .003*** )0.001( enforcement of contract 0.015*** )0.007( rule of law 0.14 0.051 0.11 0.86 0.35 0.16 0.12 0.093 r squared 0.13 0.035 0.096 0.80 0.34 0.14 0.099 0.08 adjust r squared 8.12 1.07 6.89 13.36 27.49 9.64 5.31 7.052 f test notes: standard error in bracket. *, ** and *** correspond to 0.10, 0.05, and 0.01significance levels, respectively. impact of the difference in product and services a bank offers, and the bank’s investment preferences between loans and other earning assets respectively. the ratio of total other operating income to total asset (other income) is a index for banking service quality. data are obtained directly from bankscope. results (table 5) show that higher levels of loss provision mean lower cost efficiency for banks, although this variable is statistically insignificant across all regressions. the loan toasset ratio exhibits a positive relationship with cost efficiency, indicating that for banks in our sample, loan products are more cost-efficient than other types of earning assets. this could, however, also indicate that higher market power might exist in the loan product market than other product markets (berger and master, 1997). higher total other income, however, is found to lower the cost efficiency of banks, implying that higher income from sources such as fees and commissions, could reflect a higher cost of providing such services. when we include other groups of variables in regressions, the bank-specific factors show a consistent pattern. the loss provision variable is insignificant, while the loan-to asset ratio and ratio of total other income to assets are found to be significant in most regressions except in those with macroeconomic variables or overall institution variables. 5.2. macroeconomic conditions we next consider if cost efficiency levels can vary systematically across countries due to differences in the macroeconomic environment. two macroeconomic variables are used: logarithm of per capita gdp and inflation. so, this study gives issues concerning the relationship between inflation and per capita gdp and efficiency banking system for mena region countries. both per capita gdp and inflation data are obtained from the international monetary fund’s world economic outlook (weo). per capita gdp is used to reflect the general income level. a higher income level is more likely to be associated with a more developed banking sector. our estimation shows that it can also bring higher cost efficiency. over the past two decades or so, economic growth in countries in the mena region has generally lagged behind those of the major emerging market economies in asia, latin america and central and eastern europe (nabli and véganzonès-varoudakis (2004)). figures 1 shows that real per capita gdp growth rates have picked up in the mena region over the past decade. however, in the period since 1998, emerging market economies in asia and in central and eastern europe have continued to perform significantly better, while sub-saharan africa has achieved an even more impressive acceleration in real per capita gdp growth. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 3 inflation is an indicator of macroeconomic stability, and is directly related to the interest rate levels and, thus, interest expense and revenue. macroeconomic instability would, in general, have an adverse impact on banking sector performance. a bank’s ability to manage interest rate risk under inflationary conditions can also affect its cost structure. globally, we find that inflation has a negative and significant incidence on cost efficiency in banking system. figure (2) gives an overall picture on the evolution of inflation in mena as a group and a comparative inflation performance at regional level. figure1: gdp per capita growth rate (percent) source: world development indicators database figure 2: evolution of inflation in mena (1980-2007) over the past three decades, the dynamic process of inflation is affected by a combination of global and domestic factors. at the beginning of 1980s, inflation showed a declining trend culminating ultimately in price collapse. however, at the end of 1985 the situation changed rapidly. the inflation surged extremely fast and became more volatile till mid-1990s, the ongoing geopolitical tension, as like the iranian-iraqi a wars in 1988, gulf war 1990-1991 was in cause. in the late of 1990s and early 2000s, inflation has declined from double to signal digits, reflecting improvements in the terms of trade and stronger demand management policies. the level of financial development is also crucial to bank efficiency. higher levels of financial depth could contribute to the better performance and higher efficiency levels of banks. we use bank deposits to gdp to capture the cross-country differences in financial depth. we find the relationship between cost efficiency and bank deposits to gdp to be positive. this indicates that more financial intermediation in the form of bank deposits tends to help reduce costs of bank operations. 5.3. market structure the relationship between market structure and efficiency is an important aspect that this paper explores. the usual indicator for market structure is market concentration, often expressed by a herfindahl index. there are basically two views on the relationship between market structure and bank efficiency. one view holds that concentration and restrictions generate market power and, thus, monopolistic profits. in this case, one often observes a positive relationship between concentration and profitability, which might not mean higher efficiency. the other view, however, argues that market structure is a result of competition whereby more efficient banks dominate the less efficient ones and, thus, market concentration is a result of higher efficiency. empirical studies have also found an ambiguous relationship between market concentration and bank efficiency, which, as recognized by many, indicates that simple market structure indicators, such as concentration ratios, are not good proxies of market structure (chuling chen, 2009). instead of focusing on the concentration ratios, we investigate the degree of competition in the market within which the banks are operating. this is because competition pressures might be more effective in improving efficiency, and a concentrated banking market could also be competitive and efficient. we explore whether the difference in market competition pressures can explain the variation in efficiency across countries. first, we follow panzar and rosse (1987) and estimate the reduced form revenue equations to formally test the level of competition for each country. this statistical analysis requires the estimation of the reduced form of bank revenue. the quantity and price of the equilibrium determining total revenue equilibrium depend on costs, demand and conduct. accordingly, all determinants of costs and demand must be included in banks’ revenues functions. a particular attention will be devoted to the price of factors. in the elaborated model, we preserve the linear form of the relationship between international journal of banking and finance, vol. 9, iss. 4 [2012], art. 3 dependant and independent variables. the works of claessens and laeven (2004), prasad and ghosh (2005), yuan (2006), gutiérrez de rozas (2007) and turkariss (2009), have established its theoretical basis. the reduced form of the following specification is: where it revn is the ratio of total interest revenue to total assets for bank i at time t, it intc is the total interest expenses to total deposit, lc is the ratio of personnel expense to total assets, and othc is the ratio of total other operating expenses to total assets. we also include the following variables to control for bank-specific characteristics: loan is the ratio of total loans to total assets, cap is the ratio of equity to total assets, and ta is total assets. for robustness, we also estimate models with total revenue as a dependent variable, where other operating revenue such as commission and fee income is included. in order to measure competitiveness of the banking industry, panzar and rosse (1987) define the competitiveness h measure as the sum of the elasticities of the reduced form bank revenue equations with respect to the bank's input prices. specifically, the h-statistic measures the percentage of change in the equilibrium revenue of a bank generated by a change of 1 percent in entry costs. in a case of a perfect competition, the h-statistic is 1. this situation might emerge with an oligopoly operating in a contestable market. if the market, in which banks operate, is characterized as being a monopole, then the h-statistic is inferior or equal to zero. we estimate the h -statistic for each country using both fixed effects and random effects models. the results are reported in table 4. table4: tests of conditions of competition in total interest revenues t-ratio standard error coefficient variable 2.990.27 0.83constant 22.39 0.016 0.38 log(intc) 8.66 0.027 0.23 log(ohtc) 7.18 0.021 0.15 log(lc) 6.57 0.041 0.27 log(loan) 2.770.028 0.08log(cap) 3.90 0.016 0.065 log(ta) 0.99 r squared note: all coefficients are statistically significant at or above 0.05 levels. the four specifications generally provide consistent estimates for the h -statistic for each country. most of the countries report an h -statistic between 0 and 1, which suggests that monopolistic competition best describes the level of competition in the banking sector. we then include the average of the h -statistics from the different specifications in the second stage regression to determine the effect of the market structure. we found that higher levels of competition in the market will boost the cost efficiency in various specifications of the model. 5.4. legal framework the quality of the legal framework with regard to enforcement of contracts and protection of property rights is important for banking sector efficiency. for example, in their study of financial deepening in ssa, mcdonald and schumacher (2007) find that after controlling for financial liberalization and macroeconomic variables, countries with stronger creditor rights and information sharing have deeper financial systems (chuling chen,2009). we include two variables that reflect the quality of the legal framework in our study strength of credit rights and enforcement of contracts, both obtained from the world bank’s doing business indicators. to assess the power theories of credit, we construct a measure of legal rights of creditors in these countries, the “creditor rights” index first proposed by la porta et al. (1997, 1998), for every year during this period. a score of one is assigned when each of the following rights of secured lenders is defined in laws and regulations: first, there are restrictions, such as creditor consent or minimum dividends, for a debtor to file for reorganization. second, secured creditors are able to seize their collateral after the reorganization petition is approved, i.e. there is no "automatic stay" or "asset freeze." third, secured creditors are paid first out of the proceeds of liquidating a bankrupt firm, as opposed to other creditors such as government or workers. finally, if management does not retain administration of its property pending the resolution of the reorganization. the index ranges from 0 (weak creditor rights) to 4 (strong creditor rights). (djankov, mcliesh and shleifer (2006)). the index for enforcement of contracts reflects the effectiveness of the court system in terms of the time, cost, and number of procedures involved for a plaintiff to get actual payment after filing a dispute. we obtain the ranking of each country among a total of 178 countries, where a higher number indicate a lower ranking in terms of effectiveness in enforcing contracts. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 3 the estimation results show that a better legal framework is indeed beneficial to improving cost efficiency of banks. high quality collateral and bankruptcy laws could effectively protect the banks’ rights as lenders, as well as those of borrowers, and higher efficiency in enforcing contracts could reduce costs for banks. we also estimate the impact of the legal framework while controlling for other groups of variables (table 5) and results are consistent across all specifications. 5.5. political environment the political stability and quality of public services are also influential factors in some developing countries such as mena countries. these variables are perhaps more relevant to the countries in than elsewhere, given the frequency of episodes of internal conflict. we obtain data on overall institutional quality from the world bank’s worldwide governance indicators (wgi). a higher score indicates a more stable political environment (figure 4). figure 4: political stability in mena countries the findings show that banks enjoy higher efficiency with more political stability. the coefficient is statistically significant in most specifications. to see if the results remain valid with effects from other variables, we also conduct regressions with other groups of variables (table 5). we find the results to be consistent across all specification. 6. conclusions because of deficiency in capital market and other financial institutional operations and development, mena country banks play a central role in the financial intermediation process. hence, knowing how efficient the banking sector is efficient relative to the production frontier estimated for the region did reveal the state of play across an important region straddling europe and asia. efficiency is important since the region almost totally depends on the banking intermediation for capital allocation. to understand how efficient they are, and more importantly, what determines their level of efficiency, thus, is important to help strengthen the financial intermediation function of banks, as well as the overall financial market. we find banks are operating 20 percent below the cost efficiency frontier we also find that a stable macroeconomic environment, deeper financial development, higher degrees of market competition, and stronger institutions would help improve the bank efficiency levels. the literature suggests that the more mixed financial markets with longer history operate slightly better at around 90 percent meaning the inefficiency is less than in the mena region. while acknowledging the economic fact that no system could be 100 percent efficient given the need in production sector to have slack for variations in demand and supply, we are of the view that there is room for improving the inefficiency from its current level to more like a 10 percent through careful fine-tuning of competition rules, and de-regulations. author information: saeid eisazadeh is assistant professor of economics, bu ali sina university, iran. the author may be reached via e-mail: saeedisazadeh@gmail.com. zeinab shaeri is a postgraduate student in economics at the economics faculty, bu ali sina university, iran: she may be reached via e-mail: shaeri.zeinab@gmail.com. references: ari,aisen., francisco,veiga., (2005).does political instability lead to higher inflation? a panel data analysis. imf working paper, wp/05/49. ariff m., and luc can (2008) “cost and profit efficiency of banks: a non-parametric analysis” co-author: luc can, china economic review, (2008) vol 19: 260-278. ariff, m., m bader, shamsher m. and t. hassan (2008), “cost, revenue, and profit efficiency of conventional versus islamic banks: financial ratios approach”, review of islamic economics, vol 11 (2). barth,j., caprio,g., and levine,r.,(2004). bank regulation and supervision: what works best?. journal of financial intermedition, no13(2004)pp 205-248. berger,a., mester,l.,(1997). effeciency and productivity change in the u.s. commercial banking industry: a comparision of the 1980s. research department working paper no 97-5 international journal of banking and finance, vol. 9, iss. 4 [2012], art. 3 bergestresser,d., (2001). market concentration and loan portfolios in commercial banking. pp 1-50. beck,thorsten.,hesse,heiko., (2006). bank efficiency, ownership and market structure why are interest spread so high in uganda?. wps4027, world policy research working paper 4027,october 2006. brewer,e., and e, jacsone., (2004). the risk-adjusted price-concentration relationship in banking. working paper, 4004-4035, december, pp 1-20. chen,chling., (2009). bank efficiency in sub-saharan africa middle –income countries" imf working paper, wp/09/14. caporala,g., ch, rault., r, sova., and a, sova., (2008). financila depth and economic growth in new eu members. dietrich, a.,wanzenried, g., cole, r., (2009). why are net-interest across countries so different?. institute of financial services ifz, august 29,2009. drack, p., (2000). inflation risk and portfolio allocation in the banking system. ucla, pietro garibaldi, imf. fries,s., and taci, a., (2004). cost efficiency of banks in transition: evidence from 289 banks in post-communist countries. working paper no.86, april 2004. guevara,j., and maudos, j., (2004). measuring welfare loss of market power: an application to european banks. applied economics letters, ii, p833-836. hauner,d., and peiris,sh.,(2005).bank efficiency and competition in low-income countries: the case of uganda.imf working paper, african department,wp/05/240. hesse,h.,(2006).bank efficiency, ownership and market structure: why are interest spreads so high in uganda?. world bank policy research working paper4027,october 2006. hopp,ch.,dreher,a.,(2007).do differences in institutional and legal environments explain crosscountry variations in ipo under pricing?.march 2007. kaufman,d.,a,kraay.,and m,mastruzzi.,(2009).governance indicators for 1996-2008. laeven,luc.,levine,ross., and kunt,a., (2003).regulations, market structure, institutions, and the cost of financial intermediation. finance research manager development research group, the world bank, july2003. landrom,l., (2003). good news on twelfth district banking market concentration. pp 1-8. mittal,manish.,dhade, aruna., (2007).profitability and productivity in indian banks: a comparative study. volume 1, number 2, may 2007, pp. 137-152. matthews, kent., guo, jianguang., and zhang nina.,(2007).non-performing loans and productivity in chinese banks: 1997-2006.cardiff economics working papers, october 2007. mokhtar, h.,abdullah,n., al-habshi,s.,(2006).efficiency of islamic banking in malaysia: a stochastic frontier approach. journal of economic cooperation 27, 2(2006), 37-70. nakane, marcio., weintraub, daniela.,(2005).bank privatization and productivity: evidence for brazil. journal of banking & finance 29 (2005), 2259-2289. nevine,m.,(2008).financial development. faculty of management technology, working paper no.12, july 2008. naceur,samy., goaied,mohamed.,(2002).the determinants of commercial bank interest margin and profitability: evidence from tunisia. working paper 8563. oster, alan., and lawrence antioch., (1997). measuring productivity in the australian banking sector. national australia bank. pilloff,s., and rhoades,s., (2002).structure and profitability in banking market concentration. review of industrial organization 20, pp 81-97. steinherr,a.,tukel,a., and ucer,m.,(2004). the turkish banking sector. college of europe. turk-ariss,rima.,(2008).competitive behavior in middle east and north africa banking systems. the quarterly review of economics and finance. 49(2009)693-710. yeyati,e., and micco,a., (2003). concentration and foreign penetration in latin american banking sectors: impact on competition and risk. pp 1-20. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 3 the effectiveness of bank recapitalization in japan 113the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134 the effectiveness of bank recapitalization in japan heather montgomery asian development bank, philippines abstract this study examines the effectiveness of japanese banking recapitalization policies. based on the careful reading of individual business revitalization plans submitted by several banks requesting government funds, we identified four primary goals of the capital injection plan: to increase the bank capital ratios; increase lending to avoid a credit crunch; increase the number of write-offs for non-performing loans; and encourage restructuring. using a panel of individual bank data, we empirically estimated the effectiveness of the policy. our findings suggest that capital injections are more effective for international banks than for domestic banks. the capital injections do not appear to affect lending to smes for either bank type. for international banks however, receipts of injected capital seem to relax the constraint on overall loan growth. the receipt of injected capital strengthens the capital positions of both international and regional banks, but these results do not hold up once we control for possible endogeneity when tested. jel classification: f28 key words: bank recapitalization, japan, capital injections, non-performing loans. 1. introduction in recent years, japanese taxpayers have contributed over 10 trillion yen (or 2 percent of japan’s nominal gdp) into the country’s troubled banking sector in the form of capital injections. more capital injections are likely to come in the near future. kozo yamamoto, a member of the ldp’s finance committee has gone on record stating that if the three-year limit on deferred tax assets, which triggered resona’s downfall, is applied to other banks, “there will be a flurry” of banks that will need public funds. about half the capital of japan’s big four city banks is made up of dtas. if this allowance were to be removed, all would drop below their capital adequacy requirement. this month, new legislation explicitly allowing “preventive” capital injections will become effective. the underlying question is whether or not capital injections are effective. panic and large-scale bank runs have been successfully avoided. however, fukao ijbf the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134114 (2003) predicted “the japanese government will have to nationalize most of the banking sector by 2005. capital injections will not solve the problems.” despite over 1 billion yen in capital injections to resona bank (which is japan’s 5th largest bank created through the merger of daiwa bank holdings, inc. and asahi bank), they had to receive another 2 trillion yen in may 2003, effectively nationalizing the bank. ashikaga bank (which is japan’s 10th largest regional bank) was nationalized in december of that same year despite having already received more than 135 billion yen in government capital. another 1 trillion yen was reportedly spent in “stabilizing funds” after nationalization. there is clear empirical evidence that capitalization affects the behavior of japanese banks. 1 ito and sasaki (2002) showed that banks with lower capital ratios tended to issue more subordinated debts in order to reduce overall lending. peek and rosengren (1997) found evidence of the international transmission of negative capital shocks to u.s. loan markets. in later research, peek and rosengren (2003) along with watanabe (2004), found evidence that negative shocks to bank capital result in a shift to lower quality loan portfolios. montgomery (2004) demonstrated that although both domestic and international banks are relatively insensitive to overall regulatory capital, international banks’ heavily risk-weighted assets are very sensitive to core, tier i capital. despite the large amounts of money that are at stake, there has been no quantitative research that specifically examines the effectiveness of capital injections. this study aims to inform the policy debate on how best to deal with systemic weakness in the banking sector. thus, this study holds important policy implications, not only for japan, but also for other countries facing similar banking sector problems. in the wake of the asian financial crisis of 1997, governments throughout the world have been allocating huge amounts of public funds to bail out the troubled financial sectors in those countries. based on the careful reading of the individual “business revitalization plan” submitted by several banks requesting government funds, we identified four primary goals of the capital injection plan: 1) to increase the bank capital ratios 2) to increase lending, in particular to small and medium enterprises, by avoiding a “credit crunch” 3) to increase write-offs of non-performing loans 4) to encourage bank restructuring. using a panel of individual bank data, we empirically estimated the effectiveness of the japanese government policy of public fund injection in achieving the first three of these stated goals. 1 in addition to the studies cited above, readers may see the following related papers: baba (1995), honda (2002), horiuchi and shimizu (1998), ueda (1993), and yoshikawa and ike (1994). 115the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134 section 2 will briefly outline the most recent history of japan’s banking sector crisis where section 3 will aim to catalogue policy response, focusing on the government’s decision to inject public funds into the banking sector. section 4 will explain the mechanisms by which capital injections may influence bank behaviour. lastly, section 5 will describe the data and empirical specification as well as the results of our regression analysis, ending with section 6’s closing remarks. 2. japan’s banking sector in crisis although warning signs had been evident much earlier2 , weakness in the japanese financial sector reached a crisis in 1997. the first to fall was sanyo securities, a mid-sized securities company that folded on november 3, 1997. the next failure was even larger. hokkaido takushoku bank failed on november 17, 19973 , costing the dic 3,411.3 billion yen, more than the 2,514 billion yen that had been used in all the dic’s previous history. one week later, on november 24, 1997, yamaichi securities, one of japan’s oldest and largest brokerage firms, followed suit. only 2 days later, tokuyo city bank, a small regional bank, collapsed on november 26, 1997. the following year, two other “top 20” banks were nationalized. the longterm credit bank of japan was placed under state stewardship on october 23, 1998. that very same year, nippon credit bank followed on december 13. the dic acquired all outstanding shares of both banks and provided credit to them to continue operating.4 all of these institutions had previously been considered “too big to close”. the three banks were ranked among the nations “top 20” banks, which dominated the lending markets. yamaichi was one of the top 4 securities houses, which had previously accounted for over 70% of the securities business. these failures marked the end of the traditional “convoy system” safety net. the news led to a sell off in bank shares on the stock market and the emergence of the so-called “japan premium” in overseas lending markets. policy response: 1998-1999: in february 1998, the diet responded to the financial crisis with the creation of the financial function stabilization plan.5 2 several small financial institutions failed between 1992 and 1997. by the end of 1997, sixteen (16) credit cooperatives, along with shinkin and regional banks, had failed. mergers had been arranged with healthier institutions under the so-called “convoy system”. then, in october 1995, it was reported that all 7 of japan’s jusen (home mortgage companies) were completely insolvent. under a controversial resolution scheme, these losses were also largely passed on to large financial institutions. 3 actual grant and asset purchase by the dic took place in october of 1998. 4actual grant and asset purchase by the dic took place in august and october of 1999. 5 this included two laws to stabilize the financial system, including the financial function stabilization law. the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134116 under this plan, 30 trillion yen was allocated to the deposit insurance corporation; 17 trillion yen for a special operations account which was to supplement the already established general account in order to provide full deposit protection of insolvent banks; and 13 trillion yen to a newly established financial crisis management account. the financial crisis management account funds went toward a capital injection of all 19 of the top city, trust, and long-term credit banks at the close of fiscal year 1997 (march 1998). in total, 1.8 trillion yen in capital was given out in the form of purchases of preferred stock, subordinated debt, or as a subordinated loan (see table 1 for details). many banks were reluctant to apply for the funds for fear of the message it would send to the market. as a result, all banks were pressured to apply for the same amount of capital. even tokyo-mitsubishi bank, which did not wish to apply for a capital injection, was pressured to apply for the standard 100 billion yen6 . regional banks yokohama bank, ashikaga bank and hokuriku bank applied for smaller capital injections. since subordinated debt, subordinated loans, and preferred stock count toward tier ii7 capital in calculating the capital to risk weighted asset ratio reported to the bank for international settlements, this helped many banks clear the 8% capital to risk weighted asset ratio required under the basel accord. even with the capital injection, many banks faced difficulty in meeting the 8% capital requirement, so authorities also implemented several other changes. accounting standards were relaxed in fiscal year 1997 in order to boost the banks’ reported capital.8 in april 1998 (the start of fiscal year for 1998), the “law to ensure the soundness of financial institutions”9 which is equivalent to the prompt corrective action (pca) measures in force in the united states since the late 6 tokyo-mitsubishi is the only bank to have repaid the public funds in full. 7 dated (more than 5 years) subordinated debt, subordinated loans, and preferred stock count as tier ii capital. perpetual preferred stock and subordinated debt count as tier i capital. in recent years, some short term (more than 2 years but less than 5 years) subordinated debt has been counted as tier iii capital. 8 banks were given the option of using either fair value accounting, the lower of book or market value method, or book value. this meant that banks did not necessarily have to report losses on securities held in their trading account on their balance sheets. secondly, banks were allowed to count 45% of revalued land holdings toward tier ii capital. although land prices were substantially lower than during the bubble period, most land held on the banks’ balance sheets still carried a market value much higher than the historical cost. 9 this law has two main components. first, the law defines a self-assessment process whereby banks value assets themselves according to well-defined guidelines. these findings are subject to review by external auditors and bank examiners. these new “risk asset” categories are broader than the nonperforming loan categories used by the japanese bankers association since 1993 (see appendix for details). secondly, the law specifies capital thresholds under which regulators can order banks to take remedial action ranging from a forced reduction in the number of branches to liquidation. 10 the financial revitalization plan included 4 new laws, two of which were the “financial function reconstruction law” and the “early strengthening law”. the “financial function reconstruction law” is officially called the law concerning emergency measures for the reconstruction of the function of the financial system. the “early strengthening law” is officially called the law concerning emergency measures for the early strengthening of the financial function. 117the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134 1980s, were introduced. in june 1998, the financial supervisory agency (fsa) was established. the fsa took over the supervision of banks, securities firms, and insurance companies from the ministry of finance. the fsa also took over the supervision of shinkin banks from the regional financial bureaus and the supervision of credit cooperatives from the prefectural governments. during fall 1998, the fsa conducted full-scale, on-site bank examinations of all the major banks. the agency concluded that the self-assessment of asset quality undertaken by the banks in march 1998 was based on overly optimistic assumptions and that both the major banks and regional banks had significantly understated their nonperforming loans. in response to these findings, in october 1998, after months of deliberation in a special “financial diet session”, the diet approved another plan to deal with the problems in the banking sector. the financial revitalization plan10 doubled the amount of funds allocated to dealing with the crisis to 60 trillion yen. the 60 trillion was earmarked for three separate accounts within the dic. the special operations account, which was to supplement the already established general account in order to provide full deposit protection of insolvent banks, received 17 trillion of the 60 trillion total. the financial reconstruction account, which received 18 trillion yen, was to be used for the purchase and collection of assets in liquidating or nationalizing insolvent banks. as reported in figure 21, about 1/ 3 of the funds in this account have been used to date. finally, the early strengthening account replaced the financial crisis management account. the early strengthening account, to be used in recapitalizing weak, but solvent banks, received the largest endowment of 25 trillion yen. the early strengthening account funds were used in a second round of capital injections in march of 1999 (the close of fiscal year 1998). fourteen of the largest city, long term credit and trust banks, as well as yokohama bank, received a total of 7.5 trillion yen ($75 billion), 4 times that of the first capital injection in march of 1998. as in 1998, the capital injection took the form of purchases by the dic of preferred shares or subordinated debts or loans issued by the banks. unlike the capital injection of 1998, the amount of the capital injection varied by bank and reflected the conditions of individual banks. to qualify for the capital injection, the frc required each bank to submit a restructuring plan, including raising new capital from the private sector, which would be subject to quarterly review11 . 11 the financial revitalization plan included 4 new laws, two of which were the “financial function reconstruction law” and the “early strengthening law”. the “financial function reconstruction law” is officially called the law concerning emergency measures for the reconstruction of the function of the financial system and the “early strengthening law” is officially called the law concerning emergency measures for the early strengthening of the financial function. if not satisfied, the fsa could convert its holdings of preferred stocks to common stocks after a certain grace period, the length of which is determined by the strength of the bank, and as largest shareholder could put pressure on management. the bank of tokyo mitsubishi did not apply for capital injection and instead made public its intention to pay off the subordinated debt issued to the government in march 1998. the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134118 in addition to the capital injections, regulations on consolidated balance sheet reporting were tightened in december 1998. prior to that time, banks were required to consolidate in their financial reporting only those subsidiaries and affiliates in which they had more than a 50 percent stake and a 20 percent stake respectively. under the new law, subsidiaries of banks and any company of which the bank group (a keiretsu with which the bank is associated) has more than a 40 percent stake has to consolidate in the financial reporting of the banks on a line by line basis. thus, affiliation was more determined by influence than just strictly share holdings. banks’ affiliates and any company of which a bank or bank group has more than a 15 percent stake and whose decisions are controlled by the bank were required to consolidate using the “equity method” (not line by line). consolidated reporting requirements applied to non-performing loan estimates as well. prior to december 1998 banks had been able to use “related companies” to clean up their balance sheets. banks could set up related companies that were neither subsidiaries nor affiliates and then transfer their non-performing loans to these related companies at the above market value. the banks had a meagre share of less than 5% in these related companies, along with positions in the firms inside the banks’ keiretsu with whom the banks also had interlocking shares. tightening of financial regulation continued in fiscal year 1999. in april 1999, the fsa extended the prompt corrective action (pca) framework to purely domestic banks without international operations. the fsa then conducted an inspection of regional banks in the fall of 1999. as a result of this inspection, the fsa recommended the merger of hanshin bank and midori bank into minato bank. four other regional banks namihaya, kofuku, kokumin, and tokyo sowa were found to be insolvent. they were to be operated by public administrators as a bridge bank until successor institutions were found. two other regional banks, niigata-chuo bank1 2 and hokkaido bank, were ordered to increase their capital in order to meet the 4% capital adequacy requirement. finally, mark-to-market accounting on all securities, including investment account securities, will be required as of fiscal year 2001. this will force banks to realize any hidden gains or losses on the equities held on their balance sheets, probably giving banks incentives to sell off unprofitable investments rather than keeping the shares as a way to cement business ties. to assist banks in this transition, the government is establishing an entity to purchase stocks held by the banks. the most recent trend in the japanese banking sector has been consolidation. the old “top 20” banks have been consolidated into 4 large groups: the mizuho financial group, sumitomo-mitsui financial group, tokyomitsubishi financial group, and ufj group. 12 niigata-chuo bank later failed. see appendix for details. 119the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134 the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134120 3. capital injections and bank behaviour although bank managers may care about capital ratios for various reasons, capital ratios affect bank behaviour primarily because of regulation. japanese banks face different regulations on capital depending upon whether they are internationally active or purely a domestic bank. international banks: international banks are required to maintain a capital to risk-weighted asset ratio of at least 8%. this ratio is required not only of japanese banks, but under the basle accord of 1998, of most internationally active banks worldwide. bis = tieri + tierii + tieriii – goodwil + rwa (1) table 2: summary statistics table 3: panel estimates (fixed effects model) – international banks, dependent *** represents statistical significance at 1%, ** at 5%, * at 10% level 121the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134 tier 1, or core capital, is basically shareholder’s equity with some adjustment for goodwill. tier 1 capital includes perpetual preferred stock. tier 2 capital in japan includes up to 45% of the latent capital gains banks hold on equities, 45% of any revaluation of land held on their booksiii, up to 1.25% of general loan loss provisions, dated preferred stock, and dated subordinated debt dated at more than 5 years. dated preferred stock and subordinated debt are counted at full value up to 50% of the value of tier 1 capital. tier 3 capital includes short-term (more than 2 years but less than 5 years) subordinated debt. risk weighted assets (rwa) are calculated as a weighted sum of assets held by the bank. “riskless” assets such as government bonds, receive a 0% weighting, while “risky” assets such as loans, corporate bonds, and equities receive a 100% weighting. mortgages and local government bonds fall in between. domestic banks: domestic banks in japan face a more lenient requirement of a 4% “mof” (ministry of finance) ratio. mof = tieri + tierii (mof) – goodwill + rwa(mof) (2) most components of the mof ratio are calculated similarly to the bis ratio for international banks, but tier 2 capital does not include latent capital gains on securities. short-term, subordinated debt is not counted as tier 3 capital. in order to meet the capital adequacy requirements discussed above, banks can either adjust risk-weighted assets (the denominator) or capital (the numerator). it is generally easier and less costly for banks to adjust their risk-weighted assets. thus, capital constrained banks may be inclined to substitute out of heavily weighted “risky” assets such as loans and equities into lower weighted assets such as government bonds. conversely, banks receiving a capital injection from the government may be inclined to shift their portfolios by shifting out of government bonds into loans. 4. empirical analysis this section presents an empirical analysis of the effectiveness of the capital injections made to banks in march of 1998 and 1999. data: a panel of data from 106 banks’ balance sheets and income statements for fiscal years 1990-1999 is used to estimate the effectiveness of the capital injections. regional and international banks are analyzed separately to preserve the homogeneity of the sample. several banks are excluded from the analysis due to nationalization, failure, or because they were established mid-sample (see 13 post-provision operating profits include deductions for other kinds of provisioning as well, such as write-down of sovereign risk loans or transfers to reserve for other credit losses, but these figures are much smaller than the sum of transfer to reserve for loan losses, loan charge-offs, and losses on sales of loans to the ccpc. the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134122 appendix 1 for details). mergers are accounted for by treating the merged bank as one entity for the entire sample period. loans (kashidashikin goukei) are the sum of domestic loans to all industries plus international loans and trust account loans as reported in the annual yukashoken hokokusho reports. we use three different measures of capital strength in the empirical analysis: 1. the official regulatory capital ratio (the mof ratio for domestic banks, bis ratio for internationally active banks: jikoshihon hiritsu) 2. the ratio of capital as reported on the balance sheet (shihon) to total assets (shisan) 3. the ratio of “core” tier i capital to total assets. non-performing loans are measured using the japanese bankers association definition (see appendix 2 for a detailed description of the various measures of non-performing assets), loans to bankrupt borrowers (hatansaki). this category is the strictest definition of non-performing loans and thus underestimates total non-performing assets, but we select it since it is the only measure of nonperforming loans that has been reported by all banks since reporting was required in 1993. the definition of this category of loans has not changed substantially as is the case for other classes of bad assets. written-off bad debt is defined as only direct write-offs (kashidashikin kyuufukin shoukyaku). direct write-offs include only non-performing assets that are completely removed from the balance sheet by either write-off or selling the loan at a loss to the ccpc (cooperative credit purchasing company). indirect write-offs, which also include transfers of funds to the allowance for loan losses on the banks’ balance sheets, are largely determined by the financial services agency’s (fsa) “self assessment of asset quality”, but are also to some extent dictated by past performance. our measure of profits is operating profits (keijyorieki). in order to cover bad debt write-offs, banks have taken large hits to their post-provision operating profits13 . all the major banks have reported negative post-provision operating profits every year since fiscal year 1994. they have been able to compensate for this to some extent by selling off equity holdings (realizing their latent gains on equities), allowing them to report positive pre-tax profits14 in some years. however, operating profits, which are calculated before provisions, write-offs, latent capital gains realizations, and taxes, are the best measure of a banks current performance. national and regional (prefectural) gdp are included to control for the 14 pre-tax profits are also adjusted for profits and losses on investments in trusts, write-downs of equity securities, and profits and losses on disposal of fixed assets, but these items are much smaller than the profits on sales of equity securities. 15 the regulatory capital ratio specification is estimated using arellano and bond’s (1991) generalized method of moments estimator since in some regressions we use lagged regulatory capital (the dependent variable) as a predetermined variable on the right hand side. . 123the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134 national and regional business cycle. with the exception of the mof and bis ratios, balance sheet data is reported at book value and on an unconsolidated basis. in fiscal year 1997, many large banks began reporting on a consolidated basis and all banks are now required to do so, but unconsolidated data is used in order to construct a continuous time series. gdp data is reported at current prices. data was compiled from the nikkei needs company data set, bureau van dijk electoronic publishing bankscope data set, and publicly available data on national gdp, prefectural level gdp, and the amount of capital injections. empirical specification: we investigated the effect of the capital injections on three dependent variables: growth in total lending, growth in lending to small and medium enterprises, and the regulatory capital ratio15 (bis or mof ratio) with the following specification (equation 1): (3) y i,t represents the dependent variable: growth in total lending, growth in lending to small and medium enterprises and the regulatory capital ratio in time t for bank i vector x i,t –1 denotes bank specific factors that influence loan growth such as gdp growth (regional gdp for the regional banks), change in operating profits to total assets, and change in bad loan write offs to total assets vector z i,t –1 comprises variables that indicate a bank’s capital position, including the amount of capital injection. as described above, we use three different measures of overall capital strength: capital to total assets, regulatory capital to risk-weighted assets and “core” tier i capital to total assets. to look specifically at the effect of the capital injections, we interact these measures of capital strength with a dummy for banks and years in which capital was injected. we also do a separate estimation including the actual amount of capital injected as a percentage of the banks’ total assets. 5. results international banks: ols estimation results for the panel of international banks are reported in tables 3-5. looking at table 3, we see that total lending is sensitive to the total and core capital measure, indicating that relatively well (poorly) capitalized banks tend to grow loans more quickly (slowly). however, the coefficient estimate on capital interacted with a dummy variable for capital 16 hoshi (2001) first documented the correlation between bank lending to the real estate sector during the bubble period and low bank capital during the banking crisis a decade later. watanabe (2004) applies this finding by using the shift in bank lending to the real estate sector as an instrumental variable for bank capital. we would like to thank professor hidehiko ichimura, university college london, for suggesting the interaction with changes in real estate prices. the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134124 injection is statistically speaking, significantly negative, indicating that loan growth is less sensitive to capitalization for those banks that received a capital injection. the overall coefficient is still positive, but less so. the actual amount of capital injected (columns 4-6 of table 3) does not seem to affect total loan growth. looking in particular at lending to small and medium enterprises (table 4), we see that loans to sme’s are again sensitive to total and core capital ratios, table 4: panel estimates (fixed effects model) – international banks, dependent variable = % change in sme lending table 5: panel estimates (fixed effects model) – international banks, dependent variable = bis ratio 125the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134 indicating that relatively well (poorly) capitalized banks tend to grow sme lending relatively quickly (slowly). the interaction term between capital ratios and a dummy for receipt of a capital injection is not statistically significant: receiving injected capital does not significantly reduce the sensitivity of sme loans to capitalization. there also does not appear to be any significant effect on sme lending of the actual amount of capital injected. regulatory capital on the other hand, is significantly influenced by the amount of capital injection (table 5). an injection of capital leads to a higher capital ratio in the following period. these ols results do not take into account the possible endogeneity of the amount of capital injected into each bank with the dependent variables (total loan growth, growth in lending to small and medium enterprises, and the regulatory capital ratio). to address this issue, we used the ratio of lending to the real estate industry to total domestic lending in 1990 – just before the burst of japan’s “bubble economy” – multiplied by the ratio of the decline in the regional land prices since that time as an instrument for the amount of capital injection. 16 the results of the estimates for international banks using instrumental variables are reported in tables 9-11. for lending, the results using the instrumental variables approach largely confirm our findings using ols. receipt of injected capital significantly reduced the sensitivity of total lending to capitalization, but neither total lending nor partial lending to the sme sector is sensitive to the actual amount of capital injected. however, the findings for the effect of the capital injections on regulatory capital are quite different using the instrumental variables approach. after controlling for possible endogeneity, we find no statistically significant impact of capital injections on regulatory capital in the following period. table 6: panel estimates (fixed effect model) – regional banks, dependent variable = % change in total lending the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134126 table 7: panel estimates (fixed effects model) – regional banks, dependent variable = % change in sme lending table 8: panel estimates (fixed effects model) – regional banks, dependent variable = bis ratio regional banks: ols estimation results for the panel of regional banks are reported in tables 6-8. in contrast to the international banks, neither total loan growth nor lending to sme’s by regional banks are at all affected by the injection of capital (table 6 and 7). regulatory capital is again significantly influenced by the amount of capital injection (table 8). however, as for the international banks, this result disappears once we control for possible endogeneity using instrumental variables (table 14). 127 table 9: iv estimates – international banks, dependent variable = % change in total lending table 10 : iv estimates – international banks, dependent variable = % change in sme lending the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134128 table 13: iv estimates – regional banks, dependent variable = % change in sme lending table 11: iv estimates – international banks, dependent variable = bis ratio table 12: iv estimates – regional banks, dependent variable = % change in total lending 129 table 14: iv estimates – regional banks, dependent variable = bis ratio *** represents statistical signifigance at 1%, ** at 5%, * at 10% level 6. conclusions our findings using ols on a panel of international and domestic banks suggest that capital injections are more effective for international banks than for domestic banks. the capital injections do not appear to effect lending to sme’s for either bank type, but for international banks, receipt of injected capital seems to relax the constraint that capitalization makes on overall loan growth. ols results suggest that the receipt of injected capital strengthens the capital position of both international and regional banks, but these results do not hold up once we control for possible endogeneity using an instrumental variables approach. acknowledgment: this article is based on one of three best prize winning papers selected by the editors of the journal at the fma-asian finance association 2005 17th conference, malaysia. we like to thank the editors for the suggestions for revisions which are incorporated in this published article. all errors are the responsibility of the author. author statement: heather montgomery is a staff of the asian development bank, makati, the philippines. this project was made possible by the bank providing data for this research. the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134130 references baba, n., (1995). kinnen no ginko kashidashi no teimei ni kansuru jissho bunseki (empirical analysis on low bank lendings in recent years). mimeo. institute of monetary and economic studies, bank of japan. fukao (2003) “financial sector profitability and double-gearing” in structural impediments to growth in japan blomstrom, magnus, jennifer corbett, fumio hayashi, and anil kashyap, editors. university of chicago press. honda, y., (2002). the effects of the basel accord on bank credit: the case of japan. applied economics 34: 1233–1239. horiuchi, a., shimizu, k., (1998). the deterioration of bank balance sheets in japan: risk taking and recapitalization. pacific basin finance journal 6: 1– 26. hoshi, t., (2001). what happened to japanese banks? bank of japan’s monetary and economic studies, february: 1-29. ito, t., sasaki, y., (2002). impacts of the basel capital standard on japanese banks’ behavior. journal of japanese international economies 16: 372–397. kroszner, r. and rajan, r., (1994). is the glass-steagall act justified? a study of the u.s. experience with universal banking before 1933. american economic review 844, 810-832. montgomery, h., (2004). the effect of the basel accord on bank portfolios in japan. journal of japanese international economies. peek, j., rosengren, e., (1997). the international transmission of financial shocks: the case of japan. american economic review. 87(no. 4): 495–505. peek, j., rosengren, e., (2003). unnatural selection: perverse incentives and the misallocation of credit in japan. nber working paper no. 9643. ueda, k., (1993). heisei fukyo wo koete (overcoming the heisei recession). seisaku koso forum report. tokyo. watanabe, w. (2004). do negative shocks to bank capital cause flight to quality? : evidence from the japanese financial crisis. unpublished mimeo, osaka university. yoshikawa, h., ike, t., (1994). chusho kigyo ni taisuru ginkou ni yoru kashishiburi ni tsuite keizai-bunseki: seisaku kenkyu no shiten (credit crunch for small and medium-size firms: policy analysis). working paper 1. economic planning agency, tokyo. 131the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134 appendix 1: banks included in the analysis regional banks fy1990 fy2002 aichi akita aomori ashikaga awa bank of fukuoka bank of ikeda bank of iwate bank of kansai bank of kinkibank of kyoto bank of nagoya bank of okinawa bank of saga bank of the ryukyus biwako chiba chiba kogyo chikuho chugoku chukyo daisan daishi daito ehime eighteenthbank fukui fukuoka chuo fukuoka city fukushima gifu gunma hachijuni higashi-nippon higo hiroshima bank of kinki data available through fy1999 hiroshima sogo hokkaido hokkoku hokuetsu hokuriku howa hyakugo hyakujushi iyo joyo juroku kagawa kagoshima kanto keiyo kita nippon kiyo kumamoto family kyushu michinoku mie minami-nippon miyazaki miyazaki-taiyo musashino nagano nanto nishi-nippon north pacific ogaki kyoritsu oita san-in godo sapporo senshu setouchi shiga shikoku shimizu shinwa shizuoka shokusan suruga taiko tochigi toho tohoku tokushima tokyo tomin tomato towa toyama yamagata yamaguchi yamanashi chuo. international banks fy1990 – fy1999 asahi bank of tokyo mitsubishi daiichi kangyo daiwa fuji industrial bank of japan mitsubishi trust mitsui trust nippon trust sakura sanwa sumitomo trust tokai toyo trust yasuda trust the following banks are excluded from the 18 established on april 1, 1957 as hypotec bank of japan, ltd. in 1970, its name changed to nippon fudosan bank and to nippon credit bank ltd in 1977. 19 in february 2001 the failed namihaya bank was absorbed by daiwa bank and kinki osaka bank. 20 the bank was established in 1926 as kofuku sogo bank ltd. in february 1, 1989, upon converting to a regional ii bank, its name changed to kofuku bank, ltd. after being closed, kofuku bank was purchased by the asia recovery fund and reopened on february 26, 2001 as kansai sawayaka bank, ltd. 21 yachiyo was originally established in 1924 as credit union. in 1991, its name changed to yachiyo bank. 22 minato bank was formed from the merger of midori bank and hanshin bank on april 1, 1999. 23 niigata chuo was eventually absorbed by six other regional banks: higashi-nippon, gunma bank, towa bank, taiko bank, daishi bank and hyakujushi bank. 24 sendai bank was originally established on july 5, 1951. 25 tokyo sogo bank converted to a regional ii bank in 1989 and its name changed to tokyo sowa bank. 26 the merged bank was first named the kyowa saitama bank. the bank name was changed to asahi bank on september 21, 1992. the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134132 analysis because they were closed or nationalized during the sample period between fy1982-fy1999. hokkaido takushoku bank failed on october 17, 1997. the nippon credit bank18 was nationalized on december 13, 1998 and in january 2001 reopened as aozora bank. the long term credit bank of japan was nationalized on october 23, 1998 and reopened on june 5, 2000 as shinsei bank. namihaya bank, which was formed on october 1, 1998 from the merger of bank of naniwa and fukutoku bank, was closed on august 6, 199919 hanwa bank failed in november 1996. kofuku bank , which absorbed kyoto kyoei bank on october 26, 1998, closed on may 21, 1999. kokumin bank20 failed on april 11, 1999, and was absorbed by yachiyo bank21 on august 14, 2000. minato bank22 was established toward the end of the sample on september 1, 1999. niigatachuo bank23 failed in october 1999. tokuyo city bank failed on november 26, 1997, and was absorbed by sendai bank 24 on november 24, 1998. tokyo sowa bank, established in 1950 as a “sogo” bank25, was closed on june 11, 1999. in addition, hokuto bank is excluded because it was established mid-sample in 1993. the bank was first established in 1895 as masuda bank and changed its name to hokuto bank after absorbing akita akebono bank on april 1, 1993. data for asahi bank,26 which was formed by the merger of kyowa bank and saitama bank on april 1, 1991, is backdated by combining data from the two bank balance sheets. the same technique is used to backdate data for tokyo-mitsubishi bank, which was formed by the merger of bank of tokyo bank and mistubishi bank on april 1, 1996 and for sakura bank, which was established on april 1990 through the merger of mitsui bank ltd. and taiyo kobe bank ltd. in order to include sendai bank in the sample, the last fiscal year of data is not used. several other banks in the sample dai-ichi kangyo bank, joyo bank, minami-nippon bank, and tokai bank were also formed as the result of mergers, although they occurred before the sample period, so no data adjustment was necessary. dai-ichi kangyo was established on october 1, 1971 (pre-sample) through the merger of the dai-ichi bank ltd. and the nippon kangyo bank ltd. joyo bank ltd. was established on july 30, 1935 following the merger of tokiwa bank and goyu bank. tokai bank was established in 1941 as the result of the merger between aichi bank, nagoya bank and ito bank. minami-nippon bank, a domestic bank, was established in 1943 through the merger of 2 local banking institutes in kagoshima prefecture. recent mergers forming the mizuho financial group (a holding company of dai-ichi kangyo bank, fuji bank and industrial bank of japan), the sumitomo mitsui banking corporation (formed by the merger of sumitomo bank and sakura bank in april 2001), chuo mitsui trust bank (formed by the merger of chuo trust and mitsui trust in april 2000), and the mitsubishi tokyo financial group (formed in april 2001 by tokyo mitsubishi bank and mitsubishi trust; soon to be joined by nippon trust and toyo trust) occurred post-sample and thus do not require any data adjustments. 133the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152 appendix 2: japanese bankers association risk management loans risk management loans (as defined by the japanese bankers association) have been reported by individual banks since fiscal year 1993. the categories used by the jba include: a. restructured loans: the least dangerous category of loans, restructured loans were originally defined only as loans for which the interest rate had been lowered, but in fiscal year 1997, the definition was expanded to include loans for which any contract condition has been amended and loans to corporations undergoing reorganization. beginning in 1994, restructured loans were reported by the largest banks. reporting requirements were extended to regional banks in 1996. b. 3-month overdue loans: 3-month, overdue loans, or 3pdl (3-month, past due loans), are defined as loans in arrears by more than 3 months, but less than 6 months. overdue loans originally included loans on which interest payments were past due by more than six months (see below); but, in fiscal year 1997, this stricter category of overdue loans (loans on which interest payments are past-due by more than 90 days) was added. reporting of 3-month, overdue loans has been required of all banks since fiscal year 1998. c. overdue loans: overdue loans, or pdl (past due loans), are loans on which interest payments are past due by more than 6 months. since 1993, the largest banks have reported these loans, but the regional banks had only begun to report this category of loans in 1996. d. defaulted loans: defaulted loans, or lbb (loans to borrowers in legal bankruptcy), carry the strictest definition. these are loans to companies in the legal process of declaring bankruptcy. all the banks (city, trust, long-term credit, and regional) since fiscal year 1993 report this category of loans. in march 1999, the definition of defaulted loans was expanded to include loans for which principal or interest payments have not been made in a “long time”, for which repayment is judged “unlikely”. end notes the financial revitalization plan also legislated the merger of the resolution and collection bank (rcb), which had been overseeing the liquidation of the assets of failed credit cooperatives, with the housing loan administration corporation (hlac), which had been dealing with the remaining bad assets of the jusens into the resolution and collection corporation (rcc), which is a public asset management company. details on this merger are available in figure 33. the rcc the international journal of banking and finance, vol. 5.(number 1): 2007: 113-134134 had an expanded mandate allowing it to purchase bad loans from solvent financial institutions as well as failed banks. under the financial revitalization plan, the financial revitalization commission (frc) was established under the prime minister’s office to oversee bank restructuring. restructuring followed one of two courses: the bridge bank scheme or nationalization. under the bridge bank scheme, the troubled bank was to be operated by public administrators as a bridge bank until a successor bank could be found. this system was used in closing smaller banks such as the koufuku bank, kokumin bank, tokyo sowa bank, and namihaya bank, all of which went into bankruptcy in 1999. the nationalization scheme was used for larger bank failures, such as the long term credit bank of japan and nippon credit. in these cases, the banks were temporarily nationalized, with the dic purchasing all shares, operating the bank until it was ready to resume business as a new bank. the frc was due to complete its task by 2001. it closed on january 6, 2001. 2dkb, fuji bank, ibj, and yasuda trust formed a holding company in september 2000. they later merged into the mizuho financial group in january 2003. with assets of 14 billion yen, it is now the world’s largest financial institution. the sumitomo-mitsui financial group, formed by the merger of sumitomo bank and sakura bank in april 2001, is now the world’s second largest bank in terms of assets. this merger is of particular interest because the two main banks represent different keiretsu groups (the sumitomo and mitsui keiretsu). the world’s third largest financial institution is also a japanese institution. tokyo-mitsubishi bank established the mitsubishi-tokyo financial group (mtfg) in april 2001, along with the merger of three trust banks (mitsubishi, nippon, and tokyo trust) in october that very same year. the ufj group formed a holding company in april 2001 that merged sanwa, tokai, and toyo trust in april 2002. daiwa and asahi had also merged into resona bank in march of 2003, but resona has since then failed.iii until fiscal year 2001 when accounting standards were strengthened, land holdings were recorded at book value on the balance sheet and banks were able to choose between the book value or fair value (lower of book or market) method for recording the value investment account equity holdings on their balance sheets. banks that chose the fair value method could count 45% of unrealised gains toward tier 2 capital. international journal of banking and finance 1-1-2007 the effectiveness of bank recapitalization in japan heather montgomery recommended citation financial instability, uncertainty and bank lending behavior 74 the international journal of banking and finance, volume 9 (number 4) 2012: pages 74-95 financial instability, uncertainty and bank lending behavior vigneshwara swamy and s. sreejesh indian school of business-hyderabad, india ____________________________________________________________ abstract “why do banks squeeze their lending activity” is an oft-repeated question during the times of financial crisis. this study examines an emerging economy’s banking system, and contributes to the evolving body of literature on the topic by providing answers to what causes the sluggish bank credit during times of recession. by employing cointegration technique, the study shows that bank credit has a significant positive relationship with the borrowing activities of debt users of the banks, hence, as the contrary an inverse relationship with investment activity is evident during financial crisis. accordingly, we suggest that banks could increase their lending by increasing the borrowings rapidly either from the central banks or from government supported long term lending institutions during recessionary periods. key words: time series models, financial markets, interest rates, bank lending, financial crisis, credit declines jel classification: c22, d53, e43, e51, g21 _____________________________________________ 1. introduction the 2007-8 global financial crisis also termed as ‘the great recession’ led to a grave banking panic and threw most of the economies of the world into severe recession. that crisis is attributed as having been caused by several factors. but only few of the factors are associated with the housing and credit markets, the two commonly-cited factors. insinuated causes include the insolvency of homeowners to meet their mounting mortgage payments, unusually high levels of personal and corporate debt levels, foolhardy financial product innovation, the collapse of vital financial institutions and awful errors of judgment by credit rating agencies in the rating of structured products. macroeconomic factors have also been added: monetary 75 policy, excessive liberalisation, global trade imbalances and ineffective government regulation. without refuting the impairing effects of the above factors, we believe that the origins of the global financial crisis lie in the rational profit-seeking behavior of banks. indeed, we consider that reason in the course of economic expansion is a vital cause behind the procyclical financial infirmity of any economic system given bank’s heightened sensitivity to a financial shock. the global financial crisis has spread across the world through a range of financial as well as real economy channels. cross-border bank lending has also been one of the major financial channels through which stresses in the global financial system are and were spread to other emerging economies. while for some economies in europe, the transmission conduit for the losses on account of toxic assets was obvious and thereby increased the fragility of bank balance sheets. however, for some asian economies such as that of india, the story was a bit different as the exposure of their domestic banks was lot less limited to such international toxic assets. yet even these economies with little exposure did experience a substantial slow down. hence, toxic assets alone cannot explain the virility of the effect. the connection of a recession to banking crisis is an intensely problematic research issue. the consequent fall in bank lending experienced in both the exposed or little exposed economies has key implications for growth. however, the sag in credit supply puts upward pressure on interest rate spreads, and thereby leads to an inordinate fall in lending than one might see in a typical recession. what makes the banks squeeze their lending activity during the crisis? it is an oft-repeated question, which deserves re-examination to find credible economic reason(s) applicable across diverse banking systems across the globe. the available literature, though throws up some interesting insights. we attempt to find some more answers from a rather relatively conservative banking system like that of an emerging economy in india. indian banking similar to much of asian banking is characterized by sizeable proportion of ownership under government control, and is distinct from the free market and widely-held banks of the anglo-saxon model. this paper is one among the relatively small family of works that underline the relationship between bank behavior and financial instability. it aims to analyze how the financial crisis passed through the banks’ lending behavior. our endeavor in this study is to international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 76 understand the impact of financial uncertainty and instability on banks’ behavior on credits more specifically the effect of the crisis on the lending behaviour of the banks. we begin by presenting in section 2, the theoretical framework illustrating the recent approaches on financial instability placing the banking system at the epicenter of analysis. we analyze more specifically how financial instability affects the bank’s lending behavior during times of economic shocks or crisis. the methodology involving the data and its sources and research design explaining the empirical framework including how to estimate the impact of factors as is done in section 3. the results of the analyses of the findings are presented in section 4. the conclusion and policy implications are offered in section 5. 2. financial instability and bank lending behavior 2.1 theoretical framework global financial system saw an extraordinarily high economic growth with abundant liquidity from 2001 to 2007. it is established empirically that the origins of the global financial crisis can be traced to the low interest rate policies adopted by the federal reserve and other central banks after the burst of the dot com bubble in october, 1999. an exhaustive summary of the events preceding and accompanying the global financial crisis is offered in allen and carletti (2010), brunnermeir (2009), greenlaw et al. (2008) and taylor (2008). the sudden outbreak of financial crisis in september 2008 (the worst month) affected banks across the globe with uncertainty and resulted in instability in the markets for several years. banks reacted to the crisis by reducing their lending exposure (ivanshina and scharfstein, 2008). this behavior is contrary because banks would be hesitant to cut their lending too drastically in view of their enduring relationships with the borrowers. if such relationships terminated suddenly, it would prove to be too costly in furthering their business in future (rajan, 1992; ongena, 1999). further, ralph de haas and neeltje van horen (2009) have found that, during a financial crisis, arranging banks retain large portions of loans and forge more concentrated syndicates, signifying an heightened need to screen and monitor borrowers. few economists building their analysis on a variety of analytical tools and hypotheses have scrutinized the association between bank behavior and financial uncertainty. one school of thought emphasizes the crucial role of uncertainty and confidence in the emergence of an endogenous mechanism of financial instability impelled by the dynamics of asset prices and 77 banking behavior. the most archetypal and leading work adopting this approach was the “financial instability hypothesis” of hyman p. minsky which, basing on the financial concept of economic fluctuations states that economy’s inherent tendency to transform itself into an unstable financially fragile system is dependent on the naive interplay of the profit-seeking behavior of economic agents. in the minskyan model of financial instability, the profit– seeking behavior of banks in an uncertain decision-making environment leads them to such financial practices that stem out to a situation of escalating financial fragility. of late, macroeconomic models grounded on asymmetric information also single out the role played by banks at the core of the assessment of financial instability and contend that the crux of financial crises is in the vulnerabilities of the banking sector. mishkin (1999a and 1999b) argues that increase in information asymmetry spawns ex ante a cumulative risk of adverse selection and produces ex-post a proliferation in moral hazard, which is coped by limiting credits by the financial intermediaries. as asymmetries of information are ubiquitous in financial markets, any crisis that escalates the asymmetries of information like impairing of banking or non-banking intermediaries’ balance sheets, escalating interest rates, fall in asset prices and compounding uncertainty would and must cause in a curtailment of credits. another school of thought emphasizes that the balance-sheet exposures (allen et al., 2002) like maturity mismatches, currency mismatches, capital structure problems and solvency issues could contribute to a currency and banking crisis. an empirical study by calomiris and wilson (2004) look into the causal link from the escalating credit risk by way of decreased bank funding to lower credit supply. demirgüçkunt, detragiache and gupta (2006) notice a similar kind of evidence for an extensive country sample. besides limiting the risk of their asset portfolio, the authors observe that banks quite often build up their capital buffer to insulate depositors from credit risk. because of such ‘deleveraging’, bank lending tends to moderate considerably during the times of financial crises (de haas and van lelyveld, 2006). 2.2 financial soundness one of the important sources of vulnerability that can lead to a financial crisis can be the weakness (such as a high level of short-term debt) in the financial structure of the economy i.e., the composition and the size of the assets and liabilities on the balance sheet. a financial crisis follows when the demand for financial assets of one or more sectors plummet and consequently the banking system fails to meet the outflows or may be unable to attract new international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 78 financing or roll over existing short-term liabilities. in this direction, financial soundness (table-1) matters much during the financial crisis because it gives some indication of how likely it is that financial problems would be transmitted into the real economy (by, for example) a reduction in the supply of loans. table 1: core financial soundness indicators of selected countries (percent) australia france uk usa russia china india brazil south africa capital adequacy ratio [car] 2005 10.2 11.3 12.8 12.9 16.0 2.5 12.8 17.9 12.3 2006 10.3 10.9 12.9 13.0 14.9 4.9 12.3 18.9 12.3 2007 10.1 10.2 12.6 12.8 15.5 8.4 12.3 18.7 12.8 2008 11.3 10.5 12.9 12.8 16.8 12.0 13.0 18.2 13.0 2009 11.9 12.4 14.8 14.3 20.9 11.4 13.2 18.8 14.1 2010 11.4 12.3 15.9 15.3 18.1 12.2 13.6 17.8 14.9 non-performing assets [npa] 2005 0.6 3.5 1.0 0.7 2.6 8.6 5.2 3.5 1.8 2006 0.6 3.0 0.9 0.8 2.4 7.1 3.3 3.5 1.1 2007 0.6 2.7 0.9 1.4 2.5 6.2 2.5 3.0 1.4 2008 1.3 2.8 1.6 3.0 3.8 2.4 2.3 3.1 3.9 2009 2.0 3.6 3.5 5.4 9.5 1.6 2.3 4.2 5.9 2010 2.2 4.2 4.0 4.9 8.2 1.1 2.4 3.1 5.8 provisions to npa s 2005 17.6 … 54.0 154.8 176.9 24.8 60.3 179.7 59.4 2006 17.6 … 54.6 134.8 170.8 34.3 58.9 179.9 54.5 2007 18.3 … … 91.7 144.0 39.2 56.1 181.9 44.9 2008 21.9 70.0 38.1 74.4 118.4 116.4 52.6 189.0 31.4 2009 22.6 63.2 41.1 57.7 95.8 155.0 52.1 156.7 29.6 2010 22.0 62.3 35.4 64.2 103.7 218.3 51.5 171.1 32.6 return on assets [roa] 2005 1.8 0.6 0.8 1.8 3.2 0.6 0.9 3.0 1.2 2006 1.7 0.6 0.5 1.8 3.3 0.9 0.9 2.7 1.4 2007 1.6 0.4 0.4 1.2 3.0 0.9 0.9 2.9 1.4 2008 0.9 0.0 -0.4 -0.1 1.8 1.0 1.0 1.4 2.1 2009 1.0 0.4 0.1 -0.1 0.7 0.9 1.1 1.9 0.9 2010 1.2 0.6 0.2 0.9 1.9 1.0 1.1 2.1 1.0 return on equity [roe] 2005 25.6 11.8 11.8 17.8 24.2 15.1 13.3 29.8 15.2 2006 27.8 14.0 8.9 17.2 26.3 14.9 12.7 27.6 18.3 2007 30.2 9.8 6.2 11.2 22.7 16.7 13.2 28.9 18.1 2008 18.9 -1.0 -10.3 -1.6 13.3 17.1 12.5 14.9 28.7 2009 17.4 8.2 2.6 -0.6 4.9 16.2 13.1 20.4 15.8 79 2010 20.5 13.3 3.9 8.2 12.5 17.5 12.5 21.7 14.7 data source: imf – all countries fsi data. 2.3 trends in bank credit bank credit has shown a robust upward trend in most emerging economies until around 2007. one of the focal issues is the extent to which there was a decline in credit growth during the recession period and how it affected credit supply or demand. a decline in credit supply would imply that the impact of the crisis on the financial sector has swelled the effects of the very large cyclical downturn. on the contrary, if demand effects were cogent, this would indicate that efforts to supply financing and support the operation of the financial sector have been successful in boosting credit supply and mitigating the adverse effects of the crisis in developed financial markets. growth in domestic credit in select economies particularly during the crisis period has nosedived in almost all the economies though the extent may vary (figure 1). however, it is interesting that the experiences of brics countries are quite different form that of other developed countries in view of the nature of their economies. figure 1: domestic credits of select economies over 2006-10 (consolidated balance sheet of the banking sector) source: imf dataset: principal global indicators. it is but natural to have our curiosity to understand the lending behaviour of the banks in such emerging economies, which showed some sense of resilience to financial crisis in view 6 international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 http://epublications.bond.edu.au/ijbf/vol9/iss4/6 80 of their strong domestic demand, led growth. we examine the trend of bank credit in india, which is representative of a rather domestic focused banking sector when compared to other developed markets. the trend of bank credit in india even during the crisis period has experienced an upward trajectory despite huge constraining factors (figure 2). figure 2: trend of bank credits in india over 2007-11 source: figure developed based on the data from reserve bank of india publications banking sector being an integral part of the economy in ensuring the efficient transmission of the funds, it has a close relationship with the other macro-economic factors that play a vital part in the economic development. despite the downward movement of some of the economic indicators like the imports and exports, the bank credit has continued to show rising trend in view of the strong domestic demand led growth (figure 3). figure 3: indian economic indicators over 2007-10 data source: international financial statistics (ifs) of imf. further, the core financial selector indicators for india like; capital adequacy ratio (car), capital adequacy ratio–tier-1, gross non-performing assets (gnpas) to total loans, net non-performing assets (nnpas) to total loans and return on equity (roe) have 81 experienced downward pressure during the recession period (figure-4). on the contrary, liquid assets to total assets ratio has moved upwards indicating the tendency of the banks to hold cash during the times of recession instead of investing in loans or investment products. figure 4: core financial sector indicators for india over 2008-10 data source: international financial statistics (ifs) of imf interest rates (benchmark prime lending rate), money market rate and the discount rates) which have significant impact on the lending activity showed downward movement in the indian banking scenario (figure 5). figure-5: interest rates in india over 2007-10 data source: international financial statistics (ifs) of imf of course, banks claim that sluggish bank lending was due to a fall in demand but they have tightened the terms of credit on which borrowers can access funds. however, given the backdrop of the above discussed understanding of the behavior of the banking sector during the recession period, we try to find answer for our specific question, how was the lending behavior of the banks during the recession and which was the strongly correlated determinant for bank credits. 8 international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 http://epublications.bond.edu.au/ijbf/vol9/iss4/6 82 3. empirical framework and estimation we use data on bank behavior in india to provide fresh insights into how banks respond to financial crises especially in the area of credit supply. the framework of our empirical analysis is based on a robust database and well-established techniques. 3.1 data and the key variables the weekly data on commercial banks in india for the study period has been sourced from the robust database of reserve bank of india (various issues of statistical tables relating to banks in india and report of trend and progress of banking in india). table 2: definitions for key variables in our models variable description bank credit lnbc logarithm of bank credit (total of outstanding credit for all the scheduled commercial banks in india) aggregate deposits lnad logarithm of aggregate deposits (total of outstanding aggregate deposits held by all the scheduled commercial banks in india) investments lninvest logarithm of investments (total of outstanding investments by all the scheduled commercial banks in india) money at call and short notice lnmatcal logarithm of money at call and short notice (total of all money at call and short notice held by all the scheduled commercial banks in india) borrowings lnborrow logarithm of borrowings (total of all outstanding borrowings by all the scheduled commercial banks in india) bank nifty lnbnifty logarithm of bank nifty index of nse which is considered to be most representative index for understanding the market performance of banks in india lending rates lnbplr logarithm of bplr (benchmark prime lending rates) rates which represent the average levels of bank lending rates in india cash-deposit ratio cashdr ratio of cash held by banks to their aggregate deposits investmentdeposit ratio idr ratio of outstanding investments to aggregated deposits of banks credit-deposit ratio cdr ratio of outstanding credit to aggregate deposit levels of banks 83 the choice of the period is made in line with the need of the study to compare and analyze the impact of financial crisis on the bank lending activity. accordingly, we have set three distinct comparable time horizons keeping in mind the availability of the weekly data. phase-1 covers the period from dec 2006 to july 2008 to represent the pre-recession (boom/normal) period, phase-2 covers the period from august 2008 to march 2010 to represent the recession period and finally the phase-3 includes the period from april 2010 to march 2011 to capture the recovery period. further, the variables used for the analysis include important determinants of bank behavior and are detailed in table 2. credit to deposit ratio (cdr) is one of the indicators that is used to measure / reflect the bank/s’ efficiency in credit delivery. a careful look at its trend gives a broader trend of the banks’ credit supply activity. though this ratio involves the outstanding levels, it captures the inclusive picture of the credit delivery activity. figure 6 captures the volatile movement of cdr during the study period. during the recession period, the cdr has experienced an uncertain downward tumble and has found an upward pitch during the subsequent recovery period. figure 6: trend of credit to deposit ratio over 2007-11 source: reserve bank of india database. we look at the trend of investment to deposit ratio (idr) during the study period. figure-7 presents the interesting movement of idr wherein we notice that the movement is almost opposite to that of cdr during the recesssion period which indicates that banks were hesitant towards credit delivery and turned towards safe investment of their avaailable funds instead of the riskier lending activity. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 84 figure 7: trend of investment to deposit ratio in india over 2006-11 source: reserve bank of india database we take a look at the liquidity management of the banks during the study period and find that cash to deposit ratio which indicates the cash holding levels of the banks was sliding down during the recession period (figure 8). this denotes the tight funds mangement scenario during the recession period due to the impact of the crisis. figure 8: cash to deposit ratio in india over 2006-11 source: reserve bank of india database 3.2 the model bank credit has direct relationship with the predictor variables such as aggregate deposits, borrowings, investments, money at call and short notice, bank nifty and lending rates. accordingly, we frame the following quadratic equation. yt = α + β1x1t + …………… + βnxnt + µ (1) accordingly, bank credit can be better explained and estimated with the following version of equation. 85 bc = ƒ [ad, invest, matcal, borrow, bnifty, bplr] + µ (2) due to potential nonlinearities, the natural logarithms of the regressors are considered accordingly, when we log-transform this model we obtain: lnbc = α + lnad + lninvest + lnmatcal + lnborrow + lnbnifty + +lnbplr + µ (3) we expect the borrowings to have a close positive relationship with the bank credit during the recession in view of the tight liquidity and decreasing deposits mobilization scenario. further, we also predict that as the banks are guided by their profit seeking behaviour they tend to curtail their lending activity and try to invest their funds in the assured investments instead of venturing into the risky activity of lending. 3.3 methodology given that we are dealing with time series data, the possibility of non-stationarity of the variables cannot be ruled out. we perform stationarity test on the variables that are included in our analysis to ensure that the results from the analysis are not spurious. for this purpose, augmented dickey fuller (hereafter, adf) test and phillips and perron (hereafter, pp) (1988) tests are conducted to know the stationarity of the variables. (4) ( ) (5) ) further, (6) yt is a random walk with a drift with linear time trend if γ=0. the adf test mentioned above assumes that the errors are statistically independent and have a constant variance. in case of pp test, the assumption is relaxed, allows the error disturbances to be weakly dependent, and heterogeneously distributed. this can be written as: international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 86 (7) in both the tests the null hypothesis is that the series is non-stationary (possess a unit root) and if the calculated value exceeds the critical value (based on mackinnon, 1996 for adf and pp test), the null hypothesis may be rejected implying the stationary characteristics of the data series. the adf test is a parametric auto regression to arima structure of the errors in the test regression, but the pp test corrects for serial correlation and heteroscedasticity in the errors. in adf test, schwarz information criteria (sic) have been used to select the appropriate lag length, whereas in pp test we have used the newey-west using bartlet kernel method. we use a cointegration framework to identify systematic interaction effects between the identified determinants of lending behaviour of banks. accordingly, johansen's cointegration technique was employed to verify the existence of cointegration between the determinants of bank credit and other determinants as mentioned above. once the order of integration of each variable is determined in three periods, the concept of cointegration by johansen and juselius (1990) method (hereafter jj method)1 is used to examine the existence of cointegrating relationship between the determinants. this method is considered to be more robust than the engel granger procedure (based the residual). therefore, we prefer the jj method, which uses the vector auto regressive (var) model to test the number of cointegrating vectors, and the estimation is based on maximum likelihood (ml) method. following johansen (1988), johansen, and juselius (1990) var representation of column vector xt can be written as follows: tit k i itt xbzx ε+π+= − = ∑ )( 1 )( (8) where xt is column vector of n endogenous variables, z is a (n×1) vector of deterministic variables, ε is a (n × 1) vector of white noise error terms, and πi is a (n×n) matrix of coefficients. since, most of the macroeconomic time series variables are non-stationary, var of such models are generally estimated in first-difference forms. jj test provides two likelihood ratio (lr) test statistics for cointegration analysis, the trace (λtrace) statistics, and the maximum eigenvalue (λmax) statistics. the trace statistics tests 1 see philips(1991),cheung and lai(1993) and gonzala (1994). 87 the null hypothesis that the number of cointegrating relations is r against k cointegration relations, where k is the number of endogenous variables. the maximum eigenvalue test, tests the null hypothesis that there are r-cointegrating vectors against an alternative of r+1 cointegrating vectors. to determine the rank of matrix π, the test values obtained from the two test statistics are compared with the critical value from mackinnon-haug-michelis (1999). for both tests if the test statistic value is greater than the critical value, the null hypothesis of r cointegrating vectors is rejected in favor of the corresponding alternative hypothesis. 4. analysis of results tables 3, 4 and 5 present the correlations statistics of the determinants employed in the analysis for the pre-recession period, recession period and recovery period. during the prerecession period lnbc has strong positive correlations with; lnad (0.983), lninvest (0.946), lnborrow (0.811), lnbplr (0.433) at 0.01% level of significance (table-3). table 3: correlations statistics (pre-recession period) lnad lnborrow lnmat cal lninv est lnbni fty lnbp lr lnbc lnad 1 lnborrow .753** 1 lnmatcal -.256* .113 1 lninvest .977** .728** -.295** 1 lnbnifty .319** .041 -.328** .421** 1 lnbplr .514** .184 -.284** .473** .372** 1 lnbc .983** .811** -.162 .946** .215* .433** 1 **correlation is significant at the 0.01 level (2-tailed). *correlation is significant at the 0.05 level (2-tailed). table 4: correlations statistics (recession period) lnad lnborrow lnmat cal lninv est lnbpl r lnbni fty lnbc lnad 1 lnborrow -.214* 1 lnmatcal -.723** .461** 1 lninvest .980** -.241* -.725** 1 lnbplr -.863** .138 .609** -.899** 1 lnbnifty .788** -.406** -.775** .764** -.573** 1 lnbc .964** -.135 -.659** .916** -.797** .698** 1 **correlation is significant at the 0.01 level (2-tailed). *correlation is significant at the 0.05 level (2-tailed). international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 88 during the recession period lnbc has strong positive correlations with; lnad (0.964), lninvest (0.916), lnbnifty (0.698) at 0.01% level of significance and has negative correlation with; lnmatcal (-0.659) and lnbplr (-0.797) at 0.01% level of significance (table 4). during the recovery period lnbc has strong positive correlations with; lnad (0.518), lninvest (0.728), lnbplr (0.422), lnbnifty (0.426) at 0.01% level of significance and has negative correlation (table 5). table 5: correlation among variables (recovery period) lnad lnborrow lnmat cal lninv est lnbpl r lnbni fty lnbc lnad 1 lnborrow .275 1 lnmatcal .363** .179 1 lninvest .737** .357* -.004 1 lnbplr .970** .299* .378** .645** 1 lnbnifty .928** .356* .383** .621** .942** 1 lnbc .518** .292* .255 .728** .422** .426** 1 notes: **correlation is significant at the 0.01 level (2-tailed). *correlation is significant at the 0.05 level (2tailed). tables 6, 7 and 8 show the augmented dickey–fuller (adf 1979) phillip perron (pp 1988) unit-root test results. in both the tests, the null hypothesis is that the series is nonstationary (possess a unit root) and if the calculated value exceeds the critical value (based on mackinnon, 1996 for adf and pp test), the null hypothesis may be rejected implying the stationary characteristics of the data series. table-6: stationarity test results (pre-recession period) variables at level form at first difference form adf statistic pp statistic adf statistic pp statistic lnbc -0.428456 -0.431008 -8.908858* -85.32742* lnad -0.879530 -1.457736 -10.29290* -14.75445* lnborrow -0.014152 -0.084157 -8.8621298* -79.40365* lnmatcal -0.633163 -0.557301 -10.14116* -18.52506* lninvest -0.086942 -0.086942 -8.833085* -84.55829* lnbnifty -1.583966 -1.627241 -10.06231* -10.02447* lnbplr -1.585209 -1.628026 -10.06809* -10.02983* note: *significant at 0.01 level. the results revealed that all the level (first-differenced) variables are insignificant (significant) at the 0.01 percent level, indicating that all the variables during these three periods are integrated at the first degree and satisfied the condition for the cointegration test. 89 johansen (1988) and johansen and juselius (1990) developed this methodology (cointegration) for assessing long-run relationships. the methodology applies maximum likelihood procedure to determine the presence of cointegrating vectors in a set of nonstationary time series. table 7: stationarity test results (recession period) variables at level form at first difference form adf statistic pp statistic adf statistic pp statistic lnbc -0.751725 -0.609777 -12.23925* -12.09305* lnad -1.477376 -1.379730 -13.04749* -15.385588 lnborrow -0.390756 -0.390756 -9.981443* -28.56700* lnmatcal -1.232399 -1.341713 -10.120958 -23.34173* lninvest -1.478795 -1.524909 -8.854521* -12.67635* lnbnifty -0.692584 -0.647832 -9.883277* -9.883277* lnbplr -1.346230 -1.239941 -9.814503* -9.818384* note: *significant at 0.01 per cent level. table 8: stationarity test results (recovery period) variables at level form at first difference form adf statistic pp statistic adf statistic pp statistic lnbc -0.358440 -0.206328 -9.194190* -9.422216* lnad -0.556997 -0.185782 -7.233743* -10.03406* lnborrow -0.977743 -0.977743 -7.495562* -11.73249* lnmatcal -0.602755 -0.638488 -6.936360* -9.765911* lninvest -2.854436 -2.672870 -8.378243* -9.334602* lnbnifty -1.467749 -1.365785 -8.533983* -8.471221* lnbplr 0.324520 0.116450 -9.466734* -9.472437* note: *significant at 0.01 per cent level. as the number of lag orders selected can affect the number of cointegration, the appropriate lags are carefully selected with a number of multivariate diagnostic tests. the appropriate lag orders can be selected from five information criteria (i.e., the likelihood ratio, the final prediction error (fpe), the akaike information criterion (aic), the schwarz information criterion (sic), and the hannan quinn information criterion (hqic)) are first determined. for doing cointegration, the study selected lag interval of one (for all three models) as suggested by schwarz information criteria (sic). in this study, both trace and maximum eigenvalue statistics are considered in determining the number of cointegration vectors. in particular, if two test statistics show the same number of cointegration, that number is used. if two statistics do not show the same number of cointegration, the number is international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 90 selected using trace test. luintel and khan (1999) have shown that trace test is more robust than maximum eigen value criteria in testing the cointegration. having confirmed that integration of the seven series is of the same order (checked for three periods); we test whether the seven series are cointegrated over the sample period. the numbers in the tables 9, 10 and 11 show the results of the johansen test. since the johansen test is based on vector auto regressive model (var), we select one period lag for the model based on sic. starting with the null hypothesis of no cointegration among the variables, trace statistics is 220.6801, which is well above the 0.05 critical value. thus, we reject the null hypothesis of no cointegration among these variables at 0.05 and accept that there is one cointegrating equation. considering the null hypothesis of at most one co-integrating relation, based on the trace statistics of 131.3113, which is greater than 0.05 critical value of 83.93712 we reject the null hypothesis of at most one co-integrating equation at 0.05 significance level. similar is the case with the second and third cointegration relationship, where we find trace statistics is greater than the critical value thus we reject the null hypothesis and infer that there are more than three cointegrating relationship between variables. table 9: results of cointegration analyses (pre-recession period) unrestricted cointegration rank test (trace) hypothesized trace 0.05 no. of ce(s) eigenvalue statistic critical value prob.** none * 0.650551 220.6801 111.7805 0.0000 at most 1 * 0.418264 131.3113 83.93712 0.0000 at most 2 * 0.326852 85.26360 60.06141 0.0001 at most 3 * 0.313520 51.62147 40.17493 0.0024 unrestricted cointegration rank test (maximum eigenvalue) hypothesized max-eigen 0.05 no. of ce(s) eigenvalue statistic critical value prob.** none * 0.650551 89.36884 42.77219 0.0000 at most 1 * 0.418264 46.04771 36.63019 0.0030 at most 2 * 0.326852 33.64212 30.43961 0.0193 at most 3 * 0.313520 31.97512 24.15921 0.0036 max-eigenvalue test indicates 4 cointegrating eqn(s) at the 0.05 level. notes: * denotes rejection of the hypothesis at the 0.05 level. **mackinnon-haug-michelis (1999) pvalues. the result from maximum eigen statistics table also supports that there is more than three cointegrating relationship. in addition, we find that maximum eigen value is greater than the 91 critical value and hence we reject the null hypothesis of at most three cointegration vectors at 0.05 per cent level of significance. hence based on trace test we may conclude that there are more than three co-integrating equation among these variables during pre-recession. in the similar vein, the statistics in table 10 shows the cointegrating relationship between the study variables during recession. in this case, both statistics (trace and maximum eigenvalue) are showing similar results. the trace statistics of 60.93088, which is greater than the critical value (54.07904) at 5 % level of significance (see at most two). therefore, we are rejecting the null hypothesis that there are at most two cointegrating vectors and may accept the alternative hypothesis of more than two cointegrating relationship among variables. table 10: results of cointegration analyses (recession period) unrestricted cointegration rank test (trace) hypothesized trace 0.05 no. of ce(s) eigenvalue statistic critical value prob.** none * 0.457913 191.9675 134.6780 0.0000 at most 1 * 0.410861 139.9196 103.8473 0.0000 at most 2 * 0.329805 94.94674 76.97277 0.0011 at most 3 * 0.292357 60.93088 54.07904 0.0108 trace test indicates 4 cointegrating eqn(s) at the 0.05 level unrestricted cointegration rank test (maximum eigenvalue) hypothesized max-eigen 0.05 no. of ce(s) eigenvalue statistic critical value prob.** none * 0.457913 52.04788 47.07897 0.0135 at most 1 * 0.410861 44.97290 40.95680 0.0168 at most 2** 0.329805 34.01586 34.80587 0.0619 at most 3 * 0.292357 29.39427 28.58808 0.0394 max-eigenvalue test indicates 2 cointegrating eqn(s) at the 0.05 level. notes: * and ** denote rejection of the hypothesis at the 0.05 and 0.10 levels. cointegrating relationship among variables during recovery period is shown in the table 11. from these results, it is found that there is more than one cointegrating relationship (based on trace test). in case of trace test, the statistic is greater than the critical value at 0.05 level of significance and rejecting the null hypothesis of at most one cointegrating relationship. however, in case of maximum eigen value criteria it found that there is one co-integrating relationship. however, based on luintel and khan (1999) suggestion the study preferred trace test and inferred that there is more than one cointegrating relationship among study variables during recovery period. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 92 table 11: results of cointegration analyses (recovery period) unrestricted cointegration rank test (trace) hypothesized trace 0.05 no. of ce(s) eigenvalue statistic critical value prob.** none * 0.699329 173.7180 134.6780 0.0000 at most 1 * 0.539903 114.8329 103.8473 0.0077 at most 2 0.447989 76.79325 76.97277 0.0516 trace test indicates 2 cointegrating eqn(s) at the 0.05 level unrestricted cointegration rank test (maximum eigenvalue) hypothesized max-eigen 0.05 no. of ce(s) eigenvalue statistic critical value prob.** none * 0.699329 58.88519 47.07897 0.0018 at most 1 0.539903 38.03960 40.95680 0.1028 at most 2 0.447989 29.11518 34.80587 0.2042 max-eigenvalue test indicates 1 cointegrating eqn(s) at the 0.05 level notes: * denotes rejection of the hypothesis at the 0.05 level. **mackinnon-haug-michelis (1999) p-values the normalized cointegration coefficients for the three models are presented in table 12. table 12: normalized cointegrating coefficients lnbc as endogenous variable pre-recession period recession period recovery period lnad 0.848912* 2.617435* 3.384583* (0.13895) (0.79629) (0.34804) lnborrow -0.650012* 1.177967* 0.176190* (0.07507) (0.21337) (0.06679) lnmatcal 0.151451* 0.007027 -0.069528* (0.02822) (0.06319) (0.02058) lninvest 0.721629* -2.775382* -5.015145* (0.07062) (0.63232) (0.52197) lnbnifty -7.149430* 0.334245* 0.331396* (1.08430) (0.08295) (0.05755) lnbplr 22.54094* 2.087602* -0.677602* (2.87292) (0.55803) (0.29096) constant 2.684247 31.43960* (7.05854) (5.18886) notes: figures in parentheses in normalized co integrating vectors are standard errors. * denotes statistical significance at 0.01 level level. the signs of the coefficients like lnborrow and lnbnifty were found to be negative and others are positive during pre-recession (see first column in table 12). the results show that all the variables are significant. we thus infer that bank credit increases with increase in 93 lnad, lninvest, lnmatcal and lnbplr and decreases with an increase in lnborrow and lnbnifty during pre-recession period. during recession (see table 12, column 2), the results show that, except lnmatcal, all other variables are significant. for lnad the coefficient is 2.61, signifying that its 0.01 level increase during recession led to more than 2.6 per cent increase in bank credits. in the same way the coefficients for lnborrow, lninvest, lnbnifty are 1.17, -2.27(inverse relationship), 0.33 and -2.08(inverse relationship) respectively. during recovery period (see table 12 column 3), the variables (lnad, lnborrow, and lnbnifty) were found to be positively significant indicating that the increase in these variables led to a positive significant increase in bank credits during recovery period. the results show that during this period, lnmatcal, lninvest, and lnbplr had an inverse or negative relationship with bank credit, explaining the decrease in these factors. 5. conclusions this paper contributes uniquely to the evolving body of literature on bank’s lending behavior during times of financial instability/uncertainty, which mostly prevails during the times of financial crises. our findings provides answers to the oft-repeated question “what makes the banks squeeze their lending activity during the crisis?”. broadly, our results are in agreement with the stylized facts on bank behavior in the recent financial crisis, especially on reduced bank lending, increased competition for retail deposits, and reduced monetary policy effectiveness. we have found that, during the times of recession, particularly in the emerging markets of india, borrowings by banks have a significant positive relationship on the lending behavior of the banks. this is because of the fact that banks are unable to garner the much needed financial resources for lending through deposits in view of the prevailing factors of uncertainty/instability (during recession) which makes the depositors withhold their deposits. instead depositors look for other real asset investments or to hold cash till the re-appearance of the signs of financial stability in the markets. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 94 our study has also established that lending rates, even though they were downward during the recession period, were not having positively significant relationship with banks’ lending. this elucidates that the lending rates were required to be reduced drastically in order to infuse bank lending. further, results indicate that investment activity of the banks had a negatively significant relationship with bank lending. the reason is that during the times of financial instability, the banks tend to tighten their lending activity by stringent screening and rigorous monitoring of their borrowers, thereby increase their investments only if it ensures safe returns (profit-seeking behavior and also due to the fall in the state of confidence of banks and the nature of firm decision structures). this can be discerned from the fact that, while credit to deposit ratio showed a downward trend, investment to deposit ratio showed an upward trend during the recession evidencing the era of shortened lending activity of the banks. further, we also notice decreasing cash to deposit ratio during recession period which indicates the tight liquidity scenario in view of the financial instability and uncertainty. we are of the view that failure to curb this trend of sharp decline in bank lending further results in banking crisis (gentler 2010). accordingly, we suggest that the bank’s lending could be maintained at the relatively same pace during the times of recession too by increasing the borrowings rapidly either from the central banks or from government supported long term lending institutions. besides, this measure would also enable the banks to tide over the liquidity crisis that resulted out of the financial crisis. author information: vigneshwara swamy is a professor at the indian business school, hyderabad, india. he may be reached via e-mail: vswamypm@gmail.com or phone: 91-970509-6919. the co-author is s. sreejesh: e-mail: sreejeshibs@gmail.com references allen, f., and carletti, e., (2010). an overview of the crisis: causes, consequences, and solutions, international review of finance, 10:1: 1–26. brunnermeier, m., (2009). deciphering the liquidity and credit crunch 2007–08, journal of economic perspectives, 23, 77–100. calomiris, c.w., and b. wilson (2004). bank capital and portfolio management: the 1930s “capital crunch” and the scramble to shed risk, journal of business, 77(3), 421-454. cheung,y w and lai, k.s., (1993). finite sample sizes of johanson’s likelihood ratio test for cointegration, oxford bulletine for economics and statistics, 55,313-28. de haas, r.t.a., and i.p.p. van lelyveld (2006). foreign banks and credit stability in central and eastern europe: a panel data analysis, journal of banking and finance, 30, 1927-1952. mailto:vswamypm@gmail.com mailto:sreejeshibs@gmail.com 95 dickey, d. and w. fuller, (1979). distribution of the estimators for autoregressive time series with a unit root, journal of the american statistical association 84, 427-431. demirgüç-kunt, a., e., detragiache and p. gupta (2006). inside the crisis: an empirical analysis of banking systems in distress, journal of international money and finance, 25, 702-718. gertler, m., (2010). banking crises and real activity: identifying the linkages, international journal of central banking, vol. 6 no. 4, december. gonzalo, j., (1994). five alternative methods of estimating long-run equilibrium relationships, journal of econometrics, 60(1-2), 203–233. greenlaw, d., hatzius, j., kashyap, a., and shin, h., (2008). leveraged losses: lessons from the mortgage market meltdown, u.s. monetary policy forum, report no. 2: 7–59. ivashina, v. and d. scharfstein (2008). bank lending during the financial crisis of 2008, mimeo. johansen, s., (1988). statistical analysis of cointegration vectors, journal of economic dynamics and control 12, 231-254. johansen, soren and juselius, katarina, (1990). maximum likelihood estimation and inference on cointegration-with applications to the demand for money, oxford bulletin of economics and statistics, department of economics, university of oxford, vol. 52(2), pages 169-210, may. mackinnon, j. g., (1996). numerical distribution functions for unit root and cointegration tests, journal of applied econometrics, 11, 601–618. mackinnon j.g,, alfred haug, and leo michelis (1999). numerical distribution functions of likelihood ratio tests for cointegration, journal of applied econometrics, vol. 14, no. 5, 1999, pp. 563-577. mishkin, f. (1999a). global financial instability: framework, events, issues, journal of economic perspectives, 13(4), fall: 3–20. mishkin, f. (1999b). lessons from the asian crisis, journal of international money and finance, 18: 709–723. ongena, s., (1999). lending relationships, bank default, and economic activity, international journal of the economics of business, 6(2), 257-280. phillips, p.c.b., and p. perron (1988), testing for a unit root in time series regression, biometrika 7.5, 335-346. phillips, p. c. b. (1991). optimal inference in cointegrated systems, econometrica, vol.50, pp.283-306. ralph de haas and neeltje van horen, (2009). the strategic behavior of banks during a financial crisis; evidence from the syndicated loan market, european bank for reconstruction and development (ebrd), central bank of the netherlands (dnb) retrieved from; at http://mpra.ub.uni-muenchen.de/14164/ mpra paper no. 14164 rajan, r.g. (1992). insiders and outsiders: the choice between informed and arm’s-length debt, journal of finance, 47, 1367-1400. taylor, j., (2008). the financial crisis and the policy responses: an empirical analysis of what went wrong, working paper, stanford university. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 6 http://mpra.ub.uni-muenchen.de/14164/ ijbf7-marina.indd the the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 119 ijbf 1 note from editors: this paper is one of three best papers selected by a review panel of three professors at a symposium held in november, 26-28, 2008 in melbourne, australia. the symposium was funded by the australian research council grant, 2007-2009/10, for research on islamic banking and finance. islamic micro-finance programme and its impact on rural poverty alleviation1 m. miazur rahman islami bank bangladesh ltd (ibbl), bangladesh _____________________________________________________ abstract bangladesh launched a shari’ah based micro-fi nance programme in 1995 under the rural development scheme to uplift the overall socioeconomic standards of rural poor. it covered 0.52 million group members, 94% of whom are females. this paper presents fi ndings on this experiment using 1,020 responses in a sample survey. result showed that a signifi cant of clients have improved their religious observations such as prayers and fasting. results of the econometric models showed that household income, productivity of crops and livestock, expenditure and employment increased signifi cantly due to the infl uence of changed behaviour and availability of micro-fi nance. clients stated that the micro-investment had provided better organisation of their economic activities. finally, the islamic micro-investment programme appears to spur more ethical and economically desirable behaviour leading to poverty alleviation. keywords: micro-fi nance and impact, islamic fi nance, rural development fi nancing, bangladesh, behavioral changes. jel classifi cation: g21. _____________________________________________________ 1. introduction microcredit is now a well established poverty alleviating programme which is being implemented in many parts of the world to address different development activities. this is especially so in bangladesh the birthplace of the well-known initiative by nobel laureate professor rahman. besides providing credit facilities, all such institutions also provide training for skill development and self employment of the poor. unfortunately, all these institutions provide interestbased credits and the rate of interest is often exorbitantly high, to commensurate with the risk in micro-fi nance lending. ht tp :// ijb f.u um .e du .m y 120 the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 the interest rates for institutional sources vary from 15-2%, while those for the non-institutional sources range from 33 to 120 per cent (mahmood, 2006). sometimes the interest rates of non-institutional sources may even be 120-140%. mahmood also mentioned that the poor do not have any surplus production hence accumulation of capital is very diffi cult and sometimes impossible for them. on the contrary, owing to the absence of access to productive assets, lack of skills, low-level literacy, malnutrition and absence of an organization of their own, the rural poor fall into the poverty trap (jaim, 1986). besides, there are hardly any credit facilitating institutions that care about the ethical development of the rural poor to change their behaviour consistent with a disciplined approach to economic activity. business behaviour and ethics are interrelated since the disciplined approach needed for good economic management is also seen as founded in ethically consistent behaviour. considering the above mentioned factors, islami bank bangladesh limited (ibbl) launched the rural development scheme (rds) in 1995. the main objective of this scheme is to alleviate rural poverty by providing shari’ah based small capital micro investment fi nance to the agricultural and rural sector for generating employment and raising income of the rural poor. the scheme also provides welfare, moral and ethical services to the rural people of the country. presently, the scheme is being implemented through 129 branches covering 10,023 villages in 60 districts. some 0.52 million group members are covered: 94% are females. the bank has a plan to double the activities of rds within two years. however, before expanding in to the new areas, an assessment of the achievement of the project in terms of its impact on income, livelihood and moral/ethical development of the rural people is done. it may be mentioned here that although thousands of ngos are working in the country with a view to alleviating poverty but the rural poor are still in the vicious circle of poverty and they have little hope of getting out of poverty in the near future. besides, none of them care about ethical and moral development of the clients but ibbl microfi nance is giving especial emphasis to changing behaviour, which aspect is its distinctive difference. therefore, this study will also assess the moral and ethical changes of the clients and its impact of on their income and livelihood. 2. literature review of microfi nance impact studies previous studies on the impact assessment of microcredit programmes in bangladesh have been narrow in their focus. pksf (2005) studied the impact of micro-credit on the members of partner organizations of pksf and found that absolute poverty was reduced by 9% during 1991 to 2000; moderate poverty declined by 5% during 2000 to 2004. chowdhury and bhuiya (2004) examined the impact of credit programme on the bangladeshi borrowers under the brac projects and found positive impacts on human well-being, survival rate and schooling of children. amin, rai and ropa (2003) conducted a study on the ht tp :// ijb f.u um .e du .m y islamic micro-fi nance programme and its impact on rural poverty alleviation: 119-138 121 microcredit clients of grameen bank, brac and asa. they observed that the micro-credit programme was more successful to reach the poor, but less successful to reach the vulnerable poor. zaman (2001) assessed the impact of microcredit on poverty alleviation and women empowerment. he found positive impact on income, decision making ability and in reducing gender disparity. bangladesh institute of development studies (bids) conducted impact study on the microcredit borrowers under the partner organizations of pksf. this study showed positive impact on the income of microcredit participants in comparison to non-program participants (bids, 2001). khandker (2000) assessed the impact of microcredit on saving and found that microcredit increased voluntary saving, which was more pronounced in the cases of women than men. however, all of these studies simply assessed the impact of interestbased microcredit programmes which did not consider shari’ah (islami rules and regulations) related lending in investment and none of them assessed moral and ethical development of the clients. this study assessed these aspects of microfi nance undertaken by the clients of rural development scheme functioning across the country. unlike other microcredit programmes, this scheme emphasised clients’ ethical and moral development activities in its microcredit programme. therefore, this study assessed a different type of microcredit programme than the conventional ones, using modern econometric techniques. another point is that the size of the rural households in bangladesh is generally large and they have a low level of literacy. rural development is hampered due to lack of credits, lack of training, weak infrastructure and poor transport systems. the weak resource base coupled with a faster growing population is aggravating the poverty level of the country. determination of the above factors which are most relevant in explaining poverty alleviation will have important implications for refi ning microfi nance policy. this will also help to locate appropriate clients for the rds-microfi nance. therefore, the other objective was to examine the linkage between clients moral and ethical behavioural changes as well as their income and demographic and investment factors. 3. methodology and data a. conceptual framework in this study, it was observed that the benefi ciaries are mainly rural poor having very limited access to education and institutional credit and they also have low command over productive resources due to their inability to meet the collateral requirement, the rural poor cannot start up productive activities even though they may have the adequate skills for pursuing income-generating activities. because of their low level of income, they cannot even fulfi ll their consumption needs properly and ultimately they have to lead a lower quality of life. in such circumstances, microcredit programme may support the poor in reducing their poverty by creating both direct and indirect benefi ts. ht tp :// ijb f.u um .e du .m y 122 the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 improvement of livelihood factors influences income generating activities (igas): 1. age 2. education 3. asset holdings 4. land size 5. family labor 6. rural infrastructure 7. skill building training 8. morality and ethics expenditure on food purchasing ability household income igas microfinance under microcredit programmes, borrowers can take investment without providing collateral from the microcredit providers. the amount of investment they receive from the providers increases their fi nancial ability to invest more into income generating activities. microcredit programmes also provides opportunity to generate employment for the poor in the locality. by participating in income generating activities under microcredit, rural poor may have more earnings from activities that directly add to their income. figure 1. conceptual framework of microcredit programmes in poverty alleviation this increased income would ultimately increase their purchasing ability. more purchasing ability would enable them to spend more on food, thus, leading them to higher quality of living. the conceptual framework of microcredit in alleviating poverty is shown in figure 1. besides income, there are other factors which are related to well-being of the borrowers. it is imperative to enhance human capital as it is deeply related to the well-being of the borrowers. therefore, the programme provides training and non-formal education to the borrowers, hoping that training and education will enhance their level of skills in performing income-generating activities. therefore, borrowers’ age, education and experiences need to be considered carefully in choosing participants because decision-making ability may largely be infl uenced by such factors. factors like household’s asset will increase the ability of the borrowers to invest in income generating activities. household ht tp :// ijb f.u um .e du .m y islamic micro-fi nance programme and its impact on rural poverty alleviation: 119-138 123 ijij n j ijiij ixy 1 0 asset also helps in increasing income earnings for them. in addition, it is diffi cult for the poor households to hire labour for operating income generating activities. thus, they largely depend on family labour for pursuing income-generating activities. due to the microcredit programme, they have the opportunity to utilize their manpower for productive purposes. inadequacy of rural infrastructure is one of the major obstacles for development of bangladesh. rural poor have very limited access to infrastructural facility and this hampers their economic activities. initiative to establish rural infrastructures (such as rural markets, roads etc.) under the microcredit programme may help them to accelerate their economic activity. access to infrastructure facility and its proper utilisation will increase the ability of rural poor in pursuing their income generating activities (igas). training and non-formal education of borrowers can enhance their level of skills in performing income generating activities. therefore, the rds programme also provides training and non-formal education to the clients. ethics and moral development is an important factor for developing human capital which is deeply related to the well-being of the clients. as rds investment is collateral free, therefore, clients with good ethics and moral behaviour as determined by the fi eld offi cer would indirectly act as collateral to get their investment back and also to invest the borrowed money to the proper income generating activities. b. model specifi cation literature shows that different techniques such as descriptive analysis, ordinary least square (ols), weighted least square (wls), linear programming (lp), and simultaneous equation systems (ses) had been used by researchers in order to estimate the effect of microcredit on the various outcomes such as income, consumption and saving etc. khandker (2000) estimated a conditional demand equation to assess the effect of microcredit on the economic outcomes, such as saving. the researcher used ols estimation technique, using log in both sides, for this study which is as follows: (1) where, i = amount of credit taken by the borrower, y = household income, x ij = a vector of exogenous characteristics (such as age or education of household head) β and  unknown parameters, and  ij error component in the equation. in this study, ordinary least square (ols) regression was used to assess the effect of microcredit on the dependent variables such as income, saving, and expenditure. the models were specifi ed as: ht tp :// ijb f.u um .e du .m y 124 the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 16655443322110 xxixxxxy 2income of the year 2006 – income of the year clients joined in rds. 3 ethics and moral of the clients’ is defi ned in section 3.6 eamcehcsfe domdbrfmigeduppln ii 887 65430)1( c. households income model (2) where, y = amount change2 of annual income of the household, i = amount of investment taken by the borrowers in 2006, x 1 = total land size, x 2 = age of the borrowers dummy (above 40 years of age is 1 and 0 otherwise), x 3 = education dummy (up to 5 years of schooling is 1 and 0 otherwise), x 4 = number of family members engaged in income generating activities, x 5 = distance of branch from the clients places (up to 10 km is 1 and 11 to 16 km is 0), x 6 = ethics and moral3 of the clients,   and  are the coeffi cients of the variables to be estimated, and  0 constant for the equation and  1 error term for the equation. d. estimation of well-being based on the clients’ opinion the logit model was applied to fi nd out the probability level that the clients would be well-off due to the infl uence of particular explanatory variable. in the logit model, dependent variable “clients’ well-being” had two categories such as “borrowers were well-off” under the programme coded as one and otherwise coded as zero. researchers who had used the logit model to assess the effect of microcredit programmes on loan utilization, awareness towards living-standard, and women empowerment had found positive effects of microcredit (begum, 1998; and zaman, 2001). different studies used different dichotomous dependent variables in the logit model for example; begum (1998) used awareness of the borrowers; zaman (2001) used women’s empowerment. in this study, the researcher used "borrowers’ well-being" as the dependent variable which was divided into two categories: (i) borrowers were well-off, and (ii) borrowers were not well-off. the model can be represented as follows: (3) where, pi = probability that borrowers were well-off, 1pi = probability that borrowers were not well-off, edu = education dummy for the clients (up to 5 years of schooling is 1, 0 otherwise), ht tp :// ijb f.u um .e du .m y islamic micro-fi nance programme and its impact on rural poverty alleviation: 119-138 125 domagedbreduppln ii 43210 1)1( 4ethics and moral of the clients’ is defi ned in section 3.6 fmig= no of family members involved in income generating activities, dbr = distance of branch from borrower’s place (dummy, up to 10 km distance is 1 and 11 to 16 km is 0), dom = duration of membership (years), sfe = share of food expenditure to the total expenditure (%), ehc = expenditure on health care (taka), eamc= ethics and moral4 of the clients. and  0 = constant,  i = coeffi cient to be estimated and = error term. e. estimation of ethical and moral change based on the clients’ opinion the logit model was selected in this study to fi nd out the probability level that the clients would be better off due to the infl uence of particular explanatory variable. in the logit model, the dependent variable (clients’ ethical and moral development) had two categories such as clients being “ethically and morally become well-off” under the programme coded as one and otherwise coded as zero. specifi cally model can be written as: (4) where, pi = probability that clients’ were ethically and morally well-off, 1pi = probability that borrowers were not ethically and morally well-off, edu = education dummy (up to 5 years of schooling is 1, 0 otherwise), dbr = distance of branch from borrower’s place dummy (up to 10 km distance is 1 and 11 to 16 km is 0), age= age of the borrowers dummy (above 40 years of age is 1 and 0 otherwise), dom= duration of the membership (years), and  0 = constant,  i = coeffi cient to be estimated and = error term. f. estimation of ethics and moral of the clients opinions were sought from the clients about their awareness and practice of 10 different religious activities. table 2.1 shows the different religious activities of the clients. a four-point likert scale was used to evaluate the borrowers, moral and ethical development i.e. each statement had four options, which were regular, very often, very rare and not at all. the points were summed up from each 10 statements and the total score obtained by each borrower were divided by the highest score of seventy-fi ve in order to create an index of acceptability towards effectiveness of the microcredit programme. the borrowers who received scores less than 70% percent were ht tp :// ijb f.u um .e du .m y 126 the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 coded as zero otherwise coded as one. borrowers performance was considered satisfactory when they scored 70% or above. therefore, in this study, borrowers’ moral and ethics was evaluated based on the score of 70% or above. mahmud (1999) created an acceptability index towards effectiveness of adip programme and the borrower who received scores 70% or above was coded as one to indicate that they were well-off under the adip’s microcredit programme, otherwise coded as zero. begum (1998) created an awareness index and the borrowers who received a score 50% was coded as one in order to indicate that awareness level increased toward their living-standard otherwise coded as zero. logit model was chosen in this study to fi nd out the probability level that the borrowers would be morally and ethically well off due to the infl uence of particular explanatory variable. in the logit model, dependent variable ‘clients moral and ethical development’ had two categories such as ‘clients morally and ethically well-off’ coded as one and otherwise coded as zero. table 1 religious activities performed by the clients statements regular very often very rare not at all score (no.) 10 (0) 6 (4) 4 (6) 0 (10) saying prayer 405 204 170 215 know how to recite holy quran 643 351 reciting holly quran 139 164 136 655 fasting 807 79 21 87 inviting towards islamic activities 324 289 311 70 involvement with dowry 033 961 maintain parda 596 191 109 98 involve with interest 111 71 29 783 misunderstanding with husband 049 29 11 905 involvement with social activities 007 305 372 310 note: figure in the parentheses are the score for dowry, interest and misunderstanding with husband. g. justifi cation of retaining variables in the study the factors which are responsible for the impact of microcredit on the rural poor have been reviewed. girish and mehta (2003) and asanoy (2004) observed that age of the family head and family size were an important factor for economic decision such as production and consumption. latif (2002) also observed that family size had infl uence on consumption and calorie intake of rural bangladeshi households. shrestha and shivakoti (2003) mentioned that family labour was one of the important human capitals. therefore, rural farmers are largely depended on their family labour for pursuing economic activities. khandker (2000) reported that microcredit programmes had increased production, consumption, ht tp :// ijb f.u um .e du .m y islamic micro-fi nance programme and its impact on rural poverty alleviation: 119-138 127 and employment opportunity and reduced informal borrowing among the borrowers and it had also increased the ability of the poor borrowers to save regularly for building fi nancial and physical capital. asanoy (2004) indicated that educated borrowers had higher level of knowledge and skills as compared to less educated borrowers in case of performing their economic activities. latif (2002) observed that the establishment of rural infrastructure had reduced production cost, transportation cost and it had created easy access to markets for inputs and outputs. shrestha and shivakoti (2003) mentioned that physical infrastructure facilities helped in distributing inputs and output, adding value and enhancing production. it is hypothesized that all of these variables would infl uence households’ livelihood, therefore, age, education, family size, fi nancial capital, savings and asset base has been considered in this study as important factors of livelihood. h. sources of data the primary data were directly collected from the fi eld, during december 2006 to april 2007, through interviewing 1020 clients across the country. the impact of investment on ethics and moral of the rds client’s were our major interest therefore assessment was made comparing the clients’ present position (31 december 2006) with their base (information at the time of becoming member). using the stratifi ed sampling approach, the selection procedure for the rds clients considers several factors, and involves multiple stages. the fi rst step involves dividing the rds executing branches of ibbl into seven administrative zones defi ned by ibbl. in order to capture diversity, the 34 branches were randomly selected from seven administrative zones following the probability proportionate to size (pps) method (appendix-1; bangladesh map). two subsequent stratifi cations were made within each branch on the basis of time of becoming client of rds. of the selected branches, 30 rds investment clients from each branch, was randomly selected. so, the total clients for the study were 34 x 30 = 1020 (table 2). table 2 distribution of the samples sl. no. sample types no of branch sample per branch total sample 1 clients dhaka north 4 dhaka south 3 chittagong 3 comilla 5 bogra 9 khulna 8 sylhet 2 total branch 34 30 1020 ht tp :// ijb f.u um .e du .m y 128 the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 the data were collected from the clients using a set of semi-structured pre-tested questionnaire prepared for the purpose. a total of 34 well trained enumerators (one for each branch), have collected the data under the supervision of the researcher. the data were coded and entered in to the spss (statistical package for social scientists), version 12. i. analytical frame work the major concern of this study was to assess the impact of rds investment client. so, comparison was made between the status at present (up to december 2006) and at joining (joining time of the individual clients) on a cross sectional basis. in total 1020 clients were interviewed. of them, the interview of 26 clients were incomplete and hence were dropped from the analysis. the rest of the sample (994) were analysed and the conventional tabular method was used in describing and comparing the performance of the rds clients. the responses were expressed in terms of percentage/frequencies. besides, in order to meet up the objectives the collected data were analysed economically, statistically and econometrically. 4. results and discussion the study was intended to explore the true picture of ethical and moral development of the clients’ livelihood and its impact on poverty alleviation through micro-investment of the rds. the results are presented below: a. behavioural and ethical information of the respondents religious behaviour and ethical development of the clients was one of the important objectives of rds programme. therefore, clients’ present religious performances were plotted in figure 2. result shows that most of the clients do fasting, while almost 85% clients maintained parda. although, 85% clients avoid taking interest, this information is dubious as 15% clients have involvement with other interest based ngo’s. the reason of providing misinformation could be that some of the clients are not still clear about the difference between interest and profi t. b. changes in family income effort was taken under this microinvestment programme to increase clients’ family income by investing their borrowed money to different income generating activities. table 3 shows that on average clients’ income has increased by more than 33% from joining in the rds up until 2006. the highest income change (6763 tk. /year) was observed for small business which was followed by income from services (4615 tk. /year). the crop sector income has also signifi cantly increased but highest change was observed for small business which implied that clients preferred investing money into the business sector. ht tp :// ijb f.u um .e du .m y islamic micro-fi nance programme and its impact on rural poverty alleviation: 119-138 129 0 20 40 60 80 100 p er ce n t p ra y f as ti ng q ur an r ec it at io n a vo id in te re st d aw ah h iz ab religious activities yes no table 3 annual income generation of the households by sources source of income household income (tk / year) change of income level of signifi cance present joining taka per cent t-value sig. crops 13118 10993 2125 19.33 3.188 0.001 fruits selling 1629 977 652 66.73 3.641 0.000 vegetable 2109 1464 645 44.06 3.865 0.000 livestock 4070 2841 1229 43.26 5.463 0.000 poultry 1669 1195 474 39.67 2.507 0.012 small business 32061 25298 6763 26.73 4.015 0.000 labour selling 13617 11003 2614 23.76 6.509 0.000 service income 14827 10212 4615 45.19 5.564 0.000 others 9927 5751 4176 72.61 2.222 0.038 total 93027 69734 23293 33.4 c. impact of rds on clients religious activities in addition to the poverty alleviation the microinvestment program has taken initiative to increase the clients’ awareness on religious activities. therefore, in this study attempt was made to assess the infl uence of rds on clients’ awareness on religious activities. table 4 shows that a signifi cant number of clients (21%) have started praying after joining rds programme. besides, a signifi cant number of clients have started practicing some other religious activities like, reciting holly quran, inviting others to follow islamic activities and avoid taking interest. figure 2. distribution of clients by religious activities ht tp :// ijb f.u um .e du .m y 130 the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 table 4 awareness on religious activities of the clients of rds indicators at present at joining change status frequency frequency frequency percent regular prayer 779 643 136 21.15 know how to recite holy quran 643 600 43 7.17 reciting holy quran 339 273 66 24.18 fasting regularly 907 880 27 3.07 inviting towards islamic activities 924 870 54 6.21 involvement with dowry 033 wearing hizab 896 885 11 1.24 not involve with interest 883 814 69 8.48 misunderstanding with husband 089 involvement with any local crime 007 d. results of ordinary least square (ols) estimation of the model one of the prime objectives of the study was to assess the factors infl uencing households’ income, expenditures, well-being and ethical and morals of the clients. the ordinary least square (ols) technique was used to estimate the model for assuming the effect of different factors of the model. if the regression equation estimation of similar functions involve more variables, the model would have suffered from low degrees of freedom and multicollinearity problems resulting in ineffi cient estimates. therefore, multicollinearity was tested and the outcome of some independent variables not being signifi cant could be the effect of multicollinearity. the regression model was re-estimated; dropping some variables whose level of signifi cance was very low. the results were signifi cantly improved and hence have been interpreted. e. ols results for the household income model it was expected that the amount of investment taken by the borrowers would increase their household income. therefore, this study examined the infl uence of microinvestment on the households’ income. however, household income does not depend on only one factor like amount of investment. it also depends on the other socioeconomic factors. that is why, variables like borrower’s age, borrower’s educational background, involvement of family members in farming, total land size of the households, distance to branch from the borrowers’ place and ethics and moral of the clients were considered as independent variables. ht tp :// ijb f.u um .e du .m y islamic micro-fi nance programme and its impact on rural poverty alleviation: 119-138 131 table 5 shows the results of ols for the households’ income. as can be seen from the statistics, the sign of all the variables are plausible but only four of them namely, borrowers’ age, amount of investment taken by the borrowers, number of family members involved in earning and clients’ ethics and morals had positive and signifi cant infl uence on the household income. the value of r-square was 0.693. it indicates that about 69% of the total variation of the dependent variable was explained by the independent variables. it was found that the amount of investment taken by the borrowers had a positive and signifi cant infl uence on the household income. this collateral free investment-money had increased their opportunity to start up income generating activities. it had also increased their ability to invest more on the existing economic activities resulting in more earnings from their investment. all these reasons had assisted them in increasing their household income. household income would increase by 1.21 taka because of receiving additional one unit of investment by the borrower from the microinvestment providers. the study showed that the borrowers’ age variable was positively and signifi cantly related with household income. as age increases, borrowers acquire more experiences which would ultimately assist them to increase their household income level. many researchers used age as independent variable in their regression analysis in order to assess its infl uence on borrowers saving, agricultural production, household income and consumption (khandker, 2000; amin et al., 2003) and found signifi cant infl uence. this study shows that the number of earning family members’ had positive and signifi cant infl uence on their households’ income. this result implied that higher the number of earning family members, the higher the family income. this expected result is also supported by mahmood (2006). education did not show any signifi cant infl uence on household income as there was small variation of education among the clients. result showed that more than 90% of the borrowers had schooling of less than fi ve years. however, the variable had a positive sign. this indicates that education might infl uence their income level but it required more time and more educational facilities for the borrowers to realize the effect of education on income. the results showed that the distance of branch from borrowers place had no signifi cant effect on the households’ income level. however, the sign for the dummy variable (10 to 16 km distance from branch) was negative meaning that further increasing distance between branch and borrowers’ place would decrease the income earning opportunities. it is worth mentioning here that ethics and morals had positive contribution on household income. this result is not unlikely as it is assumed that the clients who are bearing good moral and ethical characters are honest and sincere enough, so they did not divert their borrowed money to any unproductive sector. besides, they are sincere enough in their activities and hence may have generated more income. ht tp :// ijb f.u um .e du .m y 132 the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 table 5 estimated ols results for the household income variables co-effi cient t-value sig constant 4.079 38.384 0.000** log of investment taken in 2006 1.210 2.612 0.019** log of total land size 0.101 0.408 0.684 log number of earning family members 1.160 1.990 0.024** distance of branch dummy (up to 10 km is 1 and 11 to 16 km is 0) -0.030 -0.115 0.108 borrower’s age dummy (above 40 years) 0.201 2.646 0.018** education dummy (up to 5 years schooling is 1 and 0 otherwise) 0.120 0.738 0.460 ethics and moral (dummy) 0.090 1.847 0.055* r-squared: 0.603 note: ** signifi cant at 1% level, and * signifi cant at 5% level. f. results of the logit model for clients’ well being the logit model was selected in this study to fi nd out the probability level that the clients would be well-off due to the infl uence of particular explanatory variable. in the logit model, dependent variable “clients’ well-being” had two categories such as “borrowers were well-off” under the program coded as one otherwise coded as zero. the rural poor had to lead a low quality of life. it is expected that the microfi nance programme would bring a positive change in the overall livingstandard of the borrowers by improving their economic activities. therefore, an attempt was made to investigate the infl uence of socioeconomic variables on the dependent variable "borrowers welfare". based on the borrowers’ perceptions, the dependent variable was coded as one if the borrowers had answered welloff under the microinvestment programme, otherwise, it was coded as zero. the logit model was used in order to fi nd out the probability level that the borrowers would be well-off due to the infl uence of a particular independent variable. table 6 shows that the duration of rds membership, number of income generating family members, share of food expenditure to total expenditure, households health expenditure and clients ethics and morals had positive and signifi cant contribution on clients well-being. table 6 shows that the wald statistic for the variable "expenditure on health" was 3.03 and it was positively and signifi cantly related to the dependent variable. this indicates that more expenditure on health care would increase the likelihood for the borrowers to be well-off. ht tp :// ijb f.u um .e du .m y islamic micro-fi nance programme and its impact on rural poverty alleviation: 119-138 133 due to the low level of purchasing ability, borrowers had to maintain a low level of expenditure indicating a lower quality of living. they had to struggle even to meet their expenditure on food items. the microfi nance programme takes the effort to increase borrowers’ ability to spend more on food items for better living. table 6 shows that wald statistic for the independent variable "share of food expenditure to the total expenditure" was 5.746, and it was found signifi cantly and positively related with the dependent variable. it may be mentioned here that ethics and moral had positive contribution on clients’ well-being. this result is not unlikely as the clients bearing good moral and ethical characters are honest enough so they did not divert their borrowed money to any unproductive sectors. besides, they are sincere enough to their activities hence were better-off. table 6 estimated results of the logit model variable (b) standard error wald statistics sig. odd ratio exp (b) constant -0.954 0.823 1.344 0.246 0.385 duration of membership 0.238 0.062 14.64 0.000** 0.788 education dummy (up to 5 yrs of schooling is 1 and 0 otherwise) 0.014 0.251 0.201 0.986 0.996 income generating family members 0.208 0.073 13.67 0.00** 0.768 age of the clients dummy (up to 40 yrs of age is 1 and 0 otherwise) -0.060 0.556 0.011 0.916 0.942 share of food expenditure to the total expenditure (%) 0.014 0.006 5.746 0.017** 1.014 health expenditure (taka) 0.017 0.020 3.031 0.054* 1.010 ethics and morals 0.164 0.243 3.476 0.050* 0.846 cox and snell r square: 0.198 -2log likelihood: 667.280 overall accuracy: 82.8 note: ** signifi cant at 1% level and * signifi cant at 5% level. g. results of the logit model for assessing clients ethics and moral it is expected that rds program would bring a positive change in the ethical and moral development of the clients. therefore, an attempt was made to investigate the infl uence of socioeconomic variables on the dependent variable "moral and ethical development". ht tp :// ijb f.u um .e du .m y 134 the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 table 7 estimated results of the logit model for ethics and moral variable (b) standard error wald statistics sig odd ratio exp (b) constant 1.101 0.447 6.070 0.014** 3.008 education dummy (up to 5 yrs of schooling is 1 and 0 otherwise) 0.633 0.165 14.076 0.000** 0.531 age of the clients dummy (up to 40 yrs of age is 1 and 0 otherwise) 1.424 0.366 15.136 0.000** 0.241 membership duration (years) 0.102 0.032 10.028 0.002** 1.108 distance of branch (up to10 km distance is 1 and 0 otherwise) -0.402 0.246 2.665 0.103 0.669 cox and snell r square: 0.178 -2log likelihood: 1241.76 overall accuracy: 72.8 note: ** signifi cant at 1% level and * signifi cant at 5% level. based on the borrowers’ perceptions, the dependent variable coded as one if the borrowers were morally and ethically well-off under the microinvestment program and otherwise it was coded as zero. the logit model was used in order to fi nd out the probability level that the borrowers would be morally and ethically well-off due to the infl uence of a particular independent variable. table 6 shows that the educational level, age and membership duration had positive and signifi cant contribution on clients’ ethics and moral development. this result implies that elderly and educated clients who have several years’ involvement with this rds programme have ethically and morally been more developed than the young and illiterate new clients. h. clients’ opinion towards micro-investment programmes clients’ opinion about the benefi t of microinvestment programmes on their skill, social and economic condition was assessed and organized them in table 8. results showed that microinvestment programme had brought positive changes in their skill and socioeconomic status. of them, self confi dence development, economic solvency, communication skill, and knowledge on business and religion practices are mentionable. i. major problems and suggestions stated by the clients like other credit programs, rds microinvestment clients also face problems participating in the microfi nance program. their problems are diverse in nature and vary depending on time, space, socioeconomic aspects, and nature of the ht tp :// ijb f.u um .e du .m y islamic micro-fi nance programme and its impact on rural poverty alleviation: 119-138 135 program. success of the programs largely depends on identifying and solving the problems on time. therefore, in order to improve the operations of the program, it is important to identify clearly the borrowers’ problems. table 9 shows that majority (90.65%) of the borrowers had problems with investment size. they mentioned that the amount of investment they had received from the micro-investment providers was inadequate for them to pursue their income generating activities (igas) smoothly. table 8 impact of rds on clients skills stated by the clients serial types of change increased no change frequency per cent frequency per cent 1 economic solvency 775 77.97 219 22.03 2 islamic practice 240 24.14 754 75.86 3 social dignity / status 211 21.23 783 78.77 4 self confi dence 800 80.48 194 19.52 5 business knowledge 720 72.43 274 27.57 6 communication skills 785 78.97 209 21.03 7 knowledge on business computation 755 75.96 239 24.04 8 confi dence on education 596 59.96 398 40.04 table 9 major problems stated by the clients serial problems per cent 1 amount of investment is very small 90.65 2 do not have any training programme 85.60 3 investment getting period is very long 84.70 4 gestation period for repaying investment is too short 78.55 5 insuffi cient time for meeting 72.00 6 there is no place for organizing meeting 70.65 7 need to produce fake buying and selling voucher 33.65 8 tenants are ignored for getting investment 26.08 9 no islamic school for their children 23.26 10 woman has no control on their borrowed money 15.66 lack of training facilities to upgrade their skills and technical and islamic knowledge is also an important problem mentioned by 86% clients. about 85% respondents mentioned that a delay in receiving investment is also a problem which was followed by very short gestation period for repaying investment. ht tp :// ijb f.u um .e du .m y 136 the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 some respondents (78 %) mentioned that some times they needed to start repaying their borrowed money even before investing the money. some clients (33.65 %) had to produce false voucher which is a clear violation of shariah. besides, there is no islamic school under this scheme which can teach islamic knowledge to the children. 5. conclusion microcredit is now a well established poverty alleviating programme which is being implemented across the world. all these institutions working in bangladesh provide interest based credit which is straightway a violation of shari’ah. these institutions also do not care about the ethical aspects of the rural poor, although it is obligatory, not only in business sector but also in all aspects of life. business and ethics should be interrelated. however, the microinvestment programme is to uplift the overall socioeconomic plight of the rural poor, which, in this case, cares about developing ethics and morals development of the clients as it can play a crucial role in alleviating poverty. this assessment highlighted the achievement of ethics and morals of the clients and its contribution on poor people’s livelihood and summary of the study fi ndings, conclusion and policy implications are presented below: • although, level of participation in religious activities by the clients has greatly been improved after joining rds programme, there is still room to improve these activities, especially knowledge about interest, its consequence, and how to get rid of it. therefore, a weekly meeting or frequent lectures may be organized regarding this issue. • results showed that most of the clients utilized their borrowed money but the reality is that not all clients invested their borrowed money to income generating activities. instead, some of them utilised their investment in house repairing, children’s marriage ceremony and furniture purchase etc. therefore, proper monitoring and supervision should be done to develop their morals and ethics so that they use their money in income generating activities. • it was observed that some fi eld supervisors do not properly practice the bai-muajjal mode; which is a clear violation of shari'ah so, necessary measures have to be taken to develop their moral and ethics so that they do not violate shari'ah. • mudaraba mode of investment may be practised with some selective good clients, even if in some cases this mode does not bring any profi t for the scheme so that the clients and person concerned would fi nd the difference between conventional (interest based) and islamic microfi nance. incorporating this mode in place of bai-muajjal mode would reduce shari'ah violation. ht tp :// ijb f.u um .e du .m y islamic micro-fi nance programme and its impact on rural poverty alleviation: 119-138 137 • poor borrowers are not aware of the modern technology. they depend much on the traditional method of farming resulting in low production. therefore, provision should be made to provide demand-led and effective training on different aspects of on-farm and off-farm activities, credit management, environmental pollution, nutrition, health care and ethical development. • it was observed that fi eld supervisors are not well trained up and hence they found it diffi cult to motivate the rural people toward rds and also to manage the rds centres. frequent training should therefore, be organised for improving the fi eld supervisors’ knowledge, skill, moral and ethical values. • rds organises weekly meetings in each centre but effectiveness of the meetings cannot be maintained due to shortage of time. therefore, duration of centre meeting time has to be increased so as to make those meetings effective. • model estimation results showed that branch distance from rds clients place (11 to 16 km) has a negative impact on clients’ income and wellbeing. although this result is not signifi cant, it is alarming. therefore, branch command area for rds programme should not be expanded beyond the 16 kilometre range. • the number of rds clients is sharply increasing higher although the average rate of dropout (11%) is also alarming. the reason for dropout is not yet known therefore, proper selection of clients is very important and monitoring should be increased to reduce the rate of dropout. author information: dr. m. miazur rahman is the director of the islami bank, bangladesh limited (ibbl). references amin s., rai a.s. and ropa g. (2003) does microcredit reach the poor and vulnerable? evidence from northern bangladesh. journal of development economics, 70, 59-82. asanoy, a. (2004). effects of agricultural credit and microfi nance on expenditure patterns in yemen (unpublished doctoral dissertation). universiti putra malaysia, selangor, malaysia. bangladesh institute of development studies (bids). (2001, october). final report on bids study on pksf’s monitoring and evaluation system (mes) dhaka. begum, r.a. (1998). a comparative study on the status of awareness and potential of credit receiver and credit non-receiver rural women in selected areas of dinazpur district of bangladesh ( unpublished master’ thesis). agricultural university, mymensingh, bangladesh. ht tp :// ijb f.u um .e du .m y 138 the international journal of banking and finance, vol. 7. number 1: 2010: 119-138 chowdhury, a. m. r., and bhuiya, a. (2004). the wider impacts of brac poverty alleviation programme in bangladesh. journal of international development 16(3), 369-386. girish, m., and mehta, p. (2003). crop diversifi cation: an empirical analysis on kangra farms of himachal pradesh, india. asia-pacifi c journal of rural development, 13 (2). latif, m.a. (2002). income, consumption and poverty impact of infrastructure development. the quarterly journal of the bangladesh institute of development studies, 28 (3), 1-36. mahmood, t. (2006). effectiveness of microcredit for poverty alleviation under agricultural intensifi cation projects in bangladesh. (unpublished doctoral dissertation). school of graduate studies, university putra malaysia, malaysia mahmud, k. t. (1999). impact of credit on income and employment of women members of savar thana under world vision project (unpulished master’s thesis). bangabandhu sheikh mujibur rahman agricultural university, gazipur, bangladesh. pksf (2005). follow up monitoring and evaluation system (mes). hb consultant limited. shrestha, s. g. and shivakoti, g. p. (2003). prominent livelihood asset pentagon within the analytical framework of irrigation system performance assessment. asia-pacifi c journal of rural development, 13(1), 60-88. zaman, h. (2001). assessing the poverty and vulnerability impact of micro-credit in bangladesh: a case study of brac. (world bank policy research working paper no-2445). washington d.c.: world bank. ht tp :// ijb f.u um .e du .m y modeling credit risk: an application of the rough set methodology the international journal of banking and finance, volume 10 (number 1), 2013: pages 34-56 modeling credit risk: an application of the rough set methodology reyes samaniego medina and maria jose vazquez cueto pablo de olavide university, spain and seville university, spain ___________________________________________________________ abstract the basel accords encourages credit entities to implement their own models for measuring financial risk. in this paper, we focus on the use of internal ratings-based (irb) models for the assessment of credit risk and, specifically, on one component that models the probability of default (pd). the traditional methods used for modeling credit risk, such as discriminant analysis and logit and probit models, start with several statistical restrictions. the rough set methodology avoids these limitations and as such is an alternative to the classic statistical methods. we apply the rough set methodology to a database of 106 companies that are applicants for credit. we obtain ratios that can best discriminate between financially sound and bankrupt companies, along with a series of decision rules that will help detect operations that are potentially in default. finally, we compare the results obtained using the rough set methodology to those obtained using classic discriminant analysis and logit models. we conclude that the rough set methodology presents better risk classification results. keywords: rating, credit risk, basel accords, rough sets jel classification: g21, g32 1. introduction the basel accords opened the way for and encouraged credit entities to implement their own models for measuring financial risks. in this paper, we focus on the use of internal ratings-based (irb) models for the assessment of credit risk and, specifically, on one of the approaches to model the probability of default (pd). 35 the traditional methods used for modeling credit risk, such as discriminant analysis and logit and probit models, start with several statistical restrictions. the rough set methodology avoids these limitations and is presented as an alternative to the classic statistical methods. the objective of our study is to apply the rough set methodology to a database composed of 106 companies that are debtors of the same financial entity to obtain the ratios that best discriminate between healthy and bankrupt companies. a second objective is to find a series of decision rules that will help detect potentially failing credit operations as a first step in modeling the probability of default. finally, we compare the results obtained using the rough set methodology to those obtained using classic discriminant analysis and logit models. we conclude that the rough set methodology presents good risk classification results. this paper is structured as follows. section 2 reviews the most significant empirical studies. section 3 introduces the theory of rough sets. in section 4, we continue with a description of the sample of companies used for the empirical study. the empirical application is described in section 5 wherein we first use the rough set methodology to determine the variables that may explain the default, and then compare the results obtained using this methodology to those obtained using classic discriminant analysis and logit models. finally, in section 6, we draw a series of conclusions. 2. literature review the models for predicting business failure and estimating the probability of default (pd) required by the basel accords have been the subject of several studies, conducted not only by academics but also by the financial sector itself. all of the theoretical effort has been focused on the modeling of the stochastic process associated with insolvency and on determining the variables that must be included in these models. among these traditional models, we can distinguish between univariate and multivariate models. univariate models examine the behavior of each variable separately to explain any insolvency. one of the classic studies using this method was conducted by beaver (1966), who found a number of financial ratios that could discriminate between healthy and bankrupt companies during the 5-year period prior to the occurrence of the actual default. other notable studies were those conducted by courtis (1978) and altman (1993). unlike the univariate models, the multivariate models combine the information provided by a set of variables. the study that pioneered this method was performed by altman in 1968, in international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 36 which he proposed a discriminant analysis that combined the information provided by 25 financial ratios. a wide variety of studies have been based on discriminant analysis, including those by dambolena (1980) and laitinen (1991). in spain, cabedo et al. (2004) presented an adaptation of the discriminant model to calculate the probability of default in companies. the model was applied to a portfolio of hypothetical borrowers from the same sector to calculate the regulatory capital according to the foundational irb method of the basel accord. the importance of this technique is demonstrated in the bibliographic review by dimitras (1996). after analyzing 158 articles on the prediction of business insolvency during the period from 1932 to 1994, dimitras concluded that discriminant analysis is the model that is most frequently used to resolve this type of problem. other authors have opted for logit and probit analysis. ohlson (1980) was the first to apply this type of technique to predict company insolvency. wilson (1997) developed the credit portfolio view model for mckinsey, establishing a discrete process with multiple periods. with this methodology, the probability of default is obtained through logit functions of indices of macroeconomic variables that, in some ways, represent the functioning of the economy (zmijewski, 1984). dimitras (1996) found that the logit model is the second most frequently used model for resolving the problem of company bankruptcy. fernández (2005), in an attempt to combine univariate and multivariate analysis, conducted an empirical study in which he used a prior univariate analysis to select those ratios with greater discriminant power within each of the categories of ratios established from among the 23 ratios initially considered.1 subsequently, he performed a logit and probit multivariate analysis to obtain scores for each company; these scores enabled a rating system to be established and default probabilities to be assigned. trucharte et al. (2002) developed a system for rating borrowers by estimating a logistic regression model that utilizes economic and financial information. the scores obtained are used to establish homogeneous categories in which the various borrowers are classified or rated and the probability of default that can be assigned to each category. more recently, altman and sabato (2007) developed a distress predictor model specifically for the sme sector and analyzed its effectiveness against that of a generic corporate model. they 1 these categories were liquidity, leverage, activity, debt cover and productivity. 37 used a logit regression model technique on panel data from more than 2,000 us firms over the period from 1994 to 2002. in parallel with these studies, other methods have been explored to overcome the restrictive hypotheses that models of statistical inference impose on the variables. these hypotheses usually do not conform to reality and distort the results obtained. for these reasons, eisenbeis (1977), ohlson (1980) and zavgren (1983) questioned the validity of the traditional models. in particular, techniques originating from the field of artificial intelligence have begun to be used; programs have been produced that are capable of generating knowledge from empirical data and then utilizing that knowledge to make inferences based on new data. within this approach, we can distinguish techniques that seek knowledge by identifying patterns in the data. among these are various classes of neural networks and other techniques that infer decision rules from the base data. the rough set methodology belongs to this last group of techniques. authors such as dimitras et al. (1998) and daubie et al. (2002) have applied this technique to the classification of commercial loans. other authors, such as ahn et al. (2000), have combined the rough set methodology with neural networks to predict company failure. in spain, various studies can similarly be found that apply the rough set methodology for the prognosis of company insolvency. segovia et al. (2003) applied this technique to the prediction of insolvency in insurance companies, and rodríguez et al. (2005) utilized it for the same purpose in a sample of smalland medium-sized enterprises (smes). 3. theoretical framework of the rough set methodology rough set theory, first proposed by z. pawlak in 1982, is considered as an appropriate tool for handling cases in which there is considerable vagueness and imprecision. more specifically, the method is effective at working with problems of multidimensional classification (pawlak et al., 1994). the basic idea rests on the indiscernibility relation that describes elements that are indistinguishable from each other. its principal objective is to find basic decision rules that enable the acquisition of new knowledge. its key concepts are discernibility, approximation, reducts and decision rules. the point of departure for the method is the existence of an information/decision table in which each element is characterized by a set of variables (attributes) and a decision variable that classifies the element into one of two or more categories. indiscernibility exists when two elements are characterized by the same properties for all variables, but the categories in which they are classified do not coincide. this is the basis of the rough set theory. in such a case, for international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 38 each class of decision or category x and for each subset b that contains variables, two sets are constructed; these sets are called the set of the lower approximation of the decision class and the set of the upper approximation of the decision class, respectively. the set of the lower approximation of decision class x with respect to the variables b, which is called bx, is given by the group of all elements that, being characterized by b, belong to class x with complete certainty. the set of the upper approximation of decision class x, which is called xb , is given by the group of elements that, based on the information b that we possess, may belong to class x, but about which we cannot be sure. the elements that are different between the two sets form the "doubtful" elements, which are those elements that, using only the information contained in b, are not known with complete certainty to belong to class x. when these different elements exist (i.e., when the difference is not zero), it is said that class x is a rough set with respect to the subset of variables b. this set can be characterized numerically by the quotient between the cardinal of the set of lower approximation and the cardinal of the set of upper approximation. this quotient is known as the "accuracy of approximation". if various decision classes exist, the sum of the cardinals of the lower approximations divided by the total of all elements is known as the "quality of approximation of the classification, by means of set b", and this is the percentage of the elements that have been correctly classified. another important aspect of this technique is the reduction of the initial table of data, which eliminates the redundant information. this process is performed using the reducts. a reduct is a minimum set of variables that conserve the same capacity for classifying the elements as the full table of information. a reduct is thus an essential part of the information and constitutes the most concise way of differentiating between the decision classes.2 the final stage of the rough set analysis is the creation of decision rules, rules that allow us to determine whether a given element belongs to particular decision classes. these rules represent knowledge and are generated by combining the reducts with the values of the data analyzed. a decision rule is a logical statement of the following type: "if particular conditions are met, then the element belongs to a particular decision class". these rules allow us to classify new elements easily..3 2 the reducts were obtained based on the equivalence classes that defined the indiscernibility relation on the set of observations. 3 for more detail on the formal mathematical aspects of the methodology, komorowski et al. (1999) may be consulted. 39 4. data and variables a. selection of the sample. we adopted the following approach for both selecting the sample and choosing the independent variables. following altman (1968), we have paired a number of healthy and bankrupt companies of similar size and sector, thus taking a sample in which the bankrupt companies represent 50% of the total. when selecting the sample, the data under consideration should be obtained for the same period of time for healthy and bankrupt companies alike. however, the companies in bankruptcy or suspension of payments tend to delay the presentation of their accounting data in the time period prior to their declarations of insolvency. to overcome this inconvenience, we have collected the accounting data from the last full year prior to the bankruptcy from the most recent data available. in the development of our model, we employed a database, provided by a spanish savings bank that contains information on companies that requested and obtained a loan from the bank. these companies were divided into two groups: healthy and bankrupt. in particular, the sample of bankrupt companies used for the analysis only included those companies with loans whose unpaid debt, whether of interest or principal, amounted to a percentage of more than 10% of the full risk accepted. the computation date was december 31, 2003. the group of healthy companies, i.e., those that did not default in the time horizon considered, was selected using the individual pairing technique, controlled by those characteristics that could affect the relationships between financial ratios and failure. each company in the failed group was matched with a healthy company of the same industry and same approximate size. in relation to the sector, the pairing was achieved at a level of four digits of the c.n.a.e. (national classification of economic activities) of 1993. the criterion adopted for pairing by size was the total assets. as a homogenizing factor for all of the companies, we controlled for the conditions that the total amount of the customer's operations with the savings bank, i.e., their live risk, should exceed 60,120 euros and that the companies should all be public limited companies (plcs), which facilitates access to their accounting statements. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 40 in total, the sample contained 106 spanish companies, 53 failed and 53 healthy, with a very diverse spread of economic activities.4 b. selection of the independent variables for the models the independent variables chosen for the construction of the models were selected from the financial statements, principally from the balance sheet and the profit and loss account, of the companies that comprise the sample. these accounting statements were extracted from the sabi database, developed by informa, s.a., which includes more than 95% of the companies that present their accounts in the mercantile register in spain. given that most of the companies that went bankrupt presented their financial statements neither in the preceding year nor in the two years prior to the date of default, we obtained the latest data available corresponding to the year prior to the company bankruptcy, as explained above. thus, the year t-1 corresponds to that of the latest available accounts. table 1: ratios considered in the analysis this table summarizes the variables that are potentially explanatory for company bankruptcy. these, in general, include ratios of liquidity, indebtedness, structure, rotation, generation of resources and profitability. liquidity ratios degree to which the company's assets that can be liquidated, in the short term, are sufficient to meet the payments required for the short-term debts contracted. r1=current assets / current liabilities r2=(quick + available assets) / current liabilities r3=available assets / current liabilities r4=(quick + available assets current liabilities) / (operating costs + personnel costs + variation provisions + other operating costs) indebtedness ratios 4 we have excluded property development and property sales companies from the analysis, as these companies have characteristics that are very peculiar and different from other companies. in the assessment of the loan application made by this type of company, the decisive factor for granting the loan is the viability of the specific project for which the loan is sought. this information is not reflected in corresponding accounting statements. medina and vazquez cueto: modeling credit risk 41 relationship between the different components of the liabilities, in the short and long term, and the company’s own funds and between the cost of the debt and the liabilities or the profits and funds generated. r5=long-term debt / net worth r6=net worth / total liabilities r7=long-term debt / (long-term debt + current liabilities) r8=financial costs / (long term debt + current liabilities) r9=financial costs / (gross profits + provision for amortization) r10=financing costs / gross profits r11=long-term debt / total liabilities structural ratios proportionality between the balance sheet items of assets and liabilities and in the composition of these items. r12=(current assets current liabilities) / total assets. r13=current assets / total assets. rotation ratio measure of the dynamism of the business activity in relation to the structure of the company. r14=(current assets – current liabilities) / (net turnover + other income from operations) resource generation ratios relationship of the self-financing capacity of the company to various accounting magnitudes. r15=(net profit/loss for period + amortization provision) / (net turnover + other income from operations) international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 42 r16=(net profit/loss for period + amortization provision) / current liabilities r17=(net profit/loss for period + amortization provision) / (long-term debt + current liabilities) r18=(net profit/loss for period + amortization provision) / total liabilities r19=(gross profits + amortization provision) / current liabilities profitability ratios comparison of the profit obtained at various levels with the resources invested r20=(operating profit/loss + financial income + profits from financial investments + exchange rate gains) / total assets r21=profit/loss from ordinary activities / total liabilities r22=pre-tax profits / net worth r23=pre-tax profits / total liabilities r24=profit/loss for the period / net worth r25=gross profits / total assets source: trujillo et al., (2004) the accounting information derived from the sample selected was subjected to a meticulous study with the aim of detecting and resolving possible anomalies or significant incidents that could distort the final analysis. those atypical companies with clear and insuperable anomalies in their accounts were excluded from the sample. for example, those companies that presented profits despite being in a situation of default were eliminated. twenty-five ratios were selected by choosing a broad set of variables that are potentially explanatory for company bankruptcy based on the frequency and efficacy with which the ratios have been used in other predictive models of company insolvency or in the analysis of banking risks. 43 the variables used include ratios of liquidity, indebtedness, structure, rotation, generation of resources and profitability. the specific ratios considered in the analysis are given in table 1. 5. empirical application a. application of the rough set mrthodology for the empirical application, the values of the 25 economic/financial ratios shown in table 1 were calculated for each of the 53 bankrupt companies for the financial year before entering into default, and a similar procedure was adopted for each matched healthy company. this process produced a table of information containing 106x25 data items. an additional column that indicates whether the company in question is in a situation of bankruptcy or health is included in the table of information. thus, we have assigned the value 0 to bankrupt companies and the value 1 to the matched healthy companies. we thus obtain an information-decision table with 106x26 data items. from these data, we determine which ratio or ratios of the 25 variables serve to explain a company’s state of default as the first step in calculating the probabilities of default. first, napierian logarithms of the values of the ratios were computed to avoid problems with the normality of the variables when applying the discriminant analysis. then, given the nature of the variables considered, we proceeded to discretize the values. this is not an essential requirement for the application of the technique, but it facilitates the interpretation of the results; it is a more consistent way to identify bankrupt or healthy companies when the values of the variables considered fall within the same range but do not coincide exactly. for this, we utilized the codification given in table 2.5 the next step is to determine the accuracy provided by the explanatory variables using the rose software. the quality of the approximation is 1.6 5 we discretized the variables by grouping them into four ranges based on the number of observations belonging to each range. for this, we utilized the rose software, provided by the institute of computing science of the poznan university of technology, and we thank the institute for making this software available to us. 6 the quality of the approximation is expressed by the ratio between the number of companies classified correctly and the total number of companies that comprise the sample. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 44 table 2: codification ranges of the variables napierian logarithms of the values of the ratios are first computed. then, the values are discretized according to the ranges identified in this table. variables codified value 0 1 2 3 r1 (-inf, 0.00434058) (0.00434058, 0.00437793) (0.00437793, 0.00467188) (0.00467188, +inf) r2 (-inf, 0.00131066) (0.00131066, 0.00215777) (0.00215777, 0.00653424) (0.00653424, +inf) r3 (-inf, 3.01154e-005) (3.01154e-005, 4.19928e-005) (4.19928e-005, 0.000818944) (0.000818944, +inf) r4 (-inf, 0.00126886) (-0.00126886, 0.000469693) (-0.000469693, 0.000412292) (-0.000412292, +inf) r5 (-inf, 0.000651859) (-0.000651859, 0.000583821) (0.000583821, 0.00176257) (0.00176257, +inf) r6 (-inf, 1.38138e005) (1.38138e-005, 0.00072438) (0.00072438, 0.00077608) (0.00077608, +inf) r7 (-inf, 1.46802e-005) (1.46802e-005, 6.811e-005) (6.811e-005, 0.000260608) (0.000260608, +inf) r8 (-inf, 4.45802e-005) (4.45802e-005, 6.1332e-005) (6.1332e-005, 0.000253627) (0.000253627, +inf) r9 (-inf, -0.00032564) (-0.00032564, 0.00129406) (0.00129406, 0.00412803) (0.00412803, +inf) r10 (-inf, -0.00032564) (-0.00032564, 0.00170486) (0.00170486, 0.00355543) (0.00355543, +inf) r11 (-inf, 0.00308829) (0.00308829, 0.00355161) (0.00355161, 0.00430769) (0.00430769, +inf) r12 (-inf, 0.00169722) (-0.00169722, 3.00724e-005) (3.00724e-005, 0.000163319) (0.000163319, +inf) r13 (-inf, 0.00131974) (0.00131974, 0.00255195) (0.00255195, 0.00265304) (0.00265304, +inf) r14 (-inf, -0.000937993) (-0.000937993, 0.000271131) (0.000271131, 0.00183654) (0.00183654, +inf) r15 (-inf, 9.4422e(9.44223e-006, (7.94461e-005, (0.000744896, 45 006) 7.94461e-005) 0.000744896) +inf) r16 (-inf, 3.03552e005) (3.03552e-005, 0.000232274) (0.000232274, 0.000290301) (0.000290301, +inf) r17 (-inf, 6.49981e006) (6.49981e-006, 0.000152804) (0.000152804, 0.000620383) (0.000620383, +inf) r18 (-inf, 6.4581e006) (6.45818e-006, 1.90195e-005) (1.90195e-005, 0.000153841) (0.000153841, +inf) r19 (-inf, 5.4507e-005) (5.45079e-005, 5.56532e-005) (5.56532e-005, 0.000473715) (0.000473715, +inf) r20 (-inf, -0.000115164) (-0.000115164, 5.14362e-005) (5.14362e-005, 0.00121164) (0.00121164, +inf) r21 (-inf, -0.000148011) (-0.000148011, 4.68013e-005) (4.68013e-005, 0.00120903) (0.00120903, +inf) r22 (-inf, 1.16457e005) (1.16457e-005, 0.000203684) (0.000203684, 0.00417805) (0.00417805, +inf) r23 (-inf, 7.0378e007) (7.0378e-007, 1.22549e-005) (1.22549e-005, 0.000117764) (0.000117764, +inf) r24 (-inf, 1.10402e005) (1.10402e-005, 0.000426896) (0.000426896, 0.00416212) (0.00416212, +inf) r25 (-inf, 1.2036e005) (1.2036e-005, 2.53061e-005) (2.53061e-005, 0.000161758) (0.000161758, +inf) inf. = infinite. next, we construct the reducts. because there are correlations between the explanatory variables, the program produces many reducts. specifically, the program produced 18,241 reducts with and between 6 to 12 variables. importantly, there are no core elements; there is no variable that is essential for the classification or is more relevant than any other. the frequency of appearance of each variable is shown in table 3. table 3: frequency of appearance of each variable in the reducts the program produced 18,241 reducts with and between 6 to 12 variables variable r1 r2 r3 r4 r5 r6 r7 frequency 23.78% 30.55% 42.46% 23.26% 26.31% 29.58% 24.89% variable r8 r9 r10 r11 r12 r13 r14 international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 46 frequency 44.01% 32.34% 48.32% 33.71% 23.26% 38.31% 28.20% variable r15 r16 r17 r18 r19 r20 r21 frequency 31.96% 31.96% 24.75% 25,60% 24.33% 28.39% 30.72% variable r22 r23 r24 r25 frequency 39.72% 37.99% 34.67% 29.52% we selected 3 of all possible reducts. the selection criteria are the following: first, the reduct should contain the smallest possible number of variables; second, the variables should present a high frequency of appearance in the different reducts; and finally, they should be formed by the smallest number of ratios for each category considered. the ratios belonging to each of the reducts are shown in table 4. table 4: selected reducts the ratios belonging to each of the reducts reducts variables 1 {r3,r10,r13,r17,r22,r25} 2 {r3,r10,r13,r14,r17,r20,r25} 3 {r3,r11,r13,r14,r15,r23}. the first 2 reducts have been chosen because they include the ratios r3, r10 and r13, the quotients that have high percentages of appearance and contain a ratio in each of the categories studied. the third reduct was chosen because it presents ratios from all of the categories. we limited our search to those reducts formed from a maximum of 7 ratios (one more than the number of categories considered). with each of the 3 reducts, decision rules were generated using the lem2 procedure;7 these are shown in table 5. the following points can be noted regarding the items in the table. we can observe that the number of rules varies from one reduct to another; we find that 24, 28 and 22 rules are required to classify 100% of the observations correctly. the first reduct is the one that 7 chan et al., (1994). 47 table 5: selected reducts and corresponding decision rules in all sub-tables below the decision rules are generated using the lem2 procedure. table 5.1: reduct 1: ratios r3, r10, r13, r17, r22, and r25 rules of classification correct classification (d1 = 0): healthy 1. (r3 = 2) & (r13 = 3) & (r17 = 1) & (r25 = 3) 9.43% 2. (r3 = 0) & (r10 = 3) 26.42% 3. (r10 = 2) & (r17 = 2) & (r22 = 1) & (r25 = 3) 3.77% 4. (r17 = 0) 33.96% 5. (r10 = 3) & (r17 = 2) & (r22 = 2) 5.66% 6. (r3 = 2) & (r13 = 1) & (r25 = 2) 11.32% 7. (r3 = 0) & (r17 = 1) 13.21% 8. (r10 = 1) & (r13 = 3) & (r22 = 3) 3.77% 9. (r3 = 0) & (r17 = 3) 5.66% 10. (r3 = 2) & (r13 = 0) & (r25 = 2) 5.66% 11. (r22 = 0) 37.74% (d1 = 1): bankrupt 12. (r10 = 1) & (r22 = 2) & (r25 = 3) 49.06% 13. (r10 = 2) & (r17 = 2) & (r22 = 2) & (r25 = 3) 16.98% 14. (r3 = 2) & (r17 = 3) & (r22 = 2) 9.43% 15. (r13 = 2) 9.43% 16. (r3 = 3) & (r17 = 3) 24.53% 17. (r3 = 2) & (r13 = 3) & (r25 = 2) 5.66% 18. (r25 = 1) 3.77% 19. (r3 = 2) & (r13 = 1) & (r25 = 3) 15.09% 20. (r3 = 0) & (r17 = 2) & (r25 = 2) 1.89% 21. (r3 = 2) & (r17 = 2) & (r22 = 3) 3.77% 22. (r3 = 1) & (r25 = 3) 7.55% 23. (r3 = 2) & (r10 = 3) & (r13 = 3) & (r17 = 2) & (r22 = 1) 1.89% 24. (r3 = 3) & (r25 = 2) 3.77% international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 48 requires the fewest rules to identify the bankrupt companies. we can see also that the first reduct is the one that has the rules with the greatest power of classification. thus, the rules 11 [(r22 = 0) => (d1 = 0)] and 12 [(r10 = 1) & (r22 = 2) & (r25 = 3) => (d1 = 1)] of the first reduct classify, respectively, 37.74% of the bankrupt companies and 49.06% of the healthy ones. no other individual rule of the second or third reducts reaches such high percentages. the above 2 rules tell us, first, that if the value of the napierian logarithm of the profitability ratio r22 (pre-tax profits / net worth) is less than 1.16457e-005, then the company must be classified as bankrupt. second, if the value of the napierian logarithm of the ratio of indebtedness r10 (financial costs /gross profits) is between 0.00032564 and 0.00170486, that of the profitability ratio r22 (pre-tax profits / net worth) is between 0.000203684 and 0.00417805 and that of the ratio r25 (gross profits / total assets) is greater than 0.000161758, the company must be classified as healthy. table 5.2: reduct 2: ratios r3, r10, r13, r14, r17, r20, and r25 rules of classification correct classification (d1 = 0): healthy 1. (r13 = 3) & (r14 = 1) & (r17 = 1) & (r25 = 3) 11.32% 2. (r10 = 3) & (r25 = 2) 22.64% 3. (r10 = 3) & (r14 = 2) 9.43% 4. (r10 = 0) 16.98% 5. (r3 = 2) & (r14 = 0) & (r17 = 2) 5.66% 6. (r10 = 2) & (r14 = 3) 1.89% 7. (r10 = 3) & (r13 = 0) 15.09% 8. (r3 = 0) & (r13 = 3) 26.42% 9. (r3 = 3) & (r13 = 0) & (r17 = 2) 1.89% 10. (r20 = 0) 32.08% 11. (r3 = 0) & (r17 = 3) 5.66% 12. (r14 = 1) & (r20 = 1) & (r25 = 2) 9.43% 13. (r13 = 0) & (r14 = 0) 16.98% 14. (r10 = 2) & (r13 = 3) & (r20 = 1) 3.77% 49 15. (r20 = 3) 1.89% 16. (r10 = 3) & (r20 = 2) 1.89% (d1 = 1): bankrupt 17. (r10 = 1) & (r17 = 3) & (r20 = 2) 28.30% 18. (r10 = 2) & (r14 = 1) & (r17 = 2) & (r20 = 2) 15.09% 19. (r3 = 2) & (r10 = 1) & (r20 = 2) 20.01% 20. (r3 = 2) & (r13 = 1) & (r25 = 3) 15.09% 21. (r13 = 2) 9.43% 22. (r10 = 1) & (r14 = 2) 32.08% 23. (r3 = 1) & (r25 = 3) 7.55% 24. (r3 = 0) & (r13 = 1) & (r17 = 2) 3.77% 25. (r3 = 2) & (r13 = 3) & (r25 = 2) 5.66% 26. (r3 = 3) & (r13 = 1) 15.09% 27. (r14 = 2) & (r17 = 3) 26.42% 28. (r3 = 2) & (r10 = 3) & (r13 = 3) & (r14 = 1) & (r17 = 2) & (r20 = 1) 1.89% table 5.3: reduct 3: ratios r3, r11, r13, r14, r15, and r23 rules of classification correct classification (d1 = 0): healthy 1. (r3 = 0) & (r11 = 2) 26.42% 2. (r11 = 3) 26.42% 3. (r15 = 1) & (r23 = 3) 3.77% 4. (r3 = 2) & (r13 = 1) & (r14 = 0) 3.77% 5. (r23 = 0) 47.17% 6. (r13 = 0) & (r23 = 1) 3.77% 7. (r11 = 2) & (r13 = 3) & (r15 = 2) & (r23 = 2) 5.66% 8. (r15 = 3) & (r23 = 2) 7.55% 9. (r14 = 0) & (r15 = 3) 5.66% international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 50 10. (r3 = 2) & (r11 = 0) & (r14 = 3) 7.55% 11. (r13 = 1) & (r23 = 1) 1.89% 12. (r11 = 0) & (r13 = 1) & (r14 = 1) & (r23 = 2) 1.89% (d1 = 1): bankrupt 13. (r13 = 1) & (r15 = 2) & (r23 = 3) 20.75% 14. (r3 = 2) & (r15 = 2) & (r23 = 3) 30.19% 15. (r3 = 2) & (r11 = 2) & (r15 = 1) & (r23 = 2) 7.55% 16. (r11 = 0) & (r14 = 2) 30.19% 17. (r3 = 3) & (r15 = 2) 30.19% 18. (r11 = 1) & (r23 = 2) 11.32% 19. (r13 = 2) 9.43% 17. (r3 = 3) & (r15 = 2) 30.19% 18. (r11 = 1) & (r23 = 2) 11.32% 19. (r13 = 2) 9.43% 20. (r3 = 1) & (r14 = 3) 1.89% 21. (r3 = 1) & (r15 = 2) 5.66% 22. (r3 = 2) & (r11 = 2) & (r13 = 3) & (r23 = 1) 1.89% to verity these classification results, we performed a cross validation with 10 passes. we present the results for each reduct in table 6. the first and third reducts are clearly more robust than the second, with respect to percentages of correct classification. the third reduct, in addition to meeting the previously imposed requirements, is the one that presents fewest type i and ii errors (13.42% and 11.18%, respectively, for the validation sample). given the above results, we are thus able to confirm that the rough set methodology leads to good results in the classification of healthy and bankrupt companies and indicates which variables are the most relevant of those considered. thus, by applying these rules to new credit operations, a bank would be able to detect possible defaults. 51 table 6: percentages of correct classification this table includes the results of a cross validation with 10 passes. reduct correct classification: bankrupt correct classification: healthy correct classification: total r3,r10,r13,r17,r22,r25 83.98% 87.90% 86.00% r3,r10,r13,r14,r17,r20,r25 66.67% 84.83% 76.27% r3,r11,r13,r14,r15,r23 86.58% 88.82% 88.82% b. comparison with the discriminant analysis as mentioned above, discriminant analysis was the first technique that was widely used. it still remains the most frequently utilized technique for measuring company insolvency. therefore, we have compared the results of the rough set methodology with those produced by the discriminant analysis. we used the same set of data to perform this comparison8. first, we utilized box's mstatistic to test the equality of the variance-covariance matrices between the two groups. from the results, we can accept this hypothesis at a 5% level of significance. the discriminant function (obtained using the ascending stepwise method, with snedecor's f criterion for entry between 0.06 and 0.09) and the statistics associated with the model are shown in table 7. table 7: discriminant functions and statistics of the models this table reports information on the discriminant analysis. the discriminant function is obtained using the ascending stepwise method, with snedecor's f criterion for entry between 0.06 and 0.09. discriminant function (discriminant canonical function with standardized coefficients) wilks' lambda p-value z = 0.554 r4 + 0.769 r6 – 0.337 r9 0.691 <0.0000 base on the avove table, we can deduce that, according to the discriminant analysis, the liquidity and indebtedness ratios are the most relevant items for determining the possible 8the napierian logarithms of the data were used, and their values were typified. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 52 insolvency of a company. the liquidity ratio positively affects a company’s probability of being classified as healthy. with respect to the ratios of indebtedness, their influence will depend on their definition. the ratio r6 (net worth / total liabilities) has a positive influence, while the ratio r9 (financial costs / (gross profits + provision for amortization)) has a negative influence on the probability of the company in question being considered healthy. table 8 gives the results of the classification with the discriminant function applied to the original sample and validated using the cross procedure. table 8: percentages of correct classification base on discriminant analysis this table reports the results of the classification with the discriminant function. the original sample is used. the results are validated using the cross procedure. correct classification: bankrupt correct classification: healthy correct classification: total original sample 71.7% 83.0% 77.4% validation sample 71.7% 83.0% 77.4% from tables 6 and 8 together, it can be deduced that the rough set methodology is a more useful tool than discriminant analysis for the classification of the defaulting companies in the database considered for our empirical application. with the correct choice of reducts, only 6 of the 25 variables considered in the analysis are required to correctly classify 100% of the original observations. furthermore, more validation samples are correctly classified, giving type i and ii errors of 13.42% and 11.18%, respectively, compared to the errors of 28.3% and 17%, respectively, that are given by applying discriminant analysis and utilizing the forward method for the twenty-five variables. 5.3. comparison with the logit analysis because logit analysis is a widely used technique for categorizing two groups, we have compared the results obtained using the rough set methodology with the results obtained using the logit model. the practical benefits of the logit methodology are that it does not require the restrictive assumptions of the discriminant analysis. 53 in the logit model, we applied a statistical forward stepwise selection procedure on the selected variables. the logit model function resulting from our sample of companies is shown in table 9. the wald test for each of the predictors is statistically significant. additionally, the logit-likelihood test is statistically significant. table 9: logit model results this table includes the details of applying the logit model. a statistical forward stepwise selection procedure is employed. coefficient (β) standard error wald gl sig. exp(β) r4 1.271 .602 4.462 1 .035 3.566 r6 4.671 1.851 6.370 1 .012 106.765 r8 -9.568 5.268 3.299 1 .069 .000 r21 12.630 5.589 5.106 1 .024 305642.138 constant .002 .492 .000 1 .996 1.002 z = 1.271 r4 + 4.671r6 – 9.568r8 + 12.630r21 + 0.002 table 10 shows the results of the designation of healthy and failed companies. based on tables 6 and 10, it can be deduced that the rough set methodology is more useful than the logit model over the sampl of companies considered. table 10: the result of the classification between healthy and failed companies (logit model). forecast group in which the company falls total failed healthy failed 81,6% 18,4% 100% healthy 26,5% 73,5% 100% overall correct percentage 77,6% international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 54 6. conclusion in this study we present an alternative methodology to the classic discriminant analysis and logit model for determining the variable(s) that serve to explain the failure of a company to meet its debt repayments as the first step in determining the probability of default (pd). basing our arguments on a sample of sound and bankrupt companies and on a set of 25 financial ratios that are potential explanatory factors for the defaults occurring in the sample, we have shown that the rough set methodology can be a valid alternative to discriminant analysis and to the logit model when there is a need to classify objects into two different classes. in addition to obtaining acceptable percentages of correct classification using rough sets, there is no need to assume any type of prior statistical behavior of the variables involved in the classification, unlike discriminant analysis, which requires normality in the distributions and equality in the variance-covariance matrices. furthermore, the variables are included as they are presented, with no need for any transformation. among the more significant advantages of this methodology is that it eliminates redundant information, and it expresses the dependencies between the variables considered and the results of the classification through decision rules for which the language is closer to that normally utilized by the experts. ___________________________________________________________ authors information: reyes samaniego medina is the corresponding author and is with pablo de olavide university, department of financial economics and accounting, carretera de utrera, km. 1 (41013) seville, spain. tel.: +34 954 34 98 45. e-mail address: rsammed@upo.es. m. josé vázquez cueto, departamento de economía aplicada iii, universidad de sevilla (41018) sevilla, spain (españa). mailto:rsammed@upo.es references altman, e.i. (1968). financial ratios, discriminate analysis and the prediction of corporate bankruptcy. journal of finance.september 589-609. altman, e.i. (1993). corporate financial distress and bankruptcy. john wiley. new york. altman, e.i., sabato, g. (2007). modelling credit risk for smes: evidence from the u.s. market. abacus, vol. 43, 3, 332-357 ahn, b.s., cho, s.s., kim, c.y. (2000).the integrated methodology of rough set theory and artificial neural network for business failure prediction. expert systems with applications, 18. basle committee on banking supervision (2001).the internal ratings-based approach. consultive document. supporting document to the new basle capital accord, january. basel committee on banking supervision (1999). credit risk modelling: current practices and applications. april. beaver, w. h. (1966). financial ratios as predictors of failure. journal of accounting research 4, 71-127 cabedo,j., reverte, j.a., tirado, j.m. (2004). riesgo de crédito y recursos propios mínimos en entidades financieras. revista europea de dirección y economía de la empresa, 13, 2. chan. c., grzymala-busse, j. (1994).on the two local inductive algorithms: prism and lem2. foundations of computing and decision sciences, 19, 185–204. comité de supervisión bancaria de basilea (2004). convergencia internacional de medidas y normas de capital. june. courtis, j.k. (1978). modelling financial ratios: categories framework. journal of business, finance and accounting, 5 (4). dambolena, i.g., khoury, s.j. (1980). ratio stability and corporate failure. the journal of finance, 35. daubie, m.; leveck, p.; meskens, n. (2002). a comparison of the rough sets and recursive partitioning induction approaches: an application to commercial loans. international transactions in operational research 9, 681-694. dimitras, a.i., zanakis, s.h., zopounidis, c. (1996). a survey of business failures with an emphasis on prediction methods and industrial applications. european journal of operational research, 90. dimitras, a.i., slowinski, r., susmaga, r, zopounidis, c. (1998). business failure prediction using rough sets. european journal of operational research, 114. eisenbeis, r.a. (1977). pitfalls in the application of discriminant analysis in business and economics. journal of finance, 32, 875-900. fernandez, j.e. (2005). corporate credit risk modeling: quantitative rating system and probability of default estimation. ssrn: http://ssrn.com/abstract=722941 komorowski, z. (1999). rough sets: a tutoriain rough-fuzzy hybridization a new trend in decision makin. s.k. pal and a. skowron, eds., 3-98, springer-verlag singapore pte ltd. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 3 56 laitinen, e.k. (1991). financial ratios and different failure processes. journal of business finance and accounting,18/5. ohlson, j.a. (1980). financial ratios and the probabilistic prediction of bankruptcy. journal of accounting research (spring) 109-131. pawlak z. (1982). rough sets. int. j. computer and information sci., 11,341-356 pawlak,z.; slowinski,r.(1994). decision analysis using rough sets. international transactions in operational research 1, 107-114. rodriguez, m., díaz, f. (2005). la teoría de los rough sets y la predicción del fracaso empresarial. diseño de un modelo para pymes. revista de la asociación española de contabilidad y administración de empresas, 74.36-39 segovia, m.j., gil, j.a., vilar, l., heras, a.j. (2003). la metodología rough set frente al análisis discriminante en la predicción de insolvencia en empresas aseguradoras. anales del instituto de actuarios españoles, 9. trujillo, a., martin, j.l. (2004). el rating y la fijación de precios en préstamos comerciales: aplicación mediante un modelo logit. revista europea de dirección y economía de la empresa, 13. nº. 2. trucharte, c., marcelo, a. (2002). un sistema de clasificación (rating) de acreditados. estabilidad financiera, 2. wilson, t. (1997). portfolio credit risk. risk magazine, sept., 111-117. zavgren, c.v.(1983). the prediction of corporate failure: the state of the art. journal of financial literature 2, 1-37. zmijewski, m.e. (1984). methodological issues related to the estimation of financial distress prediction models. studies on current econometric issues in accounting research. the international journal of banking and finance, vol. 7. number 2: 2010: 19-35 19 market meltdown and the propagation mechanism of contagion dilip k. ghosh and dipasri ghosh american university cairo and rutgers university, and california state university (fullerton) ________________________________________________________________ abstract world economy came on a tailspin because of market meltdown and the propagation of contagion triggered by recession, starting mostly from the us economy. this work highlights the grim developments in the � nancial sectors and the real sectors world-wide, and then an attempt is made to highlight the propagation mechanism of infective contagion. theoretical structures of such interconnected are showcased through various analytical vehicles. the indices of sensitivities and dispersions are measured in mathematical terms, and in that sense a new analytical framework is presented. however, empirical evaluations of the propagation mechanism remain un� nished because of the dearth of data keywords: market meltdown, contagion, subprime loans, troubled asset relief program (tarp), securitization jel classi� cation: g1, g2 ________________________________________________________________ 1. introduction from 2007 until the present time, the economy of the united states (us) and almost every other economy of the globe have been suffering a humongous recession approaching the neighborhood of the great depression of 20th century. the magnitudes and proportions of downward spirals overwhelmed the entire world. it has been recognized that the american sub-prime loans and reckless disregard for sanity and discipline in lending and liquidity management have bought the world on its knees. the question, of course, has to go beyond the root cause of recession world-wide. it has been the severe contagion of this young century spreading all over and everywhere, and everyone recognizes that. we must attempt to seek answer to the question as to how this us contagion got propagated globally. in section 2, this paper presents the statistical measures of what happened, who got infected, and how the market meltdown affected individuals, institutions, governments, regulatory bodies, and so on. then in ijbf 1 ghosh and ghosh: market meltdown and the propagation mechanism of contagion produced by the berkeley electronic press, 2010 20 the international journal of banking and finance, vol. 7. number 2: 2010: 19-35 section 3, how markets across nations are integrated (or segmented) is examined. in this context, the existing literature is revived. then exposition is given on the interconnectedness of national economies, which and attempts to show how the market meltdown made the � nancial pandemic propagated on the global scale. in section 4, highlights the dif� culty of not giving the test results of our paradigm because of the paucity of data that are needed to verify and conclude the theoretical structure on the sensitivities of dispersions of the contagion. 2. market meltdown and the us economy market meltdown has been the buzz word in the world of � nance. crisis has gripped the globe. financial architecture has collapsed everywhere. stock markets have plunged, and dow jones industrial average had come down to its 1997 level. many banks and major � nancial institutions have disappeared, and asset values of people have declined beyond belief. the worldwide scramble to appropriate wealth through “� nancial manipulation” has been the driving force behind this crisis. it is the source of economic turmoil and social devastation. totally deregulated � nancial environment characterized by extensive speculative trade has indeed wrought havoc to the framework of � nancial stability on a global scale. in the wake of the 1987 stock market meltdown, the us treasury was admonished by wall street not to meddle in � nancial markets. free of government encroachment, the new york and chicago exchanges were invited to establish their own regulatory procedures. financial deregulation in the us has created an environment of unbounded greed and deception out of which came the major tsunami via subprime mortgage loan. the subprime loans are the genesis of the current crisis. subprime loans to mortgage holders on real estates have put the world economy on a tailspin. stock markets across the nations have taken nose dives. the questions must then be: what are subprime loans and how have these loans wrought havoc around the world? let us answer these questions to start off with the issue of � nancial chaos, or as often noted in the � nancial press, as crisis, and its spillover effects in the international markets. prime refers to prime rate at which banks lend money to their most preferred corporate customers, and it is the lowest rate of interest in the credit market. other borrowers can borrow at prime plus alpha where alpha is any value above zero. obviously then subprime loan is a loan with interest cost below prime rate. although it is not necessarily below prime rate, it has been given the appearance of cheap cost of loans, and made the unworthy home buyers attracted to such seemingly low cost credit. such loans were extended to home-buyers who even had signi� cantly below average creditworthiness mostly under adjustable mortgage rates. when the adjustment started becoming effective, mortgagees began serially delinquent. the loans became non-performing, and banks started foreclosing on the homes with lien on the properties, and the credit market dried up. it is not the end of the saga; the spillover effects have become global in the era of globalization. 2 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 2 http://epublications.bond.edu.au/ijbf/vol7/iss2/2 market meltdown and the propagation mechanism of contagion: 19-35 21 the subprime mortgage problem has been in sharp rise through home foreclosures which started in the us in late 2006 and became what has been often referred to as a global � nancial crisis during 2007 and 2008. the crisis began with the bursting of the housing bubble in the us and high default rates on subprime and other adjustable rate mortgages (arm) made to higher-risk borrowers with lower income or lesser credit history than "prime" borrowers. loan incentives and a long-term trend of rising housing prices encouraged borrowers to assume mortgages, believing they would be able to re� nance at more favorable terms later. however, once housing prices started to drop moderately in 2006-2007 in many parts of the us, re� nancing became more dif� cult. defaults and foreclosure activity increased dramatically as arm interest rates reset higher. during 2007, nearly 1.3 million us housing properties were subject to foreclosure activity, up 79% versus 2006. estimate were � oating that sub-prime defaults would reach amounts somewhere between us $200-$300 billion. the mortgage lenders that retained credit risk (the risk of payment default) were the � rst to be affected, as borrowers became unable or unwilling to make payments. major banks and other � nancial institutions around the world have reported losses of approximately us $170 billion as of february 2008, as cited below. due to a form of � nancial engineering, called securitization, many mortgage lenders had passed the rights to the mortgage payments and related credit/default risk to third-party investors via mortgage-backed securities (mbs) and collateralized debt obligations (cdo). corporate, individual and institutional investors holding mbs or cdo faced signi� cant losses, as the value of the underlying mortgage assets declined. stock markets in many countries declined signi� cantly. this paper now delineates the evolving situations, mostly in the us economy, and as they are re� ected in the stock markets in the us. 2.1 subprime loans the widespread dispersion of credit risk and the unclear impact on � nancial institutions have caused lenders to reduce lending activity or to make loans at higher interest rates. similarly, the ability of corporations to obtain funds through the issuance of commercial paper was impacted. this aspect of the crisis is consistent with a credit crunch. the liquidity concerns drove central banks around the world to take action to provide funds to member banks to encourage the lending of funds to worthy borrowers and to re-invigorate the commercial paper markets. the subprime crisis also places downward pressure on economic growth, because fewer or more expensive loans decrease investment by businesses and consumer spending, which drive the economy. a separate but related dynamic is the downturn in the housing market, where a surplus inventory of homes has resulted in a signi� cant decline in new home construction and housing prices in many areas. this also places downward pressure on growth. with interest rates on a large number of subprime and other adjustable rate mortgages (arm) due 3 ghosh and ghosh: market meltdown and the propagation mechanism of contagion produced by the berkeley electronic press, 2010 22 the international journal of banking and finance, vol. 7. number 2: 2010: 19-35 to adjust upward during the 2008 period, us legislators and the us treasury department have been taking action. a systematic program to limit or defer interest rate adjustments was implemented to reduce the impact. in addition, lenders and borrowers facing defaults had been encouraged to cooperate to enable borrowers to stay in their homes. the risks to the broader economy created by the � nancial market crisis and housing market downturn were primary factors in the january 22, 2008 decision by the us federal reserve to cut interest rates and the economic stimulus package signed by president bush on february 13, 2008. both actions were designed to stimulate economic growth and inspire con� dence in the � nancial markets. the value of us sub-prime mortgages was estimated at $1.3 trillion as of march 2007, with over 7.5 million � rst-lien sub-prime mortgages outstanding. approximately 16% of subprime loans with arm were 90 days delinquent or in foreclosure proceedings as of october 2007, roughly triple the rate of 2005. by january of 2008, the delinquency rate had risen to 21%. subprime arms only represented 6.8% of the loans outstanding in the us, yet they represent 43.0% of the foreclosures started during the third quarter of 2007. a total of nearly 446,726 us household properties were subject to some sort of foreclosure action from july to september 2007, including those with prime, alt-a, and subprime loans. this was 2-fold of the 223,000 properties in the year earlier, and 34% higher than the 333,627 in the prior quarter. this increased to 527,740 during the fourth quarter of 2007, an 18% increase versus the prior quarter. for all of 2007, nearly 1.3 million properties were subject to foreclosure � lings, up 79% versus 2006. 2. 2 risks of the subprime loans as a result extensive subprime loans, a few types of risk – credit risk, liquidity risk, asset valuation risk, and risk on relying on � nancial institutions – came into being, and this 4-tuple risk shook the entire edi� ce of the � nancial world. traditionally credit risk was assumed by the banks giving mortgage loans, but in recent years, owing to innovations in securitization, credit risk was being shared more broadly with investors, because the rights to these mortgage payments had been repackaged into a variety of complex investment vehicles, generally categorized as mortgage-backed securities (mbs) or collateralized debt obligations (cdo). the following diagram (figure 1) depicts what transpired because of the securitization structure of borrowing, lending and investing. the small spikes within the circle represent the borrowers (home mortgage holders) from the banks, represented in the cluster (in the center of the circle) that created and sold mortgage-backed securities (mbs), and collateralized debt obligations (cdo). the spikes going beyond the circle represent the investors in the mbs and cdo, longer ones are the international investors and shorter ones are domestic investors. originally, the home mortgagees with subprime loans were paying mortgages, but when the reset started, they could not pay their dues, and this reality came into being when the income in� ows did not 4 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 2 http://epublications.bond.edu.au/ijbf/vol7/iss2/2 market meltdown and the propagation mechanism of contagion: 19-35 23 get to the foreign and domestic investors. all spikes became insolvent and got crushed in the windmill of dire � nancial distress. to manage their risk, mortgage originators (banks or mortgage lenders) created separate legal entities, called special-purpose entities (spe), to both assume the risk of default and issue the mbs. the banks effectively sold all the mortgage assets (that is, banking account receivables, which were the rights to receive the mortgage payments) to these spe. in turn, the spe then sold the mbs to the investors. the mortgage assets in the spe became the collateral. another variant of spe – known as structured investment vehicles (siv) – and other corporations soon realized the devastation in the markets on commercial papers, causing illiquidity an unmanageable situation. the amount of commercial paper issued as of october 18, 2007 dropped by 25%, to $888 billion, from the august 8 level. along with or because of liquidity risk and credit risk, asset market prices fell sharply, and threw the � nancial markets into chaos and disarray, where the domino effect was observed. figure 1: securitization structure of borrowing, lending and investing subprime borrowing was a major contributor to an increase in home ownership rates and the demand for housing. the overall us homeownership rate increased from 64% in 1994 (about where it was since 1980) to a peak in 2004, with an all time high of 69.2%. this demand helped fuel housing price increases and consumer spending. between 1997 and 2006, american home prices increased by 124%. some homeowners used the increased property value c a abc 5 ghosh and ghosh: market meltdown and the propagation mechanism of contagion produced by the berkeley electronic press, 2010 24 the international journal of banking and finance, vol. 7. number 2: 2010: 19-35 experienced in the housing bubble to re� nance their homes with lower interest rates and take out second mortgages against the added value to use the funds for consumer spending. us household debt as a percentage of income rose to 130% during 2007, versus 100% earlier in the decade. misrepresentation of loan application data is another contributing factor. like predatory lending, predatory borrowing became unmanageably signi� cant. it had been estimated that almost three-fourth of the delinquency had fraudulent misrepresentations on their original loan applications. mortgage fraud increased by 1411% between 1997 and 2005, according to the us treasury department report. many analysts have claimed that government policy actually encouraged the development of the subprime debacle through legislation like the community reinvestment act, which they say forces banks to lend to otherwise uncreditworthy consumers. any sane and far-sighted person can criticize the repeal of glass-steagall act as contributing to the subprime meltdown. a taxpayer-funded government bailout related to mortgages during the savings and loan crisis in late 1989 may have created a moral hazard and acted as encouragement to lenders to make similar higher risk loans. some have argued that, despite attempts by various us states to prevent the growth of a secondary market in repackaged predatory loans, the treasury department’s of� ce of the comptroller of the currency, at the insistence of national banks, struck down such attempts as violations of federal banking laws. in response to a concern that lending was not properly regulated, the house and senate are both considering bills to regulate lending practices. some industry of� cials said that federal reserve bank of new york involvement in the rescue of longterm capital management in 1998 would encourage large � nancial institutions to assume more risk, in the belief that the federal reserve would intervene on their behalf. here is a brief view of the write-down on the value of loans, mbs, cdo and bankruptcies that were already witnessed. table 1 exhibits the data. table 1: write-downs on the value of loans, mbs and cdos company business type loss (billion $) citigroup investment bank $24.1 bln merrill lynch investment bank $22.5 bln ubs ag investment bank $18.7 bln morgan stanley investment bank $10.3 bln crédit agricole bank $4.8 bln hsbc bank $17.2 bln bank of america bank $5.28 bln cibc bank $3.2 bln deutsche bank investment bank $3.1 bln barclays capital investment bank $3.1 bln (continued) 6 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 2 http://epublications.bond.edu.au/ijbf/vol7/iss2/2 market meltdown and the propagation mechanism of contagion: 19-35 25 company business type loss (billion $) bear stearns investment bank $2.6 bln rbs bank $3.5 bln washington mutual savings and loan $2.4 bln swiss re re-insurance $1.07 bln lehman brothers investment bank $2.1 bln lbbw bank $1.1 bln jp morgan chase investment bank $2.9 bln goldman sachs investment bank $1.5 bln freddie mac mortgage gse $3.6 bln credit suisse bank $3.7 bln wells fargo bank $1.4 bln wachovia bank $3.0 bln rbc bank $0.360 bln fannie mae mortgage gse $0.896 bln mbia bond insurance $3.3 bln hypo real estate bank $0.580 bln ambac financial group bond insurance $3.5 bln commerzbank bank $1.1 bln société générale investment bank $3.0 bln bnp paribas bank $0.870 bln westlb bank $1.37 bln american international group insurance $11.1 bln bayernlb bank $2.8 bln natixis bank $1.75 bln countrywide mortgage bank $1.0 bln dz bank bank $2.1 bln businesses � ling for bankruptcy business type date new century financial subprime lender april 2, 2007 american home mortgage mortgage lender august 6, 2007 sentinel management group investment fund august 17, 2007 ameriquest subprime lender august 31, 2007 netbank on-line bank september 30, 2007 terra securities securities november 28, 2007 american freedom mortgage, inc. subprime lender january 30, 2007 2.3 us government and central bank in the wake of the almost free-falling economies all around the world in general, and the economy of the us in particular, president george w. bush table 1: write-downs on the value of loans, mbs and cdos 7 ghosh and ghosh: market meltdown and the propagation mechanism of contagion produced by the berkeley electronic press, 2010 26 the international journal of banking and finance, vol. 7. number 2: 2010: 19-35 signed into law on february 13, 2008 an economic stimulus package of $168 billion, mainly in the form of income tax rebates, to help stimulate economic growth. the troubled asset relief program (tarp) which was a program of the us government to purchase assets and equity from � nancial institutions to strengthen its � nancial sector was created. it was the largest component of the government’s measures in 2008 to address the subprime mortgage crisis. tarp allowed the us treasury department to purchase or insure up to $700 billion of “troubled” assets. “troubled assets” were de� ned as “(a) residential or commercial mortgages and any securities, obligations, or other instruments that are based on or related to such mortgages, that in each case was originated or issued on or before march 14, 2008, the purchase of which the secretary determines promotes � nancial market stability; and (b) any other � nancial instrument that the secretary, after consultation with the chairman of the board of governors of the federal reserve system, determines the purchase of which is necessary to promote � nancial market stability.” next, under the new presidency of barack obama, the american recovery and reinvestment act of 2009 (arra (pub.l. 111-5) an economic stimulus package enacted by the 111th united states congress in february 2009 appeared to calm and soothe � nancial nerve centers and had indeed controlled the turbulence. the act followed other economic recovery legislation passed in the � nal year of the bush presidency including the economic stimulus act of 2008 and the emergency economic stabilization act of 2008 which created the troubled assets relief program (tarp). these measures are nominally worth $787 billion. the act includes federal tax cuts, expansion of unemployment bene� ts and other social welfare provisions, and domestic spending in education, health care, and infrastructure, including the energy sector. the act also included numerous noneconomic recovery-related items that were either part of longer-term plans. the stock markets all around the world have rebounded with still bumps here and there. european governments and central banks have followed the us lead in the containing contagion. japan has been on board. the fed and other central banks have conducted open market operations to ensure member banks have access to liquidity. these are effectively shortterm loans to member banks collateralized by government securities. central banks have also lowered the interest rates charged to member banks (called the discount rate in the us) for short-term loans. both measures effectively lubricate the � nancial system, in two key ways. first, they help provide access to funds for those entities with illiquid mortgage-backed assets. this helps lenders, spe, and siv avoid selling mortgage-backed assets at a steep loss. secondly, the available funds stimulate the commercial paper market and general economic activity. speci� c responses by central banks are included in the subprime crisis impact timeline. the fed has been utilizing the term auction facility (taf) to provide short-term loans (liquidity) to banks. the fed has increased the monthly amount of these auctions to $100 billion during march 2008, up from $60 billion in prior months. in addition, term repurchase agreements expected to cumulate to $100 billion were announced, which enhanced the ability of � nancial institutions to 8 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 2 http://epublications.bond.edu.au/ijbf/vol7/iss2/2 market meltdown and the propagation mechanism of contagion: 19-35 27 sell mortgage-backed and other debt. the fed indicated that both the taf and repurchase agreement amounts would continue and be increased as necessary. it appears that these measures have created a slow but steady rebound. 3. are international markets integrated? 3.1 conventional approach in this section how much the spill-over effect is evaluated. how did sub-prime loans spread the virus across nations? numerous studies and serious research have shown different conclusions and this paper is more in disagreement on the question as to whether international markets are integrated or segmented. black (1974), stulz (1981a, 1981b), eun and janakiramanan (1986), hietala (1989) have viewed that international capital markets are segmented, and adler (1974), adler and dumas (1983), and stapleton and subramanyam (1977), ignoring exchange risk, examined the implication of market segmentation. french, schwert, and stambaugh (1987) have presented evidence that there is a positive correlation between the conditional expected excess return on the market portfolio and the conditional variance of its return, using generalized autoregressive conditional heteroscedasticity representation, as exposited by bollerslev (1986) and engle (1982) for the excess return on the market with the standard and poor’s 500 index (s&p 500) for the market portfolio. in their sample, it appeared that foreign capital markets are signi� cantly segmented from the us capital markets. the studies of cho, eun and senbet (1986), wheatley (1988), korajczyk and vialett (1990), and harvey (1991) have provided evidence of integration by employing different asset pricing models and using monthly data. by making use of the arbitrage pricing theory (apt), gultekin, gultekin and penati (1989) have established that for � fteen years the us and the japanese markets had been more and more integrated. chan, karolyi, and stulz (1991) have attempted to reexamine and generalize the work of french, schwert, and stambaugh (1987) by assuming that (i) markets are internationally integrated, (ii) investors are optimizers in the mean-variance framework in a common numeraire currency such as the us dollar, and (iii) the aggregate relative risk aversion, , is constant, where it z is the i-th individual’s risk aversion, and it w is the wealth at time t. under these assumptions, they have begun with the following equations: (1) (2) �� �� � n i it n i it wzz 11 / � )(var)( 1 ftmttftmtt rrzrre ��� � � )cov()( 1 mtdtdtt mrzre �� � � it it 9 ghosh and ghosh: market meltdown and the propagation mechanism of contagion produced by the berkeley electronic press, 2010 28 the international journal of banking and finance, vol. 7. number 2: 2010: 19-35 where e t is the expectation operator, r mt the return on market portfolio, r ft the risk-free rate, and var t (r mt �r f ) the variance of the excess return on the market portfolio, r dt and m mt the excess return on the us domestic market portfolio and that of the world market portfolio, while cov stands for the covariance. since the return on the world market portfolio can be expressed as: (3) where w dt is the weight on the excess return on the domestic market portfolio and is r nt the excess return on the non-domestic (foreign) market portfolio, one can rewrite (2) by plugging (3) into it as follows: (4) it is obvious now that if r dt equation (4) is reduced to french et al. relation. however, if the covariance term is zero in (4), it is not the same case that french, schwert, and stambaugh (1989) had examined. chan, karolyi, and stulz (1991) parameterize equation (4) as follows: (5) (6) where h dt and h nt are the variances and h ct is the covariance, which depends upon past returns, lag structure of stochastic terms were introduced through � , and � captured the effects of infrequent trading and non-synchronous trading hours. making use of the general process, modeled after baba, engle, kraft, and kroner (1989), chan et al. reported that their “tests are … supportive of the hypothesis that markets are internationally integrated over sample period”they consider.” blenman (1991), in an interesting theoretical model, consisting of two countries in which one country’s market is perfectly competitive for everyone, including domestic and foreign investors, but the other country’s market is barrier-free for its domestic investors, but quite partitioned for foreign investors, derives a number of useful result in a uni� ed exchange rate regime. it was shown that all investors’ risky assets hold a common factor share, appropriately weighted by their tastes. when they face further restrictions, their asset demands are modi� ed to re� ect those restrictions unless the variance-covariance matrix is block diagonal, and/or the restrictions mutually cancel each other, and/or the total risk tolerance factors are all zero. along these traditional routes some more works are noteworthy. stulz (1995) discussed globalization of capital market and the cost of capital, which under a different fold extend the issue of international connectivity of markets. portes, and rey (1999) examined the determinants of cross-border equity, and pagano, roell, and zechner (2001) looked at the geography of equity listing and ntdtdtdtmt rwrwm )1( ��� � )],(cov)1()var([)( 1 ntdttdtdtdtdtt rrwrwzre ��� � � dtdtddtdctdtdcdtdtdvddt hwhwr ����� ������� �� 2211)1( � ntdtndtnntnctntncnnnvnnt hwhwr ������� �������� ��� 1221111 )1( � 10 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 2 http://epublications.bond.edu.au/ijbf/vol7/iss2/2 market meltdown and the propagation mechanism of contagion: 19-35 29 asked the question: why do companies list abroad? in his presidential address to american finance association, published in the journal of finance (2005), stulz re-examined the limits of � nancial globalization, and threw the issue of market integration or segmentation into further focus. 3.2 investment flow matrix approach following leontief’s celebrated structure, ghosh and khaksari (187) taken us to a new line of analytical framework of � ow structure. bringing that � ow structure alive, one can examine the investment � ows of investors of different countries into different capital markets existing across several nations under the assumption that individual investors are rational and they face no political risks with their investment funds. let ij a be the amount of the investment funds in market i needed to buy one unit investment in market j, x j be the total investment units in market j, and y i be the amount of funds in market i kept exogenously for other purposes. then the available funds balance can be stated as follows: ij n j iji yxax �� � �1 (7) to put (7) in matrix notation, we have (7�) where i is an nn� identity matrix, nnijaa � ][ matrix, x and y are �n element column vectors. the solution of (7�) is given by: yaix .)( 1��� (8) or (8.1) here ij� is the ij-th element of the (i-a) inverse matrix. consider now that a matrix assumes one of the following forms: � � � � � � � � � � � � � � � � � � � � nnnn aaa aa a a ....... 0.. 0.... 0......... 0........00 21 2221 11 1 , � � � � � � � � � � � � � � � � � � � � nn n in a aa aaa a .......00 .. .... .........0 ........ 222 1211 2 � , yxai �� ).( � j n j iji yx .. 1 � � � � � ij 11 ghosh and ghosh: market meltdown and the propagation mechanism of contagion produced by the berkeley electronic press, 2010 30 the international journal of banking and finance, vol. 7. number 2: 2010: 19-35 if 1aa , 1x is independent of other markets, but 2x depends on 1x , and 3x depends on 1x , and 2x , and so on. this sort decomposability of investment � ow structure de� nes the � rst market as the “key” market that drives the rest of the world market. mortgage meltdown in the us with its tentacles on other mortgage-backed securities appeared to have driven world markets down. if, on the other hand, 2aa , then n-th market is obviously the key generator of the global market movements. if a assumes the form of a 3 , we then have a number of subsets of the global markets, each subset being independent of the others. here we � nd that (1, 2), (3,4), ……., (n-1, n) markets are intrasubset integrated, and inter-subset segmented. however, if the � ow matrix is not triangular (upper or lower), the key market can still be captured through the following indices of power of dispersion and sensitivity of dispersion, originally noted by rasmussen (1956): (9.1) (9.2) the index jd de� nes the extent to which an increase in investment � ow in market j is dispersed throughout the global economy. if 1�jd , it signi� es � � � � � � � � � � � � � � � � � � � � � � � � � � � ��� nnnn nnnn aa aa aa aa aa aa a 1, ,11,1 4443 3433 2221 1211 1 ........................00 ......................00 00.............00 00.00. 00........................0 00.......................0 � � � � � � � � � � � � � �� � � � � n j n i ij n i ij j n n d 1 1 2 1 )/1( )/1( � � for nj ��1 � � � � � � � � � � � � � �� � � � � n i n j ij n j ij i n n d 1 1 2 1 )/1( )/1( � � for nj ��1 12 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 2 http://epublications.bond.edu.au/ijbf/vol7/iss2/2 market meltdown and the propagation mechanism of contagion: 19-35 31 that market j draws heavily on the system of markets (and conversely, if 1�jd ). the index id , on the other hand, measures the extent to which the system of markets draws upon market i . if 1�id , it means that the i th market is required to accommodate more investment funds than other markets corresponding to an exogenous increase in investment. these two indices serve well, but they are not without some weaknesses as they are un-weighted. to correct this de� ciency, one can modify, for instance, the sensitivity of dispersion as follows: (10) 3.3 global and national markets: the relationship assume that markets are indexed h and i (where nimh ,....,2,1;,....,2,1 �� ) such that }{}{ ih � , the balance holds for a nation, and for markets i balance holds for the world as a whole. that is, (11) for the world as a whole, and (12) for a nation, )1( ndd �� , where yxai �� ).( � � � )()( )()( dynxxaai dwdnnwnnnn ���� � � � � � � � � � � � � � � �� � � � � � � � � n i n j n j jjij n j n j jjij w i yyn yyn d 1 1 1 2 1 1 /)/1( /)/1( � � � for�� nj ��1 � � hkmmhknni !! � � mmhk nn aa � for mkh ,.......,2,1, � � )( mnmhs nw aa �� },{},{ )()()()( d s dnd h dw xxxx )()( { dh dn yy for )1( ndd �� : national markets. khkh hkhk "#��#� 1;1 !! for is the kronecker delta that is, : national markets. 13 ghosh and ghosh: market meltdown and the propagation mechanism of contagion produced by the berkeley electronic press, 2010 32 the international journal of banking and finance, vol. 7. number 2: 2010: 19-35 from (11) one can get the following: (13) since )( mn � equations in (11) yield ,yvx nn � where equation (13) can be re-expressed as follows: (14) provided the following proportionality holds: following the procedure outlined earlier, one can compute the indices of power and sensitivity of dispersions of investment � ows within this structure of global market network, and determine if markets are integrated or segmented. to support the above theoretical segments with the applications that are undertaken by other researchers, some parallel information may be given. for example, a variant of the above is empirically applied by parhizgari (2002). using equity data on the stock exchanges of seven regions and 18 countries for january 1982� october 2002, parhizgari (2002) shows that substantial degrees of convergence have been achieved across the global stock markets over the noted period. 4. concluding remarks as noted in the introduction, empirical data are scarce and scant. this section needs to be written. this study was not capable of computing the indices that were presented in section 3. it appears from the little we know that markets are substantially integrated with openness in trade transactions and more so by the portfolio investment. the countries seriously affected by the contagion are apparently created by the us and the countries with stock exchanges projecting the powers of securities trade world-wide. the extensive securitization (or the creation of these asset-backed bonds) has eliminated the incentive for the originator of the loan to be credit sensitive. thus, the loss experienced on these loans after securitization became no longer comparable to that experienced prior to securitization (called a ‘‘moral’’ hazard). several trillion dollars investment in asset-backed bonds outstanding in the us was not all from the us investors alone, a big chunk from the overseas institutions and individuals. insurance companies, money managers and banks – nmsrxx nrs d r dn ,........,1,,)( )()( ��� ! , and .1},{ nsmxx s n ��� )()( )()(1)( dnnwdnnnnndn xayaix ��� � � ,1,)(][},{ 1 nsmaivvxx wns n ���� � � )).(()( )()(1)( yvayaix wrs d r nwdnnnnndn ! ��� � � 14 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 2 http://epublications.bond.edu.au/ijbf/vol7/iss2/2 market meltdown and the propagation mechanism of contagion: 19-35 33 in the main – all reached for yields, given the excellent ratings for these bonds. what happens if air pocket is hit? the legacy of alan greenspan has been cast into doubt with senator chris dodd claiming he created the “perfect storm”, a professor at new york university and head of roubini global economics, has said that if the economy slips into recession “then you have a systemic banking crisis like we haven’t had since the 1930s”. on september 7, 2007, the wall street journal reported that alan greenspan has said that the current turmoil in the � nancial markets is in many ways “identical” to the problems in 1987 and 1998. the associated press described the current climate of the market on august 13, 2007, as one where investors were waiting for “the next shoe to drop” as problems from “an overheated housing market and an overextended consumer” are “just beginning to emerge.” marketwatch has cited several economic analysts with stifel nicolaus claiming that the problem mortgages are not limited to the subprime niche saying “the rapidly increasing scope and depth of the problems in the mortgage market suggest that the entire sector has plunged into a downward spiral similar to the subprime woes whereby each negative development feeds further deterioration”, calling it a “vicious cycle” and adding that they “continue to believe conditions will get worse”. as of november 22, 2007, analysts at a leading investment bank estimated losses on subprime cdo would be approximately us$148 billion. as of december 22, 2007, a leading business periodical estimated subprime defaults between us$200-300 billion. as of march 1, 2008 analysts from three large � nancial institutions estimated the impact would be between us$350-600 billion. alan greenspan, the former chairman of the federal reserve, stated: “the current credit crisis will come to an end when the overhang of inventories of newly built homes is largely liquidated, and home price de� ation comes to an end. that will stabilize the now-uncertain value of the home equity that acts as a buffer for all home mortgages, but most importantly for those held as collateral for residential mortgage-backed securities. very large losses will, no doubt, be taken as a consequence of the crisis. but after a period of protracted adjustment, the u.s. economy, and the world economy more generally, will be able to get back to business.” author information: dilip k. ghosh is a professor of finance, editor-in-chief of the international journal of finance. he sits in a number of editorial boards. he is the former director of trinidad and tobago stock exchange and the former kuala lumpur stock exchange professor of finance. currently, he holds hellenic bank association chair professorship. he can be contacted by e-mail: editor@tijof.org, dghosh4@msn.com, tel. 856-424-2262. dipasri ghosh is an assistant professor of finance in the mihaylo college of business and economics at california state university, fullerton. she can be contacted by e-mail: dghosh@fullerton.edu, tel. 657-278-4821. 15 ghosh and ghosh: market meltdown and the propagation mechanism of contagion produced by the berkeley electronic press, 2010 34 the international journal of banking and finance, vol. 7. number 2: 2010: 19-35 references adler, m., (1974). the cost of capital and valuation of a two-country � rm. journal of finance, 29, 119 – 132. adler, m., and dumas, b., (1983). international portfolio selection and corporate � nance: a synthesis. journal of finance, 38, 925 – 984. baba, y., engle, r.f., kraft, d. f. and kroner, k. f., (1989). multivariate simultaneous generalized arch. working paper, university of california, san diego. black, f., (1974). international capital market equilibrium with investment barriers. journal of financial economics, 1, 337352. blenman, l. p. (1991). international asset pricing with incomplete information and market segmentation. paper presented at financial management association meetings on chicago. bollerslev, t., (1986). generalized autoregressive conditional heteroscedasticity. journal of econometrics, 31, 307 – 327. chan, k. c., g. a., karolyi, , and r. m. stulz (1991). global � nancial markets and risk premium on us equity. paper presented at batterymarch finance seminar, sloan school of management. cho, c. d., c.s. eun, and l. w. senbet (1986). international arbitrage theory: an empirical investigation. journal of finance, 41, 313340. engle, r. (1982). autoregressive conditional heteroscedasticity with estimates of the variance of the united kingdom. econometrica, 50, 987-1007. eun, c.s., and s. janakiramanan (1986). a model of international asset pricing model with constraint on the foreign equity ownership. journal of finance, 41, 897 – 914. french, k. r., g. w. schewert, and r. f. stambaugh (1987). expected stock returns and volatility. journal of financial economics, 19, 3 – 30. ghosh, d. k., and s. khaksari (1993). international capital markets: integrated or segmented? international financial markets:integration, ed. stansell, s. r., blackwell, oxford uk and cambridge, usa, 354 – 366. gultekin, n. b., m. n. gultekin, a., penati (1989). capital controls and international market segmentation: the evidence from the japanese and american stock markets. journal of finance, 44, 849 – 870. harvey, c. r. (1991). the world price of covariance risk. journal of finance, 46, 111 -158. heitala, p. k., ( 1989). asset pricing in partially segmented markets: evidence from finish market. journal of finance, 44, 697 – 718. korajczyk, r., and c. vialett (1990). an empirical investigation of international asset pricing. review of financial studies, 2, 553 -585. pagano, m., a. roell, and j. zechner, (2001). the geography of equity listing: why do companies list abroad? centro studi di economia e finanza (csef). pagano, m., a. roell, and j. zechner (2001). the geography of equity listing: why do companies list abroad? centro studi di economia e finanza, (csef). 16 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 2 http://epublications.bond.edu.au/ijbf/vol7/iss2/2 market meltdown and the propagation mechanism of contagion: 19-35 35 parhizgari, a. m. (2002). the broom theory: a perspective on global equity markets. fiu manuscript. an earlier version of this paper was presented at the october 1998 annual meeting of the financial management association, chicago. portes, r., and h. rey (1999). the determinants of cross-border equity � ows. national bureau of economic research. portes, r., and h. rey (1999). the determinants of cross-border equity � ows. national bureau of economic research. rasmussen, p. n. (1956). studies in inter-sectoral relations. north-holland publishing company, amsterdam, the netherlands. stapleton, r. c., and m. g. subramanyam (1977). market imperfections, capital market equilibrium and corporation � nance. journal of finance, stultz, r. (1981a). a model of international asset pricing. journal of financial economics, 9, 383 – 406. stultz, r. (1981b). on the effects of barriers to international investing. journal of finance, 36, 923 – 934. stultz, r. (1995). globalization of capital markets and the cost of capital: the case of nestle. jacf, 8, 30-38. stultz, r. (2005 ). the limits of � nancial globalization. journal of finance, august, lx, no. 4. wheatley, s., (1988). some tests of international equity integration. journal of financial economics, 21, 177 – 212. 17 ghosh and ghosh: market meltdown and the propagation mechanism of contagion produced by the berkeley electronic press, 2010 recommended citation international journal of banking and finance 8-1-2010 market meltdown and the propagation mechanism of contagion dilip k. ghosh dipasri ghosh does corporate governance matter in iran? 27 the international journal of banking and finance, volume 9 (number 3), 2012: pages27-45 does corporate governance matter in iran? saeed ghorbani and seyed tabaie zavareh university of economic sciences, iran ___________________________________________________________ abstract in this paper, we construct a corporate governance index (g-index) based on 13 attributes, which are associated with good and bad governance to investigate the impact of corporate governance on a firm’s stock return. after correlating each of the governance attributes of 141 tehran stock exchange listed companies with their performance separately over a period of six years, we find the direction of each attribute’s correlation. after that, we compute the g-index by aggregating the individual attributes and converting each firm’s scores on attribute into the same scale. finally, these scores are summed up by subtracting negatively correlated attributes from positively correlated attributes for each firm. we find a significantly high correlation between the firm’s performance and firm’s g-index. in the next step, we made three governance-sorted portfolios – from low to high governance which we use to evaluate stock returns. we find better-governed portfolios significantly outperformed the poorly governed portfolios. we find that corporate governance score really matters in since the results show statistically significant relationship between the qualities of the corporate governance as measured by our g-index and firm’s stock return. key words: corporate governance, corporate governance index, performance, return jel classification: g34, g32, g11 ___________________________________________________________ 1. introduction to governance literature the list of corporate governance research has increased significantly during the last decade (gillan, 2006). the reason is corporate governance is presently known as an effective mechanism for exerting corporate control in the literature in order to secure sustainable higher stock return. jensen (1986) puts it, if there is lack of ability of the other corporate control mechanisms such as capital market, legal-political-regulatory system, product and factor markets, it is effective to apply internal control system headed by the board of directors, he notes. the fundamental insight from which the field of corporate governance emanates is that there are potential problems associated with the 28 separation of ownership and control that is inherent in the modern corporate form of organization (denis, 2001). although in jensen’s opinion (1993) the board of directors is the apex of the internal control system with the final responsibility for the functioning of the firm, it was around 2001 this topic attracted intense attention following the collapse of enron. the aftermath of that collapse was accompanied by enormous amounts of research and numerous corporate governance reforms including the sarbanes-oxley act of 2002 (sox) (gillan et al. 2007). since then, increasing evidence points to the association between corporate governance and firm performance (aman and nguyen, 2008). gompers et al. (2003) found that a broad index based on 24 provisions of the investor responsibility research center (irrc) giving each irrc provision equal weight was negatively correlated with firm value measured using tobin’s q. they construct a governance index (g-index) to proxy for the level of shareholder rights found in approximately 1,500 large us firms during the 1990s. their results suggest that firms with stronger shareholder rights have higher firm value, higher profits, higher sales growth, lower capital expenditures, and make fewer corporate acquisitions. in addition, they find a significant positive relation between the g-index and stock returns over the sample period. they explain that weak shareholder rights create agency conflicts, which in return lead to low firm value in the long run. bebcuk et al. (2005) construct an entrenchment index based on six factors, which are selected from 24 governance factors developed by the irrc. they find a negative relation between the index scores and the value of firm as measured by tobin’s q for us firms over 1990-2003 (bebcuk et al. 2005). using a broader index, brown and caylor (2004) has confirmed earlier that corporate governance is positively correlated with operating performance, market valuation, and dividend payout for a large sample of us firms. black et al. (2006) using emerging markets construct a corporate governance index for 515 korean companies, using a survey conducted by the korea stock exchange, he found a positive relation between corporate governance practices and market valuation, as measured by tobin’s q and the market-to-book ratio. similar results have been documented by beiner et al. (2006) in europe; ahmadjian (2006) and miyajima (2006) in japan; and by klapper and love in 25 emerging markets (2004). following the work done by brown and caylor (2004), aman and nguyen (2008) constructed a governance index based on several attributes known to be associated with good corporate international journal of banking and finance, vol. 9, iss. 3 [2012], art. 3 29 governance and showed positive correlation with firm performance. using the index to evaluate the returns on governance-sorted portfolios, they found that poorly governed firms significantly underunderperformed better-governed firms. their result might be due to the greater risk exposure of poorly governed firms. after adjusting for size and book-to-market ratio, the excess returns become insignificant across all portfolios. finally, they concluded that stock prices appear to reflect fairly the higher (lower) risk associated with poor (good) corporate governance (aman et al. 2008). brown and caylor (2006) related the corporate governance to firm’s valuation using 1,868 firms based on 51 internal and external corporate governance provisions provided by institutional investor services (iss) as of february 1, 2002. 1.1 the governance index (g-index) in this paper, we start with constructing a governance index, g-index. we use rahavard novin and tehran stock exchange data set to identify 13 attributes of corporate governance, which are known to decrease agency conflicts and thus are associated with good (poor) corporate governance. to applying this index in this market, we had to construct the index and face the other difficulty on the lack of a publicly available information on governance attributes as well as ambiguity about the effect of corporate governance measurements. the attributes cover three dimensions of corporate governance: board composition, ownership structure and disclosure policy. contrary to that of gompers et al. (2003) on anti-takeover measures the firm’s resistance to external control mechanisms, our index is considered by those authors as emphasizing the quality of the firm’s internal controls. it is due to that reason that the market for corporate control in iran is underdeveloped (ghalibaf et al., 1985). further, the exchange is not even efficient in weak form. our index is closer to the index constructed by brown et al. (2004) and aman et al. (2008). however, considering the data availability issues for this research, we have to ignore some of the corporate governance aspects and add few more. in the following sections, we review each variable and the motives for its inclusion in our index in order to justify the selection. 1.2 board structure many researchers view the board of directors as the lynchpin of corporate governance. with fiduciary obligation to shareholders and the responsibility to provide strategic direction and monitoring, the board’s role in governance is important (gillan, 2006). unfortunately, in practice, it is not clear that the average member of the board of directors has enough incentive to do the job 30 properly. management typically has a great deal of say about who the directors shall be. nothing in the law prevents manager from stacking the board with individuals sympathetic to management’s interests. historically, many of them have done much sacking (denis, 2001). a good example is enron. although enron has been ranked as one of the five best corporate boards in 2000 by chief executive magazine, we now know that enron’s board did not restrain the firm’s management from engaging in risky behavior that led to the firm’s collapse (gillan et al., 2007). amidst public attention and calls for change by various parties, many firms are undertaking reform of the board structure. the typical reform includes some combination of reducing board size, increasing the relative proportion of outside directors, assigning these outside directors only to such tasks as nominating new directors and setting executive compensation plans, separating the positions of chief executive officer and chairperson on the board and requiring that board members own stock in the firm (denis, 2001). however, it is still not clear whether small size and the more independent board are related to better performance. according to aman et al. (2008), japanese firms’ board size is negatively correlated with performance. yermack (1996) and denis et al. (1999) provide evidence that us firms with smaller boards achieve higher market values. saito (2002) confirms that board size and tobin’s q are negatively correlated in japan. elsewhere, mak et al. (2005) single out board size as the most significant factor in explaining firm value in singapore and malaysia. the interesting point is the increase in the board size and independence after sarbanes-oxley (sox), which shows the incremental emphasis on board structure (chhaochharia et al. 2005). findings of hermalin et al. (2001) show that smaller boards and greater proportions of outside directors appear to lead management teams to take actions that are more in line with shareholders’ interests. in contrast to this finding, bhagat et al. (2002) find no relationship between the proportion of independent directors and various indicators of firm performance while (rosentein et al. 1990) observe a positive market reaction to the appointment of independent directors. other reports (coles, daniel, and naveen 2008) show that the relation between firm value and board size is u-shaped. in other words, both small and large boards are good for firms. dutchin et al. (2010) also show evidence that increased percentage of independent directors may actually hurt a firm’s performance. linck, et al. (2008) found a negative relation between past firm performance and board independence. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 3 31 in this paper, we use the number of board members (board_num), non-executive board members (ne_num) and the proportion of non-executives on board (ne-rto) as a proxy for board structure. because there has been no previous research done in this regard in this market, we do not associate board structure elements with better or worse performance and market value, so we take a neutral approach in terms of predicting the direction of these factors. 1.3 ownership structure conflicts of interest between owners-managers and outside shareholders as well as those between controlling versus minority shareholders have been the subject of many studies going far back as berle et al. (1932), jensen et al. (1976) and sheifer et al. (1986). accordingly, there is relatively a large literature on the effects of ownership and corporate performance (margaritis et al., 2010). numerous studies investigate the role of ownership structure on firm performance. in particular, large shareholders, such as institutional investors, are documented to have a positive influence on valuation (mcconnell et al., 1990). by generally accepted definition, a shareholder who holds 5 per cent or more of a corporation’s common stock is considered to be a large shareholder or block holder (denis, 2001). it is believed that the holder’s share of the firm is significant enough to potentially give them both the ability and the incentive to monitor and influence what is happening in the firm. these block holders may be individuals, corporations, and/or institutional investors (denis, 2001). nesbitt (1994), smith (1996), del guercio et al. (1999) and hartzell et al. (2003) find evidence that corporate monitoring by institutional investors can constrain managers’ behaviour. sheifer et al. (1986) show that large external equity holders can mitigate agency conflicts because of their strong incentives to monitor and discipline management. cole et al. (1998) examined the stockprice performance and ownership structure. they found that changes in performance are positively associated with changes in ownership by management: it is negatively associated with changes in ownership by employee stock ownership plans. in contrast, morck et al. (2000) show that bank ownership is detrimental to firm valuation. weinstein et al. (1998) appear to suggest that bankcontrolled firms exhibit a lower profitability. family firms are a special class of large shareholders with unique incentive structures (margaritis et al., 2010). although controlling family shareholders may expropriate minority shareholders’ value (claessens et al. 2002), several studies (e.g. anderson et al. (2003); villalonga and amit (2006); maury (2006); king and santor (2008)) show that family firms especially those with large personal owners tend to outperform non-family firms. in addition, the empirical findings of maury 32 (2006) suggest that large controlling family ownership in western europe appears to benefit minority shareholders rather than harming them. large institutional investors may not, on the other hand, have incentives to monitor management (villalonga and amit, 2006) and they may even coerce with management (mcconnell et al., 1990; claessens et al. 2002; cornett et al., 2007). moreover shleifer and vishny (1986) and la porta et al. (2002) argue that equity concentration is more likely to have a positive effect on performance in situations where control by large equity holders may act as a substitute for legal protection in countries in the context of weak investor protection in less developed capital markets: they also classify continental europe in that class. in this paper we use the percentage ownership by institutional investors (inst), percentage of ownership by non-institutional large shareholders (m_rto), the percentage of ownership by family shareholders (family), and the percentage of ownership by foreign investors (frgn) to proxy for ownership structure in our index. 1.4. information and disclosure governance role of accounting information simply means the utilization of the presented reports as a mechanism to control corporate governance. good information disclosure usually represents ceo’s willing to communicate with and surrender some of her/his power to shareholders. information disclosure plays a significant role on investment decisions for both domestic and foreign institutional investors (aggarwal, et al., 2005; ferreira and matos, 2008). to exercise control and to effectively monitor management, financial contracting parties also need value-relevant information for their individual decisions. the allocation of information rights directly affects the exertion of control and therefore, ultimately, the distribution of future payouts for investors. information rights and the flow of information inside and outside of the firm consequently constitute important elements of corporate governance (wustemann, 2001). it is expected that the quality of information released by firms exhibits a positive correlation with their performance, one reason being good performers are more willing to disclose information. in addition, recent studies (durnev et al., 2004) emphasize the role of information in relation to efficient capital allocation and growth (aman et al., 2008). in this paper, we use several attributes to reflect quality of disclosure. one is the total number of announcements (an_num) disclosed annually by a firm. the total number of mid-term financial announcements (mtfi) disclosed by a firm in a year is another. the transparency rank (tr) which international journal of banking and finance, vol. 9, iss. 3 [2012], art. 3 33 encompasses both timeliness and quality of the information releases, along with the firm’s success to meet its forecasts. transparency rank is accessed from the tehran stock exchange database. several attributes are employed for investigating the firm’s respect for its shareholders. • the existence of a website, which enables the shareholders to find general information about the company, • the existence of shareholder section in firm’s website which shows the specific emphasis that the company puts for its shareholders, and • whether the shareholders’ section of the firm’s website is updated. the rest of the paper is divided into four section: methodology is in section 2 followed by results on overall analysis in sections 3 and 4. the paper ends with conclusion in section 5. 2. aggregation of individual attributes and construction of g-index we first collate the values of every variable. next, we divide each variable by the difference between its maximum and minimum and then multiply the result by 10. after that, we use the result of spearman’s regression to add those variables. the calculated value is g-index score for each variable. in this way, we will have very few missing data. (1) corporate governance variable of same scale, corporate governance variables, variables from one to 13, m= company. the regression to be used for this analsysis is as in equation (2): number of board members, = number of non-executive board members non-executive board members ratio, institutional investor’s ownership percentage major investor’s ownership percentage, foreign investor’s ownership percentage 34 family ownership percentage, the number of official reports published annually the number of financial reports published in an interim period the existence of company website the existence of shareholder’s menu in companies website the existence of up to date information for shareholders in company website company transparency rank below is the fomula (equation (3)) for the construction of g-index: = gindex score of company (m) at time (t) same scale corporate governance variables with positive correlation with performance = same scale corporate governance variables with negative correlation with performance the maximum amount related to each variable with the same scale the minimum amount related to each variable with the same scale table-1: summary statistics for the aggregate index and each separate attribute variables no. min max mean s.d board structure board-num 846 0.000 11.000 13.000 5.3049 ne-num 846 0.000 12.000 12.000 3.3108 ne-ratio 846 0.000 1.000 1.000 0.62038 capital structure inst 846 0.000 100.000 100.000 32.8084 mj 846 0.000 99.500 99.500 72.5652 frgn 846 0.000 49.960 49.960 1.8893 famaily 846 0.000 100.000 100.000 12.4527 disclosure finfo 846 0.000 63.0000 15.4066 7.45401 mtfinfo 846 0.000 31.000 7.7446 4.48176 webe 846 0.000 1.000 0.8368 0.36969 web_sh 846 0.000 1.000 0.3617 0.48077 web_update 846 0.000 1.000 0.3120 0.46360 trans_rank 846 0.000 310.000 169.4468 104.6571 gindex 846 0.000 13.000 6.4255 2.43223 international journal of banking and finance, vol. 9, iss. 3 [2012], art. 3 35 table 1 presents the summary statistics for the aggregate index and each attribute. as is evident from the measured ratios, the numbers are within expected ranges. the data transformation should enable these measures to be accurate values of the attributes. 3. relation of governance to performance table 2 presents the correlation values of the governance index and its individual components of firms’ performance. according to the previous research, performance is more accurately measured by roa (ghalibaf et al., 2008) and (namazi et al., 2009). this is the step to verify empirically the relevance of each governance attribute (and ultimately that of the index) and quantify the firm’s governance quality by providing evidence of its impact on firm performance (through its effect on agency costs). roa is measured by operating profits scaled by total assets as at march 2004. due to the discrete nature of many governance attributes, spearman rank correlation value is also provided as the measure of the strength of the relationship on performance. table-2: correlation of governance attributes with perfomance measure (roa) correlation of governance attributes with roa roa correlations pearson spearman board structure board-num 0.001 0.007 ne-num -0.122(**) -0.130(**) ne-ratio -0.109(**) -0.130(**) capital structure inst 0.000 0.025 mj -0.032 -0.052 frgn 0.010 -0.058 famaily -.098(**) -0.095(**) disclosure finfo 0.339(**) 0.376(**) mtfinfo 0.356(**) 0.411(**) webe 0.179(**) 0.231(**) web_sh 0.209(**) 0.216(**) web_update 0.249(**) 0.238(**) trans_rank -0.462(**) -0.531(**) gindex 0.38(**) 0.410(**) ** indicates significant at 0.01 level; * at 0.05 level (2-tailed). 36 in contrast to (aman et al., 2008), the results show an insignificant negative correlation with board number and performance. the reason could be that the majority of the iranian firms tend to have 5 to 7 board members to follow the minimum requirements of iranian business law. findings of yermack (1996), saito (2002), mak et al. (2005) and aman et al. (2008) have shown the existence of such a relation even though the results are different. we find that non-executive members’ number and ratio have significant negative relation to performance. one of the reasons might be lack or low expertise of such members. the other reason might be the decrease in synergy and the speed of decision making due to the increase in the number of non-executive members. outside members may be working for other than the shareholders’ interests. these managers are not usually informed of company’s daily events and are influenced largely by the executive mangers. the findings of this study are contrary to the findings of bhagat et al. (2002), and aman et al. (2008). we have also shown that the family ownership in iranian firms has a significant negative relationship to performance. other variables related to the ownership structure (i.e., institutional investors, foreign investors and major stockholders) had no significant effect on performance, which is opposite the findings of margaritis et al. (2010), anderson et al. (2003), villalonga et al. (2006) and maury (2006). these findings can be justified by knowing that the main stockholder of the majority of the iranian firms are government and the institutional investors are finally governed by the governmental sector, which is usually synonymous with lack of management expertise and productivity. moreover, the firms with major foreign investors are rare, which might make the results unraveling. private sector is not usually supported by the government, and is under heavy regulations, taxes, and tariffs. the variables related to transparency on the other hand make significant correlation with performance. transparency rank is computed by the exchange and it interestingly has a significant negative correlation with performance, which shows the reliability of this rank for users’ decisionmaking. the finding is consistent with that of aman et al. (2008) and miyajima (2006). the finding also confirms a significant positive correlation of our g-index with performance. all three dimensions of corporate governance appear to be closely related to firm performance. using international journal of banking and finance, vol. 9, iss. 3 [2012], art. 3 37 survey data, miyajima (2006) finds disclosure to be the governance dimension with the most significant association to the firm performance in japan. it can be noted that some individual attributes exhibit a higher correlation than the index with the performance measures. for instance, transparency rank has a higher correlation with roa. hence, the index, using paramteric (pearson) as well as nonparametric (spearman) correlation, provides a compact measure that is strongly correlated with performance measure. 4. returns on governance-sorted portfolios in order to control for firm-specific information which is unrelated to corporate governance, we divide firms into governance-sorted portfolios. hence, we construct three separate portfolios with strong, medium and weak corporate governance according to the firm’s g-index values (where group 1 is the strong, group 2 is the medium and group 3 is the weak portfolio on governance). we have equal number of 282 firms in each group. the governance-sorted mean ranks for each of the groups are shown in table 3. table-3: the mean rank of governance-sorted portfolios governance rank group (portfolios) n mean rank strong 1 282 462.3918 average 2 282 434.2234 weak 3 282 373.8848 total 846 stock returns are computed by taking the log difference between two consecutive monthly prices over 6 years sourced from the tehran stock exchange price database. we have used kruskal wallis test (h test), which is a nonparametric test, to check the significance of the difference in our portfolio’s return: this is appropriate as the variable is an index. the result of the analysis indicates that there is significant difference in at least one of the medians, (2, n=846) = 19.41, p=6.07e-05 (table-4). in addition, mean rank shows the higher return of the strong portfolio compared with average and weak portfolios. although the overall test is significant and the mean return (table-5) and mean rank (table-3) for strong portfolio are higher, we cannot conclude that better-governed portfolio significantly outperforms medium and weak governance-sorted portfolios untill we perform post hoc analysis. 38 table-4: kruskal wallis test var00002 chi-square 19.41936 d.f. 2 asymp. sig. 6.07e-05 a kruskal wallis test b grouping variable: grouping the parameters in table 5 shows the simple mean return of the governance-sorted portfolios along with their standard deviation. as expected, the portfolio with better-governed companies has the highest mean return (0.0809) and the portfolio grouped as the weakest from the governance standpoint has the lowest mean return (0.0210). however, post-hoc analysis is necessary to statistacally verify the findings. table-5: mean returns of governance sorted portfolios variables n mean return stv min max equal weighted portfolios strong portfolio 282 0.080887 0.193047 -0.59 0.96 average portfolio 282 0.06273 0.200624 -0.42 1.06 weak portfolio 282 0.021028 0.131211 -0.56 0.76 the post-hoc pairwise comparisons are conducted by using the mann-whitney test for determining which groups significantly have different mean return. to calculate the statistic, this test like many non-parametric tests uses the ranks of data rather than their raw values to calculate the statistic. table 6 is summary of the median ranks of governancve-sorted portfolios. group 1 (strong governance) has the highest mean rank campared with the average and the weak portfolios. also, average governance portfolio has higher rank compared with weak porfolio. so as expected, the results indicate the higher mean return of strong governance portfolio compared with average and weak governance porftolios. likewise, mean return of average governance porfolio is higher compared with weak governance portfolio. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 3 39 table-6: pair wise comparison of governance-sorted mean ranks group n mean rank sum of ranks return strong (1) 282 300.56 84758.00 average (2) 282 264.44 74572.00 total 564 average (2) 282 302.13 85199.50 weak (3) 282 262.87 74130.50 total 564 strong (1) 282 324.26 91442.50 weak (3) 282 240.74 67887.50 total 564 it can be concluded that there is a statistically significant difference between each pair (table-7). it can also be concluded that the strong governance portfolio has statistically significant higher mean return than average and weak governance portfolio. pair-wise comparison of strong and average governance portfolios results in u=34669, p=0.000; for pair wise comparison of strong and weak governance portfolios u=27985.5, p=0.000; and for pair wise comparison of average and weak governance portfolio u=34227.5, p=0.004) table-7: pair wise comparison of governance-sorted mean ranks by mann-whitney test return (strong-average) return (average-weak) return (strong-weak) mann-whitney u 34669.000 34227.500 27984.500 wilcoxon w 74572.000 74130.500 67887.500 z -2.633 -2.880 -6.115 asymptotic significance (2tailed) 0.008 0.004 0.000 our findings, though different with results of aman and nguyen, (2008), core et al. (2006) and bauer et al. (2004), is line with gompers et al. (2003) and drobetz et al. (2004). 5. conclusions corporate governance is widely associated with better performance, whether the direction of causality stems from governance towards performance, or because high performers choose to operate under better governance rules (aman et al., 2008). a large number of empirical studies 40 confirm this association using aggregate governance ratings or specific governance attributes, such as board independence, performance-related compensation, or disclosure standards. in particular, well-governed firms appear to exhibit higher stock prices (aman et al., 2008). in this article, we investigate the case of iranian firms using our own governance index constructed to reflect the firm’s board characteristics, ownership structure, quality of disclosure and respect for investors’ interests. then we correlated our index with performance. the result confirms a strong relationship between our constructed index and performance. in the next step, we used the index to make three governance-sorted portfolios whose monthly returns are analyzed for a six-year period. the results show a significant difference between the governance sorted portfolio returns. the portfolio with strong governance makes the highest return while the portfolio with weak governance has the lowest return. the findings show that the investors take the governance measures into their decision-making even though they do not actually refer to these constructs as we describe to be the attributes of corporate governance. it is worthwhile to mention that our results do not imply that governance ratings are certainly a ruler for measurement of stock return. governance scores contribute to the information publicly available to market participants like many other sources of information. governance ratings might motivate the firms to improve their governance standards. in this paper, we faced different limitations, one of which is this was the first time to construct a g-index in iran. data were very hard to gather, as there was no single database, which contains data concerning all the corporate governance attributes. other limitation might be due to the sample used, which might make the extension of the results to other periods and places problematic. for continuing research, it is proposed that the other corporate governance mechanisms and other attributes be used to make a g-index. in addition, it is advised to adjust for risk factor and use the risk-adjusted return instead of gross return. author information: syed ghorbani is an assistant professor in the department of accounting, university of economic sciences, tehran, iran, second floor, no. 17, 15-th st., gandi st, tehran, iran. he may be contacted via e-mail: saeed.ghorbani@ses.ac.ir; phone: (98 9)1239 04868. seyed mohammad ali sadri tabaie zavareh, is a postgraduate student at the department of accounting, university of economic sciences, tehran, iran: third floor, no. 200, boustan saadi st., rudaki st., tehran, iran. e-mail: ali.sadri@ses.ac.ir or phone (98 9)12238 1891. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 3 mailto:saeed.ghorbani@ses.ac.ir mailto:ali.sadri@ses.ac.ir 41 references aggarwal, r., klapper, l., and wysocki, p. d., (2005). portfolio preferences of foreign institutional investors. journal of banking and finance, 29, 2919-2949. ahmadjian, c., inoue, k., nagai, s., and wakasugi, t., (2006). jcgr corporate governance survey final report. japan corporate governance research institute. ahmadjian, i. n., (2006). corporate governance survey final report. japan corporate governance research institute . aman, h., and nguyen, p., (2008). do stock prices reflect the corporate governance quality of japanese firms. journal of japanese and international economies , 22, 647-662. anderson, r., and reeb, d., (2003). founding family ownership and fim performance: evidence from the s&p 500. journal of finance, 58 (3), 1301-1328. bauer, r., guenster, n., and otten, r., (2004). emperical evidence on corporate governance in europe: the effect on stock returns, firm value and performance. journal of asset management , 5, 91-104. bebcuk, l., cohen, a., and ferrell, a. (2005). what matters in corporate governance? harvard law school. beiner, s., drobetz, w., schmid, m., and zimmermann, h., (2006). an integrated framework of corporate governance and firm valuation. eurob, finan, manage, 13, 249-283. berle, a. a., and means, g. c., (1930). corporations and the public investor. the american economic reveiw , 20 (1), 54-71. berle, a., and means, g., (1932). the modern corporation and private property. harcourt, brace and world . bhagat, s., and black, b., (2002). board independence and long-term performance. journal of corporate governance law, 27, 231-273. black, b. s., jang, h., and kim, w., (2006). does corporate governance perdict firm's market values? evidence from korea. journal of economics & organization , 22, 366-413. broadley, d., tohmatsu, d. t., (2006). auditing and its role in corporate governance. fsi seminar on corporate governance for banks. brown, l. d., and caylor, m. l., (2006). corporate governace and firm valuation. journal of accounting and public policy, 25, 409-434. brown, l., and caylor, m., (2004). corporate governance and firm performance. georgia state univetsity. brown, r., (1962). changing audit objective and techniques. the accounting review, , 696-703. burns, j., and scapens, r., (2000). conceptualizing management accounting change: an instituational framework. management accounting research, 14 (1), 3-25. bushee, b. j., (1998). the influence of institutional investors on myopic r&d investment behavior. accounting review, 73, 305-334. bushee, b., carter, m. e., and gerakos, j., (2010). institutional investor preferences for corporate governance mechanisms. working paper, university of pennsylvania . chandler, a. d., (1990). scale and scope, the dynamics of industrial capitalism. cambidge. the belknapp press of harvard university press. 42 cheung, y.-l., jiang, p., limpaphayom, p., and lu, t., (2008). does corporate governance matter in china? china economic review, 19, 460-479. chhaochharia, v., and grinstein, y., (2005). the transformation of us corporate boards:19972003. working paper. cornell university . claessens, s., djankov, s., fan, j., and lang, l., (2002). disentangling the incentive and entrenchment effects of large shareholdings. journal of finance , 57 (6), 2741-2771. cole, r. a., and mehran, h., (1998). the effect of changes in ownership structure on performance: evidence from the thrift industry. journal of financial economics , 50, 291-317. coles, j. l., daniel, n. d., and naveen, l. boards: does one size fit all. 87, 329-356. core, j. e., guay, w., and larcker, d. f., (2001). executive equity compensation and incentives: a survey. uneversity of pennsylvania. core, j., gauy, w., and rusticus, t., (2006). does weak governance cause weak stock returns? an examination of firm operating performance and investors' expectations. journal of finance , 61, 655-687. cornett, m. m., marcus, a. j., and tehranian, h., (2008). corporate governance and pay-forperformance: the impact of earnings management. journal of financial economics, 87, 357-373. cornett, m. m., marcus, a. j., saunders, a., and tehranian, h., (2007). the impact of institutional ownership on corporate operating performance. journal of banking & finance, 31, 17711794. cremers, m. k., and nair, v. b., (2005). governance mechansims and equity prices. the journal of finance, lx (6), 2859-2894. del guercio, d., and hawkins, j., (1999). the motivation and impact of pension fund activism. journal of financial economics, 293-340. denis, d. k., (2001). twenty-five years of corporate governance research ... and counting. review of financial economics (rfe , 10, 191-212. denis, d., and sarin, a., (1999). ownership and board stuructures in publicly traded corporations. journal of financial economics, 12 (3), 187-223. diaconu, p. (n.d.). harmonization of the international accounting system. downs, a. (1957). an economic theory of democracy. new york: harper. drobetz, w., schillhofer, a., and zimmermann, h., (2004). corporate governance and expected stock returns: evidence from germany. european financial management , 10, 267-293. dutchin, r., matusaka, j. g., & ozbas, o. (2010). when are outside directors effetive. journal of financial economics , 87, 3941-3976. fama, e., (1980). agency problems and the theory of the firm. journal of political economy , 88, 134-145. fama, e. f., and jensen, m. c., (1983). agency problems and residual claims. journal of law & economics, 26 (2), 327-351. fan, j. p., and wong, j., (2005). do external auditors perform a corporate governance role in emergin markets? evidence from east asia. journal of accounting research , 43 (1), 35-72. ferreira, m. a., and matos, p., (2008). the colors of investors' money: the role of institutional investors around the world. journal of financial economics , 88, 499-533. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 3 43 ghalibaf asl, h., and nateghi, m., (1985). the value relevance of dividends, book value and earnings in tehran stock exchange. journal of financial research , 8 (22), 47-66. ghalibaf, and ahmadvand (2008). a survey on the effects of onwership structure over firm's performance. m.sc. desertation . gillan, s. l., (2006). recent developments in corporate governance:an overiew. corporate finance, 12 (12), 381-402. gillan, s. l., and martin, d. j., (2007). corporate governance post-enron: effective reforms, or closing the stable door. journal of corporate finance , 13, 929-958. goergen, m., manjon, m. c., and renneboog, l., (2008). recent developments in german corporate governance. international review of law and economics , 28, 175-193. gompers, p., ishii, j., and metrick, a., (2003). corporate governance and equity prices. quarterly journal of economics , 107-155. gulamhussen, m. a., and guerreiro, l., (2009). the influence of forein equity and board membership on coporate strategy and internal cost management in portugues banks. journal of mangement accounting research , 20, 6-17. hackenbrack, k., and nelson, m. w., (1996). auditors' incentives and their application of financial accounting standards. the accounting review , 71 (1), 43-59. hartzell, j., and starks, l., (2003). institutional investors and executive compensation. journal of finance , 58 (6), 2351-2375. haw, i. m., (2008). discussant comments for "do auditing standards improve the accounting disclosure and information environment of public companies? evidence from the emerging markets in china. the international journal of accounting (43), 170-177. hermalin, b. e., and weisbach, m. s., (2001). boards of directors as an endogenously determined institution: a survey of the economic literature. economic policy review , 9, 7-26. holderness, c. g., (2001). a survey of blockholders and corporate control. economic policy review. hutchinson, m., and gul, f. a., (2004). investment opportunity set, corporate governance practives and firm performance. journal of corporate finance , 10, 595-614. jaggi, b., leung, s., and gul, f., (2009). family control, board independence and earnings management: evidence based on hong kong firms. journal of accounting and public policy, 28, 281-300. jensen, c. m., (1986). agency costs of free cash flow, corporate finance, and takeovers. american economic review , 6, 323-329. jensen, m. c. (1993). the modern industrial revolution, exit, and the failure of internal control systems. journal of finance , 48, 831-880. jensen, m., and meckling, w., (1976). theory of the firm: managerial behavior, agency costs and ownership structure. journal of financial economics , 3, 305-360. kapopoulos, p., and lazaretou, s., (2007). corporate ownership structure and firm performance: evidence from greek firms. corporate governance: an international review , 15 (2), 144158. keay, j., (1991). the honorable company: a history of the english east india company. newyork: macmillan. kedslie, m., (1990). firm foundation: the development of professional accounting. university of hull press , 1850-1900. 44 klapper, l., and love, i., (2004). corporate governance, investor protection, and performance in emerging markets. journal of corporate finance , 10, 702-128. la porta, r., lopez-de-silanes, f., shleifer, a., and ross, v., (1997). legal determinants of external finance. journal of finance , 57, 1147-1157. lamoreaux, n. r., (1985). the great merger movement in american business, 1895-1904. cambridge universit press . linck, j. s., netter, j. m., and yang, t., (2008). the determinants of board structure. jornal of financial economics , 87, 308-328. lipton, m., and lorsch, j., (1992). a modest proposal for improved corporate governance. the bussiness lawyer , 48 (1), 59-77. mak, y., and kusnadi, y., (2005). size really matters: further evidence on the negative relationship between board size and firm value. pacific-basin finance journal , 301, 301-318. margaritis, d., and psillaki, m., (2010). capital strucutre, equity ownership and firm performance. journal of banking & finance , 34, 621-632. maury, b., (2006). family ownership and firm performance: empirical evidence from western european corporations. journal of corporate finance , 12, 321-341. mcconnell, j., and servaes, h., (1990). additional evidence on equity ownership and corporate value. journal of finan. econ. , 595-612. miyajima, h., (2006). the performance effects and determinants of corporate governance reform in japan. waseda university. mizruchi, m. s,. (2001). berle & means revisited: the governanace and power of large u.s. corporations. shareholder-value, capitalism & globalization. bad homburg: werner reimers foundation & the volkswagen foundation. morck, r., nakamura, m., shivdasani, a., (2000). banks, ownership structure, and firm value in japan. journal of business , 73, 539-566. namazi, m., and kermani, e., (2009). the effect of ownership structure over tehran stock exhange firm's performance. accounting and auditing research , 53, 83-100. namazi, m., and shoshtarian, z., (1374). the investigation of the efficiency of iran's stock exchange. journal of financial research (7 & 8), 82-104. navissi, f., & naiker, v., (2006). institutional ownership and corporate value. managerial finance , 32 (3), 247-256. nesbitt, s., (1994). long-term rewards from shareholder activism: a study of the "calpers effect". journal of applied corporate finance , 6, 75-8-. nitkin, m. r., (2007). an investigation of the association between goveranance and accrual-based earnings quality. boston university school of management. o'donovan, g., (2003). change managemet-a board culture of corporate governance. mondaq business briefing . oecd. (2004). oecd principles of corporate governance. paris: oecd. patibandla, m., (2006). equity pattern, corporate governance and performance: a study of india's corporate sector. journal of economic behavior & organization , 59, 29-44. pound, j., (1988). proxy contests and the efficiency of shareholder oversight. journal of financial economics , 20, 237-265. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 3 45 puro, m., (1984). audit firm lovvying before the financial accounting standards board: an empirical study. journal of accounting research , 22 (2), 64-646. rosentein, s., and wyatt, j., outside directors, board independence, and shareholder wealth. journal of finan. econ., 26, 175-191. ross, s. a., westerfield, r. w., and jaff, j., (2005). corporate finance (7th ed.). new york: mcgraw hill irwin. saito, t., (2002). board size and market valuation: an empirical analysis of governance mechanism for keiretsu firms. jcer econ. j. , 44, 85-111. sajadi, h. s., dastgir, m., and naseh, l., (2005). usfulness of independent auditor's report. journal of social sciences & humanities of shiraz university , 22 (2). sami, h., and zhou, h., (2008). do auditing standards improve the accounting disclosure and infomation environment of public companies? evidence from the emerging markets in china. the international journal of accounting (43), 139-169. sheifer, a., vishny, r., (1986). large shareholders and corporate control. jornal of political economy , 96, 461-488. shleifer, a., and vishny, r. w., (1996). a survey of corporate governance. cambridge: national bureau of economic research. smith, m., (1996). shareholder activism by institutional investors: evidence from calpers. journal of finance , 51, 227-252. stein, j. c., (2001). agency, information and corporate investment. nber working paper 8342. tsaia, h., and gu, z., (2007). the relationship between institutional ownership and casino firm performance. hospitality management , 26, 517-530. villalonga, b., and amit, r., (2006). how to do family ownership, control and management affect firm value. journal of financial economics , 80, 385-417. weinstein, d., and yafeh, y., (1998). on the costs of a bank centered financial system: evidence from the changing main bank relations in japan. journal of finance , 53, 635-672. weir, c., laing, d., and mcknight, p. j., (n.d.). an empirical analysis of the impact of corporate governance mechanisms on the perfomance of uk firms. working paper. wustemann, j. (2001). disclosure regimes and corporate governance. www.ssrn.com . www.wikipedia.com. (2010). wikipedia the free encyclopedia. retrieved from http://en.wikipedia.org/wiki/corporate_governance yermack, d., (1996). higher market valuation of companies with a small board of directors. journal of finance economics , 40, 185-212. zaingales, l., (1998a). corporate governance in: newman, p.(ed),. the new palgrave dictionary and the law. zhou, c., (2004). corporate governance and ipo performance. master degree fulfillment, university of calgary . zhou, h. (2007). auditing standards, increased accounting disclosurem and information asymmetry: evidence from an emerging market. journal of accounting and public policy , 26, 584-620. the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 59 ijbf patterns of debt use in small businesses: a non-parametric analysis atreya chakraborty1 and rajiv mallick2 university of massachusetts-boston and amundi investment solutions americas llc, united states ______________________________________________________ abstract this paper uses non-parametric techniques to examine patterns of debt use by small � rms and how such patterns differ across � rm categories. the methodological goal is to use the richness of the � rm level data and allow convincing presentations with minimum of assumptions. the procedures used provide easily comprehendible graphical descriptions of the data. the procedures augment what can be discerned from descriptive statistics by accounting for differential weights and allowing for clustering that is a native feature of cross-sectional data. we also investigate how � rms could bene� t if credit availability improves. though a model-based analysis would be required to provide a detailed analysis, our analysis suggests that greater credit availability will bene� t all � rms. firms with low levels of equity will be better off as their credit constraints will be less binding, while � rms with high levels of equity will bene� t from acquiring more debt. keywords: debt holding, small business, non-parametric, credit constrained jel classi� cation: g21 and c14 ______________________________________________________ 1. introduction using data from the national survey of small business finances (nssbf) (1993), this paper offers some insightful � ndings into the debt holdings of small businesses in the united states. the primary purpose of the paper is to examine 1the authors would like to thank the editors and anonymous referees for their valuable comments and suggestions. all remaining errors are our own responsibility. 2the opinions expressed here are those of the author and do not necessarily re� ect the views and opinions of amundi investment solutions americas llc. 1 chakraborty and mallick: patterns of debt use in small businesses produced by the berkeley electronic press, 2010 60 the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 patterns of use of debt by small � rms and observe how those patterns differ across � rm categories. subsequently, � rms’ debt pattern for varying levels of equity was examined. firms’ propensity to hold debt and their likelihood of being credit constrained was also studied.3 apart from observing debt patterns, we analyze the incidence of credit availability for small business. our methodological goal is to use the unusually rich and comprehensive � rm level data collected by the nssbf on small business � nances and apply nonparametric tools to analyze it.4 the tools impose minimum of assumptions on the underlying distribution of the data and let the data speak for itself. in time-series data analyses, with thousands of observations, one has a luxury of being able to examine and make inferences, for example by plotting the series as a function of time. with the same number of observations, in a cross sectional setting it is less obvious what the appropriate graphical tools should be. histograms for example, are useful for single variables, but the relationship between variables is harder to describe and discern. a two-dimensional scatter diagram on the other hand is too cluttered to be informative. instead heavy reliance is placed on cross-tabulations and linear regressions as means of summarizing the data. while scatter diagrams do not transmit information clearly, cross-tabulations and regressions tend to over summarize, rarely doing justice to the amount of information available. this paper, highlights simple non-parametric techniques for density estimation and regression to summarize data and describe relationships. these methods provide easily comprehendible graphical descriptions of the data that are informative about the problem at hand. these techniques call for little more than presentation of the data relying less on economic or econometric assumptions. the results indicated that while the non-parametric techniques used may be limited in scope, they provide important insights by revealing natural clustering in the data. this paper augments often used cross-sectional data analysis techniques with non-parametric methodology to understand the � rms’ behavior toward using debt, debt-equity relationship and variations in debt usage. the paper examines � rms’ probabilities of holding debt, being credit-constrained, applying for loans and incidence of loan approval. descriptive statistics such as mean and variance capture the central tendency and dispersion of random variables, respectively. however, descriptive statistics do not incorporate the formula that accounts for differential weights and allows for clustering that may be a natural feature of cross-sectional data. techniques like kernel density estimations provide simple mechanisms to incorporate such features of random variables in easily comprehendible visual representations. 3we de� ne debt (or credit) as the combined amount of total loans including trade credit, mortgages, notes, bonds, and capital leases. 4 we focus on data visualization to this end as several studies have used different models to estimates demand for debt (cole, 1998), incidence of loan (cole, 2010) and have found statistically signi� cant difference among different legal forms of small businesses. 2 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 4 http://epublications.bond.edu.au/ijbf/vol7/iss2/4 patterns of debt use in small businesses: a non-parametric analysis: 59-78 61 : non-parametric methods were used to examine the relationship between debt use and levels of equity. regressions are used to describe linear relationships between variables with some distributional assumptions. by contrast, kernel regressions are more akin to a cross-tabulation and devoid of causal signi� cance. it is a descriptive device that is at best preliminary to a model-based analysis. such description provides easily comprehendible ‘maps’ of the effects of differences across � rms and some distributional characteristics on � rms’ levels of debt without imposing the linearity and distributional assumptions. while kernel regression captures the relationship between variables, it provides no impression of the variability in debt use at each level of equity. contour maps allow one to gauge the variability in debt use at each level of equity, and provide the � exibility to include observations on � rms that have either no debt or equity or both. exclusion of such � rms renders the sub-sample non-random.5 therefore, any model-based analyses of small � rms’ debt that do not incorporate � rms without debt either provides inconsistent results or their conclusions relate only to the sub-sample not to all small � rms. though a model-based analysis would be required to incorporate a hypothesis that some small � rms are credit constrained, and what is the effect of such constraints on debt holdings of � rms, a simple kernel regression with dichotomous dependent variable will allows one to discern whether � rms with intrinsic preference for holding debt are more likely to be credit constrained. such regressions are akin to probit regressions and plot the probabilities. the probability plots allow one to draw simple conclusions about small businesses’ behavior toward holding incremental debt if credit availability increased due to some policy changes.6 in essence, non-parametric analyses contain a good deal less than what one would like to know. the methodology, nevertheless, allows the data to speak for itself. this paper contributes to the existing body of literature on credit usage and small business in a several important ways. first, it documents that a large segment of small businesses in the united states are non-borrowers, i.e., they have no debt in their balance sheets. empirical research indicated that small businesses with no debt grow slowly, hire fewer workers and invest less in productive capital (king and levine, 1993; rajan and zingales, 1998). for policy makers, it raises questions such as why some small businesses use no debt and what would be the economic implications. for academicians, it would be relevant to be cognizant of a segment of small businesses that refrains from borrowing therefore skewing their empirical estimates about small business debt 5as an example, leece (1999) uses cluster analysis to segment market for risky � nancial assets and his analysis reveals that previous study may have missed some important interactions in the data. 6kunz (2007) presents a rich application of choice model based on latent variable models and principal component analysis to visualize competitive market structures based in individual consumer choice data. 3 chakraborty and mallick: patterns of debt use in small businesses produced by the berkeley electronic press, 2010 62 the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 use. second, our use of non-parametric methodology provides some compelling indications of how small businesses that do borrow will generally bene� t from greater availability of credit. there are important policy implications of understanding the patterns of debt usage by small business by different sub-categories of small business owners.7 first, a better understanding of the debt usage by different ownership types of small businesses is critical for targeting businesses that are more vulnerable to changing credit conditions. since the debt usage differs across ownership types, policy intervention will be more effective when targeted toward groups of small businesses that are more vulnerable to changing credit availability. any evidence on the differences on debt usage at � rm level will help in drawing appropriate tax and transfer policies to aid such small business operators. second, similar information on small businesses can also affect the outcome of a comprehensive monetary policy (gertler and gilchrist, 1994). for example, the “credit” or “lending” view stresses the ability of monetary policy to regulate the pool of funds available to bank-dependent borrowers.8 our � ndings can be used to design monetary policies that may have a disproportionate impact on borrowers with limited or no access to capital markets. the plan of the paper is as follows. the general organization follows the main objective of the paper and provides some methodological comments along the way. section 2 brings up some salient features of small businesses that characterize small businesses’ debt and their equity holdings. section 3 provides the analysis of debt and some distributional features of credit availability. section 4 concludes. 2. debt in small businesses there are three parties to the choice of capital structure of a � rm: a) the management, b) the equity holders and c) the creditors. the capital structure of a � rm depends on the judgment that each of the groups has about the future cash � ows. in small businesses, there could be just two parties: equity holders and banks. more precisely, managers and owners constitute the equity holders and banks are the creditors. in essence, owners sell � nancial claims when � rms choose to borrow funds from banks. but, the absence of a well-formed and complete market for ownership stakes in small businesses affects � rms’ � nancial decisions and differentiates them from those � rms operating in mature capital markets. small businesses are inevitably more dependent on the ability of a single individual or a small number of individuals to raise either debt or equity than in the case of large enterprises. in this section, the relationship between legal 7see kashyap and stein (2000) for a survey. 8romer and romer (1990); kashyap, stein, and wilcox (1993); and bernanke and blinder (1992) provide discussions. 4 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 4 http://epublications.bond.edu.au/ijbf/vol7/iss2/4 patterns of debt use in small businesses: a non-parametric analysis: 59-78 63 forms of small � rms and other categories of small businesses and their debt holdings are discussed and the � rms’ debt and equity holdings and their relative positions in their � nancial growth cycles are analyzed. to understand the amount of equity debt holdings, � rst look at the legal forms of ownership. forms of ownership are related to agency considerations. the conditions for information asymmetry arise both due to less stringent auditing requirements and because a single owner can monopolize � nancial information more easily than a broad management group in a large corporation. unlike large corporations, small businesses do not allow for separation of management and ownership. small businesses such as proprietary and partnership � rms can be seen as extensions of owners, especially in matters of their � nancial matters. proprietorship provides the simplest structure with a few legal requirements. the business and the owner are treated as one entity for legal and tax purposes. the main advantage is its simplicity and � exibility, but the lack of distinction puts owners’ personal assets at risk in the event of default or failure of the business. therefore, banks judge proprietary � rms not only by � rm characteristics, but also with those of the owners. the reliance of a proprietary � rm on its owner makes it vulnerable and, therefore, not a highly regarded prospect for a potential lender. in many instances, a � rm’s creditworthiness is as good as that of its owner. while proprietary � rms provide a better recourse in case of default or failure, opacity of business activity makes them a less attractive prospect for a lender. partnership � rms have a wider � nancial base than that of proprietary � rms. though the sources of � nance available to partnership � rms are similar to those available to proprietary � rms, there is a greater pool to draw from by the virtue of multiple partners. to the extent that a partnership reduces the vulnerability and riskiness of the business by providing a broader range of managerial skills and greater � nancial stability, partnership � rms may be in a better position to raise capital than a proprietary � rm. the ease with which corporations acquire capital depends on their size, stability, quality of management, nature of the industry, track record and their investment opportunity sets. the nature of limited liability to shareholders is perceived as a disadvantage by potential lenders, and in many cases lenders will require additional collateral both from shareholders as private individuals and from the company. it is hypothesized that bank-� rm relationships matter more in credit extension to cand scorporations than to proprietary and partnership � rms. when � rms default, banks have greater recourse with proprietary or partnership � rms than with corporations. looking beyond the legal form of ownership, the study extended the analysis to some of the owner characteristics, such as race and gender. research on small business lending indicates that some owners face racial discrimination in the credit market (cavalluzo and cavalluzo, 1998). impact of such actions would be seen either in low loan approval rate for certain groups of owners or such groups may self-select not to apply for loans. dummy variables for the race of owner captures the extent of discrimination, while the selection methodology used in other studies took into account the lower incidence for loan application in 5 chakraborty and mallick: patterns of debt use in small businesses produced by the berkeley electronic press, 2010 64 the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 general (cole, 1998; cavalluzzo, cavalluzzo and wolken, 2002; blanch� ower, levine and zimmerman, 2003). to account for sexual discrimination, we could include a dummy variable for female operators, though the literature indicated that female owners are generally not discriminated in the credit market (cavalluzo and cavalluzo, 1998). while ownership structure explains the legal framework and role of some owner characteristics on debt in small business, it ignores relative positions of � rms in their � nancial growth cycle and relationship between equity and debt. the study started with a basic presumption that assets of � rms are highly correlated with the stages of growth. financial growth cycle explains � rms’ debt holdings relative to � rms’ � nancial growth cycle. firms go through � ve stages of growth, namely: formation, rapid growth, growth to maturity, maturity and decline (weston and brigham, 1981).9 despite limited agreement on the terminology, � nancial growth cycle theory provides a reasonable framework to analyze the debt of small businesses. major sources of � nance for start-up businesses are insider � nance and particularly owners’ personal resources. so this stage of growth does not see much bank debt as de� ned earlier, except in some instances usage of personal credit cards. insuf� cient resources at this stage may result in under-capitalization of � rms. as � rms grow, they gain access to venture capital on the equity side and bank � nance on the debt side. continued growth beyond formation is very likely to be � nanced by retained earnings, trade credit and bank loans – some of the only and critical sources of � nance small businesses have (sahlman, 1988; wetzel jr., 1994). rapid growth outstrips � nancial resources leading to liquidity crises and the outcome is a greater reliance on short-term � nance due to lack of long-term � nance. the unavailability of long-term debt and equity, which is de� ned as � nancial gap, will have direct effect on the � nancial characteristics of small businesses. in the presence of � nancial gap, � rms are presented with very different situations. firms have to rely more on internally generated funds. lack of long-term debt � nance forces reliance on short-term sources and reduces liquidity. similarly, lack of equity � nance results in higher levels of leverage. thus, short-term debt substitutes not only long-term debt but also equity. apart from issues of ownership and stages of growth, other factors may affect the � nancial characteristics of small businesses. neck (1977) grouped such factors as hosts, agents and environment. the host are the owner-manager, agents are various � nancial environment institutions, and the environment constitutes legal, tax and economic institutions and market conditions. hosts matter, because different owners using the same amount of � nance could produce very different 9this classi� cation does not imply that � rms necessarily go through all growth phases. maturity and decline of a � rm can take place at any point after the formation. incidentally, some of the small businesses avoid some stages by forming and � oating in the stock market at the same time. in some instances, the term growth is misleading, because growth can continue to occur during the maturity stage. 6 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 4 http://epublications.bond.edu.au/ijbf/vol7/iss2/4 patterns of debt use in small businesses: a non-parametric analysis: 59-78 65 results. agents, particularly creditors, often turn out to be a major source of external funds. the environment is so broad that it is beyond the scope of this paper. but what does come up very often is the transparency of small business activities and the extent to which small businesses are legally required to disclose their � nancial statements publicly. 3. debt patterns of small businesses the study used the data from nssbf (1993) to describe debt patterns of small businesses. non-parametric methods were applied to examine the relationship between debt and equity, and how does this relationship vary across legal forms of � rms and owners’ gender and ethnicity. the study examines a general hypothesis that some small � rms are credit constrained and some small � rms do not have any debt in their capital structure. using a simple kernel regression with a dichotomous dependent variable, the investigation attempted to predict the effect of greater credit availability on � rms that are credit constrained and those that do not have any debt. a similar analysis was conducted to understand the effect on � rms’ incidence of credit application and credit approval. this paper begins with the description of � rms’ � nancial data, and this forms the basis for tables 1 and 2. table 1: structure of the sample: summary statistics firm categories firm characteristics debt equity assets firm age proprietary 92,349 109,821 188,403 13.99 partnership 2,038,759 963,067 2,852,831 14.78 s-corporations 856,214 782,826 1,981,783 13.75 c-corporations 1,121,264 995,394 2,596,871 17.58 female owned firms 335,334 255,808 651,813 12.92 male owned firms 938,362 746,582 1,925,309 15.84 minority operated firms 200,129 216,991 449,578 11.27 majority operated firms 954,104 746,385 1,945,232 16.13 table 1 shows some � rm characteristics and their descriptive statistics for different ownership categories. there are 4,637 � rms in the nssbf, and the study chose four � rm characteristics – debt, equity, assets and age, and mapped them for different legal forms of the � rms and owners’ gender and ethnicity. the nssbf has proportional representation of each of the categories and throughout this paper debt is used as a measure of leverage; either the absolute amount of debt or the debt-asset ratio. judging by the average amount of debt, c– and 7 chakraborty and mallick: patterns of debt use in small businesses produced by the berkeley electronic press, 2010 66 the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 scorporations and proprietary � rms are less levered than partnership � rms. when it comes to equity, proprietary � rms have the highest equity followed by cand s–corporations, and partnership � rms have the least amount of equity. c–corporations are some of the oldest � rms in the sample, while other � rms’ average age is about 14 years.10 minority and female owned � rms are the youngest � rms in the sample. table 1 ranks partnership � rms the largest with the highest average amount of assets, followed by c and s–corporations. lower panels of table 1 present similar statistics for � rms by female and minority owned organisations. minority owned � rms average about 92% debt-equity or 44% debt-asset compared to 127% debt-equity and 49% debt-asset ratios for majorities. however, such disparity is not observed among business ownership differentiated by gender. such inferences drawn from table 1 point to the central tendencies and can be misleading when we compare similar statistics at greater granularity. it is expected that non-parametric methods that were used in this paper will improve upon inferences that were generally gathered from descriptive statistics. the study also examined the incidence of � rms’ holding debt and the probability of � rms’ being credit constrained. some small � rms chose not to have any debt in their capital mix, and it is easy to identify them in the present dataset. however, it was dif� cult to identify � rms that are credit constrained due to the lack of any direct measure of a credit constraint. the nssbf of 198889 and 1993 provides information to identify credit constraint � rms. a � rm is de� ned be credit constrained if the � rm replied af� rmatively to one of the two following questions: a) “with the most recent loan application did a bank turn down the loan application or has the � rm been unable to get as much as it applied for?” and b) “during the past three years, were there times when the � rm needed credit, but did not apply because it thought the application would be turned down?” on average cand scorporations are more likely to be credit constrained compared to proprietary and partnership � rms, and also more likely to have debt (see table 2). proprietary and partnership � rms are less likely to be credit constrained and to hold debt. it was noted that there are similar proportions of credit-constrained � rms irrespective of owners’ gender or race. an economy wide credit crunch may affect corporations more adversely because of their greater reliance on debt than proprietary and partnership � rms. the � nal two columns of table 2 show the percentage of � rms that applied for a loan in last three years and the percentage of � rms that had their loans approved. the incidence of holding debt and being credit-constrained separately were analyzed from the probability of applying for a loan and the application being approved for all � rms in the survey. this was done because � rms’ decision to hold debt can be seen as a separate process than their decision to apply for credit.11 10berger and udell (1995) have shown that longer relationship does impact availability of credit as it lowers the likelihood of collateral being required for lines of credit. 11see cole (2010) for an analysis of why some small businesses do not use credit. 8 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 4 http://epublications.bond.edu.au/ijbf/vol7/iss2/4 patterns of debt use in small businesses: a non-parametric analysis: 59-78 67 averages such as those in table 1 and 2 conceal as much as they reveal. let us look at examples from each table. table 1 ranks partnership � rms as the largest with the highest mean value of assets over cand scorporation � rms. more large � rms in cand s-corporation categories than in partnership category were found when examined their asset values at different percentile levels. it could be true for any descriptive statistics that we stated in table 1, and it is hoped that non-parametric methods will improve upon them. similarly, in table 2 the broad inter-categorical patterns of distribution tell us which group of � rms will be more impacted by a poor lending scenario due to their greater debt exposure. it may be concluded that corporations are more likely to be credit constrained looking at the statistic in table 2, when compared to partnership and proprietary � rms. however, any direct effects of credit crunch will be dif� cult to uncover unless the methodology allows us to examine � rms at different levels of debt. table 2: small businesses and their bank activities firm categories debt creditconstrained equity* applied** approved*** percent of � rms proprietary 0.65 0.44 0.88 0.27 0.76 partnership 0.75 0.53 0.86 0.43 0.88 s-corporations 0.83 0.61 0.82 0.52 0.84 c-corporations 0.83 0.62 0.84 0.52 0.88 female owner 0.70 0.52 0.83 0.35 0.79 male owner 0.78 0.56 0.85 0.45 0.85 minority operated firms 0.70 0.58 0.83 0.33 0.60 majority operated firms 0.78 0.55 0.85 0.45 0.88 *percentage of � rms that have positive amount of reported equity. **percentage of � rms had applied for a loan in last three years. ***percentage of � rms that had loan applications approved. figure 1 shows estimates of the distribution of total debt for four legal forms of � rms at different levels of debt, unlike the statistics of table 1. institutional features indicate that lenders have better recourse against funds lent to proprietary and partnership � rms than to sor c-corporations. corporations are characterized by limited owner liability unlike proprietary and partnership � rms. 9 chakraborty and mallick: patterns of debt use in small businesses produced by the berkeley electronic press, 2010 68 the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 it stands out that corporations have generally been in business for longer and are at later stages in their � nancial growth cycle than proprietary and partnership � rms. we expect to see different distributional attributes of debt across the � rms of these four legal categories. density functions allow us to examine distribution attributes of debt at different levels. for estimation purposes, the logarithmic transformation for most variables was chosen because the distribution of debt like most of other variables is skewed, and log transformation yields a distribution that is more symmetric and closer to the normal distribution.12 figure 1. distribution of debt across legal forms of firms the most obvious feature of figure 1 is the relative positions of debt holdings of four types of the � rms. partnership � rms and sand ccorporations have higher level of average debt and some of them have very large amount of debt as evident by fat right tail of the distribution. proprietary � rms tend to hold less debt. the kernel density plot shows their debt holdings and suggests how it is different from other three categories of legal forms of � rms. the mean level of debt of s-corporations is about 75% of that of the partnership � rms and 40% c-corporations. the density plots indicate that sand ccorporations tend to have much similar distributions of debt holdings. partnership � rms have greater number of outliers evident by the fat right tail and a signi� cant mass of � rms around its median value. 12we provide a brief note in appendix 1 on bandwidth selection in kernel estimation. � d e n s it y log of debt s-corporation. corporation proprietary partnership 0 2 4 6 8 10 12 14 16 18 20 0 .05 .1 .15 .2 .25 .3 10 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 4 http://epublications.bond.edu.au/ijbf/vol7/iss2/4 patterns of debt use in small businesses: a non-parametric analysis: 59-78 69 univariate plots of figure 1 reveal little about the relationship between two or more variables. while organizational structure theory of small � rms predicts that partnership � rms may have greater amount of debt among four legal categories of � rms because they provide a greater recourse against loan default and failure than corporations and proprietary � rms. the � rm growth cycle theory sheds light on role of relationship on borrowing as such relationships forms very gradually over extended period of time.13 we want to gauge how debt levels differ with equity for different categories of � rms using kernel regressions. debt-asset ratios for different levels of equity for each category of � rms are plotted in figure 2. for each value of equity, figure 2 shows a weighted average value of debt-asset ratios nearby. the weights are the same weights used to construct the density, but are scaled by the estimate of the density at the point. the plots of estimated regressions look similar. but, the advantage of the technique used is that the underlying data determines the shape of the function. no assumptions were made about how the data should look like – for example lie along a straight line, or along a low-order polynomial. nonparametric regression, therefore, reveals the true relationship between debt-asset ratio and level of equity. figure 2. debt-asset regressions, bandwidths are 0.88, 0.69, 0.59 and .91 the downward sloping curves in figure 2 indicate that � rms with less proportion of debt will have greater equity, and regression plots af� rm that the role of debt in the capital mix declines with increasing equity.14 13 a similar model is employed by koutmos and booth (1995). 14 the data are available from the authors upon request. � d e b ta s s e t r a ti o log of equity s-corporation c-corporation proprietary partnership 2 4 6 8 10 12 14 16 18 20 .2 .3 .4 .5 .6 .7 .8 .9 1 11 chakraborty and mallick: patterns of debt use in small businesses produced by the berkeley electronic press, 2010 70 the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 it was noted that at low equity levels the debt-asset ratio is high for all types of � rms, and then tends to decline in a two-step fashion. over the whole distribution, the curve is steeper at low levels of equity then � attens out for � rms with larger amounts of equity. debt-asset ratios could be as high as 75%, before it declines precipitously. it can be concluded that � rms that borrow tend to have at the minimum about one-third to one-half debt as a share of total assets. the impact of credit crunch would be very different for � rms if we incorporate two obvious aspects that are not being captured in figure 2: some � rms choose not to have any debt or equity in their capital structure, and some of them self-select not to apply for loans fearing denial. if all � rms have equal access to bank credit and there is no precipitous difference in the interest rates charged to � rms, then in case of a general credit crunch � rms with high levels of debt will be more adversely affected. apart from obvious omission of the � rms without debt and/or equity, the curves in figure 2 can also be misleading for giving no impression of the variability in debt holdings at each level of equity. on average � rms with low levels of equity have as much as 60 to 80% debt in their capital structure as evidenced by the regression lines in figure 2. contour maps bring to surface the variability in debt at each level of equity and vice-versa while incorporating observations with no equity and/or no debt (see figures 3a & b). figure 3a. all � rms: bivariate density contour maps; log of debt, bandwidths are 0.60 and 0.72 the graphs present estimates of the joint density of the logarithm of debt (on x-axis) and the logarithm of equity (on y-axis). while the x and y-axis are the same as in the kernel regression graphs, and the height of the graph represents the fraction of � rms at the levels of the log value of equity and log value of debt 0 5 10 15 20 0 5 10 15 20 h = 0.60996 , 0.71617 log of equity lo g of d eb t � 12 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 4 http://epublications.bond.edu.au/ijbf/vol7/iss2/4 patterns of debt use in small businesses: a non-parametric analysis: 59-78 71 represented by the co-ordinates along the base.15 figure 3b provides the joint density surface map also known as the net-map that presents a visual impression of the surface of the joint density. the joint density map gives a better picture of relative height and the concentration of mass. the visual superiority of the contour maps is the following. first, the density surface provides information about the tails of the distribution where there may be very few observations. second, one also observe that whether � rms have just debt or equity or both. the density does not fall to zero, and the presence of an open ‘hole’ or ‘cave’ indicates either � rms with some equity and no debt or vice-versa. the joint density graphs also bring to surface any clustering of � rms. it was noted that for the most part figures 3a and b are similar, and that no new caveat emerges. figure 3b. all � rms: bivariate density contour maps; debt-asset, bandwidths are 0.60 and 0.72 our next step is to incorporate a general hypothesis prevalent in the literature that credit constraints do exist in small businesses; as small businesses have varying levels of debt therefore they face varying degrees of credit constraints.16 the nssbf dataset provides an additional caveat and that is small businesses demonstrate different willingness to hold debt and some carry no debt at all. though a model-based analysis is required to provide estimates of overall impact of credit constraints on the level of debt held by � rms, the data in 15more details on contour maps can be found in härdle (1987) and silverman (1986). 16see berger and udell (2006), berger, frame and miller (2005), petersen and rajan (1994). 13 chakraborty and mallick: patterns of debt use in small businesses produced by the berkeley electronic press, 2010 72 the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 the survey allows us to visualize probability of holding debt and being credit constrained for different levels of equity. we are able to shed light on whether � rms with intrinsic preference for holding debt are more likely to be credit constrained. figure (4) shows estimates of proportion of small � rms that chose to have debt in their capital mix and � rms that are credit constrained. this is a non-parametric regression using a dependent variable which takes one for � rms with debt and zero otherwise, and one for � rms that are credit constrained and zero otherwise. the regression is akin to a probit regression, and regression lines plot the probability of holding debt and � rms being credit constrained, respectively. figure 4. all firms: incidence of debt and probability of being credit constrained, bandwidths are 0.31 and 0.39 the general inference from figure 4 is that among small � rms the likelihood of holding debt is greater than the likelihood of being credit constrained. as expected at low levels of equity, small businesses are more likely to have debt. the probability of holding debt reduces more rapidly as level their equity share increases. over 60% of � rms are credit constrained with no or limited equity, and in excess of 40% of them continue to be creditconstrained at high levels of equity. the probability of a � rm being credit constrained decreases with greater levels of equity and so does the probability of holding debt; the upper line is downward sloping at higher levels of equityasset ratios. the positioning of curves also suggests that for any equity-assets ratio the probability of holding debt is higher than the probability of being credit constrained. therefore � rms that are more likely to have debt are more likely to � p ro b a b lit y equity-asset ratio debt/no debt constrained/uncosntrained 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 .3 .4 .5 .6 .7 .8 .9 1 14 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 4 http://epublications.bond.edu.au/ijbf/vol7/iss2/4 patterns of debt use in small businesses: a non-parametric analysis: 59-78 73 be credit-constrained. figures 3a and 3b where it was included both borrowers and non-borrowers con� rms an upward sloping relationship between equity and debt. figure 4 shows that � rms with equity-asset ratio greater than 0.90 have probability of being credit constrained at 40% or greater. therefore, most small businesses will bene� t from better credit availability. first, for � rms with lower equity-assets ratios, better conditions will lower their probability of being credit constrained. second, � rms with higher equity-assets ratios can presumably lower the weighted average cost of capital by acquiring more debt for equity.17 figure 5. all firms: incidence of loan application and approval, bandwidths are 0.43 and 0.29 it is emphasized that better credit conditions may encourage more request for debt. the analysis of incidence of loan application and approval reveals that better credit environment may rarely improve the probability of loan approval, however, may increase the incidence of loan application. the process of obtaining a loan is sequential in nature. the � rst step in the process of applying for one is critical. most factors that in� uence the incidence of loan application also affect the probability of loan approval, and nssbf (1993) cites very similar reasons � rms gave for not applying for loans and banks gave for declining loan applications. � p ro b a b lit y equity-asset ratio probability of loan application probability of loan approval 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 .2 .3 .4 .5 .6 .7 .8 .9 1 17 the distribution at the tails could be misleading. at the tails the estimated density becomes smaller, and since their estimate enters into the denominator of the conditional mean, the regression function is less precisely estimated. 15 chakraborty and mallick: patterns of debt use in small businesses produced by the berkeley electronic press, 2010 74 the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 however, while examining the probability of loan approval, the probability of applying for a loan should be taken into consideration.18 figure (5) suggests that if credit conditions turn favorable, the increase in loan application approval may not be as large as the increase in number of � rms that apply for loans. 4. conclusions the main conclusion of this paper is that there are some marked differences among small businesses in their debt holdings across different ownership types. the contour maps uncover an important detail some small businesses do not carry any debt in their balance sheet and some have no equity. this becomes an important distributional feature to be recognized in any empirical work. the often-used practice of excluding such � rms may provide results that cannot be generalized. cole (2010) noted that this sub-sample of small businesses has received limited attention in academic work. it is important to recognize that several empirical strategies can be used to incorporate these � rms with no debt or equity. for example, heckman’s two stage estimation technique can ameliorate the selection bias of leaving out a subsample of data. it was also observed that greater credit availability is likely to bene� t all small businesses. however, the bene� t would come from reducing the probability of being credit constrained for � rms with no or low equity, allowing � rms to switch to less expensive debt or by increasing the probability of application for loans. on the methodological front, the study was able to view some distributional features of debt in small businesses using a large data set. though some of the substantive conclusions will require rigorous model based analyses, the study was able to use the data to draw conclusions about the impact of favorable credit conditions. by analyzing the sub-samples for different categories of � rms, we are able to shed light on how different is allocation of debt within the small business sector. it was observed that larger corporations have more debt and more equity, but are also more likely to be credit constrained. author information: submitting author, professor atreya chakraborty, university of massachusetts-boston, college of management, 100 morrissey boulevard, boston, ma 02125, united states of america, phone +1 617 287 7673, email: atreya.chakraborty@umb.edu. rajiv mallick is head of risk management at amundi investment solutions americas llc, email: rajiv. mallick@nyc.amundi.com. 18 cole (2009) suggests that a large number of � rms that do not apply for credit have substantial amount of debt in their books. 16 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 4 http://epublications.bond.edu.au/ijbf/vol7/iss2/4 patterns of debt use in small businesses: a non-parametric analysis: 59-78 75 appendix 1 bandwidth selection our purpose behind bandwidth selection is how near observations have to be to contribute to the weighted average at each point. if the bandwidth is too wide, then all observations will fall in one band, and the histogram or the regression will be an uninteresting single rectangle or one regression point. if the bandwidth is too narrow, then few of them will contain more than one reading, and the histogram will consist of a large number of same size narrow rectangles or the regression curve will be very rough. the simplest method would be to experiment with different bandwidths, say l, at each value of log of equity to calculate the average of the debt-asset ratio of � rms whose equity is within l of the value. this procedure gives equal weight to all observations near and close to the value of log of equity being considered. we can improve on this procedure using a method that gives greater weight to � rms the closer is their value to log of equity that is being considered. therefore, the estimate of the regression corresponding to a point x, m(x) is: ,),()( iii yxxwxm �� (1) where n is the sample size, x i and y i are the x and y values of observation i, and i runs from 1 to n. in the method described above, the weights w i will be zero for x i far from x and close to one otherwise. some other methods do allow all observations to contribute and simply let the weights decline with the distance between x and x i . the estimator (1) is a very general one, and is described as a kernel estimator with the following weight scheme: ,)(/)(),( � ��� ihihii xxkxxkxxw (2) where k h is the kernel, h is the bandwidth. k h is a symmetric monotone decreasing function that integrates to unity. most statistical software provide a menu of kernels to choose from. we use the epanechnikov kernel that which is described by: (3) where i is an indicator function such that i = 1 if x and x i are within h of one another, i = 0 otherwise. ),|,(|1 4 3 )( 2 1 hxxi h xx h xxk iih � � � � � � � � � � � � ��� � 17 chakraborty and mallick: patterns of debt use in small businesses produced by the berkeley electronic press, 2010 76 the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 the equations (1) (3) are used for non-parametric regressions in the paper. in some � gures, the dependent variable y is either logarithm of debt, or debt-asset ratio or simply one or zero depending on whether the � rm does or does not have debt or is credit constrained. the graphs such as in figure 2 of text are constructed by calculating (1) for 20 equally spaced values of the log of equity and plotting the results. to select bandwidth, we start with trial and error. the basic criterion behind this procedure is to select a bandwidth that appears to give enough smoothness without obscuring detail. we do experiment with crossvalidation, one of the procedures generally followed in the literature. however, we � nd that the informal methods were unlikely to be misleading, at least for the graphical purposes of this paper. for the density estimations, such as figure 1 in the text, we follow a similar procedure. at each point on the x-axis, we take into account 50 nearby observations. we obtain an estimate of density by taking a ratio of the count and of the sample size. like kernel regression, it is sensible to use a weighing arrangement that gives closer � rms greater weights. we use the following kernel function to achieve this. (4) where k h (.) integrates to unity; a condition required to generate a proper estimate of the density. equation (4) is used for both the univariate densities and also for the kernel regressions. contour maps and surface plots are calculated using similar general principles. we � rst construct a grid over the range of the two variables. second, at each point on the grid, a weighted count is made of the observation within a neighborhood of the point. we use a kernel weighting function just as we did for kernel regressions and densities. the bivariate epanechionov is given by, (5) where d i is a two element vector of deviations of x i – x and y i – y each divided by the bandwidth h. an observation for estimation is included only if it is within a circular region centered at the current point and with radius h. the density estimate at a point is given by, (6) where k(d’d) = k(d), and s is the sample variance covariance matrix of the two variables. ),()( 1 xxknxf ihh �� �� � ),1()1( 2 )( '' 2 ��� iiiii ddiddh dk � � )],()'([ )(det )( 12 2/1 iih zzszzhkn s zf ��� �� � � � 18 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 4 http://epublications.bond.edu.au/ijbf/vol7/iss2/4 patterns of debt use in small businesses: a non-parametric analysis: 59-78 77 references berger, a. n., udell, g. f., (1995). relationship lending and lines of credit in small � rm � nance. journal of business, 68, 351-382. berger, a. n., and udell, g. f., (2006). a more complete framework for sme � nance. journal of banking and finance, 30, 2945-2966. bernanke, ben s., blinder, alan s., (1992). the federal funds rate and the channels of monetary transmission. american economic review, 82, 901-21. berger, a. n., frame, s., and miller, n., (2005). credit scoring and the availability, price and risk of small business credit. journal of money, credit and banking, 37, 191-222. blanch� ower, d., levine, p., and zimmerman, d., (2003). discrimination in the small business credit market. review of economics and statistics, 84, 930-943. cavalluzzo, k. s., cavalluzzo, l. c., (1998). market structure and discrimination: the case of small businesses. journal of money, credit, and banking, 30, 771-792. cavalluzzo, k. s., cavalluzzo, l. c., and wolken, j., (2002). competition, small business � nancing, and discrimination: evidence from a new survey. journal of business 75, 641-679. cole, r. a., (1998). the importance of relationships to the availability of credit. journal of banking and finance, 22, 959-977. cole, r. a., (2008). what do we know about the capital structure of privately held � rms? evidence from the surveys of small business � nance. u.s. small business administration, of� ce of advocacy, research report no. 324, 1-52. cole, r. a., (2009). who needs credit and who gets credit? evidence from the surveys of small business � nance. u.s. small business administration, of� ce of advocacy, research report no. 348, 1-136. cole, r. a., (2010). bank credit, trade credit or no credit: evidence from the surveys of small business � nances. u.s. small business administration, of� ce of advocacy, research report no. 365, 1-48. gertler, m., gilchrist, s., (1994). monetary policy, business cycles, and the behavior of small manufacturing � rms. quarterly journal of economics, 109, 309-340. härdle, wolfgang, (1987). applied non-parametric regression. cambridge: cambridge university press. kashyap, a. k., stein, j. c., (2000). what do a million observations on banks say about the transmission of monetary policy? american economic review, 90, 407-28. kashyap, a. k. stein, j.c., wilcox, d. w., (1993). monetary policy and credit conditions: evidence from the composition of external � nance. american economic review, 83, 78-98. 19 chakraborty and mallick: patterns of debt use in small businesses produced by the berkeley electronic press, 2010 78 the international journal of banking and finance, vol. 7. number 2: 2010: 59-78 king, r., levine, r., (1993). finance, entrepreneurship, and growth: theory and evidence. journal of monetary economics, 32, 513-542. kunz, werner, (2007). visualization of competitive market structure by means of choice data. sfb 649 discussion paper 2007-032, 1-19. leece, d., (1999). applying data visualization and knowledge discovery in databases to segment the market for risky � nancial assets. managerial and decision economics, 20, 267-280. neck, p.a., (1977). small enterprise development: policies and programmes. international labour organization, geneva. petersen, m. a., rajan, r. g., (1994). the bene� ts of � rm-creditor relationships: evidence from small business data. journal of finance, 49, 3-37. rajan, r. g., zingales, l., (1998). financial dependence and growth. american economic review, 88, 559-586. romer, christina d., romer, david h., (1990). new evidence on the monetary transmission mechanism. brookings papers on economic activity 149-98. sahlman, w.a., (1988). aspects of � nancial contracting in venture capital investment. journal of applied corporate finance, 1, 23-36. silverman, b.w., (1986). density estimation for statistics and data analysis. chapman and hall. weston, j. fred and brigham, eugene f., (1981). essentials of managerial � nance. dryden press. wetzel jr., w.e., (1994). venture capital. in: bygrave. w.d. (ed.), portable mba in entrepreneurship. wiley, new york, pp. 172-194. 20 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 4 http://epublications.bond.edu.au/ijbf/vol7/iss2/4 recommended citation international journal of banking and finance 8-1-2010 patterns of debt use in small businesses: a non-parametric analysis atreya chakraborty rajiv mallick market reactions to audit committee director's gender: evidence from us-traded foreign firms the international journal of banking and finance, vol. 8, number 1, march 2011: 59-75 59 market reactions to audit committee director’s gender: evidence from us-traded foreign firms hua-wei huang national cheng kung university, taiwan, roc yun-chia yan university of new orleans, usa james m. fornaro state university of new york-college at old westbury, usa ahmed elshahat university of new orleans, bradley university, usa ________________________________________________________________ abstract this study investigates whether the appointment of a female to the audit committee of a foreign issuer in the us is positively associated with subsequent market price reaction. we hypothesize that female members on the audit committee can strengthen corporate governance by their conservative and ethical qualities. accordingly, such appointments deliver a positive message to capital market participants. in order to observe the impact of audit committee gender diversity on foreign fi rms, we include all audit committee appointments for ustraded foreign fi rms from 2002 to 2009. we fi nd that the appointment of female audit committee members has signifi cant positive cumulative abnormal returns compared to the appointment of male audit committee members. keywords: audit committee, female director, corporate governance, market price reaction, us-traded foreign fi rms jel classifi cation: g14, m42, o16 ________________________________________________________________ 1. introduction the primary purpose of this study is to investigate whether the appointment of a female to the audit committee of a foreign issuer in the us is positively associated with subsequent market price reaction. research shows that female members on the audit committee can strengthen corporate governance by their ijbf huang et al.: market reaction to audit committee gender 60 the international journal of banking and finance, vol. 8. number 1, march 2011: 59-75 conservative and ethical qualities. accordingly, when a fi rm announces the appointment of a female to its audit committee, it delivers a positive message to capital market participants. in order to observe the impact of audit committee gender diversity on foreign fi rms, we include all audit committee appointments for u.s.-traded foreign fi rms from 2002 to 2009 in our sample. over the past two decades, foreign issuers have found the us capital markets to be an important gateway for capital-raising activities. however, the attractiveness and growth of other international stock exchanges (e.g., hong kong, tokyo and london) represent signifi cant challenges to the competitive position of the american markets. indeed, the sluggish economic conditions in the country and the challenging global economic environment also present signifi cant obstacles to the competitiveness of our capital markets.1 these conditions signal the need for the us to attract new listings by foreign fi rms while retaining the existing population of foreign listings. in response to a series of fi nancial scandals in the early 2000s, the sarbanesoxley act (sox) was signed into law in 2002 to restore investor confi dence in the capital markets and fi nancial reporting in general. the provisions of sox encompass four major areas: business ethics, internal controls and procedures, governance and independence, and reporting and disclosures. although particular sections of sox focus on strengthening the corporate governance of public fi rms, evidence suggests that these strict requirements and onerous compliance costs present barriers to foreign fi rms since sox does not provide them with a general exemption from its provisions. for example, hostak, karaoglu, lys and yang (2009) note that foreign fi rms, which voluntarily deregistered from exchanges, had weaker corporate governance and experienced, a signifi cantly negative stock market reaction when sox was passed. these fi rms also suffered a signifi cant price decline when they announced their intention to deregister. given this environment, balakrishnan (2007) suggests that the future challenge for the sec is to balance two competing goals which include: (1) protecting investors by setting sound standards of corporate governance, and (2) enhancing the attractiveness of the capital markets, particularly with respect to foreign issuers. sox includes several provisions that focus on strengthening the role of the audit committee as a critical monitor over the fi nancial reporting process. in particular, section 301 (sox 301) states that the audit committee ‘‘shall be directly responsible for the appointment, compensation, and oversight’’ of the auditor. under sox 301, audit committee members are expected to be concerned about their personal reputations and ensure that the external auditors have fulfi lled their responsibilities. moreover, audit committee members are required to be independent and are prohibited from accepting compensatory or advisory fees from the fi rm. eaddy et al. (2003) indicate that the audit committee requirements raise signifi cant problems for most foreign issuers because they deal 1 as of 2003, over 1,300 foreign companies from almost 60 countries fi led disclosure reports with the sec, compared with approximately 400 companies from less than 30 countries in 1990 (eaddy, faciszewski, fitz & van lowe, 2003). as of 2008, foreign listings had declined to 1,024 companies from 50 countries (deloitte, 2010). international journal of banking and finance, vol. 8, iss. 1 [2011], art. 4 market reactions to audit committee director's gender: evidence from us-traded foreign firms: 59-75 61 with technical structures, relationships and processes rather than with ultimate results, as do most of the disclosure requirements. again, given the lack of a general exemption for public foreign issuers, it is important to understand how the characteristics of audit committee members affect foreign fi rms. prior research (e.g., mcmullen & raghunandan, 1996; klein, 2002; bedard, chtorou & courteau, 2004; defond, hann & hu, 2005) has explored the following characteristics of audit committee members: independence, meeting frequency, size, and fi nancial expertise. moreover, specifi c research concerning the effect of gender diversity of audit committee members has recently emerged. ittonen, miettinen and vahamaa (2009) fi nd evidence that fi rms with female representation on the audit committee reduces the inherent risk of misstatements. their results have implications for external auditing since they also fi nd that gender diversity is associated with lower audit fees. huse and solberg (2006) fi nd that female directors are better prepared for board meetings than male directors which results in improved board behaviour and effectiveness. these benefi ts should also be realized by audit committees with female membership. research on ethics fi nds that women are less likely to engage in unethical behaviour in the workplace to obtain fi nancial rewards (bernardi & arnold, 1997; betz, o’connell & shepard, 1989). ruegger and king (1992) fi nd that females are more ethical than males in their perceptions of ethical conduct in business situations. moreover, related studies (e.g., krishnan & parsons, 2008) fi nd that earnings quality is positively associated with gender diversity in senior management, that females are more risk averse than men, and are less likely to engage in unethical behaviour in the workplace. we fi nd that the appointment of female audit committee members in u.s.-traded foreign fi rms has signifi cantly positive cumulative abnormal market returns compared to the appointment of male audit committee members. our fi ndings contribute to the extant literature in several ways. first, the results confi rm that share prices react favourably to the appointment of females to the audit committee. this reinforces the belief that gender diversity enhances the perceptions of market participants with respect to the audit committee’s independence and the overall performance of the governance function. second, this study has implications for global investors and other users of fi nancial information who are contemplating future investments in the us capital markets. if the sec’s major goal is to improve the quality of corporate fi nancial reporting, it should encourage gender diversity on audit committees. finally, this study contributes to the emerging literature that examines the effects of audit committee gender diversity. we provide additional evidence from the market participants’ perspective that the appointment of female audit committee members contributes to shareholder value by improving the reliability of fi nancial reporting through enhanced monitoring. the remaining sections of this paper are organized as follows. section 2 outlines the related literature and hypotheses. section 3 provides an explanation of the research method and data. in section 4, is provided the empirical tests and explanations of the results are presented. section 5 is the conclusion. huang et al.: market reaction to audit committee gender 62 the international journal of banking and finance, vol. 8. number 1, march 2011: 59-75 2. related literature and hypothesis development 2.1 gender research in corporate governance traditional gender studies in the accounting literature have focused on the dissatisfaction of women in the workplace and the lack of high-level female directors2 (e.g., pillsbury, capozzoli and ciampa, 1989; neidermeyer, tuten and neidermeyer, 2003). behavioural studies in business have shown that women in the accounting profession are generally more ethical than men when performing accounting tasks. for example, schaefer and welker (1994) indicate that cpas disciplined for work-related ethics violations are more likely to be males who practice in small fi rms. bernardi and arnold (1997) fi nd that female managers are at a signifi cantly higher average level of moral development than male managers in the big six accounting fi rms. cohen, pant and sharp (1998) indicate that women have consistently different ethical evaluations, intentions and orientation than men. more specifi cally, women consider questionable actions to be less ethical than men, and indicate that they would be less likely to engage in certain actions than men. neidermeyer et al. (2003) suggest that women believe that the practice of low-balling audit fees is less acceptable to women than to men, and women also agree signifi cantly more often that low-balling is a violation of the "independence in appearance" clause of the aicpa code of professional conduct.3 chung and monroe (2001) fi nd signifi cant gender differences with respect to information processing abilities and that females exercise more accurate audit judgment and decision-making abilities than males when faced with complex accounting decision tasks. due to these unique characteristics of women, researchers have recently begun to investigate how gender diversity infl uences the effectiveness and effi ciency of corporate governance, particularly with respect to the board of directors and audit committees. for example, carter, d’souza, simkins and simpson (2008) indicate that gender diversity has a positive effect on fi nancial performance primarily through the audit function of the board of directors. ittonen et al. (2009) suggest that female representation on audit committees may decrease audit fees by affecting the auditor’s assessment of the integrity of the fi nancial reporting process, as well as by reducing the inherent risk of misstatements. gul, srinidhi and tsui (2008) fi nd that boards with female directors are more likely to demand higher monitoring in the form of more audit effort. levi, li and zhang (2008) document that the bid premium over the 2 although a substantial number of women have entered the public accounting profession, only 3.7% of the partners in the big eight accounting fi rms are women (pillsbury et al., 1989). 3 two parts of the code cover principles and rules. the principles are goal-oriented and represent guidelines that address members’ responsibilities, public interest, integrity, objectivity and independence, due care, and the scope and nature of services to the public, clients and fellow practitioners. the rules are more detailed enforceable applications that support the principles. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 4 market reactions to audit committee director's gender: evidence from us-traded foreign firms: 59-75 63 preannouncement target share price is statistically and economically smaller when there is a larger proportion of women on the target company’s board, provided that the female directors are independent appointees. abbott, parker and presley (2009) study the relationship between board diversity and fi nancial restatements and fi nd that a female board presence is indicative of a more control-conscious ceo and a more vigilant board of directors. in summary, extant gender studies document the positive infl uence of female representation on the corporate governance function. 2.2 market reaction to the appointment of audit committee members investors respond to important announcements and events that are believed to impact a fi rm’s value at a point in time. the appointment of a new audit committee member can be viewed as critical information to market participants. previous studies document a signifi cant increase in the market’s reaction to earnings reports subsequent to the formation of the audit committee (wild, 1996). davidson, xie and xu (2004) fi nd signifi cantly positive stock price reaction when newly appointed members of audit committees have fi nancial expertise. further, defond et al. (2005) document a positive market reaction to the appointment of accounting experts assigned to audit committees. these studies demonstrate the existence of information content in the capital markets for audit committee related announcements. given this discussion and the positive infl uence of female representation on corporate governance, we expect that the appointment of a female to the audit committee provides signifi cant information content regarding: (1) enhanced oversight over the fi rm’s fi nancial reporting process, (2) more effective supervision over the external auditor, and (3) expected improvements in the quality of the fi rm’s fi nancial reporting to market participants. we further hypothesize that this signal simultaneously induces favourable market price reactions as follows: hypothesis 1a: market reaction in response to the appointment of a female audit committee member is positive. hypothesis 1b: market reaction in response to the appointment of a male audit committee member is neutral/negative. hypothesis 2a: market reaction is positively associated with the number of female audit committee members appointed. hypothesis 2b: market reaction is neutrally/negatively associated with the number of male audit committee members appointed. huang et al.: market reaction to audit committee gender 64 the international journal of banking and finance, vol. 8. number 1, march 2011: 59-75 3. research method, data and variables 3.1 model specifi cation to examine our hypotheses, the following ols regression models are employed (e.g. chen & church, 1996; defond et al. 2005; choi, kim, liu & simunic, 2008): (1) where: car = cumulative abnormal market returns; chprice = changes in stock price prior to the issuance of the auditor’s report; lnmve = natural log of the market value of equity; leverage = total debt divided by total assets; lnregime = the strictness of a country’s legal regime, measured by the natural log of wingate’s 1997 litigation index for each country; lngdp = natural log of gross domestic product (gdp) per capita (in thousands of us dollars); fdi = foreign direct investment scaled by gdp; lnequity = the importance of each country’s equity market, measured by the natural log of extent to which each country’s fi rms rely on equity fi nancing (the data are extracted from choi and wong, 2005); discl = a country’s disclosure level measured by the center for international financial analysis and research (cifar) index; tech = 1 if a technology fi rm (defi ned as sic= 3570–3579, 3670– 3679, and 7370–7379), else 0; listann = 1 if the fi rm is traded on amex, nasdaq or nyse, else 0; and female = 1 if a female audit committee member is appointed, else 0. (2) car = + 1chprice + 2lnmve + 3leverage + 4lnregime + 5lngdp + 6fdi+ 7lnequity + 8discl + 9tech + 10listann + 11female + car = + 1chprice + 2lnmve + 3leverage + 4lnregime + 5lngdp + 6fdi+ 7lnequity + 8discl + 9tech + 10listann + 11fsize + 12 msize + + 6fdi+ 7lnequity international journal of banking and finance, vol. 8, iss. 1 [2011], art. 4 market reactions to audit committee director's gender: evidence from us-traded foreign firms: 59-75 65 where: fsize = number of female audit committee members appointed; msize = number of male audit committee members appointed; and the other variables are the same as in model (1). our main tests consist of a traditional event study that examines: (1) three-day (−1 to +1); (2) seven-day (−3 to +3), and (3) eleven-day (-5 to +5) car around the appointment disclosure dates of the new audit committee members. in the ols regression models, car is regressed on variables of interest representing gender diversity, such as female, fsize, and msize. we control for the potential effects of the size of the company and total market environment by using the following three variables: chprice, lnmve, and leverage as in previous studies (chen & church, 1996; defond et al., 2005). country-specifi c litigation and governance variables, including lnregime, lngdp, fdi, lnequity and discl, are controlled as in choi et al. (2008). as in choi et al. (2008), each country has one litigation index, gdp, fdi, equity index, and disclosure level index. therefore, us-traded foreign fi rms from the same country will be assigned the same governance indices. the gdp and fdi are different by years, but the litigation index, equity index, and disclosure level index are the same across the sample years due to data availability. 3.2 sample selection panel a of table 1 provides a summary of the sample selection process. the initial sample of 476 audit committee appointments (from 2002 to 2009) by foreign fi rms were obtained from the auditanalytics database. to clearly identify that the car is driven by male or female appointments separately, we excluded 182 companies that appointed both female and male members on the same day. after matching with the compustat and crsp databases, the missing fi nancial data reduces the available observations to 107 for our regression analysis. there are no delisted fi rms in our fi nal sample. panel b shows that the selected fi rms are distributed among 14 countries/economic entities. the highest representation is by canadian fi rms (56) followed by fi rms from the united kingdom (20) and china (12). 4. results 4.1 descriptive statistics table 2 provides descriptive statistics on the full sample tested in this study.4 we use eventus to calculate the car for all of our 107 sample fi rms. the mean and median car in all three event windows are negative. sixty-three per cent of our 4 extreme observations are winsorized at 5% and 95% to reduce the potential infl uences of outliers. huang et al.: market reaction to audit committee gender 66 the international journal of banking and finance, vol. 8. number 1, march 2011: 59-75 selected fi rms were traded on amex, nasdaq or nyse (listann). ten per cent of our sample fi rms reported the appointment of a female audit committee member (female). table 1: sample selection process and composition panel a: sample selection process number of new appointments of audit committee members: january 1, 2002 december 31, 2009 476 less: companies that appointed both female and male members on the same day* -182 sample used for restatement analysis 294 less: companies missing compustat or crsp fi nancial data -187 sample used for regression analysis 107 * our sample consists of companies that appointed only either (1) female or (2) male members. thus we exclude companies that appointed both female and male members on the same day from our analysis. panel b: sample composition by country/economic entity country n argentina 2 australia 2 canada 56 china 12 costa rica 2 france 3 hong kong 2 ireland 1 netherlands 1 netherlands antilles 2 panama 1 russia 1 singapore 2 united kingdom 20 total 107 also, the mean value of msize (1.02) is higher than that of fsize (0.13). the mean (74.38) and median (73.00) discl of foreign fi rms are lower than us fi rms (86), as expected. the mean (1.80) and median (1.65) lnregime of our sample fi rms both fall within a normal range.5 5 choi et al., 2008 suggest that the value of lnregime in a country with a strong (weak) legal regime is 2.40 (1.50). international journal of banking and finance, vol. 8, iss. 1 [2011], art. 4 market reactions to audit committee director's gender: evidence from us-traded foreign firms: 59-75 67 t ab le 2 : d es cr ip ti ve s ta ti st ic s of g en de r a na ly si s s tu dy n = 10 7 (1, + 1) c a r (24 0, -2 ) c a r (3, + 3) c a r (24 0, -4 ) c a r (5, + 5) c a r (24 0, -6 ) c a r l nm v e l e v e r a g e l nr e g im e m ea n -1 .4 4% -3 .2 0% -1 .8 3% -1 .1 1% -2 .4 1% -4 .6 3% 20 .4 0 0. 51 1. 80 s td . d ev ia ti on 0. 04 0. 28 0. 07 0. 10 0. 11 0. 30 1. 76 0. 25 0. 28 p er ce nt il es 25 -3 .4 0% -1 8. 10 % -5 .0 0% -5 .0 0% -8 .0 0% -1 7. 40 % 19 .2 0 0. 30 1. 65 m ed ia n 50 -1 .0 0% -0 .7 0% -0 .5 0% -0 .5 0% -0 .8 0% -1 .4 0% 20 .3 0 0. 50 1. 65 75 1. 40 % 12 .1 0% 2. 60 % 2. 60 % 4. 00 % 11 .0 0% 21 .7 0 0. 71 2. 09 n = 10 7 l ng d p f d i l ne qu it y d is c l t ec h l is t a n n f e m a l e f s iz e m s iz e m ea n 2. 67 0. 21 -1 .6 9 74 .3 8 0. 16 0. 63 0. 10 0. 13 1. 02 s td . d ev ia ti on 0. 94 0. 13 0. 46 7. 21 0. 37 0. 49 0. 31 0. 41 0. 58 p er ce nt il es 25 2. 40 0. 19 -1 .7 0 73 .0 0 0. 00 0. 00 0. 00 0. 00 1. 00 m ed ia n 50 2. 40 0. 19 -1 .7 0 73 .0 0 0. 00 1. 00 0. 00 0. 00 1. 00 75 3. 60 0. 19 -1 .7 0 80 .0 0 0. 00 1. 00 0. 00 0. 00 1. 00 n ot es : c a r = c um ul at iv e ab no rm al m ar ke t re tu rn s; c h p r ic e = c ha ng es i n st oc k pr ic e pr io r to t he i ss ua nc e of t he a ud it or ’s r ep or t; l n m v e = n at ur al lo g of m ar ke t va lu e of e qu it y; l e v e r a g e = t ot al a ss et s le ss b oo k va lu e di vi de d by t ot al a ss et s; l n r e g im e = t he s tr ic tn es s of a c ou nt ry ’s l eg al r eg im e, m ea su re d as n at ur al l og o f w in ga te ’s 1 99 7 li ti ga ti on i nd ex f or e ac h co un tr y; l n g d p = g ro ss d om es ti c pr od uc t (g d p ) pe r ca pi ta i n th ou sa nd s of u s do ll ar s; f d i= f or ei gn d ir ec t in ve st m en t sc al ed b y g d p ; l n e q u it y = t he i m po rt an ce o f ea ch c ou nt ry ’s e qu it y m ar ke t, m ea su re d by t he e xt en t to w hi ch ea ch c ou nt ry ’s fi r m s re ly o n eq ui ty fi n an ci ng ( th e da ta a re e xt ra ct ed f ro m c ho i an d w on g, 2 00 5) ; d is c l = a c ou nt ry ’s d is cl os ur e le ve l m ea su re d by t he c en te r fo r in te rn at io na l f in an ci al a na ly si s an d r es ea rc h (c if a r ) in de x; t e c h = 1 if a t ec hn ol og y fi rm , de fi ne d as s ic = 3 57 0– 35 79 , 36 70 –3 67 9, a nd 73 70 –7 37 9; l is t a n n = 1 i f th e fi rm i s tr ad ed o n a m e x , n a s d a q o r n y s e , el se 0 ; f e m a l e = 1 i f a fe m al e au di t co m m it te e m em be r is a pp oi nt ed , ot he rw is e= 0; f s iz e = n um be r of f em al e au di t co m m it te e m em be rs a pp oi nt ed ; m s iz e = n um be r of m al e au di t co m m it te e m em be rs a pp oi nt ed . huang et al.: market reaction to audit committee gender 68 the international journal of banking and finance, vol. 8. number 1, march 2011: 59-75 table 3: car analysis on gender difference of audit committee appoinments car (-1, +1) car (-3, +3) car (-5, +5) female mean (n=11) -1.55% 1.42% 1.92% female median (n=11) -2.30% 2.60% 3.70% male mean (n=96) -1.43% -2.21% -2.91% male median (n=96) -0.95% -1.05% -1.55% independent t-test t value 0.204 0.382 0.731 p-value 0.937 0.083 0.171 non-parametric test mann-whitney u 509 357 380 wilcoxon w 575 5013 5036 z -0.20 -1.76 -1.52 p-value 0.841 0.078 0.129 notes: where car = cumulative abnormal market returns. table 3 presents the car analysis for both male and female groups. the mean and median of car in event windows (-3, +3) and (-5, +5) are positive for the appointment of female audit committee members. conversely, the mean and median of car are negative in all three event windows for the appointment of male audit committee members. the difference in car between male and female groups is signifi cant (p<0.10, two-tailed) in window (-3, +3). accordingly, the empirical results provide some evidence that the market reacted positively (negatively) to the appointment of female (male) committee members. overall, however, the univariate tests provide weak evidence to support our hypotheses. table 4 shows the correlation matrix of our variables and reports signifi cant results in window (-3, +3). both female and fsize are positively correlated with car and signifi cant at the conventional level (p<.10, twotailed). in window (-5, +5), fsize is also positively correlated with car and signifi cant (at p<.10, two-tailed). for male appointments, msize is negatively correlated with car (-3, +3) and signifi cant at the conventional level (p<.05, two-tailed). accordingly, the correlation analysis provides consistent evidence to support our major hypotheses international journal of banking and finance, vol. 8, iss. 1 [2011], art. 4 market reactions to audit committee director's gender: evidence from us-traded foreign firms: 59-75 69 t ab le 4 : c or re la ti on s m at ri x n = 10 7 l nm v e l e v e r a g e l nr e g im e l ng d p f d i l ne qu it y d is c l t ec h l is t a n n f e m a l e f s iz e m s iz e c a r ( -1 , + 1) 0. 19 -0 .0 5 -0 .0 9 -0 .0 3 -0 .0 5 -0 .0 1 -0 .0 7 -0 .1 1 0. 21 -0 .0 1 0. 00 -0 .1 3 pva lu e 0. 05 0. 58 0. 38 0. 73 0. 59 0. 91 0. 47 0. 25 0. 03 0. 94 0. 98 0. 19 c a r ( -3 , + 3) 0. 16 0. 06 0. 04 0. 03 0. 00 0. 02 -0 .0 1 0. 03 0. 22 0. 17 0. 18 -0 .2 2 pva lu e 0. 11 0. 53 0. 68 0. 76 0. 98 0. 83 0. 93 0. 78 0. 02 0. 08 0. 07 0. 02 c a r ( -5 , + 5) 0. 18 0. 15 -0 .0 5 -0 .0 3 -0 .0 9 -0 .0 5 -0 .0 6 0. 09 0. 12 0. 13 0. 16 -0 .1 2 pva lu e 0. 06 0. 12 0. 64 0. 73 0. 37 0. 58 0. 53 0. 34 0. 20 0. 19 0. 09 0. 23 n o te s: v ar ia bl es a re d efi n ed i n t ab le 2 . . huang et al.: market reaction to audit committee gender 70 the international journal of banking and finance, vol. 8. number 1, march 2011: 59-75 4.2 ols regression results table 5 reports the ols regression results for the primary tests.6 several control variables are signifi cant in window (-5, +5), including chprice, fdi, discl, and tech. as expected, car is positively associated with the fi rms’ prior market returns (chprice). technology fi rms with higher industrial risks experience greater market reactions after the appointments of female directors. this is perhaps due to the positive infl uences of female representation on the fi nancial reporting process and the quality of fi nancial reporting, particularly in fi rms exposed to higher risks. the coeffi cients of fdi and discl are both negatively signifi cant in event window car (-5, +5), which suggests that a country’s foreign direct investment (scaled by total gdp) and disclosure level reduce the impact of the appointment of female audit committee members on the market price of foreign fi rms. a higher disclosure level (discl) is found to reduce the volatility of market reactions to the announcements of female appointments. the results of our variables of interest show that the appointment of female audit committee members has information content to the stock market in two event windows after controlling for the effects of cross-country variables. the coeffi cient of female is positive and signifi cant (+0.050, p<.05, twotailed) in event window car (-3, +3) and (+0.077, p<.05, two-tailed) in event window car (-5, +5). these fi ndings support our major hypotheses that the appointment of female audit committee members in u.s.-traded foreign fi rms has signifi cant positive cumulative abnormal market returns compared to the appointment of male audit committee members. the coeffi cient of fsize is also positive and signifi cant (+0.070, p<.05, two-tailed) in event window car (-5, +5). the evidence shows that the greater the number of female audit committee members appointed, the greater the positive market return incurred. conversely, the coeffi cients for msize are negative and signifi cant or insignifi cant in all event windows. this indicates that the number of male audit committee members appointed has a negative or no signifi cant impact on the company’s market price. this empirical evidence consistently supports our major hypotheses. 4.3 additional analysis: gender diversity and restatements if the presence of female audit committee members enhances audit committee effectiveness, it is reasonable to expect that such fi rms would experience fewer fi nancial restatements subsequent to the appointment of female audit committee members. we investigated the association between audit committee gender diversity and fi nancial restatements for the 294 fi rms indicated in panel a of table 1 by dividing our sample fi rms into: (1) female audit committee member appointment group (n=23), and (2) male audit committee member appointment group (n=271). we found that only 1 of the 23 foreign fi rms (4.3%) restated their fi nancial statements in the year subsequent to the appointment of female audit committee members. 6 multicollinearity should not be an issue since all vifs are less than 2.0. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 4 market reactions to audit committee director's gender: evidence from us-traded foreign firms: 59-75 71 t ab le 5 : o l s r eg re ss io n r es ul ts o n g en de r a na ly si s c a r ( -1 , + 1) c a r ( -3 , + 3) c a r ( -5 , + 5) m od el 1 m od el 2 m od el 1 m od el 2 m od el 1 m od el 2 c oe ffi c ie nt pva lu e c oe ffi c ie nt pva lu e c oe ffi c ie nt pva lu e c oe ffi c ie nt pva lu e c oe ffi c ie nt pva lu e c oe ffi c ie nt pva lu e c on st an t 0. 47 4 0. 08 5 0. 54 2 0. 05 1 0. 32 9 0. 25 3 0. 39 2 0. 17 9 0. 74 6 0. 13 3 0. 80 3 0. 11 0 c h p r ic e 0. 04 2 0. 06 6 0. 04 2 0. 06 7 0. 03 2 0. 14 7 0. 03 2 0. 14 6 0. 07 5 0. 05 0 0. 07 6 0. 04 6 l nm v e 0. 00 3 0. 35 6 0. 00 3 0. 36 3 0. 00 2 0. 52 4 0. 00 2 0. 52 9 0. 00 5 0. 45 8 0. 00 5 0. 49 4 l e v e r a g e -0 .0 21 0. 42 1 -0 .0 29 0. 28 1 0. 02 8 0. 32 4 0. 01 9 0. 50 9 0. 07 8 0. 11 0 0. 07 3 0. 14 1 l nr e g im e -0 .0 20 0. 57 8 -0 .0 26 0. 47 2 0. 02 7 0. 47 2 0. 02 0 0. 59 3 0. 01 2 0. 85 8 0. 00 3 0. 96 8 l ng d p 0. 01 2 0. 58 5 0. 01 3 0. 53 0 0. 01 4 0. 54 9 0. 01 5 0. 51 7 0. 02 8 0. 47 1 0. 03 0 0. 43 9 f d i -0 .3 62 0. 03 6 -0 .3 85 0. 02 6 -0 .2 17 0. 22 9 -0 .2 40 0. 18 3 -0 .5 48 0. 07 8 -0 .5 74 0. 06 5 l ne qu it y 0. 11 7 0. 05 7 0. 12 6 0. 04 1 0. 08 4 0. 19 6 0. 09 2 0. 15 4 0. 17 5 0. 11 7 0. 18 4 0. 10 0 d is c l -0 .0 04 0. 19 1 -0 .0 04 0. 16 1 -0 .0 04 0. 12 9 -0 .0 05 0. 11 7 -0 .0 08 0. 09 8 -0 .0 09 0. 08 7 t ec h 0. 00 0 0. 99 0 0. 00 0 0. 98 0 0. 01 8 0. 33 4 0. 01 8 0. 33 7 0. 05 7 0. 07 9 0. 06 0 0. 06 4 l is t a n n 0. 01 0 0. 43 6 0. 00 7 0. 59 3 0. 03 0 0. 03 3 0. 02 7 0. 05 5 0. 03 0 0. 21 5 0. 03 2 0. 19 5 f e m a l e 0. 01 4 0. 49 7 0. 05 0 0. 02 7 0. 07 7 0. 04 6 f s iz e -0 .0 04 0. 76 4 0. 02 8 0. 13 5 0. 07 0 0. 03 2 m s iz e -0 .0 15 0. 09 5 -0 .0 15 0. 25 8 0. 00 2 0. 94 1 a dj r 2 4. 1% 5. 1% 6. 0% 6. 7% 7. 8% 8. 6% f 1. 42 1. 47 1. 61 1. 63 1. 82 1. 83 pva lu e 0. 17 8 0. 14 8 0. 10 7 0. 09 5 0. 06 2 0. 05 4 n 10 7 10 7 10 7 10 7 10 7 10 7 huang et al.: market reaction to audit committee gender 72 the international journal of banking and finance, vol. 8. number 1, march 2011: 59-75 however, 23 of the 271 u.s.-traded foreign fi rms (8.4%) restated their fi nancial statements in the year subsequent to the appointment of male audit committee members. accordingly, the percentage of restatements in the female group is signifi cantly lower than that of the male group. 4.4 sensitivity analysis the market reaction of fi nancial fi rms may be different from other industries due to their unique characteristics. to control for this potential effect, an additional dummy control variable, finance (fi nance fi rms, defi ned as sic= 60006999), were included in the two ols regression models. the results show that finance is insignifi cant. the results of gender variables remain consistent with the results presented in table 5. we also designed a 1 by 1 matched sample based on the fi rms’ sic code as well as size and re-tested our mean and median differences of car between female and male groups in table 3. the difference in car between the male and the female groups is consistently signifi cant (p<.05, two-tailed) in window (-3, +3) and (p<.10, two-tailed) in window (-5, +5). 5. conclusion researchers have recently begun to examine the important impact of gender diversity on audit committee effectiveness (two recent studies are huse & solberg, 2006; ittonen et al. 2009). this is particularly important given the enhanced role of the audit committee function under sox. also, the increased importance of foreign issuers to the american capital markets and concerns over competition from other exchanges (balakrishnan, 2007) suggests that it is extremely important to understand how the characteristics of audit committees affect the quality of fi nancial reporting of foreign fi rms. in this study, we investigated the relationship between the appointment of female and male audit committee members and related market-price reaction for foreign fi rms. we found that the appointment of female audit committee members has signifi cant positive cumulative abnormal returns compared to the appointment of male audit committee members. furthermore, the greater the number of female audit committee members appointed, the greater the positive market returns incurred. our evidence reinforces the belief that gender diversity increases investors’ confi dence in a fi rm’s audit committee and its ability to provide effective oversight over the fi nancial reporting process as well as the external audit. these results have implications for global investors who seek to invest in the u.s. capital markets. these fi ndings also suggest that, if the sec’s major goal is to improve the quality of corporate fi nancial reporting, it should further stress the importance of female presence on the audit committee. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 4 market reactions to audit committee director's gender: evidence from us-traded foreign firms: 59-75 73 5.1 limitations and avenues for future research the following limitations apply to this paper. first, this study focused on the audit committee gender diversity of foreign fi rms traded in the us. the limited sample unavoidably reduced the power of the models and the signifi cance of our variables. second, we only examined the market reactions in three event windows following the appointment of a female audit committee member. the three event windows may not fully refl ect the overall impact of this type of event. future studies can explore whether audit committee gender diversity affects the fi rm’s corporate governance by facilitating better earnings quality for us-traded foreign fi rms. author information: hua-wei huang and james m. fornaro are associate professors; yun-chia yan and ahmed elshahat are assistant professors in their respective universities. their e-mails are: hwawei7@yahoo.com.tw; yunchiayan@gmail.com; fornaroj@oldwestbury.edu; and aelshahat@bradley. edu. references abbott, l., parker, s. & presley, t. (2009). female board presence and the likelihood of fi nancial restatement. american accounting association annual meeting.http://aaahq.org/am2009/abstract.cfm ?submission id= 3177. balakrishnan, j. (2007). sarbanes-oxley and foreign issuers: new developments. asialaw, october: 1-9. bedard, j., chtorou, s. m. & courteau, l. (2004). the effect of audit committee expertise, independence, and activity on aggressive earnings management. auditing: a journal of practice and theory, 23(2), 13-35. bernardi, r. a. & arnold, d. f. (1997). an examination of moral development within public accounting by gender, staff level, and fi rm. contemporary accounting research, 14(4), 653–668. betz, m., o’connell, l. & shepard, j. m. (1989). gender differences in proclivity for unethical behavior. journal of business ethics, 8(5), 321-324. carter, d., d’souza, f., simkins, b. & simpson, g. (2008). the diversity of corporate board committees and fi nancial performance. working paper. available at ssrn: http://ssrn.com/abstract=1106698. chen, k. c. w. & church, b. k. (1996). going concern opinions and the market’s reaction to bankruptcy fi lings. the accounting review, 71(1), 117-129. choi, j. & wong, t. (2005). auditors’ governance functions and legal environments: an international investigation. cig working paper series. center for institutions and governance: chinese university of hong kong. huang et al.: market reaction to audit committee gender 74 the international journal of banking and finance, vol. 8. number 1, march 2011: 59-75 choi, j., kim, j., liu, x. & simunic, d. (2008). audit pricing, legal liability regimes, and big 4 premiums: theory and cross-country evidence. contempora ry accounting research, 25(1), 55-99. cohen, j., pant, l. & sharp, d. (1998). the effect of gender and academic discipline diversity on the ethical evaluations, ethical intentions and ethical orientation of potential public accounting recruits. accounting horizons, 12(3), 250-271. chung, j. & monroe, g. (2001). a research note on the effects of gender and task complexity on an audit judgment. behavioral research in accounting, 13, 111-125. davidson, w., xie, b. & xu, w. (2004). market reaction to voluntary announcements of audit committee appointments: the effect of fi nancial expertise. journal of accounting and public policy, 23(4), 279-293. defond, m. , hann, r. n. & hu, x. (2005). d oes the market value fi nancial expertise on audit committees of boards of directors? journal of accounting research. 43(2), 153-193. deloitte touche tohmatsu. (2010). statistical database relevant to international fi nancial reporting. ias plus. available at: http://www.iasplus.com/stats/ stats.htm. eaddy, w. r., faciszewski, s., fitz, j.d. & vanlowe, r.s. (2003). sarbanesoxley one year later: the impact on foreign issuers. association of corporate counsel annual meeting. gul, f. a., srinidhi, b. & tsui, j. (2008). board diversity and the demand for higher audit effort. working paper. available at: http://ssrn.com/ abstract=1359450. hostak, p., karaoglu, e., lys, t. & yang, y. g. (2009). an examination of the impact of the sarbanes-oxley act on the attractiveness of u.s. capital markets for foreign fi rms. working paper series. available at: http:// papers.ssrn.com/sol3/papers.cfm?abstract_id=956020. huse, m. & solberg, a. g. (2006). gender-related boardroom dynamics: how scandinavian women make and can make contributions on corporate boards. women in management review, 21(2), 113-130. ittonen, k., miettinen, j. & vähämaa, s. (2009). does female representation in audit committees affect audit fees? working paper series. available at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1314243. klein, a. (2002). audit committee, board of director characteristics, and earnings management. journal of accounting & economics, 33(3), 375-400. krishnan, g. v. & parsons, l. m. (2008). getting to the bottom line: an exploration of gender and earnings quality. journal of business ethics, 78, 65–76. levi, m. d., li, k. & zhang, f. (2008). mergers and acquisitions: the role of gender. working paper. available at ssrn: http://ssrn.com/ abstract=1123735. mcmullen, d. a. & raghunandan, k. (1996). enhancing audit committee effectiveness. journal of accountancy, 182 (2), 79-81. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 4 market reactions to audit committee director's gender: evidence from us-traded foreign firms: 59-75 75 neidermeyer, p. e., tuten, t. l. & neidermeyer, a. a. (2003). gender differences in auditors’ attitudes towards lowballing: implications for future practice. women in management review, 18(8), 406-413. pillsbury, c. m., capozzoli, l. & ciampa, a. (1989). a synthesis of research studies regarding the upward mobility of women in public accounting. accounting horiz ons, 3(1), 63-71. ruegger, d. & king, e. w. (1992). a study of the effect of age and gender upon student business ethics. journal of business ethics, 11(3), 179-186. schaefer, j. & welker, r. (1994). distinguishing characteristics of certifi ed public accountants disciplined for unprofessional behavior. journal of accounting and public policy, 13(2), 97-119. wild, j. (1996). the audit committee and earnings quality. journal of accounting, auditing & finance, 11(2), 247-276. wingate, m. (1997). an examination of cultural infl uence on audit environment. research in accounting regulation, 11(supplement), 129-48. huang et al.: market reaction to audit committee gender ijbf7-marina.indd the international journal of banking and finance, vol. 7. number 1: 2010: 1-18 1 ijbf profit sharing ratios in mudaraba contracts revisited* zubair hasan inceif: the global university of islamic finance, malaysia _____________________________________________________ abstract this paper examines three interlinked issues: first, what is the current state of profi t sharing in islamic banking, that is, is the division of profi t between the banks and the depositors satisfactory? second, can the profi t sharing in a two-tier mudaraba contract give the same rate of return to depositors as the bank receives from the investment of their deposits in business? finally, can the central bank use in some ways the profi t sharing ratio along with the rate of interest as an instrument for credit control in a dual banking system? the answer to the fi rst two questions is in the negative. to the third, a tentative response is yes. the paper also suggests a policy tool the central banks can presumably use for controlling credit, more so in view of the recurring fi nancial crises like the one emanating from the us that the world is facing today. the tool may in addition help improve the link between the banks and depositors by adopting an iniquitous distribution of profi ts. keywords: islamic banking, two-tier mudaraba, profi t sharing ratio, division of profi t credit control. jel classifi cation: g12. _____________________________________________________ 1. introduction in an earlier article (hasan, 1985), i had explained the juridical position in mudaraba contracts on the profi t sharing ratio (psr) between the borrowing *the views expressed in this paper are those of the author; not necessarily of inceif, where he works. note from editors: this paper is one of three best papers selected by a review panel of 3 professors at a symposium held in november, 26-28, 2008 in melbourne, australia. the symposium was funded by the australian research council grant, 2007-2009/10, for research on islamic banking and finance. ht tp :// ijb f.u um .e du .m y 2 the international journal of banking and finance, vol. 7. number 1: 2010: 1-18 fi rms and the islamic banks that were assumed operating in competition with their mainstream interest-based counterparts in a dual banking system. that assumption became a reality when malaysia allowed mainstream banks to open islamic windows.1 the paper identifi ed the main determinants of the psr as (i) the expected rate of profi t, r, on investment, (ii) the proportion of borrowed funds in total capital fi rms employed in business, (iii) the market rate of interest r i and (iv) the risk premium . the paper dealt with the issue both at the macro and micro levels and showed that in principle, the islamic system had superiority over interest bearing mainstream banks in matters of returns and stability. the main elements of that work were incorporated later in a comprehensive discussion on mudaraba (hasan, 2002) which dealt at some length with reasons of its relative unpopularity in modern times and suggested ways to overcome the diffi culties. in the present paper we shall desist from going over the material and areas already covered in these writings, more so because they have already become a familiar part of the knowledge on the subject.2 however, some misgivings and several new developments in the area have prompted the current revisit. siddiqui (2008) presents a critical appraisal of the main theoretical models developed in the area of islamic banking over the years. his survey especially highlights striking similarities the model of anwar (1987) has with that of sargent (1979), and notes that former has merely replaced the rate of interest r with a rate of profi t ө to make the latter look islamic (p.250-251). because anwar assumes ө as known, kө in his model, according to shamim, just works as the rate of interest r and that is “why all of the model appears to be similar to the conventional classical or keynesian model” (p.254). another paper that has received a major part of shamim’s attention is of mohsin khan and abbas mirakhor (1989) where he provided clarifi cation to support some of the positions the authors had taken. in conclusion, siddiqui felt that a number of questions related to the islamic monetary system remain unanswered, but more signifi cant, he thought, is the failure of islamic banking to use the profi t and loss sharing modes of fi nancing on the assets side of the balance sheet, which could alone necessitate any meaningful change in monetary policy and the tools used for its implementation. there seems to be no diffi culty in agreeing with him on the point. 1the policy has of late been modifi ed. mainstream banks are now encouraged to have exclusive islamic subsidiaries instead of windows. the subsidiaries fall under the governance of the islamic banking act 1983 while windows are covered by the bafia of 1992, conventional banks. however, there is yet no legal bar on opening the windows. 2the 1985 and 2002 articles have often appeared as references in the writings on islamic fi nance. both have been downloaded or their abstracts accessed on the internet at a combined average exceeding three a day over the past 12 months ending august 2008. web: <http//logec.repec.org/ra5/pha42htm>. ht tp :// ijb f.u um .e du .m y profi t sharing ratios in mudaraba contracts revisited: 1-18 3 finally, in the process of surveying the literature, two of my writings alluded to above dealing, among other things, with the sharing of profi t ratios attracted siddiqui's attention, more than what i feel they deserved. however, his enlightening comments made me sit up and think afresh on the issue of using the ratios. the present paper has three basic objectives. (1) to have a look at the sharing ratio theory and the way it is being currently used in islamic banking. this is an addition to my earlier deliberations on the subject and is taken up in the following sections 2 and 3. (2) to examine if there would not be any difference between the rate of profi t a bank may earn on investing customers’ deposits in business on the one hand; and the rate of profi t it could allow to them on their money on the other. this discussion is in response to a point siddiqui has raised on my review in defense of khan and mirakhor (1989). section 4 is devoted to reinforce my position. in section 5, the discussion extends to a more important issue: the adequacy of return banks provide to the depositors on investment. do the prevalent profi t sharing ratios result in a fair distribution of profi t between the bankers and the depositors in islamic fi nance and if not what can be done to remedy the situation? (3) to review whether the central bank could use the sharing of profi t ratio, as is at times suggested, for controlling credit, assuming that islamic banks can and do create credit. i have discussed this issue in a recent paper (hasan, 2008) and make on it some more observations here in section 6. finally, section 7 contains a few concluding remarks. 2. profi t sharing theory the initial theoretical models of interest free banking were based on the view that "no risk, no gain" alone was the principle in islam for organizing banking operations. the claim got inspiration presumably from the early days of islam when mudaraba was the dominant mode for fi nancing specifi c business projects or trading partnerships. that the claim was only partially true has already been demonstrated (hasan, 2005). there can be areas such as leasing or mark-up pricing where gain can arise without virtually involving any risk in an islamic contract. it may be mentioned that the notion of profi t sharing pervaded even conventional business organizations, let alone islamic fi nance. for example, mainstream economics now sees profi t in sharing profi t with labour to the extent it helps maintain industrial peace. in partnership contracts also, it allows profi t sharing ratios for some of the participants to differ from their loss sharing ratios as in mudaraba.3 the partnership contracts defi ne profi t sharing ratio as the one in which profi ts or losses of a business are shared as set out in the agreement. the 3in fact, mudaraba was a pre-islamic mode of profi t-sharing fi nance that fl ourished as a dominant form of business organization around thirteenth century in the muslim lands. ht tp :// ijb f.u um .e du .m y 4 the international journal of banking and finance, vol. 7. number 1: 2010: 1-18 4profi t sharing ratio – dictionary defi nition of pofi t sharing. 5paraphrasing bank negara malaysia, mudaraba is an agreement made between a party who provides the capital and the other an entrepreneur – who is thus enabled to carry out business projects on the basis of sharing profi t in pre-agreed ratios. however, losses, if any, are borne solely by the provider of funds. bank negara malaysia http://www.bnm. gov.my/index.php?ch=174&pg=469&ac=383. 6it is this classical puritan model of mudaraba that underlies the discussion in many writings on the subject including that of shamim. ratios are usually expressed as a percentage of the total profi ts each partner will get. in some agreements there is a fi rst charge on profi ts, the remainder is then distributed according to the profi t sharing ratios the agreement contains. the profi t sharing ratios are in general proportionate to capital contributions of the partners but that need not always be the case; the agreement may specify a different ratio for any of the partners.4 thus, there are resemblances between mudaraba on the one hand and modern partnership contracts on the other. however, differences between them, especially because of the different treatment of the interest factor are much more signifi cant. for example, in conventional partnerships, the profi t and loss sharing ratios of partners are mostly the same as their capital contributions but in mudaraba the two are invariably different. also, the non-intervention of the fi nancier (bank) in the management of business is a mudaraba imperative, but in conventional partnerships no partner can automatically be excluded from participation in managing a fi rm’s business unless he agrees to be a sleeping partner. mudaraba is a contract in which a fi nancier, say a bank, provides funds to an entrepreneur (fi rm) for investing in a business venture to share profi ts in an agreed proportion, the loss falling on capital alone.5 this view implies what we may call a pure mudaraba model where the fi nancier is assumed to provide the entire capital to an empty handed entrepreneur; the model fi ts well even today to small partnership businesses to undertake specifi c projects6. but the modern economic scene is dominated by large corporations that have long eclipsed small proprietary businesses. likewise, banks have almost completely replaced personal fi nancing of the earlier era with institutional arrangements (an important point to understand that profi t share is different from interest). what realistically fi ts in the present situations is the model of what we can term as mixed mudaraba, where the bank is an outside fi nancier providing fund to running businesses on a profi t sharing basis. corporations operate mostly with their owner shareholders’ money supplemented by bank fi nance, if need be. banks likewise fi nance many and varied sort of businesses simultaneously. in a mixed mudaraba model  fi rst mooted in siddiqi (1975)  the bank provides  fraction of total capital k invested in a business. thus, borrowed amount of money l divided by k equals where operates both as the loss sharing ratio for the bank as well as the leverage measure for a fi rm. it makes the business owners’ portion in capital equal to (1 k. of course, losses if any will be shared between the fi rm and the bank in the same ratios as are their capital contributions i.e. (1-) and  respectively. ht tp :// ijb f.u um .e du .m y profi t sharing ratios in mudaraba contracts revisited: 1-18 5 profi t sharing applies to earnings that are allocable to the part of capital k a bank provides to the fi rm. thus, if p were distributable profi ts, p would be allocable to bank fi nance. it is this part of profi t which is the subject matter for sharing with the fi rm. negotiations between them lead to the decision that a fraction of this, say *, will go to the bank and the remaining (1-*) the fi rm will retain for entrepreneurial services it rendered to make bank money earn a return. it is easy to see what goes to the bank is a smaller fraction, say , of total profi t p than *. for, * p, the bank’s profi t share, divided by p would equal * . in  = *  both * and  being less than 1, their product must be smaller than either of them. the derivation of  allows the treatment of the ratio issue at the macro level and helps construction of models to show that profi t sharing ratio is a function of the variables identifi ed earlier i.e. the expected rate of profi t r on capital k, the proportion of borrowings in it, the market rate of interest r i, and the risk premium we have shown earlier that the sharing ratio for bank would be as under (hasan, 1985): (1) thus, in a competitive setting the sharing ratio  at the macro level varies inversely with profi t expectations r and directly with the remaining three determinants  and r i . we shall use this result in the following sections. 3. profi t sharing in practice many banking companies, notably in pakistan, have been successful in mobilizing large amounts of money from the people in the form of deposits and publicize their profi t sharing ratios as well. to illustrate, for the rhb-islamic mudaraba is a term deposit “based on the concept of profi t sharing. under this concept, customers will provide the capital for the bank to invest for a fi xed duration. the profi t earned from the investment will be shared as dividend between the customers and the bank in the predetermined profi t sharing ratios”. investment accounts are classifi ed as general where the bank is free to decide the use of funds; and as special where the customer has specifi c avenues to choose from; other rules of the game remaining the same. the minimum initial deposit size for the general investment account is rm 5000 for one month or rm 2500 for two months; after which the deposit could be invested for specifi c tenure ranging from 1 to 60 months. for special investment accounts the required minimum deposit amount is rm 100,000 for infl exible duration of 365 days. the profi t sharing ratios are claimed to be the result of negotiations between the parties. the claim is fallacious; banks invariably use standard form contracts; signing on blank spaces is neither expressive of free will nor of negotiation. )r( r i ht tp :// ijb f.u um .e du .m y 6 the international journal of banking and finance, vol. 7. number 1: 2010: 1-18 habib bank of pakistan announces each quarter two sets of profi t sharing rates for the depositors: (i) declared for the preceding quarter and (ii) the probable ones for the next. it fi rst apportions gross revenue into 30% for the bank and 70% for the depositors. it may be presumed in the absence of required information that profi t allocable for distribution is divided between various deposits categories on the basis of weight assigned to each. profi t rates are then calculated and declared for separate categories.7 table 1 provides the types of deposits, and the profi t rates as well as the weighting system for the quarter ending june 30, 2008. table 1 habib bank, profi t distribution for the quarter ending june 30, 2008 savings million rupees profi t rates % weights assigned less than 1 5.20 0.65 1----------10 5.50 0.67 10-----------50 5.75 0.70 more than 50 5.90 0.72 term deposits duration - - 7 days - 0.72 1 month 6.0 0.75 3 months 7.5 0.80 6 months 8.5 0.90 1 year 10.2 1.10 3 years 9.5 1.25 5 years - 1.30 equity fund - 1.50 source: constructed from the data reported at the website of the bank. a perusal of the data on the ratios obtained from the websites of four other islamic banks including rhb, standard chartered saadiq, meezan and dawood revealed the same pattern as of al habib shown in table 1.8 some interesting common features  albeit varying in details  of banks’ sharing profi ts with their depositors are as follows: 7the website does not provide explanation as to why psr is applied to gross revenue in the fi rst instance, how net profi t going to depositors is calculated or what is the basis of assigning relative weights to different types of deposits. 8the remaining four banks have patterns closely similar to that of habib bank. tables for them are not produced for that reason. ht tp :// ijb f.u um .e du .m y profi t sharing ratios in mudaraba contracts revisited: 1-18 7 deposits are accepted in investment or saving accounts; the investment category is further divided into general and special. the profi t sharing ratio varies from 30% to 40% for the bank; the other part is allocated to the depositors. some banks apply the ratio in the fi rst instance to gross revenue and then distribute profi t among the depositors allocable to their share of gross revenue. the distribution of profi t among the depositors in the pool varies with category; savings receiving lower proportion than investments. within the category, rates of profi t mostly move up with the amount and duration of deposits. we could not obtain information on how individual banks arrive at these rates. banks included in the study all claim that the profi t sharing ratio is the result of negotiations with the depositors. one is not sure if depositors, especially the smaller ones really have negotiating power and get opportunity to exercise it; or they simply sign on the dotted lines in the bank documents. the loss if any is borne by the depositors pro rata; the assumption being that banks have no money of their own to invest or keep it distinctly separate from that of the depositors in matters of investment. the pure mudaraba model is implied operating which is most unlikely in modern times. in any case, the question is from where does the profi t to be distributed among the depositors come from? suffi cient data is not available on the uses of funds side of islamic banks, so we cannot answer this question. the information on the sharing of profi t ratios between the banks and the fi rms they invest in the money of their depositors is all the more scanty9. however, this brings us face to face with the doubts siddiqui has raised in his paper (p.247) with reference to my comments on khan and mirakhor (1989, section 2, pp. 86-87). 9the malaysian economic report 2008/2009 revealed two interesting features of islamic banking in the country. firstly, among the deposits, the substantial (26%) and the highest growth (47.8) category of deposits over the two years is unspecifi ed as ‘others’. secondly the bulk of fi nancing (60%) goes to the household sector signifying the dominance of fi xed return murahaba in microfi nancing and expanding credit card business; other sectors  agriculture manufacturing trade and insurance – put together receive the remaining 40% (the new straits times supplement, 30 august 2008: p.7). ht tp :// ijb f.u um .e du .m y 8 the international journal of banking and finance, vol. 7. number 1: 2010: 1-18 4. profi t rates equality: untenable modern banks serve as intermediaries between the providers of funds, the depositors on the one hand and their users, mostly the business fi rms on the other (and governments). the profi t banks distribute among the depositors comes from the investment of deposit money in business ventures the fi rms undertake. the relationship concerning profi t sharing between banks and fi rms is of the same sort as between them and their depositors in a two-tier model. here, we fi rst have a contract between the banks and the depositors which eventually gives the latter a rate of return r b on their deposits (d b ). secondly, there is a contract between the banks and business fi rms –the entrepreneurs – who borrow from the banks. the banks ultimately get a rate of return (r) on loans they advance. palpably, r b must be an increasing function of r operating through the profi t sharing ratios agreed upon between the banks and fi rms on the one hand and the banks and depositors on the other. khan and mirakhor used this two-tier model for their work making a number of assumptions: explicit and implicit. one of these assumptions that i missed to take note of in my comments, says siddiqui (p.247), is that a bank in mudaraba fi nance has no money of its own to invest.10 even if the bank has, we may further assume that it keeps the account for investments of its own money entirely separate from the investments of clients’ deposits. but even the extension of the assumption makes little difference to our argument. in the business of a fi rm the contribution of the bank, even if consisting solely of customers’ investment deposits, will still be a part i.e. a proportion, say , of the fi rms’ total capital k. the profi t allocable to bank fi nance will only be p of the fi rms’ aggregate profi t p. of p,*p alone will be available to the bank, giving us: (2) if the sharing ratio between the bank and the depositors were μ, the depositors would get: (3) b * d p r b * b d pμ rμ r 0 < μ < 1 10in fact, there was no lapse on the point the assumption was explicitly stated in our comments (hasan, 1991: p.85) furthermore, even if f b and d b were taken as equal in khan and mirakhor, their defense siddiqui bases on that equality (p.247), would not hold; r b will still be less than r. see equations (2) and (3). ht tp :// ijb f.u um .e du .m y profi t sharing ratios in mudaraba contracts revisited: 1-18 9 clearly, r b ≠ r. it will be so even if  = 1 as siddiqui suggests. since the entire structure of khan and mirakhor rests on the untenable equality of rates, r b = r assumption, little remains defensible in that structure. siddiqui makes another effort to salvage the equality of rates proposition. he argued that there was no need to use mudaraba on the assets side of the argument (p. 247). the assumption of a musharakah contract between the bank and the fi rms, he feels, would have kept intact the model of khan and mirakhor. however, in that case too, the bank would get from the fi rms onlyp from total profi t and r would equal p/d b . the depositors will have to leave p to the bank, making r b again less than r. thus, the result would be no different even if we use musharkah in place of mudaraba on the asset side of the balance sheet; our 1985 position on profi t rates thus remains intact in either case. in fact, our comments indicated many other blemishes in the khan and mirakhor model some of which siddiqui has also endorsed. 5. the equity question the equality of rates is not a pertinent question in islamic fi nance. rather, pertinent is the question of fairness concerning the return the customers receive on their deposits in islamic banking. for, fairness and justice is what islam essentially stands for (in all human dealings). sharing of profi t is mandatory in islamic banking but its essence is what such sharing results in? in other words, what it gives to the depositors compared to the owners (shareholders) of bank in the mixed mudaraba contracts. let me preface the discussion on the point with an observation. history bears evidence that fi nancing has ever been an instrument in the hands of the rich used against the poor for exploitation and oppression. in the past, the fl ow of funds in the form of loans was from the rich toward the poor; interest rates were kept high rather exorbitant even as most of borrowings were for consumption purposes. the poor suffered. today, the fl ow of funds with banks operating as intermediaries is from the poor toward the rich as the major part of national savings comes from the lower and middle income groups; it is pooled in provident and pension funds or insurance premiums. the funds so pooled go through the banks to the rich business tycoons of the community owning and controlling big businesses. the rates of interest are kept low; cheap money policy dominates modern economies. thus, the multitude of depositors relatively poor from the lower rungs of society is cut off from high returns their savings help businesses earn by low interest rates they get. infl ation is the order of the day that reduces their real value further. to make this binding, laws have cropped up over 200 years to make the small lender (perhaps on profi t shares) to become punishable if he/she lends money, since bank is the sole lender today. cheap money policy makes the ht tp :// ijb f.u um .e du .m y 10 the international journal of banking and finance, vol. 7. number 1: 2010: 1-18 rich available deposit funds just for a song to magnify their profi ts via leverage.11 non-bank fi nancial institution shareholders bring in about 40%-70% of capital and only rely on debt for the rest. but the banks take the deposits as bank debt from small depositors to the tune of possible 90%! thus, the free enterprise system was unjust and exploitative of the poor in the past; and so it is today, thanks to the institution of interest. the return to depositors in islamic banking has made the situation no better; rather it seems to worsen it further. the situation has to be blamed mainly on the use of standard form contracts. the rates of return depositors are normally getting are not much different from the corresponding interest rates see table 1 offered by the conventional banks on customer deposits.12 “the bank for international settlement reports that commercial banks earn anywhere from 11% to 3% per year as roe going to their shareholders (the profi t to bank owners after covering all costs). if there is true profi t sharing in the promotion of profi tshared banking as is so earnestly being shoved into the muslim conscience, is a rate of profi t share that is equal to the conventional “interest rate” of roughly half the 11%-31% going to the (bank) shareholders just? common sense (leave alone the justice as in qur’an) would suggest that profi t share is totally misunderstood [in islamic fi nance].”13 the concept of profi t sharing loses meaning if it is divorced from the basic norms of justice and fair play shar’ah insists on promoting. how can this be done is a complex question but the key presumably lies in central banks exercising some control over the profi t sharing ratios. classical jurists mostly conceived of mudaraba in its puritan form where the entrepreneur (mudarib) was empty handed of all money for business coming from the fi nancier (the rub-ul mal): they could not foresee the modern day fi nance. the pronouncements on the magnitude of profi t sharing ratio for the fi nancier were based on this view of mudaraba. varying ratios were considered allowable by different schools and scholars, the malikis allowing up to 50% for 11to illustrate, the bank in new delhi where i maintain a nre account offered me on 1 october, 2008 a long-term deposit plan for 10 years. return of capital was guaranteed with the promise of a 10% to 12% percent annual return arising from a structured investment; the offi cer explained that the structure was composed of government securities, infrastructure lending and equities; the annual yield being 30, 35, and 30 to 38% respectively. the return offered being around a third of what the bank would get on my money, looked unfair but i was told that it as a ‘take it or leave it affair’. the infl ation running in double digits, was otherwise also a losing game. i declined the offer and invested the money in real estate. 12this author raised this point in a internal seminar inceif held on 28 august, 2008 on settlement of islamic finance and banking disputes: issues and solutions where bankers and shari’ah scholars were the panelists. the justifi cation a banker who also presented an excellent paper to form the basis for discussion provided was that islamic banks, unlike their mainstream competitors, do not impose any penalties on depositors for premature withdrawals of their money. one is not sure how signifi cant are such withdrawals. 13i endorse this observation an unknown referee made in comments on the earlier draft of this paper. ht tp :// ijb f.u um .e du .m y profi t sharing ratios in mudaraba contracts revisited: 1-18 11 the fi nancier (hasan, 1985). such high sharing ratios could be considered just in that version of mudaraba and the tiny scale of business operations. in mixed mudaraba of today and large scales of operations, a 30-40 range for profi t share can be shown as exploitative of the depositors. let us assume that in a project, the total bank fi nance is $10,000 (k). of this the bank provides 4,000 (k b ) or 2/5 =  of k and uses deposits worth $6000 (k d ). the psr, or * for the bank, is 30% and the project ends with a profi t p of rm 4,000. under mixed mudaraba rules 40% of profi t (p) = $1,600 is accrued to the bank on k b and the remaining rm 2,400 to the depositors. now, of the latter amount 30% or $720 will be the profi t share for the bank; the remaining $1,680 will be available for distribution among the depositors. thus, the total profi t going to the bank would equal 1,600 + 720 = 2320 giving it a return on k b equal to 58%, while the depositors will get on (k d ).just 28%. the gap between the two returns equals 30% even as the risk exposure for both was the same. the reason is the high profi t sharing ratio for the bank. what sharing ratio will keep the gap reasonable, perhaps 10%? we explore below if a solution were available. presumably, one must target the leverage gain available to banks in islamic fi nance as a control variable14. the recent failures of such giant fi nancial institutions as lehman, merrill lynch and the aig seemed to be stoking the same sort of fear of debt in the us and europe as in japan during the 1990s. the facts of the japanese case broadly were these. the nikkei index peaked at 38,916 on 29 december, 1989 at the end of a fi ve year orgy of debt-fuelled speculation centered largely on the real estate market. during the fat years, banks lent against property in the confi dent expectations that prices would never fall, but from 1990 share prices started a 13 year decline, punctuated by sharp rallies. the japanese were undone by the deception of rising leverage gains (as was the origin of the global financial crisis 2008-09). soon the lure became the demon of ‘de-leveraging’ staring them in the face: they had to pay the price.15 leverage must all the more be a matter for concern in islamic fi nance as islam, apart from placing emphasis on equitable distribution, does not encourage borrowing in principle. companies that have too much borrowing i.e. a debt ratio of more than 33% of their stock market value stand out of bonds. such criterion means that shari’ah-compliant investors are to steer clear of highly leveraged banks, conventional or islamic.16 indeed, it is time for the world as a whole to be wary of loans. 14it seems to me that the lure of unbridled leverage gains may have something to do with the cloud that taqi usmani recently caste on sukuk and the malaysian court declared bba contracts banks executed invalid. both are permissible in islamic law; documentation spelling out the conditions went wrong. 15the story is being repeated in the us. the sub-prime debacle has certainly put the economy into recession that may continue for years despite the massive $700 billion bailout effort. there has already been a loss of 71,000 and 159, 000 jobs in august and september 2008 respectively. a separate labor department survey of households put the unemployment rate at 6.1%, a fi ve year high [toi, p.22]. 16see ‘faith-based fi nance’. the economist (print edition) sep. 4, 2008. ht tp :// ijb f.u um .e du .m y 12 the international journal of banking and finance, vol. 7. number 1: 2010: 1-18 going back to our main argument, let us postulate that the central bank allows an islamic bank the leverage gain not to exceed 10% additional to the rate of profi t on total capital k (= k b + k d ) it had invested in business. beyond that all profi t is to go back to the deposit holders. now, in the above illustration, the return on capital employed is [(4000/10000)100] or 40 per cent. so, the bank can have under the constraint a maximum of 40 + 10 = 50% return on its portion of capital (4,000), i.e. it cannot have more than $ 2,000 as profi t. of this 1,600 has already accrued on its capital. so, it will have another rm 400 from the remaining profi t to fi ll the allowable gap. this leaves $1,600 for distribution among the depositors that would give them a return on k d of 33.3 per cent. the constraint would thus reduce the profi t sharing ratio  for the bank from the original 30% to 16.67% in an ex post adjustment.17 we may thus state that the islamic banks are free to negotiate the psr with the depositors subject to the provision that their leverage gain will not exceed by more than % over the rate on k. let us fi x the rule using the symbols in our illustration. the rate of profi t on capital employed r = p/k and the maximum leverage gain allowed to the bank is . the upper limit for return on k b = k, therefore, is r + . the profi t allocable to the bank is p but it also gets  fraction of profi t allocable to deposits, i.e. of (1 –) p. thus, we may set up: (4) notice that is equal to k b / k d : an alternative expression for leverage. it is obvious that for any given values of r and the psr would vary directly with . thus,  can be a policy variable that the central bank of a country can use for mandatory ex post adjustment of the psr in islamic fi nance to enforce fairness in the distribution of mudaraba profi ts. 6. profi t sharing ratio and credit control i had argued in an earlier paper (hasan, 2008) that in principle credit creation (and destruction) by banks is, within confi nes, an economic imperative for frictionless running and stability of an economy. in principle there presumably is no shari’ah provision denying islamic banks to participate in the process. siddiqui in his paper under reference (section 2.2; pp. 237-238) provides a neat 1 . r toreduces this r kk p 1 p * b * 1 17the bank gets 1600 as profi t accrual on its capital and the remaining 400 from 2400 allocable to depositors. thus,= 400/2400 = 16.67%. ht tp :// ijb f.u um .e du .m y profi t sharing ratios in mudaraba contracts revisited: 1-18 13 summary of the earlier literature on the point for us. he seems to go with the view that the absence of rate of interest from the scene in the islamic system of fi nance does not reduce the tools of monetary policy because the profi t sharing rates (ratios) can serve as a replacement in contrast, i had expressed the opinion that profi t sharing ratio is apparently not a price for credit like the rate of interest and cannot, therefore, take its place in the central banks’ arsenal of credit control. i shall now discuss the implications if attempts were made to use the sharing ratio for the purpose. the exercise has signifi cance because in most countries a dual monetary system with mainstream commercial banks operating along with islamic banks is in operation but it is regulated by a unitary policy. the use of profi t sharing ratio as a monetary policy tool raises some ticklish questions. for instance, which of the profi t sharing ratios  or *  is to be the target variable? here, the choice is not diffi cult; monetary policy being a macro level matter palpably wins the day. under mudaraba rules,  is essentially a matter of negotiations between the parties which we know is not the case with the rate of interest. the size ofcould differ from customer to customer within a bank as also between banks. more complex questions include if in a dual banking system both the bank rate and psr manipulations will be needed for simultaneous use or in isolation of one another? if used together, could the two be moved in the same direction or will have identical impact? figure 1 may help answer some such questions. it has two sections x and z. in section x, we have shown the relationship of profi t sharing ratio with expected profi t rate, treating ß. rate, i =  (r i + ) is a constant (hasan, 1985).18 section z relates to mainstream macroeconomic variables and their interrelationships; it is divided into four quadrants a, b, c and d. in a we show the usual inverse relationship between interest rate r i and growth in output via the is curve. quadrant b sees interest rate in a negative relationship with the expected rate of profi t r. in c we show investment having positive correlation with profi t expectations while d relates savings (= i) to growth. the solid line rectangle shows the various variables in a state of stable equilibrium that can be disturbed due to a change in any of them. if monetary policy uses (lowers) rate of interest to bring about the change (to cure recession) a whole process of adjustments is set in motion shown by the direction of arrows along the broken line path until a new equilibrium is established. the readjustment process is very complicated and immediate consequences of any change a monetary policy move may initiate are diffi cult to predict or control. figure 1 provides a rather schematic demonstration. the corner points of each rectangle show that interest rates, profi t expectations, savings = investment and growth in output can simultaneously have values compatible with one another depicting a harmonious and stable state of macroeconomic equilibrium. 18refer to equation (1) above for ready reference. ht tp :// ijb f.u um .e du .m y 14 the international journal of banking and finance, vol. 7. number 1: 2010: 1-18 however, it does not help specify and explain the sequence of events or pace of change as also the implications of happenings during transition from one state of equilibrium to another. however, the fi gure still provides some useful insights on the issue under discussion. putting x and z sections of the figure together we can venture the opinion that the use of profi t sharing ratio for credit control will not be inconsistent with the simultaneous employment of the interest rate policy as both tools are related to the expected profi t rates in the same direction. the rate of interest r i and (= *) have a positive correlation thus, even if the central bank uses only the interest rate as a policy tool, the market is likely to readjust  for new mudaraba contracts to match the altered rate of interest i p is b a i0 i0 s0 i1 i1 s1 expected profit rate (r) output (y) r0 r1 0 y0 y1 i0 s0 c d i1 s1 i s i=s investment (i) savings (s) z figure 1. relationship between profit (p) and investment (i) via interest x ß = ( ri + ) profit sharing ratio 1 2 profit rate (r) 0 figure 1. relationship between profi t (p) and investment (i) via interest ht tp :// ijb f.u um .e du .m y profi t sharing ratios in mudaraba contracts revisited: 1-18 15 rate of interest. in a dual fi nancial system like the one operating in malaysia (and most countries), this result may be of value and signifi cance; if interest rate applicable to mainstream banks were for instance raised to curb infl ation, the profi t sharing ratios will appropriately increase and the credit creation activities of islamic banks, provided they indulge in it, will automatically be curbed. even as the principle and relationships are clear enough, a central bank may face many cobwebs to clear as the questions like the ones we raised earlier will have to be answered to ensure the operational effectiveness of the instrument. for example, the constantmay change due to a change in r i , and  individually or in combination and they may change in the same or opposite directions. it would be diffi cult to predict such changes or assess their impact. thus, the overall impact of a change in  may be quite hazardous to visualize. for its signifi cance and complexities, this can be a worthwhile area for research in islamic fi nance. presumably, it would be easier and more effective to use as explained above in place of  as a policy variable. 7. conclusion this paper has shown that the current use of mudaraba contracts in islamic banking is beset with confusion and ambiguities. even a cursory look at the prevalent profi t sharing schemes (for example as in pakistan), especially on how the ratios are settled and weights assigned to different categories of deposits with reference to amount and time period involved, needs scrutiny and control. a shari’ah issue involved in the matter seems to be this: if a one month deposit were not withdrawn after the expiry of its tenure but is renewed as investment  principal + profi t earned  for another month, will such renewal not attract islamic injunctions against interest? the bankers’ response to the query at the inceif organized seminar referred to earlier was that the reinvestment of principal plus profi t is under a new contract and therefore the question is out of place. however, renewal under interest fi nance is under a de facto new contract too; renewal requires the consent of the depositor and the rate of interest may be varied. a study of profi t rates the depositors are getting in islamic banks gives the impression that, but for their commitment to faith, the believers might choose conventional banks if risk factor were taken into consideration. the suggestion is to ensure fairness in the division of profi t between the banks and the depositors that seems currently missing in the mixed mudarabah contracts.19 this paper suggests for consideration a policy variable  the maximum enhancement over 19some suggested in this context that banks may build return equalization reserves out of profi ts to smoothen rates differences over time. but what has it to do with the issue of equity? narrowing differences does not improve the level of rates. rather it raises the ticklish question of distributive justice between the present and the future generations of depositors. ht tp :// ijb f.u um .e du .m y 16 the international journal of banking and finance, vol. 7. number 1: 2010: 1-18 the rate of return r the banks get on total investment k. moreover, islamic banks and conventional banks involved in islamic fi nance may be required to publish in their periodic fi nancial statements profi t rates the banks earn on their equity side by side the rates allowed on deposits to improve information and transparency. how free are the negotiations between banks and the depositors, especially the smaller ones, may also have to be looked into. the use of standard contract forms that the depositors have no option but to sign looks patently un-islamic. table 1 suggests that smaller size deposits are presumably attracting less attention of islamic banks with reference to returns and facilities provided. should they not differ in this matter from the mainstream banks? is it possible to empower depositors by organizing them in some sort of councils for collective negotiations on psrs? such councils may be organized at the level of individual banks and have an apex body federating them. possibly, legal basis for such organizational structures can be created and perhaps the central bank may play a role in the matter? in view of the current fi nancial crisis  the worst capitalism has faced since 1930  central banks must somehow put a tab on the lure for leverage gains. otherwise, the massive bailout exercise now underway in the us,20 even if successful, will certainly not be the last. this adds weight to our suggestion of employing as a constraint for manipulating  in the case of islamic banks along with the rate of interest for the mainstream institutions to control credit in a dual banking system as the one operating in malaysia. however, it is a complicated matter involving many imponderables. much research is needed before making a decision including the changes legal framework dealing with banking would require. finally, the paper has as a minor point demonstrated in reply to shamim ahmad siddiqui that even if we use musharaka contract on the asset side of the fi nancing operations, the rate of return a bank would earn on investment of deposits in business will not be the same as the depositors will eventually receive from the bank; it would invariably be smaller. author information: zubair hasan, is a professor of islamic economics and finance, in the inceif: the global university of education in islamic finance, kuala lumpur, malaysia. e-mail: zubair@inceif.org; phone: + 603 2781 4185 (o). 20the bailout equity purchase of giants in trouble like the aig went up to 80%. at the european financial crisis summit in paris on october 4, 2008 called to seek a coordinated response to the deepening credit crunch, the italian prime minister declared: “i want the message to go out from this meeting today: no sound and solvent bank should be allowed to fall because of a lack of liquidity” (times of india, p.22). all this smacks of what people in india call a nationalization program. ht tp :// ijb f.u um .e du .m y profi t sharing ratios in mudaraba contracts revisited: 1-18 17 references alternative profi t sharing ratios. retrieved from: http://www.standardchartered. com.my/islamic-banking/personal-banking/generalinvestment. anwar muhammad. (1987). modeling interest free economy: a study in macroeconomics and development. the international institute of islamic thought, herndon, va, usa. bank al habib ltd. retrieved from: http://www.bankalhabib.com/ib_declaration ofprofi t.php. bank negara malaysia islamic banking: concepts and products. retrieved from: http://www.bnm,gov.my/index.php?ch=174&pg=469&ac=383. hasan, zubair (1985). determination of profi t and loss sharing ratios in interestfree business fi nance. journal of research in islamic economics, 3(1), 13-27. hasan, zubair (1991). the fi nancial system and monetary policy in an islamic economy by mohsin khan and abbas mirakhor (comments), jkau: islamic economics, 3, 85-93. hasan, zubair (2002). mudarabah as a mode of fi nancing in islamic banking: theory, practice, and problems. middle east business and economic review, 14 (2), 41-53. hasan, zubair (2005). islamic banking at the crossroads: theory versus practice. in munawar iqbal and rodney wilson (eds.), islamic perspectives on wealth creation (pp. 3-20).uk : wilson, edinburgh university press. hasan, zubair (2008). credit creation and control: an unresolved issue in islamic banking. international journal of islamic and middle eastern finance and management, 1 (1), 69-81. inceif ( 2008, august). settlement of islamic fi nance and banking disputes: issues and solutions. panel discussion. kuala lumpur: bank negara auditorium. khan, mohsin, and mirakhor, abbas, (1989). the fi nancial system and monetary policy in an islamic economy. jkau: islamic economics, 1(1), 39-57. lin chung-chen, chang juin-jen and lai, ching-chong (2002): profi t sharing as a worker discipline device. economic modelling, 19( 5), 815-828. malaysia home loan: profi t distribution. retrieved from: http://www.money3. com.my/malaysiahomeloanmarket/article/profi tdistribution.aspx. meezan bank: karobari munafa account: profi t calculation and distribution method. retrieved from: http://www.meezanbank.com/pages.aspx? i page id=238 . profi t in profi t sharing: time. retrieved from: http://www.time.com/magzine/ article/0,9171,790142,00.html. range of islamic banking products. retrieved from: http://www.bnm.gov.my/ index.php?ch=174&pg=467&ac=370. rhb islamic: investment categories. retrieved from: http://www.rhbislamic bank.com.my/index.asp?fuseaction=deposits.investment. ht tp :// ijb f.u um .e du .m y 18 the international journal of banking and finance, vol. 7. number 1: 2010: 1-18 siddiqi, m. n. (1978). (urdu) islamic publications, lahore. faith-based fi nance. (2008, september). the economist. (2008. august 30). the new straits times(supplement), pp. 7. (2008, october 5).times of india (tio). siddiqui, shamim ahmad (2008, april). an evaluation of research on monetary policy and stability of the islamic economic system. paper presented at the ghair soodi bankari 7th international conference in islamic economics, 235-270. standard chartered bank malaysia: general investment account alternative profi t sharing ratioos. retrieved from http://www. standardehartered. com.my/islamic-banking/personal-banking/general-investment. ht tp :// ijb f.u um .e du .m y asian derivative markets: research issues 1the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 asian derivative markets: research issues shamsher m. and taufiq h. universiti putra malaysia abstract theory suggests that the introduction of derivative market in a market with spot trading completes the market-based price discovery process. there are 19 derivative markets in asia. derivative trades in such markets help to hedge away the risk of price changes in the spot markets at very low costs. commodity futures instruments are also needed to hedge away price changes in real sector just as financial futures does this function for the financial sector. in this paper, we examine the current status of selected commodity and financial derivative markets in asia. it suggests that the more industrial/advanced economies have developed liquid commodities markets, and few of them have also developed active financial derivative markets. but for most of the 19 or so emerging markets in asia, the development of derivative markets is still at an early stage. jel classification: e44 & g15 key words: derivative markets, futures markets, price risk, commodities futures, asian markets. 1. introduction the development of the modern day auction markets for shares and bonds in spot markets can be traced to financial developments in the early 20th century in the then industrial economies. commodity futures markets also developed in parallel that was needed for the management of price changes in commodities heavily demanded in industrial economies of the last century. modern finance theories suggest that sufficiently well-structured and price-efficient spot and futures markets are necessary for providing funds needed for economic development of any country and for risk transfer emanating from price risk, the so-called basis risk when prices of assets change over time ahead of their usage. ijbf 1 the international journal of banking and finance, vol. 5.(number 1): 2007: 1-252 such markets help to price the risk of capital usages, and also provide means to transfer risk in derivative trading be it in the real sector or the financial sector. efficient capital raising and allocation of financial resources are an integral part of economic development. it is now well established that securities markets, particularly the derivative markets, are not casinos but are essential for the orderly growth and development of sustainable economic activities since the contracts traded in such markets enable economic agents to manage price risk. markets also fulfill the essential role for improving corporate governance, disclosure standards, transparency in the market place and to spur accounting standards development. the creation of a global financial system has made the risks of a contagion and systemic failures much more likely. the impact of the asian financial crisis 1997-98 was felt far beyond asia, which provides evidence of greater integration of world financial markets. integration of global capital markets, particularly emerging markets, requires greater uniformity of legal, regulatory, and operational procedures. capital market, especially in an emerging economy, is undergoing rapid transformation, brought about by both technological innovations and globalization of the market place. market capitalization and trading volume of emerging markets have increased significantly during the last 15 years. world market capitalization increased almost nine times from $3.38 trillion (1993) to $26.52 trillion 1998 to a figure close to 35 trillion estimated for 2006; for the us markets, it increased seven fold from $1.90 trillion to $12.93 trillion. the most dramatic increase was that of 23-fold growth for developing markets from $0.083 trillion to $1.91 trillion (see table 1) for emerging markets. the rise in trading volume provides evidence of increased liquidity in the markets and increased necessity of risk management in these markets. table 1: capitalization and trading volume of world, us and emerging markets source: aggarwal r., (2001). mcdonough school of business, georgetown university. 2 3the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 the growth in emerging markets has been due to several factors, including privatizations, participation of foreign institutional investors, increase in the domestic investor base, and increased initial public offers, ipos, in these markets. put simply, nations are taking steps to lower the cost of capital to spur economic activities by fast-tracking financial market development for share and bond trading. however, emerging markets now need to start addressing second-generation capital market development issues. these are: developing financial intermediaries that have professionals with financial sector skills; enhancing domestic institutional investor base in addition to foreign investors; providing self-regulatory organizations; and developing mechanisms for investor protection. more importantly, risk management at all levels of the financial structure is extremely important, but is severely lacking in emerging markets due first to the scarcity of derivative instruments and second, where the derivative markets are already functioning, because of failure to generate liquidity. this is the subject of this paper to provide an overview to the asia’s status of derivative market development. there are about 55 organized derivative exchanges in the world for trading some 112 financial derivatives while there are 95 spot markets trading some 150,000 common stocks and unknown number of bond listing. parallel to this is the long-established commodity markets that have existed for almost 100 years especially in the industrial economies. in asia, the developed economies (hong kong, japan, korea, and singapore) have developed viable financial derivative markets during the post-1980s. some emerging markets (india, malaysia, and thailand) have also organized in the 1990s fast-developing financial derivative markets for interest rates, currency, futures and options. derivative market process (commodity as well as financial derivative) plays an important role for managing systematic risk in the market place. developed economies have nurtured commodity futures over more than a hundred years that assists even today to transfer the price-change risk (basis risk) to willing parties to bear these risk. similarly, since 1972, trading in financial derivatives in some 35 large and efficient derivative markets in developed countries has become liquid, and pricechange risk of financial assets are transferred effectively to willing parties at very low costs of about 0.5-1 percent. as for the asian derivative markets, these have mixed experience even in the second largest economy, japan, and are severely underdeveloped in several emerging markets while some of them are very liquid. the attitude of regulators is also quite discouraging for the nurturing of derivative markets, given the volatility and high-profile cases of failures of derivative hedge activities in several asian countries, which have increased the political costs of promoting such markets because of the bad publicity from these cases. 3 the international journal of banking and finance, vol. 5.(number 1): 2007: 1-254 the lessons to be learned from several such failures in emerging markets are not clearly and cogently analyzed, thus further adding to the bad public perception of the need for derivative markets. because each failure has its own peculiar characteristics it is difficult to provide standardized reasons for failures of all cases. commodity futures have been around for a long time to address the issue in agricultural and industrial sectors. however financial derivative are recent developments and the regional financial derivative landscape has changed dramatically in the years since 1997. the rest of the paper is organized as follows. in sections 2 to 6, we describe the evolution of futures markets in selected asian countries. the main theme is to draw some conclusions about the organization, trade, size and liquidity of these markets from an analysis of 6 selected markets. the chapter ends with a discussion in section 8. 2. indian commodity and futures markets agriculture is a significant sector in which 70 percent of india’s population finds a living. the commodity futures market in today’s more industrial india is almost as old as in the us and uk. the indian experience, however, is much older as references to such markets can be traced to kautilya’s arthasastra, a document dating at least two millenia. the first organized futures market in modern history was established in 1875 as the bombay cotton trade association trading in cotton futures contracts. oil seeds and food grains contracts followed soon. a large number of commodity exchanges trading futures contracts in several commodities like cotton, groundnut, groundnut oil, raw jute, jute goods, castor seed, wheat, rice, sugar, precious metals like gold and silver were flourishing throughout the country in the first half of the 20th century. futures trading was prohibited during the second world war. after 1947, especially in the second half of the 1950s and the first half of 1960s, commodity futures trading picked up. however, due to persistent shortages of real commodities during the early and mid-sixties due to the governments till then focusing on industrializing the country, futures trading in most of the commodities was prohibited. considering the importance of agriculture in india’s gdp and the genuine requirements of promoting sound commodity futures markets in the country, a number of major reforms have been undertaken since the early 1990s. in this regard, the most important changes that have taken place in the commodity area were the removal of prohibition on futures trading in a large number of commodities and the facilitation of setting up of modern, de-mutualised exchanges by the government of india. these two initiatives together have spurred commodity futures markets in terms of both participation and standards of practices. 5the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 however there are still a number of obstacles in fully exploiting the opportunities available due to lack of knowledge and accessibility of traders to the market. liquidity constraint is one of the major problems of the indian commodity futures market. there are many commodity futures exchanges, which are small in size and face liquidity problem. due to this, the market participants, especially hedgers, are not able to trade in these commodity exchanges. according to mansing (2004) the removal of restrictions on regional commodity exchanges, which are functioning in a fragmented and disorganized way, are important to achieve greater volumes in trading. the financial futures trading in the two key financial markets is found in bombay and in kulkata. for some decades, a form of futures contract (badla) was traded in these markets, which was briefly banned for several years. badla, with minor changes have been reintroduced in the year 2000, and is providing share-based hedging activities, a resurgence of interest. other new instruments are also traded, but the futures trading is not that well developed. badla is an indigenous carry-forward system invented on the bombay stock exchange as a solution to the perpetual lack of liquidity in the secondary market. trading involved buying stocks with borrowed money with the stock exchange acting as an intermediary at an interest rate determined by the demand for the underlying stock and a maturity not greater than 70 days. like a traditional futures contract, badla is a form of leverage; unlike futures, the broker-not the buyer or seller—is responsible for the maintenance of the market-to-market margin. badla contracts were banned by the o “stock exchange board of india” stock exchange board of india (sebi) in 1993 effectively from march 1994, table 2: exchange value of trading of indian commodity futures notes: there are several other commodity futures exchange but their size is very small and these are facing liquidity problems. * 1usd = 45 rupees; 10 crores equal million. shamsher and taufiq: asian derivative markets the international journal of banking and finance, vol. 5.(number 1): 2007: 1-256 amid complaints from foreign investors as promoting excessive speculation, with the expectation that it would be replaced by a futures-and-options exchange. after the ban took effect, trading volume on bse fell dramatically. statistics show (berkman and eleswarapu, 1998) that compared to the first two months of 1994, the average daily trading volume for badla stocks was 72 percent lower for the period april to december 1994. however promised futures-and-options exchange was not established and badla contracts were legalized again in 1996 (with a carry-forward limit of rs 20 crore – rs 2 million per broker) and banned again on july 2, 2001, following the introduction of futures contracts in 2000. the futures contracts are now shaping up, and it will take some years before these gain liquidity and hedge efficiency. 3. korean futures and option market south korean economy is an industrial economy. contribution of the agriculture sector in the economy is minimal. there is no dominant agricultural product, even long time before, in korea that needed to be hedged for risk. therefore commodity futures market is not very well known among the korean investors. in the ten years since 1995, the trading activity on the korea stock exchange, kse, has led to huge increases in trading of derivatives. since the kse launched futures, kospi 200 index, in may 1996 and options on the stock index in july 1997, the traded volume has increased exponentially, creating huge liquidity. the options contract now ranks as one of world’s biggest listed derivatives, measured in terms of both volume and value. just eight years after the debut of these markets, the average daily volume was 210,000 futures and 11,500,000 options. more than 900 million kospi 200 futures and options were traded in 2001. already 1.5 billion contracts have been traded in the first half of 2002 (both sides of a transaction treated as one contract). trading volume rose 120 percent in 2000 and 330 percent in 2001. on a combined basis, the kospi 200 futures and options now outrank the s&p 500 and the nikkei 225. according to data published by bloomberg, the notional value of trading in the kospi contracts was $6.9 trillion in the first half of 2002: $2.9 trillion was the number for s&p 500 in the us. the nikkei 225 futures and options were 15 times smaller in notional values and 100 times smaller in volume than the korean equivalents. there are several key factors to be considered for this degree of success of derivative markets in korea. first, the root cause is the entry of households into the market. korean traders have an appetite and an enthusiasm for trading that cannot be found anywhere else in the world. trading in the financial market is now part of modern korean culture, and there is a widespread knowledge of the fundamental mechanisms of the futures and options especially after the devastating financial crisis that shook the economy in the late 1990s and cost us$ 54 billion to repair the derailed 7the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 economy. at any time during day or night, there are educational programs on the television about technical analysis and trading strategies for stocks and derivatives. trading the market is part of the everyday life of the people and comes as a natural behavior. second key factor is the small size of these contracts. in mid-october 2002, the value of a single futures contract on the kospi 200 was approximately 42.5 million won, which at the then exchange rate was equivalent to $34,000, making it roughly comparable in size to the e-mini s&p 500. the options contract is even smaller, representing one-fifth of the value of the futures contract. a unique characteristic of this market is the preponderance of local retail participants. individual investors account for two-thirds of overall trading volume. the third key factor is the formidable penetration of the internet in korean society, which has allowed online trading to develop very rapidly since 1999. the number of people with internet access rose from 1.6 million in 1997 to 10.9 million in 1999 to 24.4 million in 2001. the number continues to grow and in 2006 there are more accounts than there are households. the ratio of online securities trading to total securities trading has followed similar pattern, moving from 1.9 percent in 1998 to 19.0 percent in 1999, 46.6 percent in 2000 and 52.3 percent in 2001. as of 2001, among the 44 local brokers, 38 were running online trading business and of these, five were exclusively online. these figures make korea the world leader in online securities trading. one reason for the popularity of online trading is the low transaction cost. foreign investors have progressively increased their presence in the kospi 200 futures and options markets. they represent currently about 7 percent of the total futures and options trading, up from 2 percent in 1999. foreign investors include the proprietary trading desks of large european and us investment banks along with us options market makers. for any other derivatives exchanges, volume like that would be cause for celebration. but for kfe, it is a sign that the kospi market may be entering a new, more mature phase, with lower growth and less speculative trading. but possibly this is because of greater appreciation of the risk management aspects of futures and options. this may be seen in the wide divergence between the open and total volume, which indicates the speculative nature of the trade. future development is more likely to be towards more hedge-interest. 4. hong-kong futures and option market after being ceded by china to the british under the treaty of nanking in 1842, the colony of hong kong quickly became a regional centre for financial and commercial services based particularly around the hong kong and shanghai bank and merchant companies such as jardine matheson. in 1841, there were only 7,500 chinese inhabitants and a handful of foreigners, but by 1859, the the international journal of banking and finance, vol. 5.(number 1): 2007: 1-258 chinese community was over 85,000 augmented by about 1,600 foreigners. the economy was closely linked to commercial activities dominated by shipping, banking and trade of merchant companies. gradually there was increasing diversification to services and retail outlets to meet the needs of the local population, and also shipbuilding and maintenance linked to the presence of the british naval and merchant shipping. there was some industrial expansion in the nineteenth century; notably sugar refining, cement and ice factories among the foreign sector, alongside smaller-scale local workshop manufactures. the mainland territory of hong kong was ceded to british rule by two further treaties in this period; kowloon in 1860 and the new territories in 1898. the industrialization of hong kong is usually dated from the time of embargoes of the 1950s. certainly, hong kong’s prosperity could no longer depend on china trade. however, industry emerged in the nineteenth century and began to expand in the inter-war periods. nevertheless, industrialization accelerated after 1945 with the inflow of refugees from china, entrepreneurs and also capital fleeing the civil war on the mainland. the most prominent example is immigration from shanghai, table 3: trading activities for kospi 200 futures contract from 1996-2005 (million krw*) 1usd = 936 korean won. table 4: trading activities for kospi 200 option contract from 1997-2006 (million krw) * 1usd = 936 korean won at end 2006. 9the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 1 economic history of hong kong by catherine r. schenk, university of glasgow, encyclopedia. which created the cotton spinning industry in the colony. hong kong’s industry was founded in the textile sector in the 1950s before gradually diversifying in the 1960s to clothing, electronics, plastics and other labor-intensive production mainly for export. while manufacturing was moved out of the island during the 1980s and 1990s, with the rise of china as production centre, there was a surge in the service sector. this transformation of the structure of hong kong’s economy from manufacturing to services was dramatic.1 most remarkably it was accomplished without faltering growth rates and with an average unemployment rate of only 2.5 percent. since the late 1960s, singapore and hong kong have been competing with each other to be the next largest international financial center after tokyo. while singapore had a head start in 1968 when the asian-dollar market was first introduced, hong kong was lagging behind because of its moratorium on banking licenses arising from the banking crisis in 1965 (jao, 1997). hong kong’s privileged location in the northeast asia makes it a gateway to china. moreover hong kong is situated at appropriate time zones that allows 24-hour continuous trading of foreign exchange and gold when the two markets in new york and london are closed. with the onset of globalization in the 1990s, hong kong now relies more and more on centrifugal forces for the success as international financial centre. such forces have attracted multinational corporations to set up their operational headquarters for their funding needs, treasury operations as well as risk management. in this regard, hong kong stock market was started in 1891 to accomplish the objective of international financial location for northeast asia and to boost economic growth. it is one of the most established stock markets in the world, classified by the international finance corporation (ifc) as a developed market. in hong kong, there is a strong presence of the world’s major financial institutions: 167 foreign banks including 85 of world’s top 100 banks, 323 overseas securities and commodity trading companies, 122 overseas insurers, and 1,182 unit trusts and mutual funds. hong kong has the second largest stock market in asia and it is the 7th largest in the world. as at end 2006, hong kong market capitalization was over about us$ 500 billion. therefore risk management tools (both in commodity and financial products) are well established in hong-kong. the hong kong futures exchange (hkfe) started as a commodity futures exchange in 1976 (formerly called hong kong commodity exchange) with modest volume. in may 1986, it shifted to index futures trading with the introduction of hang seng index futures. now the hang seng index futures is the flagship product for the exchange. in march 1993, hkfe entered into option trading with the launching of hang seng index options. hkfe achieved another milestone when a brand new currency futures product called the ”rolling forex” was officially launched in november 1995. the international journal of banking and finance, vol. 5.(number 1): 2007: 1-2510 hkfe has a link with the philadelphia stock exchange for the trading of currency option products at hkfe during asian trading hours. in addition, hkfe also has an agreement with the new york mercantile exchange to trade its precious metal and energy products in hong kong. on the chinese front, hkfe maintains a close relationship with the chinese securities regulatory commission and the chinese commodity exchange. unfortunately, the enthusiasm outstripped the risk management measures. the rudimentary state of the latter was exposed by the october 1987 stock market crash. a number of futures brokers collapsed and the government had to organize the rescue of the futures market. in subsequent years, the futures exchange built up its volume and reputation. table 5: trading volume of financial derivative in hong-kong exchange source: hong kong exchange, 2006. 11the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 table 5 clearly indicates that only few futures and option contract are trading with reasonably high volume. however, most of the futures and option contract are facing serious liquidity. out of 17 contracts, only six contracts are actively traded in the hong kong exchange. trading on only three contracts namely the index futures, index option and stock option steadily increased (more than 200%) over the six year time from 2000 to 2006. however, in august 1998 the government intervened in the market due to the alleged speculation against the hong kong dollar. in the following month, numerous measures were introduced to restrict derivatives trading. perhaps due to regulatory measures, trading volumes of derivative contracts in hong kong’s exchange have not risen in line with the global growth trend. new product launches have seen limited success, and trading continues to be dominated by index futures products and stock options only. 5. china futures and option market two years after the death of mao zedong in 1976, it became apparent to many leaders that economic reform was necessary to move away from central planning to engage the masses in economic activities by bringing the capitalists to exploit the cheap labor supply. during his tenure as china’s leader, mao had encouraged social movements such as the great leap forward and the cultural revolution which had had as their bases ideologies such as serving the people and maintaining the class struggle. by 1978 chinese leaders were searching for a solution to serious economic problems inherited by hua guofeng, the man who had succeeded mao zedong, and by deng shiaopeng, who ousted hua. the issue of economic reform became politicized as the issue was used as a means by deng to attain the leadership of the chinese communist party. therefore the initial form the new economy took was not a radical one. china was still a state in which the central government retained the dominant power in economic resource allocation and responsible local officials worked for the interest of the units under their control, as stated by an observer. one important aspect of chinese economic reform was the decision of china to join the world economy. deng xiaoping and his allies hoped to engineer this 1979 resolution in two ways: by expanding foreign trade, and by encouraging foreign companies to invest in chinese enterprises. this policy denoted the “open policy” was a drastic departure from the policies of mao zedong and, in fact, from centuries of chinese political culture. the implementation of the open policy was so successful that by 1988 the leaders of the ccp were encouraged to create a new program called the coastal development strategy. moreover, by involving more overseas investors, importing 2 in this brochure china refers to the mainland of the people’s republic of china, excluding hong kong sar, macao sar and taiwan region. the international journal of banking and finance, vol. 5.(number 1): 2007: 1-2512 both capital and raw materials, and exporting china’s cheap excess labor power, the new policy was one of export-led growth or export-oriented industrialization. it was explicitly modeled on the experiences of taiwan and the other asian ‘small dragons’. due to this reform, chinese government started to develop the chinese financial market from late-1990s. china’s spot and futures markets2 have been developing since the 1980s, in tandem with the market reforms and the liberalization of the country’s economy. in july 1981, issuance of treasury bonds was resumed in china. six years later, a secondary market for the trading of treasury bonds was established. by the mid-1980s, enterprises started to issue corporate bonds and shares to the public. in december 1990, the shanghai stock exchange and shenzhen stock exchange were established. china securities regulatory commission (csrc) is the sole national securities regulatory body. it was mandated in october 1992 to regulate all of china’s securities and futures markets. however, despite all these reforms, the stock exchange is a branch of the government, unlike the stock exchanges of most countries. at present, china’s securities market offers seven types of financial instruments: a shares,1 b shares, 2 treasury bonds, treasury bond repurchases, corporate bonds, convertible bonds and securities investment funds. initially the two stock exchanges traded bond futures in very high volumes. unfortunately, risk management procedures were not very adequate – margins were as little as one per cent of contract price in early 1995. these difficulties came to a head in a scandal involving chaotic trading of bond futures contracts in the shanghai stock exchange, one of approximately 10 exchanges trading bond futures at that time. one of largest securities firms incurred huge losses when its short-selling strategy went disastrously wrong, to the point of threatening the stability of the entire financial system. shortly thereafter, the government shut down the bond futures markets and scaled back trading in commodity futures. that experience has been very important in having a bearing on the reopening of any trading in financial futures contracts. therefore no financial derivatives have been permitted since, although there is in mid-2000s discussion of a possible index futures contract. there are several factors to be discussed regarding the failure of bond futures and subsequent restriction on the introduction of financial futures. for the most part, china’s futures markets are closed to foreign investors and brokerage firms. in addition, chinese futures companies, which number around 200 or so, serve primarily as agents, and are not involved in broader business dealings. stock and fund brokerages cannot currently participate in futures transactions. in december 2001, china securities regulatory commission announced that the shanghai futures exchange was preparing to launch longawaited stock index futures. but he gave no timeframe in the announcement! 3 common shares denominated in rmb. 4 domestically listed common shares quoted in us dollars or hk dollars. 13the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 to avoid a repeat of the events in the mid-1990s, china will have to develop a stable regulatory system for its futures market backed up by vigorous enforcement. this too is an area that raises the concerns of some. according to leong, “the infrastructure on the futures side is comparatively rudimentary on the mainland” and “pertinent rules and law still await china securities regulatory commission’s formulation.” in this area, hong kong could play a crucial advisory role, according to katherine wang, spokeswoman for hong kong’s market regulator, the securities and futures commission. but in the case of commodity futures, china already is quite advanced in trading and also in regulating the market mechanism. chinese economy traditionally is based on agriculture. even after economic reforms, which increased industrial contribution to the gdp, the bulk of the people still depend on agricultural economy. due to the wto agreement (china already signed the wto agreement in 2002) agri-product prices are subject to international competition. therefore, chinese regulatory authorities are keen to develop agriculture futures faster than financial futures. the first domestic commodity futures market was zhengzhou grain wholesale market, which introduced its first futures contract based on physical delivery in october 1990. during the years 1990-93, 50 futures exchanges in china commenced operations. to put this in context, that was equivalent to almost the number of futures exchanges in the world combined. trading volume exploded, but the same product was often traded on a number of exchanges resulting in confusion; contracts were not standardized and were often traded on unreasonable terms. the exchanges were operating without appropriate and coordinated management (in particular, risks management) systems. the market was, to say the least, disorderly. in the october of 1994, the state council of people’s republic of china regulated and merged these exchanges into 15. the 15 futures exchanges were regulated again and restructured into three by the state council in 1998. the three futures exchanges are: shanghai futures exchange (shfe), zhengzhou commodity exchange (zce), and dalian commodity exchange (dce). today, the futures market has three exchanges and 180 brokerage firms, compared with 50 exchanges and 1,000 brokerage firms at the peak in 1994. the traded products have been downsized from 35 to 12 products. the fia statistics also reflects the emergence of china as a major player in the global commodity futures market. only two of the three commodity exchanges in the mainland china, namely the dalian commodities exchange and the shanghai futures exchange, are recorded in the fia statistics. both exchanges showed impressive growth in 2003. the former recorded a 54.9 percent rise in its turnover while the latter more than doubled its turnover (229 percent). the dalian commodities exchange even overtook the chicago board of trade (cbot) to become the third largest commodity exchange in the global ranking. the international journal of banking and finance, vol. 5.(number 1): 2007: 1-2514 one of the reasons for this impressive performance is due to the small size of the contracts: soybean contract is 7 percent the size of the cbot contract for the same product. zce has contracts on wheat, mung bean, etc., while dce offers contracts on soybean, soybean meal, etc. shfe provides futures contracts on other commodities such as copper, aluminum, rubber, etc. by and large, the trading in agricultural products tends to dominate. the most actively traded products are soybean, wheat and copper. the bar graphs in figure 1 shows the trading summary of the futures in 2003. the increase in china’s futures trading no doubt reflects the country’s surging consumption of physical commodities to support its rapid economic growth. there was a period in the early 1990s, when the financial futures trading was permitted in over the counter trades. after spectacular failures of hedges in some cases, a total ban has been imposed on futures trading in financial instruments. nevertheless, some degrees of private deals are still made in financial futures trades. no statistic is available to report on this. china’s competitive advantage to attract foreign capital for its rapidly developing economy will be lost if it does not have a risk-management instrument like financial futures. china’s financial futures market was set up in 1990 and experienced a quick expansion until the mid-1990s. however, due to the loopholes in the regulatory scheme it led to irregularities like fraud and strong speculation forcing the government to close these markets. in addition, there are other barriers to reintroduce financial futures market in china. for example, the majorfigure 1: trading volume and turnover of chinese futures contracts in 2003 source: china securities regulatory commission, april 2004 15the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 ity of chinese investors do not understand the significance of the market; the relevant laws and regulations are not gazetted yet; the immaturity of china’s market oriented economy need to be perfected. 6. malaysian futures and option market malaysian futures market started in the 1980s when the kuala lumpur commodity exchange (klce) was organized to meet the needs for a price discovery and hedging mechanism within the region to improve the marketing efficiency of palm oil. the tin futures were done in london for some long time before, although most tin exports emanated in malaysia, which failed to develop a futures market on this commodity. the first futures contract traded on the klce was palm oil which was launched on october 23, 1980. a rubber futures contract was introduced in 1983, which was followed by tin futures in october 1987, cocoa in august 1988 and refined, bleached palm olein contract in february 1990. due to the heavy short selling tactics of speculators, massive defaults followed in february and march 1984 and a series of crises led to the collapse of the three and a half year old crude palm oil futures contract. the authorities undertook a major revamp of the exchange in 1985 to correct the structural weaknesses that were exposed in that episode. among others, a new clearinghouse was set up and a new commodities trading act, 1985, was introduced. however since 1990 (after recovery from the economic recession of 1986-89), the government changed the direction of the country’s economic policy from agriculture based economy to industrial economy. as a result, most commodity investor lost their interest on futures market. most of the commodity futures, except the palm oil futures, could not survive. the following table shows the trend of the all commodity futures contracts from 1989 to 1992. table 6 clearly shows that introduction of different commodity futures failed to sustain due to liquidity constraint, except for crude palm oil futures contract. among the reasons cited for the lack of interest in the market is liquidity which caused problems when participants wanted to leave the market at the price they wanted. another reason is the lack of government commitment due to the heavy concentration given to the financial market development. another reason was the denomination of the contract in us dollar, which authorities had justified in terms of attracting foreign participants but failed to do so. since 1990 under the new economic policy, regulators obviously became committed to fostering the growth of an effective capital market to support the financial system. indeed, the government’s goal of achieving the status of developed country by the year 2020 meant that a rapidly expanding capital market is essential for funding, and risk management. it also promotes private enterprise by providing a convenient means of raising capital for corporate investment and expansion. integral to the growth of capital market is the development of futures market. futures provide hedging and asset allocation facilities, which allows investors to the international journal of banking and finance, vol. 5.(number 1): 2007: 1-2516 table 6: overview of malaysian commodities contracts hold larger debt and equity positions and, as a result, also enhances the liquidity of these underlying markets. an active futures market that complemented the spot capital market was seen as enabling investors to hedge or adjust their positions and thus make them more willing to take larger holdings, thereby reducing risks of large capital investments. the stock index futures contract was introduced on the kuala-lumpur option & financial futures exchange (kloffe) on the 15th december 1995. the contract is recognized as fkli futures contract which is based on the 100 kuala-lumpur composite index (klci) stocks. subsequently the three-month klibor futures (on interest rate) was introduced under the kloffe. due to the lack of coordination between financial futures and commodity futures, kloffe and klce merged together and became one exchange, which is called malaysian derivative exchange (mdex). table 7 indicates that introduction of various financial futures and option contracts in malaysia had also limited success except for the index futures and cpo futures contracts. one of the possible reasons is that soon after the introduction stock index futures contract experienced serious financial meltdown, which obviously affected the pricing behavior of newly introduced financial futures contracts. capital control regulation was introduced to stem the economic meltdown in 1997-8 and also to control the interest rate movement that affected the growth of the financial futures contract. it can be noted that korean stock index and interest source: bursa malaysia website. table 7: trading volume of malaysian financial futures and option contracts source: bursa malaysia website. 17the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 rate futures were introduced at the same time. however korean financial futures are considered as one of the highest-ranking financial futures in the world whereas in terms of trading volume, financial futures contract in malaysia is one of the lowest ranking futures market in the world as well as in the region. the following table 8 shows the ranking of the futures market in the world and in the region. 7. futures and options markets in japan japan’s futures exchanges (commodity products only) date to the 1700s and preceded financial securities markets by nearly a century and a half. this pattern is generally the reverse of that in europe, england, and the united states, where securities markets usually predated commodity markets. spot, or cash, trading in rice, the most widely used staple food in japan, dates from the early 1700s and forward contracting in rice on the osaka rice exchange was legally recognized table 8: asia-pacific derivatives exchanges ranked by contract volume source: futures industry association, zhengzhou commodity exchange and dalian commodity exchange. the international journal of banking and finance, vol. 5.(number 1): 2007: 1-2518 in 1730. though there were as many as eight commodity exchanges in major japanese population centers, the osaka market was the largest. there were also japanese markets for edible oils, cotton, and precious metals, though their trading volume was small in comparison with that for rice. today there are varieties of commodity futures trading in tokyo commodity exchange, which is well organized, and regulated by the different regulatory authorities. the tokyo commodity exchange (tocom) was established on november 1, 1984, as an amalgamation of the tokyo textile exchange, the tokyo rubber exchange and the tokyo gold exchange. currently gold futures and options, silver, platinum, palladium, aluminum, gasoline, kerosene, crude oil, gas oil and rubber futures are actively traded in tocom. in addition, having successfully launched options on gold futures in may 2004, tocom are planning lpg (liquefied petroleum gas) futures. since its inception, tocom has experienced a remarkable expansion in its market size. tocom’s trading volume soared from 4 million contracts in 1985 to 87 million contracts in 2003. in terms of trading value, tocom is now the world’s largest futures market for platinum, gasoline and kerosene. various authorities regulate japanese commodity futures. in japan, jurisdiction over futures transactions is divided between the ministry of finance (bonds and shares), the ministry of international trade and industry (fibers and precious metals), and the ministry of agriculture, forestry and fisheries (grains). the majority of trading in commodities exchanges in japan is in precious metals (gold, silver and platinum). the commodities exchange law has been amended 13 times since its enactment in 1950. major amendments were made in 1967 and 1975. an advisory body to the ministry of international trade and industry came out with a report during 1986 recommending another revision to the law. revisions were recommended in definitions of objectives, commodities and listing in the law. the revision would also permit the listing of “all things suitable for listing on a commodities future exchange (for example commodity indexes).” indeed the japanese commodity market is dominated by domestic retail investors. the foreign participation is extremely limited in the commodity futures market. in fact, it is the commodity futures markets that account for the bulk of futures trading in japan. these markets technically are not closed to foreigners, but they operate on such different principles that very few foreign firms wish to participate. even local institutions shy away from these markets, mainly because they are geared to small-scale speculative trading and lack important customer protections. the clearest sign of change is at tocom, which in 2004-2005 has made extraordinary changes to the ways it operates. for foreign brokers, the most 19the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 important development has been the creation of a true clearinghouse. in the past, tocom, like other commodity futures exchanges, processed transactions but did not act as a central counterparty, and every firm that traded on the exchange was therefore exposed to the risk that the other side of the trade would not fulfill its obligations. this was a major hindrance to foreign brokers and commercial users of the exchange. that changed in june, 2005 when the exchange implemented a new clearinghouse structure and began providing a guarantee on every trade, just as clearinghouses do on european and us futures exchanges. if ever a firm defaulted, the clearinghouse first would seek to cover the firm’s obligations out of its margins and what the exchange calls the special deposit for default. if this were not sufficient, the exchange could then turn to its own funds, including a security reserve fund against default that holds 1.23 billion yen ($11.5 million) in 2005. tocom also radically overhauled its technology policy in 2003 and agreed to allow independent software vendors (isv) to make connections directly to the exchange. “the introduction of isvs will improve service for overseas clients,” according to chiaka furui, director of international business derivatives and it business at yutaka shoji, a leading japanese futures brokerage. tocom’s changes have created a groundswell of interest, not only within the futures community, but also among traders and brokers as far away as chicago. other than tocom, japan has six other commodity futures exchanges, for example tokyo grain exchange, kansai agricultural commodities exchange (kanex), cubu commodity exchange and none of them has announced plans to establish a clearinghouse or reach out to the isv community. most of these exchanges are geared primarily to the domestic market place, and liquidity comes primarily from retail speculators and brokerage firms trading for their own accounts. though commodity futures exchanges are quite large in terms of contract volume, however financial derivatives volumes are relatively lower. restrictions imposed by regulators, sometimes on the basis of misunderstanding of the impact of derivatives, have tended to keep the financial derivatives market small. securities firms have not been able to promote derivative products to the mass retail market. the institutional fragmentation of the market has probably not helped its development. table 9 shows the statistics on japanese futures and option contracts. only topix futures trading volume increased substantially (over 300 percent) over 10 years period (from 1997 to 2006). whereas topix option, equity option and government bond futures increased very little over 10 year’s period. the government has launched a modernization plan for both the commodity and financial futures markets. one of the catalysts for this initiative is the recognition by the government and the futures industry that the current structure will not sustain. the legal structure governing commodity futures trading is being the international journal of banking and finance, vol. 5.(number 1): 2007: 1-2520 table 9: trading activities for futures and option contract, 1997-2006 (yen million) source: tokyo stock exchange and tokyo international financial futures exchange. 1usd = 110 yen in 2006. aligned closer to global standards. at the end of 2004, fixed commission rates expired. this is widely expected to cause a drastic reduction in revenues for japanese brokers. commission rates could fall by more than 50 percent. as a result, the competition among the brokers will be fierce, and it will be survival of the fittest. so there will be a need for strict compliance and surveillance under the new legal environment in order to make sure that the competition is fair. 8. futures and derivative market in singapore due to its strategic location and well-developed infrastructure, singapore traditionally had been the trade and financial services centre servicing a everincreasing stretch in the region. the country’s rapid development was closely linked to the government’s efficient financial management. conservative fiscal and monetary policies generated high savings, which, along with high levels of foreign investment, allowed growth without the accumulation of external debt. in the 1970s, the government identified financial services as a key source of growth and provided incentives for its development: it has paid of by growing the financial sector as provider of some 25 percent of the gdp today compared to its share of 5 percent some 30 years ago. by the 1980s, the focus was on further diversification, upgrading, and automation of financial services. emphasis was placed on the development of private investment portfolio management, securities trading, capital market activities, foreign exchange and futures trading, and promotion of more sophisticated and specialized fee-based activities. consequently, by the mid-1980s, it was the third most important financial center in asia after tokyo and hong kong. the financial services sector, having sustained double digit growth over the previous decades, accounted for some 23 percent of gdp and employed approximately 9 percent of the labor force. 21the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 inaugurated in 1973, the stock exchange of singapore was governed by a committee comprising four elected stockbroker members and five appointed nonbroker members. in late 1988, the 327 companies listed on the main board of the exchange were classified into six groups: industrial and commercial, finance, hotel, property, plantation (farming), and mining. the exchange underwent a major, prolonged reorganization following the december 1985 collapse of a singaporean company, pan electric, which revealed a massive web of forward share dealings based on borrowed money. the collapse resulted in a tighter regulation of the financial futures market and the securities industry. in 1986 the securities industry council was established to advise the minister for finance on all matters relating to the securities industry. the exchange is now called sinogapore exchange, and it is demutualised. singapore also expanded other international financial markets in the late 1980s. trading in gold futures originally was undertaken in the gold exchange of singapore, which was established in 1978 and reorganized in 1983. the scope of its activities was widened to include financial futures trading, and it was renamed the singapore international monetary exchange (simex). starting in 1984, the financial futures market secured a mutual offset arrangement between simex and the chicago mercantile exchange, which allowed contracts executed on one exchange to be offset on the other without additional transactional cost for market participants. it was a smart move that helped establish a fast expanding futures market henceforth. the linkage was the first of its kind in the world and greatly facilitated roundthe-clock trading in futures contracts. in 1988 six futures contracts were traded: international gold futures; the eurodollar time deposit interest rate; the nikkei table 10: trading activities for futures and derivatives, 1990-2006 (units) source: singapore exchange. the international journal of banking and finance, vol. 5.(number 1): 2007: 1-2522 average stock index; and three currency exchange rates on us dollar/west german deutsche mark, the us dollar/japanese yen, and the us dollar/british pound sterling. trading volume on the simex had grown steadily. table 10 shows the average volume of trade for 16 financial contracts traded in simex, and now in the singapore exchange. from the 16 futures contracts, four contracts, namely, nikki 225 futures, msci taiwan index futures, msci singapore futures and euroyen libor futures are the most successful in terms of volume of trade. note that these contracts are risk management tools for foreign investors, and are speculative instruments for others. the trading volume of all futures contracts increased from 5.6 million units in 1990 to 26 million units in 2005: a five-fold increase. trading volume of options increased only from 75,180 units in 1997 to 158,467 units in 2005: a two-fold increase. however, during the 1997-1999 periods, the volume of trade was high indicating hedging activities during the asian financial crisis period. 9. conclusion and policy discussion thus technology, as it has throughout the history of man, has dictated once again the fundamental and revolutionary change in our social structure and reshaped both the political and economic landscape of our planet. although it is impossible to completely perceive the ultimate consequences of this technological march around us, we know with certainty that its influence to date has been felt nowhere more than in financial markets. our separate financial existence has been transformed into one inter-related, interdependent world economy. this globalization of markets did not happen overnight. the first decisive action by financial markets occurred two-and-a-half decades ago with the suspension of dollar convertibility into gold and the abandonment of the system of fixed exchange rates. globalization and implementation of wto agreements provide numerous advantages to emerging markets; however, it increases volatility and uncertainty of commodity prices as well as prices of financial instruments. the creation of a global market has made the risk of a contagion and systemic failures much more likely. therefore development of well-organized derivative market (commodity as well as financial derivatives) is an important policy issue for an emerging market. this policy issue can be divided into two parts: policy for commodity derivative market and policy for financial derivative market. commodity derivative exchanges are, and have been very important institutions in commodity pricing and have been prominent in the developed countries including japan for over a century or more. in europe and asia too, commodity derivatives are booming. price volatility for raw materials has increased for many reasons including the recent changes within the common agricultural policy of the european union and the wto which have increased 23the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 price risks for both the primary producers and the processing industry as well. on the other hand, most of asian emerging countries are transforming their agriculture-based economies to industrial based economies. however, still agricultural production is contributing a significant portion to gdp. therefore asian emerging markets require agricultural commodity futures contract (based on the leading commodity a particular country produces, for example, copra in philippine, oil palm in malaysia and so forth) at the same time they also require different raw industrial materials futures market in countries with industry dominance. but it is very costly for asian emerging countries to introduce many commodity futures contracts. liquidity constraint, lack of regulatory framework and financial constraints are seen to be most prominent problems. therefore, in that situation, japan, singapore and hong-kong can be seen as a common commodity derivative market in which other emerging countries can participate. for example, japan commodity futures exchanges provide a wide range of commodity futures (from gold futures to energy futures). according to the himawari’s mogi (japanese brokerage house through which foreign customer can participate in the japanese commodity futures market), “all over asia we are seeing increasing reliance on the market to set prices for commodities”. china has foreign currency issues, singapore does not have enough liquidity in its commodity futures contracts, and korea only has securities futures. so japan is really the best place for hedging commodity risk in the asian time zone. the region’s financial derivatives have checkered experience in performance, given the dominance of speculators and the small size of the economies to truly reduce the transaction costs. most of the asia-pacific equity markets are reasonably well developed. asian trading venues have undergone multiple layers of transformation in the past decade and will continue to evolve as the world compresses trades through technological innovation. general market liberalization, conflicting regulatory requirements, broker consortiums, and electronic communication networks have all played their parts in asia changing the trading scene. whereas market volatility remains high in asia, but the opportunities for diversification and risk management opportunities are good if there are futures markets that are efficient with liquidity. therefore equity based financial futures and option markets must be developed in this region. some financial derivatives exchanges in the region have shown outstanding growth, for example, singapore, south korea, hong kong, and japan. study of their success does reveal one obvious factor that can be applied universally: the contracts must be relevant to the economy! from this review of regional experience, it seems that by far the most important success factor for a financial futures and option is liquidity. malaysian index futures and hong-kong financial futures are examples of the liquidity problem due to the government policy shift during several crises. however the the international journal of banking and finance, vol. 5.(number 1): 2007: 1-2524 secondary market for other financial instrument, for example, government bond market, corporate bond market, warrant market, options, etc. are not well developed in asia’s emerging countries. therefore security commissions and other government agencies should give more attention to develop the secondary markets in other than equity by diversifying into debt, etc. markets before the introduction futures market. advice to form common markets with larger economies would provide the liquidity lacking in smaller economies. acknowledgment: we wish to thank the journal’s editors for suggestions for changes, which are incorporated in this published version of a paper submitted to the journal for review. all remaining errors are the responsibility of the authors. author statement: the authors are respectively professor and senior lecturer in the department of accounting and finance, university putra malaysia, serdang, selangor, malaysia. both authors’ research interests span capital market studies as well as studies of banking institutions in malaysia. reference aggarwal, r., (2001). “integrating emerging market countries into the global financial system: regulatory infrastructure covering financial markets.” prepared for the brookings-wharton papers on financial services: 4th annual conference, january 11-12, 2001, the brookings institution, washington, d.c. alexander balfour. february (1995). “bogged-down in bombay.” euromoney. issue 310. p. 96. berkman, h., and eleswarapu, v.r., (1998). “short term trader and liquidity:a test using bombay stock exchange data”, journal of finance economics 47: 339-355. joost m.e. pennings, j m.e and meulenberg, m.t.g. (1999), “the financial industry’s challenge of developing commodity derivatives,” ofor paper number 99-01, may 1999. mansingh l., (2004). “volume of commodity futures trading to overtake stock market trade volumes”. keynotes speech at seminar jointly organized by ficci (federation of indian chamber of commerce) and ncdex (national commodity & derivatives exchange ltd). melamed, l., (1994). “derivatives and the asian market”, international futures seminar seoul, south korea december 15, 1994 nair, c.k.g., (2001a). “ market dynamics and regulatory concerns” nse news, june 2001. nair, c.k.g., (2001b). “ securities market and commodity derivative market – “rush vs slow growth”, nse news, december 2001. nair, c.k.g., (2004). “commodity futures markets in india: ready for “take off”? nse news, june 2004. ng, beoy kui, (1987). “an overview of the foreign exchange markets in the 25the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25 seacen countries,” seacen centre. foreign exchange markets in the seacen centre. kuala lumpur: the south east asian central banks (seacen) research and training center. ng, beoy kui, (1996). “bank liquidity management and the implementation of exchange rate policy in singapore,” in lim chong yah (ed.), economic policy and management in singapore, singapore: addison-wesley publishing co., pp 253-274. ng, beoy kui, (1996). “the role of the central provident fund in social development, stabilization and restructuring: experience from singapore,” in new zealand journal of business 18 (no.1): 39-51. ramaswami, b, ravi s. and chopra s.d (2003). “risk management in agriculture”, discussion in economics (03-08), indian statistical institute, delhi, planning unit. sahadevan k.g., (2002). “derivative and price risk management; a study of agricultural commodity futures in india”, institute of management, lucknow. susan, t (2003), “ agricultural commodity markets in india; policy issues for growth,” indira gandhi institute for development research, mumbai. wu, friedrich, (1997). “hong kong and singapore: a tale of two asian business hubs,” journal of asian business 13 (no. 2). . international journal of banking and finance 1-1-2007 asian derivative markets: research issues m. shamsher h. taufiq recommended citation new evidence on the effect of cboe options listing on the volatility of new york listed stocks 59the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 ijbf new evidence on the effect of cboe options listing on the volatility of new york listed stocks khelifa mazouz aston university, united kingdom abstract this paper re-examines the evidence on how the listing of options impacts on underlying stock’s volatility by taking into consideration the possible presence of a learning effect, along with the impact of the very endogenous nature of the options listing decision itself. our analyses are centred on both the portfolio approach as well as the individual stock approach applied on the sample of optioned stocks with a matched control sample. the results show that the individual stock approach yielded accurate results, as it is amenable to both the sign and the statistical significance test of variance change. however, unlike the individual stock approach, the more frequently applied portfolio approach relies more on the sign rather than the statistical significance. based on these analyses, we found no evidence of the cboe-option listing effect’s presence on the volatility of the underlying stocks in the new york stock exchange. jel classification: g14, g20 & g21 key words: financial institutions, cost efficiency, profitability, firm size. 1. introduction whether listing options on an underlying stock stabilises or destabilises the underlying share’s market price has been of interest to both regulatory bodies and market players since the initiation of the cboe’s1 first stock option contract. the topic is a continuous source of research focus for academic literature. several theoretical models have been developed along with the adoption of different, empirical methodologies, investigating the implications of stock options trading regarding the pricing of the underlying shares. 1 the first option contract was listed and traded on the chicago board options exchange (cboe) on the 26th of april 1973. at a much later date, options were also listed and traded within the new york stock exchange. the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8260 academic theory generally focuses on the concepts of market completeness and market efficiency to evaluate the effect derivative-securities trading has on the existing spot financial markets. the market can be made more complete if the newly introduced financial instruments accomplish certain objectives that could not have been accomplished by existing ones. moreover, the market can become more efficient if the innovation can accomplish the same end as the existing products but at a lower transaction cost. in a complete market – with trivial transaction costs and trading constraints – stock options are regarded as redundant securities, as their payoffs can be synthesised by combining a position in the underlying stock with risk-free lending or borrowing.2 however, in the real world, where dynamic trading strategies replicating option payoffs are infinitely costly, introducing options can be useful in attaining efficiency in competitive equilibrium by expanding the set of contingencies covered by marketed assets (ross, 1976). this gain in efficiency may still be jeopardised if the options listing makes the underlying stock more volatile (chaudhury and elfakhani, 1997). theoretically, the impact of options listing on the underlying stock is controversial depending upon the assumptions that were initially made regarding the underlying market structure (harris, 1989). options may convey a certain demand for securities and remove the uncertainty regarding the cost of obtaining an option-like payoff. the revealed information from the option trading by investors may affect the volatility of the underlying stocks (grossman, 1988). similarly, dybvig and ingersoll (1982) argued that the introduction of options would destroy traditional mean-variance pricing if market participants had a heterogeneous, utility function. the dybvig and ingersoll argument implied that the expected returns on both the options and their underlying stocks would differ from those predicted by capm. cox (1976) claimed that in the context of futures trading, derivative securities attracted an additional group of traders that based their decisions on different information sets than those used by stock market participants before the introduction of derivatives. thus, the stability or instability of the stock market after the options trading would depend on the accuracy of the new information. for example, a stabilising effect is produced if new information provided a better prediction of the spot price and vice versa. the volatility of the underlying stocks may also increase or decrease depending on whether options attract hedgers (gorton and pennacch, 1993) or speculators (stein, 1987). further, ma and rao (1988) contended that there existed a dual impact of options on the volatility of the underlying stocks. they argued that volatile stocks become more stable after listing because of the hedging behaviour of the uninformed traders. this resulted in stable stocks becoming more volatile after listing because of the speculative behaviour of the informed traders. 2 the first option contract was listed and traded on the chicago board options exchange (cboe) on the 26th of april 1973. at a much later date, options were also listed and traded within the new york stock exchange. 61the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 these theoretical positions show that the impact of options listing on the volatility of the underlying stocks is ambiguous. because of this theoretical debate, the question about whether options trading stabilises or destabilises the market for the underlying assets is probably best addressed empirically. using data from the u.s. market, the evidence from previous empirical studies is still mixed. a number of empirical studies trennepohl & dukes (1979), whiteside et al. (1983), klemkosky and manes (1980), bollen (1998), and mayhew and mihov (2000) showed that listing options had no significant effect on the volatility of the underlying stocks. however, bansal et al. (1989), conrad (1989), skinner (1989) and damodran and lim (1991) provided empirical evidence that listing options reduced the volatility of the underlying stocks. there was at least one study (rao and ma, 1987) with evidence of a significant number of stocks experiencing higher volatility after options were traded. this paper seeks to further examine the analysis of ma and rao’s findings (1988). more volatile stocks become less volatile because of the hedging effect of the uninformed traders, where as less volatile stocks should become more volatile because of the speculative behaviour of the informed traders. unlike other studies in literature, we will attempt to obtain what we deem to be the cleanest possible sample by removing suspected factors that may cause a bias in our results. we employ a control sample methodology to account for the possibility that any change in the volatility of the underlying stocks after the options introduction is due to market-wide conditions and/or to the endogenous nature of the decision to list options. the possible effect of options listing is examined using both a portfolio approach (which is widely used in literature) along with an individual stock approach (which we believe is more likely to provide accurate results).3 the rest of the paper is structured as follows. section 2 will discuss our research motivation. section 3 will introduce our sample selection criteria along with some detail of the empirical methodology applied. in section 4, a description of the data set and presentation of the empirical results will be provided. lastly, section 5 will offer our concluding remarks, specifically stating that the findings indicate no destabilisation of pricing as a result of the options listing effect. 2. literature on option listing effect 2.1. the non-uniform effect of options listing according to ma and rao (1988), listing stock options do not have a uniform impact on the volatility of the underlying security. they argued that options 3 bollen (1998) used the individual stock approach to examine the impact of options listing on the residual variance of the underlying stocks. he employed the approach with the assumptions of no beta changes and that the residuals are normally distributed. in this study, the individual stock approach is used to examine the impact of options listing on the total risk and on the residual risk, without the assumptions of bollen. the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8262 listing would increase the volatility of the stocks that were originally volatile, i.e., those stocks traded by uninformed traders. listing would decrease the volatility of those stocks that were originally stable. the rationale behind their argument was that highly volatile stocks exhibited such volatility because of the large number of uninformed traders making transactions. the availability of the options market as an efficient hedging vehicle would attract the uninformed traders from the stock market to the options market. the ultimate result was a decrease in the volatility of the underlying stocks. stable stocks are presumably less volatile because informed investors trade in them. high leverage and low transaction costs make options a useful vehicle for informed traders to capitalise on their information. their speculative actions will in turn increase the noise in the stock market; thus, informed traders are more likely to trade in the options markets rather than the stock market. the resulting random noise in the options market is transmitted back to the stock market leading to an increase in the volatility of those stocks that were originally stable. 2.2. learning effect hypothesis whiteside et al. (1983) and detemple and jorion (1990) argued that the options listing effect might not be stable over time. market players might discover novel and diverse uses for listed options as time passes. this discovery might include new trading strategies involving options. it is also believed that the options listing effect might change over time following any future changes in regulations that govern the use of trading strategies involving options. 2.3. the endogenous problem of options listing decision another argument advanced by skinner (1989) was that options listing was not a random event since options exchanges adopted stocks for listing that were rated highly on various attributes such as investors’ interests, trading activity, and price volatility. thus, if options exchanges deliberately listed options on stocks with mean-reversing volatility, one would expect the volatility to revert to its mean some time after the option initiation. despite this attempt, damodaran and lim (1991) and kamar et al. (1995) addressed the issue, claiming that the existing research had not adequately accounted for the endogeneity bias, which is the result of decisions made by the exchanges and regulators. they further argued that as long as the selection criteria changed with the changing market characteristics in the market, changes would coincide with option listings, even if the option listing itself had no impact on the underlying market. thus, if the proposition of ma and rao (1988) held, it could then be expected that the result of the impact of options listing on the underlying stock’s volatility would depend on which type of stocks (stable or volatile) are dominant. if the sample chosen for a study included more volatile stocks than stable ones, the researcher would be more likely to conclude that options listing stabilises the 63the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 market. the opposite conclusion would be made if the selected sample contained more stable stocks than volatile ones. however, if by chance a researcher selected a sample in which the decrease in volatility of volatile stocks is equal to the increase in volatility of the stable stocks, then a non-volatility effect of options listing would be the conclusion. mayhew and mihov (2000) provided evidence that this criterion was changing over time, meaning that exchanges were moving from listing more to less volatile stocks. this change in the listing requirements again implied that we would expect, on average, options listing to have a different impact on the volatility of the underlying stocks in different sub-periods during which the sample is selected. thus, we believe the answer as to whether options listing stabilises, destabilises, or has no impact on the volatility of the underlying stock lies in testing whether it is ma and rao’s arguments and/or the endogenous nature of the options listing decisions that are responsible for the dependence of the empirical results on the sample type, sample size, and time horizon of the empirical study; or it is really that options listing is volatility-neutral with optioned stocks having similar behaviour to other stocks with similar characteristics. 3. sample selection criteria, data, and methodology 3.1 sample selection criteria the aforementioned statements posed the idea that we were more interested in obtaining the cleanest possible sample rather than the biggest possible sample. section 2 has shown why it might not be important to use the biggest possible sample in the study of options listing impact. this sub-section will further illustrate the way in which we construct what we determine to be the cleanest possible sample. in order to obtain the cleanest possible sample, we attempted to avoid any suspected problems that might cause bias in our results. we define at least four possible sources of bias that are linked to the sample selection. these might include market completeness, trading location, thin trading problems, and control sample. the market completeness hypothesis can partly explain why including stocks with cross-listed options in the sample may cause a bias in the research results. the possible market completeness bias in the empirical literature on options listing was first recognised by detemple and jorion (1990). they examined the effect of option introductions on stock prices. they found significant increases in the total price for those options listed between 1973 and 1982 with no significant effect for the options introduced after 1982. the authors attributed the absence of the options effect after 1982 to the introduction of options on s and p 500 futures on april 1982, which may have completed the market. this fact was ignored by a number of authors, including ma and rao (1986, 1988), skinner (1989), damodaran and lim (1991), fendenia and grammatikos (1992), and bollen (1998). the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8264 consider a given stock that has options listed on two different option exchanges. including that stock in the sample on the basis of the latter option listing date might cause a problem, as the earlier listing for this stock in the market had probably been completed. even if the stock was included on the basis of the earlier date, there is still a possibility of a sample selection bias if the cross-listing dates were close enough (i.e., the latter listing date is included in the analysis). the bias in this latter case might result from differences in the microstructure of the option exchanges. although there are no theoretical arguments why options should have a differential impact based on various trading locations, several authors (including damodaran and lim (1991) and bollen (1998) suspected the trading location problem and they separated between stocks from different option exchanges and those from different stock exchanges. being cautious, we accounted for the bias possibility that options have differential impacts in various trading locations, so we analysed only one stock market and one option market. similar to chaudhury and elfakhani (1997), precautions were also taken to avoid thin trading problems. the thin trading problem might be minimal in the earlier periods of options listing as exchanges listed mostly large and well-known stocks. however, the problem may be important in more recent periods being there is evidence of exchanges moving towards the listing of small and lowtrading volume stocks (mayhew and mihov, 2000). thus, to avoid thin trading problems in our sample, we dropped all the events that do not have a complete data set.4 because we attempted to compare the behaviour of the optioned stocks with those non-optioned stocks from the same industry, we also considered the availability of data for the best-matched pair of option and stock as another selection criterion. market completeness, trading locations, and thin trading problems should also be taken into consideration when selecting control stocks from the same trading location as the optioned stocks. 3.2 testable hypotheses driven by this aim, we test the following hypotheses: h0: on average, there is no effect of options listing effect on the volatility of the underlying stocks. the failure to reject h0 does not necessarily suggest that option listing is volatilityneutral. some stocks may experience increases while others may experience decreases in the volatility; thus the overall effect may be neutral. 4 chaudhury and elfakhani (1997) argued that to avoid the thin trading problem, those stocks with more than 20 missing, daily-price observations on either side of the options listing window should be excluded. 65the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 thus, we test the following hypothesis: h1: options listing have dual effect on the volatility of the underlying stocks. they increase the volatility of the relatively volatile stocks because of the hedging effect of uninformed traders, and they decrease the volatility of the relatively stable stocks because of the speculative effect of the informed traders (ma & rao; 1988). again the failure to reject h1 does not necessarily suggest that the proposition of ma & rao (1988) holds only for option listed stocks. in other words, the possibility that stable stocks in one period become more volatile in the next period while volatile stocks in one period become less volatile in the next period may be a market-wide phenomenon. this leads us to: h2: hypothesis h1 holds not only for optioned stocks, but also for nonoptioned stocks with similar characteristics. if h2 is accepted, we can argue that options listing does not affect the volatility of the underlying stocks, otherwise a further investigation would be required. 3.3 returns and volatility measures selecting the control sample and testing the above stated hypotheses require the calculation of daily returns. following hamill et al. (2002), we calculated the daily returns for stocks and the market index using the following formulas: y i,t = ln[(p i,t +d i,t ) /p i,t-1 ] (1) x m,,t =ln(p m,t )-ln(p m,t-1 ) (2) y i,t and x m,t are the returns of stock i and market index m at time t; p i,t is the price of security i on day t; d i,t is the total dividend paid during period t; p i,t is the price of security i at the end of the day t-1; and p m,t and p m,t-1 are the prices of the market index on day t and day t-1. the, following ma & rao (1988) and kumar et al. (1995), we calculated the variance of returns for the underlying stocks, in both preand post-listing periods, by using the following variance formulas: the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8266 5 as a reminder, the indicator ‘b’ means before listing and the indicator ‘a’ means afterlisting. 6 the model construction principle is similar to kumar et al. (1995) and mendenhall & fehrs (1999). is the variance of the return of stock i before options listing; is the variance of return of stock i after the option listing date; and yib and yia are the mean of returns in the pre-listing period and in the post-listing period for stock i, respectively. similar to others, we use the single index model to estimate the residual risk for each stock i. the single index model is: is the residual risk for a given stock i; and takes two values depending on the estimation period of the single index model. if the model is estimated in the pre-listing period (day -200 to day -1), then becomes . however, if the model is estimated in the post-listing period (day 0 to day +200), then becomes . 3.4 methodology portfolio approach: to begin the analysis of the effects of options listing on the underlying equity’s volatility and for the empirical convenience, we constructed a portfolio of 234 observations. the portfolio contained 117 of the observations that represent the pre-listing return variance for each stock i that is included in our sample . the other 117 are the post-listing return variance for the same stocks . we then created the following regression model:6 indicator l takes two values: b before option listing, and a after option listing; dumi is a dummy variable, taking a value of zero before options listing and unity after the options listing; η is the random error term of equation (6); α0is the average daily pre-listing variance; and (α0 +α1 ) is the average daily post-listing variance. if options listing increases or decreases the return variance of the underlying stock, we would expect the coefficient on the dummy variable α1to be positive (negative) and statistically significant. otherwise, the equation would support hypothesis h0. then, we checked the results obtained from the equation (6) by applying the same equation to the portfolio of residual variances of the underlying stocks included in our sample as follows: 67the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 again, it is the dummy variable’s coefficient γι that indicates the direction and the significance of the change in the volatility of the underlying stocks. if γι is positive (negative) and statistically significant, equation (7) suggests that there is a significant increase (decrease) in the residual variance after options listing. otherwise, hypothesis h0 is supported. following skinner (1987) and kumar et al. (1995), we checked the results using the wilcoxon signed rank test (wsrt). wsrt is used to test whether the return variance series (residual variance series) before options listing and the return variance series (residual variance series) after options listing are the same. accepting or rejecting hypothesis h0 based only on the above tests does not actually end the task as it is argued that options listing may have a differential effect on different stocks (hypothesis h1). we use the same equation used by ma and rao (1988) to test hypothesis h1. ma & rao (1988) estimated the following equation: they argued that if options listing reduced the volatility of the more volatile stocks because of the hedging effect while increasing the volatility of stable stocks because of the speculation effect, we would then expect coefficient n 1 to be positive and coefficient n 2 to be negative. thus, if the expectation of ma & rao from equation (8) holds for our sample, then there is a possibility for the presence of a non-uniform effect of options listing. but, even if h1 was confirmed, there could be still the possibility that stocks with similar characteristics to our selected optioned stocks would show the same behaviour, and the entire story would then be explained by the market-wide conditions (hypothesis h2). therefore, unless h2 is rejected, there is no options listing impact on the volatility of the underlying stocks. the best way to test h2 is to possibly use a control sample methodology. control sample methodology: control sample methodology is not new in the area of options listing. bollen (1998) used the control sample methodology to test whether the behaviour of the residual variance between the preand post-options listing periods was an industry-wide phenomenon. mayhew and mihov (2000) also used a similar approach to test whether the behaviour of the variance of return between preand post-options listing periods could be explained by the endogenous nature of the options listing decisions. however, control sample methodology has never been used to test whether the ma and rao (1988) proposition could be an industry-wide phenomenon. the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8268 before employing the control sample methodology to achieve the above stated purpose, it is first important to illustrate the way in which the control sample is constructed. when constructing the control sample, we undertook the following procedure: first, we matched each optioned stock with a group of potential control stocks selected from the same industry sector which, (i) had options listed neither before nor a year after the option listing date, and (ii) had complete daily price data available. (these selection requirements were used to avoid the possible biases arising from different trading locations, market completeness, and thin trading.) then, the best control stock for each option-listed stock (among the group of potential control stocks) was selected by employing a ranking approach (see mendenhall and fehrs, 1999) on pre-listing average size, pre-listing return variance, and pre-listing residual variance. the ranking approach was used three times. the first time, we considered a given optioned stock’s pre-listing average size as a benchmark. we then ranked the potential control stocks according to how close their pre-listing average size was to the benchmark. we associated number 1 to the potential control stock with the closest pre-listing average size to the benchmark, number 2 with the potential control sample with the second closest prelisting average size to the benchmark, and so on. we applied the ranking approach for the second (and the third) time by using each optioned stock’s pre-listing return variance (residual variance) as a return variance (residual variance) benchmark, ranking the potential control stocks according to how close their pre-listing return variance (pre-listing residual variance) was to the return variance (residual variance) benchmark. we then associated the ranking with numbers starting from 1 for the potential control stock with the closest pre-listing return variance (residual variance) to the return variance (residual variance) benchmark, number 2 to the potential control stock with the second closest pre-listing return variance (residual variance) to the return variance (residual variance) benchmark, and so on. employing the ranking approach three times would end up with each potential control stock, of a given optioned stock, having three numbers associated with it. the first number indicated how close the potential stock’s average pre-listing size was to the pre-listing size of the optioned stock. the second (third) number indicated how close the pre-listing return variance (residual variance) of the potential control stock was to that of the optioned stock. for each optioned stock, we selected the best control stock, from its potential control stocks, as being that stock with the smallest sum of ranks. the selected control sample is the used to test the hypothesis h2 in the following manner: first, we employed the same portfolio approach on the control sample to test if the volatility behaviour between preand post-options listing of the sample of optioned stocks and that of non-optioned stocks was similar or different. we 69the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 then checked the results using another alternative portfolio approach which combined the analysis of both the sample of optioned stocks and that of non-optioned stocks. this alternative portfolio approach used both a t-test and a mann-whitney test to check whether the change (if any) in the volatility of the optioned stocks was the same (or different) as the one (if any) of the control sample. represents the series of the change (post-listing minus prelisting) in the return variance (residual variance) of each optioned stock i and each control stock j; and dumi,j is a dummy variable, which takes a value of 1 if the stock is optioned and a value of 0 if the stock is not optioned. it is the sign of coefficient d1, e1 that indicates whether the change in the volatility of the optioned stock is higher or lower than the change in the volatility of the control sample. a t-test was then used to check the significance of the difference. as a further check of the results, we performed a mann-whitney test to see whether the change in the volatility of options listing was significantly different from the change in the volatility of the control sample. we then checked the ma andrao (1988) proposition that options listing had a dual effect dependent upon the volatility of the stocks when using equation (8). if all the above tests showed that there was a similar change (if any) between preand post-options listing periods in the volatility for both samples, and if ma and rao’s equation (8) held in a similar way for both samples, then we could more confidently accept hypothesis h2. accepting h2 means that options listing had no impact on the volatility of the underlying stocks. individual stock approach: although the portfolio approach is widely used in the literature of options listing, we argue that the individual stock approach is more likely to provide accurate results. however, using only the sign of the change (or the ratio of the change) as an indicator for an increase or decrease in the volatility without checking the significance of the change may provide us with misleading results. only a few authors have used the individual stock approach to test the possible volatility impact of options listing. the individual stock approach is pursued using merton's (1980) variance approximation. merton (1980) argued that the var(rt), where var(rt) is the variance of the return r at time t, approximately equals e(rt2), where e(rt2) is the expected value of the squared return at time t, for short measurement intervals.7 using this argument, we examined the impact of options listing on the individual stocks’ total risk and residual risk by applying the following equations to each stock: the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8270 is the squared daily returns of the stock i in the 400-day window,, is squared daily residual, as estimated from the single index model, for each stock i; and the sign of the dummy coefficien indicates the direction of the change in the return variance (residual variance) of the security i. a t-test can then determine the significance of the change. we also used the wilcoxon signed ranks test to examine whether daily pre-listing variances and daily post-listing variances for each stock (both from the sample of optioned stocks and the control sample) belonged to the same distribution. 4. data and empirical results 4.1. data the cboe provided us with a list of option-listed stocks along with their corresponding dates of listing. the list involves all the stocks for which options have been introduced, within all the us options exchanges, from april 1973 to march 2001. in order to avoid the problem of trading location together with the issue of market completeness associated with the sample selection bias, we only included those new york stock exchange (nyse) listed stocks with options contracts first introduced on the chicago board option exchange (cboe) whose listing were found nowhere else within a year. there are 358 stocks that fulfil the above requirements. using the daily price data from datastream for the standard & poor’s 500 index (s&p500), which is used as a proxy for the market index, we examined only those stocks with options listing after january 1980 (298 stocks). to be selected, a stock must have complete, daily, dividend-adjusted price data for 400 observations around the options listing dates (200 observations before and 200 days after options listing). this restriction is necessary to avoid thin trading problems. data are then available for 142 stocks. our sample shrinks to 117 stocks when the control sample is being considered. the potential control stocks were identified by the first four digits of their compustat sic code. the first four digits of the sic code provided for each optioned stocks a list of matching stocks within the same trading location and industry sector as this particular optioned stock. in order to avoid the biases that might be linked to the control sample selection, the potential control stock was rejected if it had an option listed either earlier, or less than one year later, than the sample stock’s option introduction. the control stock is also rejected if it does 7 skinner (1989) also used merton’s (1980) approximation to estimate the preand post-option listing 71the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 not have a complete set of 400 daily price observations. then, a ranking approach (described in section 3.5.2) based on the pre-listing average size, pre-listing variance, and pre-listing residual variance was used to select the best control stock from the group of potential control stocks, for each option-listed stock. the mean and the median of pre-listing size return variance and residual variance for both the sample of optioned stocks and that of the control sample are reported in table 1. 4.2 empirical results the optioned stocks results: panel a of table 2 shows the summary statistic for the mean and median change along with the percentage (number) of stocks which show decreases and increases in the return variance and residual variance, between the preand post-options listing periods. the sample of optioned stocks shows a mean increase of 3.59% and a median increase of 3.23%. based on the fisher sign test of the difference between the postand pre-listing return volatilities, 51.28% (60 out of 117) of the stocks experienced an increase, while 48.71% (57 out 117) showed decreases in the return variance after options listing. panel a also shows that the residual risk has decreased by an average of 2.85% and a median of 4.98%. again, based on the sign of the change in the residual risk, 53.8% (63 in number) of the stocks showed a decrease and 46.2% (54 in number) experienced an increase in the residual risk after the options listing. to test the significance of the change in the return variance and its components, panel b of table 2 shows the ordinary least square (ols) estimation of equations (6) and (7) when they are applied to the sample of optioned stocks. panel b shows that the dummy coefficient of equation (6) takes a positive value, suggesting a possible average increase in the volatility of the underlying stocks after the options listing. however, a t-statistic of only 0.314 (p-value = 0.754) associated with the coefficient αι suggests that the increase is not significantly different from zero. the negative value associated with the dummy coefficient γι of equation (7) implies a possible decrease in the average residual risk, but the ttable 1: the characteristics of the sample of the optioned stocks and the control sample which were selected using the ranking approach, in the pre-options listing periods. areturn variances are calculated using the standard formula: where yi,t is the return of stock i at time t; yib is the mean return of stock i in the prelisting period; and n is the number of observations. the reported return variances are multiplied by 104. bresidual variances are the variances of the error terms obtained from the application of the single index market model on each stock in the pre-listing period. the residual variances reported here are multiplied by 104. the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8272 statistic of 0.042 (p-value = 0.967) again indicates that this decrease is not statistically significant at the conventional confidence levels. as a further check of the results, panel c of table 2 shows the result from the wilcoxon signed ranks test (wsrt). wsrt fails to reject the hypothesis that pre-listing return variances, residual variances, and betas are drawn from the same distribution as post-listing return variances, residual variances, and betas, respectively. the z-scores (and p-values) from the wsrt that is associated with testing the significance of the change in return and residual variances are, -0.271 (0.787) and -0.366 (0.715), respectively. thus, the results reported from the application of the portfolio approach seem to support the null hypothesis h0. we argue that the results could be suspicious as our sample contains nearly an equal number of stocks that showed table 2: summary statistic, regression technique, and non-parametric tests are used to test the possible impact of options listing on the volatility (variance of returns and residual variance) of options listing. * indicates the significance at the 1% level a the change is calculated as the difference between the post-and pre-listing variance; the reported mean and median change are multiplied by 104. b the significance is based on fisher signed test c is the variance of return of each stock i, taking two values. one is the return variance beforee options listing, and the other is the return variance after the option listing. dumi is a dummy variable that equals to zero in the pre-listing period and unity in the post-listing period. d is the residual variance of each stock i and it takes two values from each stock. one is the prelisting residual variance and the other is post-listing residual variance. e wilcoxon signed rank test examined whether the pre-listing variances and the post-listing variances of the sample of optioned stocks belong to the same distribution. f denotes the difference between post-listing and pre-listing return variance of each stock i; is the change in the residual variance between postand pre-listing periods for each stock i; and refers to the residual variance of stock i in the pre-listing period. 73the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 increase and those which experienced decrease in the return variance (and residual variance) after options listing. thus, there is a possibility that options have a dual effect on the volatility of the underlying stocks. these effects are balanced out and averaged to zero. panel d of table 2 shows the results of the ols estimation of equation (8). the results from the ols estimation of equation (8) are quite different from the expectations of ma and rao (1988). similarly to ma and rao (1988), the coefficient n 1 from equation (8) is positive, equalling 1.050, and is statistically significant at the 1% level. this suggests a positive and statistically significant relationship between the change in the return variance and the change in the residual variance. however, in contrast to the findings of ma and rao (1988), we discovered no statistically significant relationships between the change in the volatility and the residual variance in the underlying stocks (n 2 = 9.05e-05, pvalue = 0.9489).8 we believe that these results are still logical for a number of possible reasons. first, ma and rao examined the early period (starting from 1973 and ending in 1984). we examined the more recent period (starting from 1980 until 2000). because options listing is an endogenous decision and there is evidence that the listing criteria are changing over time (see mayhew and mihov, 2000), it is possible, therefore, that the characteristics of the stocks included in our sample differ from those included in ma and rao’s (1988) sample. also, the learning effect can play an important role here. furthermore, it is possible that the market completeness hypothesis proposed by detemple and jorion (1988) holds. this means that the assumptions of ma and rao may hold in the earlier periods of options listing (1970s), but it does not hold for more recent periods (1980s and 1990s) as the market becomes complete. above all, there is also and possibility of further questioning the assumptions that were set forward by ma and rao (1988). even following ma and rao’s argument that options attract uninformed traders for hedging purposes and attract informed traders for speculation purposes, it is still possible that the residual error of any stock (stable or volatile) would either increase or decrease depending on how much stabilisation is gained from hedging and how much destabilisation is caused by speculation. further, ma and rao’s argument was based on the possible migration of informed and uninformed traders from the stock market to the options market. however, the availability of options as a hedging vehicle could actually attract more risk averse investors (those who believe that investing in stocks is too risky) to the stock market (figlewski, 1986). this attraction, in turn, may further increase the volatility of those stocks that were originally more volatile (i.e., more uninformed traders making transactions). also, following 8 we also separately examined the period between 1980 and 1984, and the results are still similar to those of the entire sample. however, we admit that the number of stocks included in this period is very small (6 companies). the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8274 cox’s (1976) argument, options might attract a new group of traders to the options market and the volatility effect will depend on the accuracy of the information set of this new group of investors. however, even after rejecting the hypothesis h2, which suggests a dual listing effect, we are still interested whether the ups and downs in the volatility of the stocks can be explained by the endogenous nature of the options listing decision and/or industry-wide conditions. the control samples results: panel a of table 3 shows the summary statistics of the average and median volatility change in the control sample after options listing. it also shows the percentage (a number) of the stocks that experienced increases and decreases in volatility based on the fisher sign test. after options listing, the control sample shows that both the mean and the median table 3: summary statistic, regression technique and non-parametric test to test the possible change in the volatility of the control stocks between preand post-options listing periods. * and ** indicate significance at 1% and 5% levels, respectively. a the change is calculated as the difference between post-and pre-listing variance. the reported mean and median changes are multiplied by 104. b the significance is based on the fisher signed test c is the variance of return of non-optioned stock i and it takes two values. one is the return variance before options listing and the other is the return variance after the option listing. dumi is a dummy variable which equals two zero in the pre-listing period and unity in the post-listing period. d is the residual variance of non-optioned stock i and it takes two values from each stock. one is the pre-listing residual variance and the other is post-listing residual variance. e wilcoxon signed rank test examined whether the pre-listing variances and the post-listing variances of the control sample belong to the same distribution. f denotes the difference between post-listing and pre-listing return variance of non-optioned stock i; is the change in the residual variance between postand pre-listing periods for nonoptioned stock i; and refers to the residual variance of non-optioned stock i in the pre-listing period. 75the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 of the return volatility increased by 9.41% and 4.62%, respectively. similarly, after options listing, the mean and the median of the residual risk increased by 5.54% and 3.30%, respectively. 40.17% (47 in number) of the stocks’ total volatility had decreased after options listing, with the remaining 59.83% (70 in number) experiencing total volatility increases after options listing. similarly, the 46.15% (54 in number) of stocks experiencing a decrease in the residual volatility is a little lower than 53.75% (63 in number) of those with an increased residual volatility after the options listing. panel b of table 3 shows the results of the ols estimation of equations (6) and (7) applied to the control sample. the dummy coefficient α 1 of equation (6) equals 4.68e-05 with a t-statistic of 0.776 (p-value = 0.439). the dummy coefficient γ 1 from equation (7) is 2.45e-05 with a t-statistic equalling 0.449 (pvalue of 0.654). thus, the results from equations (6) and (7) suggest that there is no statistically significant change in the total and the residual volatility between the preand post-options listing periods. similar to the analysis applied on the sample of optioned stocks, the results of the possible change in the volatility of the control sample between the preand post-options listing periods are also checked using a wilcoxon signed rank test (wsrt). panel b of table 3 reports the results. wsrt supports the regression results of no volatility change suggested by equations (6) and (7). the z-score from the wrst equals –1.829 (p = 0.067) for the total volatility change between the preand post-options listing period and is significant only at the 10% level. table 4: comparison between the sample of the optioned stocks and the control sample in terms of the change in the return variances and residual variances between preand post-options listing periods. a denotes the variance change between preand post-listing periods. takes two values. one is , which is the variance change between preand post-listing for the optioned stock i. the other one is , which is the variance change between preand post-listing periods for the control stock j. dumi,j is a dummy variable that takes the value of unity if the stock is optioned and value of zero if the stock is optioned. b refers to the change in the residual variance. also takes two values. one is the change in the residual variance of the optioned stock i, and the other is the change in the residual variance of the control stock j. c mann-whitney test compares whether the series of the changes in the variance (and residual variance) of the sample of optioned stocks and the control stocks are the same. the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8276 also, the z-score of –1.293 (p-value = 0.196) fails to reject the hypothesis that the residual variance series of the preand post-options listing periods belong to the same distribution. the ols estimation results of equation (8) on the control sample are reported in panel c of table 3. the application of equation (8) to the control sample provides us with similar results to the application of the same equation on the sample of optioned stocks. however, this time the coefficient of the n2 residual error is positive and statistically significant at the 1% level. results from the combined analysis: while the above section treated the sample of optioned stocks and the control sample separately, suggesting that the behaviour of both samples are very similar, this section combines the analysis of both samples, checking the results. table 4, panel a reports the results of the estimation of equations (11) and (12), which test whether the change in the sample of optioned stocks and that of the control sample are the same. the dummy coefficient of equation (11), d 1 = -2.52e-05, takes a negative value, suggesting that the increase in the total volatility of the sample of optioned stocks is less than that for the control sample. however, a t-statistic of –0.431 (p-value = 0.667) associated with d 1 suggests that the difference in the total volatility change between the two samples is not significantly different from zero. similar results are also concluded when equation (12) is estimated. the negative sign of the dummy coefficient e 1 (equals –3.96e-05) suggests that the residual variance increases more for the control sample than for the sample of optioned stocks. however, this difference is not statistically significant, as the t-statistic associated with the coefficient e1 is only –0.757 (p-value = 0.450). a mann-whitney test also supports the above results. panel c of table 4 shows z-scores and p-values resulted from mann-whitney test comparison of the total variance change between the sample of optioned stocks and the control sample, which are –0.922 and 0.356, respectively. similarly, the z-score and p-value resulting from the mann-whitney test (which compares the samples residual variance change) are –1.432 and 0.152, respectively. to make the results of the above portfolio approach comparable to those of the individual stocks approach, as reported below, we repeat all the above reported tests using squared daily returns (squared daily residuals) as a proxy for the daily return (residual) variance. the results are very similar and are not reported in detail. we also applied the portfolio approach in two different sub-periods (the period of 1980s and that of 1990s), and the results in both sub-periods are still supportive to the null hypothesis of no volatility change are caused by options listing.9 results from the individual stock approach: so far the results from the portfolio approach fails to reject the null hypothesis h0, which suggests that options 9 the null hypothesis of no variance change is also supported when we divide our sample into a number of arbitrary chosen sub-samples. 77the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 are volatility neutral. in the portfolio approach, the volatility effect of options listing depends on a large extent to the sign, rather the statistical significance of the change in the individual stocks’ variances. dependence only upon the sign can be a misleading approach as the volatility of any stock (or even the index) in the market will either show a positive or a negative relative change when estimated across two different periods (even with the same length). thus, it is possible that the results reported above are simply a chance event. the options listing may actually cause a significant change in the volatility of the individual optioned stock, but the changes in the volatility of the individual non-optioned stock counter-part is not significant. it just so happens that the average and the median of the changes are the same. this possibility is examined by the application of the individual stock approach. the results are presented in table 5. before describing the results, it is important to pinpoint that the conventional 1% and 5% significance levels are used to indicate the statistical significance. table 5 shows that, based on the statistical significance of dummy coefficient, of equation (13), 66.67% (78 in number) of the optioned stocks did not show a statistically significant change in the return variance after options listing; whereas only 23 (19.65%) stocks showed a significant increase, with 16 (13.67%) experiencing a significant decrease in the return variance after options listing. similar results are also indicated by the wsrt, where 64.89% (75 in number) of the stocks showed no change, while 21.36% (25 in number) showed an increase with 14.53% (17 in number) experiencing decreases in the return variance after options listing. table 5 also shows the results of the application of the equation (14) and the wsrt on the residual variance of the individual optioned stocks included in our table 5: the application of individual stock approach to test the significance of the change in the volatility of the underlying stocks between preand post-options listing periods. a, b the number and percentage of the increases and decreases are significant at the 1% and 5% levels. c equation (13) regresses the daily squared returns of each stock on the dummy variable, taking the value of zero before options listing and unity after options listing. the coefficient of the dummy variable measures the change in the variance between the two periods. if the coefficient is statistically significant, the change in the variance is also statistically significant. d wilcoxon signed ranks test (wsrt) compares the change of the daily squared returns of each stock between the preand post-listing periods. the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8278 sample. the null hypothesis of no change in the residual variance of the individual optioned stocks is rejected 69.23% (81 out of 117) and 64.10% (75 in number) of the time based on the statistical significance of the dummy coefficient, , of equation (14) and the wsrt, respectively. equation (14) indicated that 17% (20 in number) of the stocks experienced a statistically significant increase while 13.67% of the stocks showed a significant decrease in the residual variance after the options introduction. similarly, the wsrt indicates that 19.65% (23 in number) of the stocks showed a statistically significant increase while 16.23% (19 out 117) showed a statistically significant decrease in their residual variance after options listing. the results of the individual stock approach on the control sample are broadly similar to those reported from the individual stock approach on the sample of optioned stocks. the results are also reported in table 5. based on the dummy coefficient, of the equation (13), 26% (31 in number) of the control sample’s stocks show statistically significant changes in the return variance after options listing. among these stocks, 18.8% (22 in number) showed a significant increase with only 7.7% (9 in number) showing a significant decrease in the return variance after listing. however, the majority of the stocks in the control sample, 73.5% (86 in number), experienced no statistically significant change in the variance after options listing. the wsrt also fails to reject the null hypothesis that there is no change in the individual stock return variance, 62.39% (73/117) of the time. based on the wrst, only 19.65% (23 of 117) of stocks showed a significant increase while 17.94% (21of 117) showed a significant decrease in the return variance after options listing. similar results are indicated while equation (14) is applied on the test of the null hypothesis that there is no change in the residual variance of the individual stocks included in the control sample after options listing. based on the statistical significance of the dummy coefficient, of equation (14), the null hypothesis of no change in the residual variance of the individual stocks after options listing is rejected 76.2% (89 out of 117) of the time. the remaining 23.8% (28 out 117) of cases are made up of 17% (20 of 117) of stocks with significant increases in the variance after options listing and 6.8% (8 out 117) with significant decreases in the residual variance after options listing. again, the wsrt also supports the regression results, suggesting that 63.25% (74 out 117) of the stocks do not show a significant change in the residual variance after options listing. the wrst suggests that there are 22.22% (26 in number) of stocks that show a significant increase while 14.53% (17 in number) of stocks experience a significant decrease in the residual variance after options listing. we also applied the individual stock approach on the sub-sample of the stocks that were listed in the 1980s and that of the stocks listed in the 1990s. the results also support the null hypothesis of no volatility change caused by options listing.10 10 similar results are also found when the individual stock approach is applied on a number of arbitrary. 79the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 therefore, the individual stock approach strongly supports the null hypothesis h0 that options listing is volatility-neutral. this is not only because the majority of optioned stocks in our sample did not show a statistically significant change in the return variance and the residual variance after the options introduction, but, also because the behaviour of the return variance (and the residual variance) between the preand the post-options listing periods is nearly the same for both the optioned stocks and their control stocks. being that we used the period starting from the 1980s on, the argument of market completeness can be used against our findings, i.e., the market became complete after 1970s, and therefore, did not show an impact thereafter. however, if the market has become complete and options listing do no longer have an effect on the underlying stocks’ volatility, then the issue is no longer of concern to the regulatory agencies and market participants. furthermore, there are still other studies, such as that by bollen (1998) and mayhew and mihov (2000), which covered the entire period (starting from 1973 to 1996) of options listing and found no evidence of options listing on the volatility of the underlying stocks. both bollen and mayhew dan mihov also showed that the change of the volatility of the optioned stocks is very close to that of the control sample in different sub-periods. 5. summary and conclusions for some time, the possible effect of options listing on the volatility of the underlying stocks has been of great concern by the popular press, regulatory bodies, traders, and academics. because of the ambiguous nature of the theoretical debate in the area, the question whether options stabilise or destabilise the volatility of the underlying stocks is presumably best addressed empirically. the empirical evidence in the area is in itself mixed – some studies document increases, others decreases, and some argue for no change in the volatility of the underlying stocks after options listing. in 1988, ma and rao provided a theoretical model along with empirical results, which indicated the possibility that options listing may have a dual effect on the volatility of the underlying stocks, reducing the volatility of the more volatile stocks and increasing that of the stable stocks. ma and rao’s argument, particularly when it is combined with the endogenous nature of the options listing decisions, suggests that the mean (and even the median) may not be a correct measure of the volatility change between the preand post-options listing periods. this study had two purposes. the first was to provide new empirical evidence on the area of options listing. the second was to re-examine the argument proposed by ma and rao. in other words, our second purpose was to test whether ma & rao's argument held only for the optioned stocks or whether it also applied to nonoptioned stocks with similar characteristics. driven by these objectives, we undertook the following analysis. we first selected the cleanest possible sample of optioned stocks and matched it with another comparable sample of non-optioned the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8280 stocks. then, using a portfolio approach that involved both regression techniques and non-parametric tests to examine the significance of the change in the volatility (both total and residual) of the underlying stocks. the portfolio approach was applied on both samples, separately and as a combination. we then estimated the ma & rao model on both samples. finally, we employed the individual stock approach, which we believed to be more accurate on all the stocks in both the sample of optioned stocks and the control sample. the results of the application of the portfolio approach suggest that there is no statistically significant change in the volatility of the underlying stocks, either the optioned and non-optioned ones, after options listing. we also find no evidence of the dual effect of options listing in that the results of the application of the ma and rao model, where our samples differ from ma and rao’s expectation. we do not interpret our results as spurious for at least two reasons. first, we believe that the assumptions made by ma and rao are questionable. second, market completeness could be a reason for the difference between our results and those of ma and rao. as a further check for the possibility of a chance event, which may be resulted from the application of the portfolio approach, we examined the statistical significance of the change in the volatility of each of the underlying stocks in both samples. the individual stock approach supports the null hypothesis that options listing are volatility-neutral. this is not only because the majority of the optioned stocks show no significant volatility change after options listing, but also because the behaviour of the volatility of individual stocks in both sample is very similar. acknowledgement: the author acknowledges the useful comments of an anonymous reviewer and editorial help of the editorial desk on the submitted version of the paper. the author’s e-mail contact is k.mazouz@aston.ac.uk; phone: (4412) 1204 3041. author statement: khelifa mazouz is an academic staff at the aston business school, birmingham, united kingdom. the anonymous reviewers’ comments are acknowledged with thanks. the editors of the journal made significant changes to the original version of the paper, for which we record our sincere appreciation. the errors are the responsibility of the author. references bansal, v., pruitt, s., and wei, k., 1989. an empirical reexamination of the impact of cboe option initiation on the volatility and trading volume of the underlying equities: 19731986. financial review 24, 19-29. 81the international journal of banking and finance, vol. 5.(number 1): 2007: 59-82 black, f., and scholes, m., 1973. the pricing of options and corporate liabilities. journal of political economy, 637-654. bollen, n., 1998. a note on the impact of options on stock return volatility. journal of banking & finance 22, 1181-1191. chaudhury, m. and elfakhani, s., 1997. listing of put options: is there any volatility effect?. review of financial economics 6, 57-75. conrad, j., 1989. the price effect of options introduction. journal of finance 44, 487-498. cox, c., 1976. futures trading and market information. journal of political economy 48, 1215-1237. damodran, a., and lim, j., 1991. the effects of options listing on the underlying stocks’ return processes. journal of banking and finance 15, 647-664. detemple, j., and jorion, p., 1988. options listing and stock returns. fist boston working paper no. 89-13, columbia university. detemple, j., and jorion, p., 1990. option listing and stock return: an empirical analysis. journal of banking & finance 14, 781-801. dybvig, p., and ingersoll, j., 1982. mean-variance theory in complete markets. journal of business 55, 233-251. fendenia, m., and grammatikos, t., 1992. options trading and the bid-ask spread of the underlying stocks. journal of business 65, 335-351. figlewski, s., 1986. the interaction between derivative securities on financial instruments and the underlying cash markets: an overview. journal of accounting, auditing & finance, 299-318. gorton, g., and pennacchi, g., 1993. security baskets and index-linked securities. journal of business 66, 1-28. grossman, s., 1988. an analysis of the implication for stock and return price discovery in stock, futures, and options markets. journal of futures markets 16, 353-387. hamill, p., opong, k., and mcgregor, p., 2002. equity option in the uk: a comparison of market-based research methodologies. journal of empirical finance 9, 91-108. harris, l., 1989. s&p500 cash stock price volatilities. journal of finance 44, 1155-1175. kemkosky, c., and manes, s., 1980. the impact of options on the underlying securities. journal of portfolio management (winter), 12-18. kumar, r., sarin, a., and shastri, k., 1995. the impact of index options on the underlying stocks: the evidence from the listing of nikkei stock average options. pacific-basin finance journal 3, 303-317. ma, c., and rao, r., 1988. information asymmetry and option trading. the financial review 23, 39-51. mayhew, s., and mihov, v., 2000. another look at option listing effects. working paper, http://www.ssrn.com the international journal of banking and finance, vol. 5.(number 1): 2007: 59-8282 mendenhall, r., and fehrs, d., 1999. option listing and stock price response to earnings announcements. journal of accounting & economics 27, 57-87. merton, r., 1980. on estimating the expected return on the market: an explanatory investigation. journal of financial economics 8, 323-361. ross, s., 1976. options and efficiency. quarterly journal of economics 90, 7589. skinner, d., 1989. options markets and stock return volatility. journal of financial economics 23, 61-78. stein, j., 1987. information externalities and welfare reducing speculation. journal of political economy 95, 1123-1145. tennepohl, g., and dukes, w., 1979. cboe options and stock volatility. review of business and economic research 14, 49-60. whiteside, m., marry, m., dukes, p., and dunne, p., 1983. short term impact of option trading on underlying securities. journal of financial research 6, 313-321. 37 international journal of banking and finance 1-1-2007 new evidence on the effect of cboe options listing on the volatility of new york listed stocks khelifa mazouz recommended citation ijbf7-marina.indd the international journal of banking and finance, vol. 7. number 1: 2010: 79-98 79 ijbf testing the performance of asset pricing models in different economic and interest rate regimes using individual stock returns ann marie hibbert and edward r. lawrence west virginia university morgantown and florida international university, united states _____________________________________________________ abstract using return data for all stocks continuously traded on the nyse over the period july 1963 to december 2006, we tested the performance of the two-moment capital asset pricing model (capm) and the fama french three-factor model in explaining individual stock returns. we found the performance of fama french three-factor model to be marginally better than the capm. we further test the models for the signifi cance and stability of parameters in the bull/bear periods and the federal increasing/decreasing interest rate periods and found the performance of the two models comparable. keywords: capm, three-factor model, asset pricing, bear-bull periods, interest rate regimes jel category: g12, g30. _____________________________________________________ 1. introduction sharpe’s two-moment capital asset pricing model is the model most widely used to obtain the discount rate (required rate of return or the cost of equity capital). graham and harvey (2001) survey a sample of 392 fi rms and fi nd that “capm is by far the most popular method of estimating the cost of equity capital: 73.5% of respondents always or almost always use the capm”. even though practitioners use asset pricing models to predict the required return on individual assets, most researchers have used returns on portfolios to test different asset pricing models.1 1the formation of portfolios in asset pricing tests was introduced initially by researchers such as blume (1970), friend and blume (1970) and black, jensen and scholes (1972) and further enhanced by fama and macbeth (1973) to improve the precision of estimated betas for use in cross-sectional regression analysis. ht tp :// ijb f.u um .e du .m y 80 the international journal of banking and finance, vol. 7. number 1: 2010: 79-98 using portfolio returns, researchers fi nd the performance of capm less promising as compared to its most prominent rival, the fama french three-factor model. in this paper we use individual stock returns data to test the performance of capm and the fama french three-factor model and found that contrary to the highly superior performance of fama french three factor model using portfolio returns data, when tested on individual stock return data, the fama french threefactor model performs marginally better in explaining the stock returns and the proportion of stocks that have a signifi cant alpha is comparable for both models. we further investigated the signifi cance and stability of the parameters of capm and fama french three-factor model under changing economic and interest rate cycles and found that unlike the return on the market portfolio which is signifi cant over the entire period, the signifi cance of smb and hml varies with both economic cycles and interest rate cycles. over the last four decades, several studies2 have appeared in the literature that empirically demonstrate that sharpe’s (1963, 1964) two-moment capital asset pricing model does not fully explain the asset pricing mechanism. in general, researchers found four shortcomings in the capm; namely, the model is not a good fi t to the actual rates of return data because of very low coeffi cient of determination, the intercept term is statistically signifi cant signaling specifi cation error problem, the model overestimates (underestimates) the discount rates for low (high) beta stocks and beta is unstable over time. one alternative to the capm that has received a great deal of attention in the fi nance literature is the fama french three-factor model. fama and french (1993) developed a threefactor model that explains the average returns of investment opportunities better than any of the previous models. whereas the central theme of the capm is that the return on a market portfolio is suffi cient to explain asset returns, the threefactor model postulates that in addition to the loading on a market portfolio, loadings on two additional replicating portfolios, smb, the difference between the rates of return on a portfolio of small stocks and large stocks and hml, the difference between the rates of return on a portfolio of high-book-to-market, and a portfolio of low-book-to-market stocks, are needed to explain the returns on assets. fama and french (1996) test their three-factor model on portfolios constructed based on the market value and book value of stocks. they found that, not only is the average coeffi cient of determination (r2) for the fama french three-factor model close to one, but the constant term is insignifi cant as well, suggesting that the model does not suffer from misspecifi cation error. the high r2 and the insignifi cance of the constant term suggest that the fama french three-factor model does not suffer from the problem of underor overestimation of excess returns. 2friend and blume (1970), black (1972), black, jensen, and scholes (1972), miller and scholes (1972), blume and friend (1973), blume and husick (1973), fama and macbeth (1973), basu (1977), reinganum (1981), litzenberger and ramaswamy (1979), banz (1981), gibbons (1982), stambaugh (1982), shanken (1987), fama and french (1992), kothari, shanken, and sloan (1995) and many others. ht tp :// ijb f.u um .e du .m y testing the performance of asset pricing models in different economic and interest rate regimes: 79-98 81 3there is ample evidence reported in the literature indicating that the widely used twomoment capital asset pricing model (capm) shows signifi cantly different results in bear and bull market periods (see, for example, black (1972), levy (1974) chen (1982) whitelaw (2000), perez-quiros and timmermann (2000), and ang and chen (2002)). the cross sectional superiority of the fama french three-factor model over the capm is already academically established and started with the fama and french (1992) claim that capm as a model is “dead”. in a recent paper, lawrence, geppert and prakash. (2007) compared the performance of the twomoment capm, the three-moment capm and the fama french three-factor model using the fama-french 25 portfolio data. based on the time series and the cross sectional tests, they found that the fama french three-factor model outperforms the other models. in this paper we do not compare capm and fama and french three-factor model cross-sectionally as the question of interest here is not if the risk premiums are priced. we tested the two models in time series regressions to investigate the predictive powers of the models using individual stock returns. using individual stock returns, we also tested the stability of the parameters of the two asset pricing models. fama and french (1996) did not test whether the parameters of their three-factor model depend on the market conditions. since most of the models for portfolio selection and allocation of long-term resources (capital budgeting) use asset pricing models to compute the investors’ required rate of return and/or the cost of capital, any inherent instability of the parameters3 in changing market conditions may result in an incorrect decision. therefore, it becomes imperative to search for the model that remains largely immune to the changing market conditions. using individual stock returns, in this paper, we test the stability of the parameters of capm and fama and french three-factor model in the bear and bull market periods. there has been a plethora of empirical studies on the effect of federal discount rate change announcements on the asset prices (waud (1970), cook and hahn (1988), smirlock and yawitz (1985), jensen and mercer (2002)). there seems to be no empirical study that has specifi cally examined the effect of interest changes on the parameters of asset pricing models. the federal (fed) monetary policies are designed to infl uence the overall economy and the fed regularly use the discount rates to revive (restrict) the slowing (growing) economy by reducing (increasing) the discount rates. though the discount rate changes are used to trigger changes in the macroeconomic variables such as overall output, employment and infl ation, the most prominent and direct effect of the discount rate changes is felt in the fi nancial markets through the changes in asset prices and their returns. if this is so, then the discount rate changes should affect the parameters of asset pricing models as well. according to waud (1970), the stock market reacts positively to discount rate decreases and negatively to rate increases. cook and hahn (1988) and smirlock and yawitz (1985) found negative short-term market reaction to discount rate increases and vice versa. jensen and johnson (1995) fi nd evidence that the long-term stock market performance is correlated with changes in the fed discount rate. jensen, merces ht tp :// ijb f.u um .e du .m y 82 the international journal of banking and finance, vol. 7. number 1: 2010: 79-98 and johnson (1996) claim that the monetary environment infl uences investor’s required returns. jensen, johnson and bauman (1997) provide evidence regarding the relevance of monetary conditions for asset pricing. bernanke and kuttner (2005) found strong and consistent response of stock markets to the unexpected changes in the fed interest rates. these studies clearly document the infl uence of fed interest rate regimes on the security prices and their returns; however none of the studies so far have studied the effect of the fed interest rate changes on the parameters of the asset pricing models. in this paper we made an attempt to fi ll this gap. we tested the two asset pricing models in the chronologically delineated non-overlapping (such as bear and bull periods4 and the up and down interest rate regimes) market periods. success of an asset pricing model should necessarily be gauged on how well it explains the returns on single assets. our fi rst contribution was to show that the superior performance of the three-factor model is largely in explaining portfolio returns and not the stock returns. we performed time series analysis of the performance of the two models using both stock and portfolio return data over the 522 months, from july 1963 to december 2006. for portfolio returns we found that the average r2 of the fama french three-factor model is a convincing 18% more than that of the capm. however, when these models are used on the individual stock returns, the differential average r2 falls to 5% and for those stocks where both models perform exceptional the increment is only 3%. furthermore, the proportion of stocks that have a signifi cant alpha is comparable for both models; 7% in the fama french three-factor model and 11% in capm. our second contribution was an investigation of the signifi cance and stability of the smb and hml under changing economic and interest rate cycles. the period of our study is conducive to such an investigation since over the period there have been a number of both bull/bear markets as well as a large number of increasing/decreasing discount rate periods. we found that unlike the return on the market portfolio which is signifi cant over the entire period, the signifi cance of the other two factors varies with both economic cycles and interest rate cycles. in the bull and bear periods, both smb and hml are signifi cant in nearly all of the 25 fama french portfolios but the signifi cance of both smb and hml reduces for individual stocks; smb is signifi cant in 60% of stocks in bull periods and 45% of stocks in bear periods whereas hml is signifi cant in 64% of the stocks in bull periods and 70% of stocks in the bear periods. similar to our fi nding for bull/bear market periods, both smb and hml are signifi cant in nearly all portfolios for the increasing and decreasing interest rate time periods. 4bull and bear markets are measured from the highest closing value on an index to the lowest closing value on an index, and then back again. the defi nition of a bull or bear market is that during a bull market, the market must rise by at least 40%, preferably to a new high in the market, and the market must decline by at least 15% during a bear market. this defi nition fi ts in the “popular investment text” market defi nition of bull and bear market as defi ned by fabozzi and francis (1977). ht tp :// ijb f.u um .e du .m y testing the performance of asset pricing models in different economic and interest rate regimes: 79-98 83 however smb is signifi cant in 54% of stocks in increasing interest rate periods and 53% in the decreasing interest rate periods whereas hml is signifi cant in 69% of the stocks in the increasing interest rate time periods and is signifi cant in 60% of the stocks in the decreasing interest rate periods. our results indicate that the parameters for smb and hml are signifi cant for most of the portfolios returns but they are not signifi cant for the individual stock returns. also, the fama french three-factor model shows weaker results in the bear periods and in the increasing interest rate regimes. with respect to the stability of parameters we found the two models comparable. in the bull/bear periods, we found that the parameter for the market is different in 9% of the stocks using capm and 3% of the stocks using the fama french three-factor model but the parameters for smb and hml are different in respectively 9% and 8% of the stocks. in the fed increasing and decreasing interest rate regimes, the parameter for the market remains nearly the same for the two models, 7% for capm and 8% for the three-factor model while the differences in the parameters for smb and hml are 5% and 3% respectively. the layout of the paper is as follows: in section 2 we briefl y discuss capm and the fama and french three-factor model. in section 3 we provide data and methodology. section 4 has the empirical results. the conclusions are in section 5. 2. capm and fama french three-factor model under the assumptions for the capm, the market portfolio is effi cient and there is a risk free rate available to all investors. the following pricing relationship of the security market line (sml) holds for all individual assets and their portfolios: e[r i ] = r  +  i,m (e[r mt ] r  ) (1) where r i denotes the return on any portfolio or asset i, r m is the return on some proxy of the market portfolio and  i,m = cov (r i r m ) / var (r m ). the above sml relationship allows a test of the capm using the following excess return market model regression equation: r it r  =  i +  i,m (r mt r  ) +  it (2) taking expectations in the above market model we get: e[r i ] r  =  i +  i,m (e[r mt ] r  ) (3) comparing equation 3 with the sml equation 1, we see that capm imposes the restriction that the intercept  i is not signifi cantly different from zero and the coeffi cient on the excess market return (the beta coeffi cient) is statistically signifi cant. ht tp :// ijb f.u um .e du .m y 84 the international journal of banking and finance, vol. 7. number 1: 2010: 79-98 in the late 70s and 80s a number of anomalies concerning certain fi rm specifi c characteristics that seem to have explanatory power for the cross-section of returns beyond the market beta of the capm were reported. for example, basu (1977) provides evidence that when common stocks are sorted on earnings-price ratios, future returns on high e/p stocks are higher than predicted by the capm and banz (1981) document a size effect where low market capitalization fi rms have higher sample mean returns than would be expected if the market portfolio was mean-variance effi cient. other researchers document a leverage effect and a role for the ratio of the book value of a fi rm’s equity to its market value, (be/ me).5 fama and french (1992) investigate the joint role of all these variables by including all of them in their fama-macbeth style cross-sectional regression using portfolios formed fi rst on size and then on betas. using a sample of monthly returns for non-fi nancial fi rms on nyse, amex and nasdaq from 1962-1989, they fi nd that beta does not explain the cross-section of average stock returns, there is a negative relation between size and return, book-to-market equity is signifi cantly positively related to average returns and the combination of size and book-to-market equity seem to absorb the roles of leverage and e/p ratio. they conclude that the two dimensions of risk which are priced are proxied by size and the ratio of book value of equity to market value of equity. fama and french (1996) also report similar fi ndings using the time-series regression approach applied to portfolios of stocks sorted on price ratios. the evidence provided by fama and french (1992) started the claims that capm as a model is “dead”. this has however been countered by other researches who consider fama and french results to be spurious and the result of data mining, (kothari et al. 1995). in using time series regressions to test if the factors in the three-factor model are suffi cient to explain asset returns, the following model is used. r it r ƒt =  i +  i (r mt r t ) + s i smb t + h i hml t +  it (4) if the three-factor model holds, then all three-factor coeffi cients are signifi cantly different from zero and the intercept is not signifi cantly different from zero. the three-factor model is now widely used in empirical research that requires a model of expected returns. it has been used in event studies to test for abnormal performance (loughran and ritter (1995); mitchell and stafford (2000) as well as models that study mutual fund performance (carhart, 1997). however, todate there is no theory underlying this model. 3. data and methodology a. data the data for this study consisted of all fi rms with monthly return data on crisp from july, 1963 to december 2006. monthly value-weighted market return, 5bhandari (1988) found that high debt-equity ratios are associated with returns that are too high relative to their market betas and rosenberg, reid and lanstein (1985) documented that stocks with high book-to-market equity ratios have higher average returns than predicted by their betas. ht tp :// ijb f.u um .e du .m y testing the performance of asset pricing models in different economic and interest rate regimes: 79-98 85 ttt bear itt bull i tt bear itt bull i ftmtt bear iftmtt bull i t bear it bull iftit tftmtt bear iftmtt bull i t bear it bull iftit hmlbbhhmlbbh smbbbssmbbbs rrbbrrbb bbabbrrfbbff rrbbrrbb bbabbarrcapmbb 1 1 1 1:3 1 )1(: return on the benchmark portfolios, hml, smb and the monthly risk-free rate of return for the sample period are obtained from kenneth french’s website. we also obtained monthly value-weighted return on the 25 fama-french portfolios which are the intersection of 5-size sort and 5-be/me sort from kennethfrench’s website. table 1 provides summary statistics of the data. we included only those stocks that have been continuously traded over the sample period, a total of 245 stocks. over the sample period the mean monthly excess return on the market is 0.476% which is similar to the value of 0.47% that was reported by fama and french (2006) for their july 1963 to december 2004 period. b. individual asset returns we performed time series analysis on each of the individual stocks and each of the 25 fama french portfolios using the following two models: camp : r it r ƒt = a i +  i (r mt r t ) +  t ff3f : r it r t = a i +  i (r mt r t ) + s i smb + h i hml +  t (5) in the above models we tested for the signifi cance of the coeffi cient of determination of capm and ff3f. in addition, we also test if the signifi cance of the intercept is close to zero and ,  i s i and h i are signifi cantly different from zero. c. stability tests over different market conditions we investigate the stability of the parameters in capm and the fama french three-factor model over bear/bull economic cycles and the fed interest rate cycles. similar to the models used by fabozzi and francis (1977) we extend the capm and the three-factor model to include dummy variables for bull/bear market conditions and increasing/decreasing interest rate periods. the extended models that we use to test the stability of the parameters over bull/bear market conditions are: (6) ht tp :// ijb f.u um .e du .m y 86 the international journal of banking and finance, vol. 7. number 1: 2010: 79-98 p an el a – m ar k et d at a m ea n m ax im um m in im um s ta nd ar d d ev ia ti on r f 0. 47 0 1. 35 0 0. 06 0 0. 22 5 m kt 0. 94 6 16 .5 60 -2 2. 53 0 4. 36 5 im b 0. 24 8 21 .8 70 -1 6. 58 0 3. 22 1 m l 0. 45 3 13 .7 10 -1 2. 66 0 2. 90 1 p an el b – s to ck s n m ea n m ax im um m in im um s ta nd ar d d ev ia ti on 24 5 1. 26 7 2. 46 6 0. 43 8 0. 30 7 p an el c – f f 25 p or tf ol io s b oo k to m ar ke t e qu it y (b e /m e ) qu in ti le s s iz e l ow 2 3 4 h ig h l ow 2 3 4 h ig h m ea n r et ur n s ta nd ar d d ev ia ti on s m al l 0. 71 1 1. 29 7 1. 33 7 1. 54 6 1. 66 0 8. 12 5 6. 92 9 5. 92 5 5. 54 6 5. 83 5 2 0. 87 8 1. 14 1 1. 41 1 1. 45 8 1. 52 4 7. 35 6 5. 95 8 5. 30 3 5. 09 4 5. 65 4 3 0. 88 9 1. 20 5 1. 21 0 1. 33 4 1. 50 6 6. 74 5 5. 37 9 4. 85 0 4. 67 5 5. 31 4 4 0. 99 8 0. 99 4 1. 22 2 1. 33 4 1. 37 4 5. 97 0 5. 07 7 4. 78 3 4. 60 6 5. 24 4 b ig 0. 87 9 0. 96 8 0. 98 2 1. 06 6 1. 07 4 4. 71 3 4. 47 8 4. 24 3 4. 16 8 4. 72 3 t hi s ta bl e pr ov id es s um m ar y st at is ti cs o f th e da ta w e us e in t hi s st ud y. p an el a g iv es t he m ea n, m ax im um , m in im um a nd s ta nd ar d d ev ia ti on o f th e m on th ly r is kfr ee r at e of r et ur n (r ) , t he m on th ly r et ur n on t he m ar ke t po rt fo li o (m kt ) an d th e tw o ad di ti on al f am af re nc h fa ct or s, s m b a nd h m l; p an el b p re se nt s th e re sp ec ti ve s ta ti st ic s of t he r et ur ns o f al l st oc ks c on ti nu ou sl y tr ad ed o ve r th e sa m pl e pe ri od , ju ly 1 96 3 to d ec em be r 20 06 a nd i n pa ne l c w e pr ov id e th e m ea n an d s ta nd ar d d ev ia ti on o f th e m on th ly v al ue w ei gh te d re tu rn o n th e 25 f am af re nc h po rt fo li os f or m ed b as ed o n th e in te rs ec ti on of t he 5 s iz e an d 5 bo ok -t om ar ke t eq ui ty q ui nt il es . t ab le 1 s u m m a ry s ta ti st ic s ht tp :// ijb f.u um .e du .m y testing the performance of asset pricing models in different economic and interest rate regimes: 79-98 87 ttt decr itt incr i tt decr itt incr i ftmtt decr iftmtt incr i t decr it incr iftit tftmtt decr iftmtt incr i t decr it incr iftit hmldrhhmldrh smbdrssmbdrs rrdrrrdr dradrrrfdrff rrdrrrdr dradrarrcapmdr 1 1 1 1:3 1 )1(: bb is a dummy variable which has a value of “1” for months that are part of bull market periods and zero otherwise. we used. similar models to test the stability of the parameters over different discount rate periods, by including a dummy variable, dr which takes a value of “1” for months when the discount rate is increasing and zero otherwise: (7) we fi rst tested the signifi cance of the market factor in explaining individual stock returns over different market conditions using models capmbb and capmdr. then, using stock data, we investigate the stability of the additional factors in the fama french threefactor model by estimating models ff3fbb and ff3fdr. specifi cally, our null hypotheses are: (8) (9) 3. empirical results a. individual asset returns vs. portfolio returns panel a of table 2 provides regression results of the capm and the fama french three-factor model for the individual stocks in our sample and panel b provides similar results for the fama french 25 portfolios. for portfolio returns, our results are similar to those of fama and french (1993, 1995, 1996). for the capm the r2 ranges from a low of 58% to a high of 87% with an average of 73%; while for the fama french three-factor model, the lowest r2 is 79%, the highest is 95% and the average r2 is a convincing 91%. in addition, whereas the intercept is signifi cant in 15 of the portfolios when capm is used, this number is reduced to 8 using the fama french three-factor model. bear i bull i bear i bull i bear i bull i hhh ssh h : : : 03 02 01 decr i incr i decr i incr i decr i incr i hhh ssh h : : : 06 05 04 ht tp :// ijb f.u um .e du .m y 88 the international journal of banking and finance, vol. 7. number 1: 2010: 79-98   ttitiftmtiiftit hmlhsmbsrrarrfff  :3   tftmtiiftit rrarrcapm  : table 2 comparison of time series regressions of stocks vs. fama french 25 size/ beme portfolios in this table we report the regression results for the following two models: in panel a, we provide the results for the time series regression with the dependent variable being the monthly stock return on each of the 245 stocks continuously traded over the sample period of july 1963 to december 2006. in panel b we present the regression results for the 25 size/beme portfolios. number and proportion of stocks (portfolios) with signifi cant parameters model ave. r2 max r2 min r2   s h panel a: stocks capm 22% 60% 3% 26 245 11% 100% ff3f 27% 63% 4% 17 245 164 191 7% 100% 67% 78% panel b: 25 size/beme portfolios capm 73% 87% 58% 15 25 60% 100% ff3f 91% 95% 79% 8 25 25 25 32% 100% 100% 100% the results for the individual stocks reported in panel a are less convincing. here the range of r2 is similar for the two models, 3% to 60% for the capm and 4% to 63% for the three-factor model. also, unlike the results for the fama french 25 portfolios in panel b where we get an 18% improvement in average r2 using the fama french three-factor model versus the capm model, with individual stocks, the difference in average r2 is only 5%. in addition, for capm, the intercept is signifi cant in 11% of the stocks compared to 7% for the three-factor model. whereas, smb and hml are signifi cant in explaining ht tp :// ijb f.u um .e du .m y testing the performance of asset pricing models in different economic and interest rate regimes: 79-98 89 the returns of all 25 portfolios, they are signifi cant only in 67% (smb) and 78% (hml) of the stocks. together these results provided evidence that the improvements in the fama french three-factor model over the capm is in explaining portfolio returns than individual stock returns. b. bull/bear markets in appendix 1, we provided the start and end date of the bull/bear periods and a summary of the total number of months for each. in table 3 we present the results of the regression models when the capm and the fama french threefactor models are extended to include dummies for the bull/bear months. comparison of panels a and b shows that only the market factor is consistently signifi cant in explaining both stock and portfolio returns during the bull and bear market periods. in the bear period, both smb and hml are signifi cant in all the portfolios but in bull periods, smb is signifi cant in 96% of portfolios whereas hml is signifi cant in 88% of the portfolios. the signifi cance of both smb and hml reduces for individual stocks; smb is signifi cant in 60% of stocks in the bull period and 45% of stocks in the bear period whereas hml is signifi cant in 64% of stocks in the bull period and 70% of stocks in the bear period. the results indicated that parameters for smb and hml are signifi cant for most of the portfolio returns but they are not signifi cant for nearly half of the individual stock returns. the three-factor model shows weaker results in the bear periods where the parameter for smb is insignifi cant for 55% of the stocks. c. increasing/decreasing interest rates in appendix 2, we provided the start and end date of the increasing and decreasing interest rate periods and a summary of the total number of months for each. over the sample period, the number of months during which the interest rates was increasing is approximately equal to the number of months when interest rate was decreasing (266 vs. 256). ht tp :// ijb f.u um .e du .m y 90 the international journal of banking and finance, vol. 7. number 1: 2010: 79-98 n um be r an d p ro po rt io n of s to ck s (p or tf ol io s) w it h s ig ni fi ca nt p ar am et er s m od el a ve . r 2 m ax r 2 m in r 2  b u l l  b e a r b u l l b e a r sb u l l sb e a r hb u l l hb e a r p an el a : s to ck s c a p m b b 22 % 61 % 3% 10 27 24 5 24 4 4% 11 % 10 0% 10 0% f f 3f b b 28 % 63 % 5% 13 20 24 4 24 5 14 7 11 0 15 6 17 1 5% 8% 10 0% 10 0% 60 % 45 % 64 % 70 % p an el b : 25 s iz e/ b e m e p or tf ol io s c a p m b b 73 % 87 % 59 % 4 8 25 25 16 % 32 % 10 0% 10 0% f f 3f b b 91 % 95 % 79 % 7 3 25 25 24 25 22 25 28 % 12 % 10 0% 10 0% 96 % 10 0% 88 % 10 0% t ab le 3 t im e s er ie s r eg re ss io n s o f s to ck s a n d f a m a f re n ch 2 5 s iz e/ b e m e p o rt fo li o s w it h d u m m ie s fo r b u ll /b ea r m a rk et s     t t t b e a r i t t b u ll i t t b e a r i t t b u ll i ft m t t b e a r i ft m t t b u ll i t b e a r i t b u ll i ft it hm l b b h hm l b b h sm b b b s sm b b b s r r b b r r b b b b b b r r f b b f f                      ) 1( ) 1( ) 1( ) 1( : 3     t ft m t t b e a r i ft m t t b u ll i t b e a r i t b u ll i ft it r r b b r r b b b b b b r r c a p m b b                ) 1( ) 1( : in t hi s ta bl e w e re po rt t he r eg re ss io n re su lt s fo r th e fo ll ow in g tw o m od el s: b b i s a du m m y va ri ab le w hi ch i s 1 fo r th e m on th s th at a re p ar t of b ul l m ar ke t pe ri od a nd 0 o th er w is e. i n p an el a w e pr ov id e re su lt s fo r th e ti m e se ri es re gr es si on w it h th e de pe nd en t v ar ia bl e be in g th e m on th ly s to ck r et ur n on e ac h of th e 24 5 st oc ks c on ti nu ou sl y tr ad ed o ve r th e sa m pl e pe ri od o f ju ly 1 96 3 to d ec em be r 20 06 . i n p an el b w e pr es en t th e re gr es si on r es ul ts f or t he 2 5 s iz e/ b e m e p or tf ol io s. ht tp :// ijb f.u um .e du .m y testing the performance of asset pricing models in different economic and interest rate regimes: 79-98 91 n um be r an d p ro po rt io n of s to ck s (p or tf ol io s) w it h s ig ni fi ca nt p ar am et er s m od el a ve . r 2 m ax r 2 m in r 2  in c r  d e c r i n c r d e c r si n c r sd e c r hi n c r hd e c r p an el a : s to ck s c a p m d r 22 % 61 % 3% 23 38 24 5 24 4 9% 16 % 10 0% 10 0% f f 3f d r 27 % 63 % 4% 20 20 24 5 24 3 13 2 12 9 17 0 14 8 8% 8% 10 0% 99 % 54 % 53 % 69 % 60 % p an el b : 25 s iz e/ b e m e p or tf ol io s c a p m d r 73 % 87 % 59 % 3 13 25 25 12 % 52 % 10 0% 10 0% f f 3f d r 91 % 95 % 79 % 1 6 25 25 24 25 24 25 4% 24 % 10 0% 10 0% 96 % 10 0% 96 % 10 0% t ab le 4 t im e s er ie s r eg re ss io n s o f s to ck s a n d f a m a f re n ch 2 5 s iz e/ b e m e p o rt fo li o s w it h d u m m ie s fo r in cr ea si n g /d ec re a si n g i n te re st r a te r eg im es     t ft m t t d e c r i ft m t t in c r i t d e c r i t in c r i ft it r r d r r r d r d r d r r r c a p m d r                ) 1( ) 1( :     t t t d e c r i t t in c r i t t d e c r i t t in c r i ft m t t d e c r i ft m t t in c r i t d e c r i t in c r i ft it hm l d r h hm l d r h sm b d r s sm b d r s r r d r r r d r d r d r r r f d r f f                      ) 1( ) 1( ) 1( ) 1( : 3 d r is a d um m y va ri ab le w hi ch is 1 f or m on th s in w hi ch th e di sc ou nt r at e is in cr ea si ng a nd 0 o th er w is e. i n p an el a w e pr ov id e re su lt s fo r th e ti m e se ri es re gr es si on w it h th e de pe nd en t va ri ab le b ei ng t he m on th ly s to ck r et ur n on e ac h of t he 2 45 s to ck s co nt in uo us ly t ra de d ov er t he s am pl e pe ri od o f ju ly 19 63 t o d ec em be r 20 06 . i n p an el b w e pr es en t th e re gr es si on r es ul ts f or t he 2 5 s iz e/ b e m e p or tf ol io s. in t hi s ta bl e w e re po rt t he r eg re ss io n re su lt s fo r th e fo ll ow in g tw o m od el s: ht tp :// ijb f.u um .e du .m y 92 the international journal of banking and finance, vol. 7. number 1: 2010: 79-98 in table 4, we present the results of the regression models when the capm and the fama french three-factor models are extended to include dummies for the increasing/decreasing interest rate periods. comparison of panels a and b shows that only the market factor is consistently signifi cant in explaining both stock and portfolio returns during the different interest rate periods. similar to our fi nding for bull/bear market periods, both smb and hml are signifi cant in almost all portfolios for the increasing and decreasing interest rate time periods. however, for individual stocks, the signifi cance of both smb and hml reduces signifi cantly. smb is signifi cant in 54% of stocks in increasing interest rate periods and 53% in the decreasing interest rate periods. hml is signifi cant in number and proportion of stocks (portfolios) with different coeffi cients model βbull/bear sbull/sbear hbull/hbear βincr/decr sincr/sdecr hincr/hdecr panel a: stocks capmbb 23 9% ff3fbb 7 22 19 3% 9% 8% capmdr 18 7% ff3fdr 19 8 8% 3% panel b: 25 size/beme portfolios capmbb 10 40% ff3fbb 4 0 12 16% 0% 48% capmdr 0 0% ff3fdr 2 7 7 8% 28% 28% table 5 test of equivalence of slopes in this table we report results for the tests of equivalence of slopes for each of the models previously reported and described in the texts with dummies included for bull/bear market periods and increasing/decreasing interest rate periods respectively. in panel a we present the results of the test of slope coeffi cients for the individual stocks and in panel b we report the results for the 25 size/beme portfolios. the results are based on f-statistics signifi cant at the 1% level. ht tp :// ijb f.u um .e du .m y testing the performance of asset pricing models in different economic and interest rate regimes: 79-98 93 69% of the stocks in the increasing interest rate time periods and is signifi cant in 60% of the stocks in the decreasing interest rate periods. in table 5, we provide the results for the f-tests of equivalence of the slope coeffi cients for each of the two models in the bear/bull and the fed interest rate increasing/decreasing time periods. for capm, beta is different in 9% of the stocks in the bear/bull periods and 18% of the stocks in the increasing/ decreasing interest rate periods. for the fama french three-factor model, beta is different in 3% of stocks in the bear/bull periods and 8% of stocks in the increasing/decreasing interest rate periods. the parameters for smb and hml are different in respectively 9% and 8% stocks in the bear/ bull time periods and respectively 5% and 3% stocks in the increasing/decreasing interest rate periods. the differences in the parameters are more prominent in the portfolio returns where for the bear/bull market period, beta values are different for 40% of the portfolios for capm and 16% of the portfolios for the three-factor model. for the increasing/decreasing interest rate periods there is no difference in the beta values for capm but the beta values are different for 8% of portfolios for the three-factor model. the parameter for smb is not different for any of the 25 portfolios in the bear/bull periods. it is different for 28% portfolios in the increasing/decreasing interest rate periods. the parameter for hml is different in 48% of the portfolios in the bear/bull periods and 28% of the portfolios in the increasing/decreasing interest rate periods. panels a and b shows that the market factor is generally more stable than smb and hml in explaining both stocks and portfolio returns over different market conditions. 4. conclusions in practice, asset pricing models are used to compute the expected returns of individual assets. these returns are then used in the computation of fundamental price of stock by investors and the net present value of projects by corporate managers. even though asset pricing models are used for the individual assets they are invariably tested using portfolio return data to avoid the problem of errors in variables. though capm is inarguably the most used model by practitioners, it performs poorly when tested against the fama french three-factor model using portfolio return data. in this paper we tested the performance of capm and the fama french three-factor model using individual stock return data and fi nd that the fama french three-factor model performs marginally better than the capm. we also testwd the stability of parameters of the two models in the economic conditions (bear and bull periods and the federal increasing and decreasing interest rate regimes) and found the two models comparable. ht tp :// ijb f.u um .e du .m y 94 the international journal of banking and finance, vol. 7. number 1: 2010: 79-98 author information: submitting author, edward r. lawrence, assistant professor of finance, college of business administration, florida international university, miami, fl 33199. phone: (305)348-0082. email: elawrence@fl u. edu ann marie hibbert is an assistant professor of finance, west virginia university, wv 26506-6025, tel (304)293-2447. email: annmarie.hibbert@ mail.wvu.edu. appendix 1 bull/bear periods from july 1963 to december 2006 time period bull/bear months july 1, 1963 to february 9, 1966 bull 31 february 9, 1966 to october 7, 1966 bear 8 october 7, 1966 to november 29, 1968 bull 26 november 29, 1968 to may 26, 1970 bear 18 may 26, 1970 to january 11, 1973 bull 32 january 11, 1973 to december 6, 1974 bear 23 december 6, 1974 to september 21, 1976 bull 21 september 21, 1976 to february 28, 1978 bear 17 february 28, 1978 to april 27, 1981 bull 38 april 27, 1981 to august 12, 1982 bear 15 august 12, 1982 to august 25, 1987 bull 61 august 25, 1987 to october 19, 1987 bear 2 october 19, 1987 to july 16, 1990 bull 33 july 16, 1990 to october 11, 1990 bear 3 october 11, 1990 to july 17, 1998 bull 93 july 17, 1998 to october 5, 1998 bear 3 october 5, 1998 to january 14, 2000 bull 15 january 14, 2000 to october 9, 2002 bear 33 october 9, 2002 to december 31, 2006 bull 51 total bear 133; total bull 139 ht tp :// ijb f.u um .e du .m y testing the performance of asset pricing models in different economic and interest rate regimes: 79-98 95 appendix 2 periods of increasing/decreasing interest rates from july 1963 to december 2006 time period series number of months july, 1963 to march, 1967 increasing 47 april, 1967 to october, 1967 decreasing 7 november, 1967 to july, 1968 increasing 9 august, 1968 to november, 1968 decreasing 4 december, 1968 to october, 1970 increasing 23 november, 1970 to june, 1971 decreasing 8 july, 1971 to october, 1971 increasing 4 november, 1971 to december, 1972 decreasing 14 january, 1973 to november, 1974 increasing 23 december, 1974 to july, 1977 decreasing 32 august, 1977 to april, 1980 increasing 33 may, 1980 to august, 1980 decreasing 4 september, 1980 to october, 1981 increasing 14 november, 1981 to march, 1984 decreasing 29 april, 1984 to october, 1984 increasing 7 november, 1984 to august, 1987 decreasing 34 september, 1987 to november, 1990 increasing 39 december, 1990 to april, 1994 decreasing 41 may, 1994 to december, 1995 increasing 20 january, 1996 to july, 1999 decreasing 43 august, 1999 to december, 2000 increasing 17 january, 2001 to may, 2004 decreasing 41 june, 2004 to december, 2006 increasing 31 total increasing 266 total decreasing 256 ht tp :// ijb f.u um .e du .m y 96 the international journal of banking and finance, vol. 7. number 1: 2010: 79-98 references ang, a., and chen, j., (2002). asymmetric correlations of equity portfolios. journal of financial economics, 63, 443-494. banz, r.w., (1981). the relationship between return and market value of common stocks. journal of financial economics, 9, 3-18. basu, s., (1977). investment performance of common stocks in relation to their price-earnings ratios: a test of the effi cient market hypothesis. the journal of finance, 32, 663-682. bernanke, b., and kuttner, k.n., (2005). what explains the stock market's reaction to federal reserve policy? the journal of finance, 60, 12211257. bhandari, l.c., (1988). debt/equity ratio and expected common stock returns: empirical evidence. the journal of finance, 43, 507-528. black, f., (1972). capital market equilibrium with restricted borrowing. journal of business, 45, 444-455. black, f., jensen, m.c., and scholes, m., (1972). the capital asset pricing model: some empirical tests. praeger, new york. blume, m.e., (1970). portfolio theory: a step toward its practical application. journal of business, 43, 152-173. blume, m.e., and friend, i., (1973). a new look at the capital asset pricing model. journal of finance, 28, 19-33. blume, m.e., and husick, f., (1973). price, beta and exchange listing. journal of finance 28, 283-299. carhart, m.m., (1997). on persistence in mutual fund performance. the journal of finance, 52, 57-82. chen, s.n., (1982). an examination of risk return relationship in bull and bear markets using time varying betas. journal of financial and quantitative analysis, 17, 265–286. cook, t., and hahn, t., (1988). the information content of discount rate announcements and their effect on market interest rates. journal of money, credit and banking, 20, 167–180. fabozzi, f.j., and francis, j.c., (1977). stability tests for alphas and betas over bull and bear market conditions. the journal of finance, 32, 1093-1099. fama, e.f., and french, k.r., (1992). the cross-section of expected stock returns. the journal of finance, 47, 427-465. fama, e.f., and french, k.r., (1993). common risk factors in the returns on stocks and bonds. journal of financial economics, 33, 3-56. fama, e.f., and french, k.r., (1995). size and book-to-market factors in earnings and returns. the journal of finance, 50, 131-155. fama, e.f., and french, k.r., (1996). multifactor explanations of asset pricing anomalies. the journal of finance, 51, 55-84. fama, e.f., and french, k.r., (2006). the value premium and the capm. the journal of finance, 61, 2163-2185. ht tp :// ijb f.u um .e du .m y testing the performance of asset pricing models in different economic and interest rate regimes: 79-98 97 fama, e.f., and macbeth, j.d., (1973). risk, return, and equilibrium: empirical tests. the journal of political economy, 81, 607-636. friend, i., and blume, m., (1970). measurement of portfolio performance under uncertainty. american economic review, 60, 561-575. gibbons, m.r., (1982). multivariate tests of fi nancial models. journal of financial economics, 10, 3-27. graham, j.r., and harvey, c.r., (2001). the theory and practice of corporate fi nance: evidence from the fi eld. journal of financial economics, 60, 187-243. jensen, g.r., and johnson, r.r., (1995). discount rate changes and security returns in the us, 1962-1991. journal of banking and finance, 19, 79-95. jensen, g.r., johnson, r.r., and bauman, w.s., (1997). federal reserve monetary policy and industry stock returns. journal of business finance & accounting, 24, 629-644. jensen, g.r., and mercer, j.m., (2002). monetary policy and the cross-section of expected stock returns. the journal of financial research, 25, 125-139. jensen, g.r., mercer, j.m., and johnson, r.r., (1996). business conditions, monetary policy, and expected security returns. journal of financial economics, 40, 213-237. kothari, s.p., shanken, j., and sloan, r.g., (1995). another look at the crosssection of expected stock returns. the journal of finance, 50, 185-224. lawrence, e.r., geppert, j., and prakash, a.j., (2007). asset pricing models: a comparison. applied financial economics, 17, 933-940. levy, r.a., (1974). beta coeffi cients as predictors of return. financial analysts journal, 30, 61-9. litzenberger, r.h., and ramaswamy, k., (1979). the effects of dividends on common stock prices: theory and empirical evidence. journal of financial economics, 7, 163-195. loughran, t., and ritter, j.r., (1995). the new issues puzzle. the journal of finance, 50, 23-51. miller, m.h., and scholes, m., (1972). rates of return in relation to risk: a reexamination of some recent fi ndings. in michael c. jensen (ed.), studies in the theory of capital market (pp. 47-78). new york: praeger. mitchell, m.l., and stafford, e., (2000). managerial decisions and long-term stock price performance. journal of business, 73, 287. perez-quiros, g., and timmermann, a., (2000). firm size and cyclical variations in stock returns. the journal of finance, 55, 1229-1262. reinganum, m.r., (1981). a new empirical perspective on the capm. journal of financial and quantitative analysis, 16, 439-462. rosenberg, b., reid, k., and lanstein, r., (1985). persuasive evidence of market ineffi ciency. journal of portfolio management , 11, 9–17. shanken, j., (1987). multivariate proxies and asset pricing relations: living with the roll critique. journal of financial economics, 18, 91-110. ht tp :// ijb f.u um .e du .m y 98 the international journal of banking and finance, vol. 7. number 1: 2010: 79-98 sharpe, w.f., (1963). a simplifi ed model for portfolio analysis. management science (pre-1986) , 9, 277-293. sharpe, w.f., (1964). capital asset prices: a theory of market equilibrium under conditions of risk. the journal of finance, 19, 425-442. smirlock, m., and yawitz, j., (1985). asset returns, discount rate changes, and market effi ciency. the journal of finance, 40, 1141-1158. stambaugh, r.f., (1982). on the exclusion of assets from tests of the twoparameter model: a sensitivity analysis. journal of financial economics, 10, 237-268. waud, r.n., (1970). public interpretation of federal reserve discount rate changes: evidence on the ‘announcement effect’. econometrica, 38, 231–250. whitelaw, r.f., (2000). stock market risk and return: an equilibrium approach. the review of financial studies, 13, 521-547. ht tp :// ijb f.u um .e du .m y the efficiency of trading halts: emerging market evidence the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 125 the efficiency of trading halts: emerging market evidence obiyathulla i. bacha, mohamed eskandar s. a. rashid and roslily ramlee internacional islamic university malaysia abstract this paper reports new findings on the price effect from trading halts both voluntary and mandatory over 2000-04 in an emerging share market, malaysia. based on our overall sample, trading halts lead to positive price reaction, increased volume, and increased volatility. we found evidence of information leakage resulting in a significant difference between voluntary and mandatory halts as well as the type of news released during halts to warrant such an impact. the duration of the halt has an isolated impact and is largely inconsequential. the frequency of halts does not seem to matter. keywords: trading halts, price, efficiency, and malaysia jel classification: g14 1. introduction the efficacy of trading halts (or trading suspensions) in overcoming informational asymmetries remains a controversy. being that it is a temporary suspension of trading in a stock, a trading halt is essentially a signal by the exchange that disequilibrium exists or is expected to exist. that disequilibrium may be the result of an order imbalance or more likely due to pending news. the halt therefore is a “time-out” for the adjustment of prices during real or perceived, temporary disequilibria. though numerous studies have been undertaken in various markets, there appear to be no consensus on the usefulness of trading halts/suspensions. the objective of a trading halt, whether voluntary (one initiated by the firm) or mandatory (one imposed by regulators), is to always ensure a fair access to information and price formation. therein lies the debate of whether that objective is achieved or not. supporters of trading halts propose that calling a halt to trading just prior to the voluntary or forced release of critical information enhances price formation/discovery by ensuring equitable dissemination and assimilation. often denoted as the price efficiency hypothesis of trading halts1, the rationale is that since a trading suspension gives time for investors to better digest the impact of news, price dispersions should be smaller, implying lower volatility 1 see chen h, chen h, and valerios, n(2003). ijbf 126 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 and a higher degree of price efficiency upon reopening. opponents of trading halts using the “learning through trading models” would argue that since a halt prevents trading during the suspension period, excess volatility is not necessarily being avoided, but rather, merely postponed. based on the logic of “learning through trading models”, reopening prices would be noisy. this implies reduced pricing efficiency and increased volatility. just as the arguments have been diametric, empirical evidence thus far have been mixed. most empirical work on firm-specific trading halts had focused on the three key variables most likely to be affected by a halt: • excess returns • price/returns volatility • trading volume. while the behavioral findings of these three variables, post-halt, provide a mixed picture, studies that had refined the analysis had produced interesting results. indeed, the mixed evidence begs for refinement. it appears that factors driving the halt may be just as important as the halt itself in explaining the post-halt behavior of the three variables. our examination of existing literature points to at least four parameters associated with a trading halt that could throw additional light on the evidence. firstly, we will examine the reason for the halt/suspension on whether it was voluntarily or mandatorily imposed. secondly, we will take a look at the type of news or information released during the halt, whether it constitutes good, bad, or neutral news. thirdly, we will also take into account the duration of the halt. lastly, the fourth parameter would be the frequency of the stock in question, whether the halt constitutes a first (single) halt or one of several(multiple) halts. while earlier studies had focused on the impact of the halt/suspension alone, later studies had brought to bear at least one or more of the above parameters. 1.1 motivation trading halts by definition are ‘disruptive’ events that are almost always associated with events that have the potential to cause abrupt or extreme moves. while marketwide, trading halts are indeed rare and unpredictable events2, whereas firm specific trading halts are not. indeed, as emerging markets have expanded their market capitalization, firm specific halts have been more frequent. there have been more than 300 trading suspensions over the last five-years at bursa malaysia. despite the fact that trading halts appear to be a fairly common occurrence, we are unaware of any systematic studies of the phenomenon in malaysia. this absence provides the motivation and justification for this study. as is the case with other markets, trading halts or temporary suspensions of listed stocks in bursa malaysia may be voluntary or mandatorily imposed. the administrative framework for temporary suspensions is outlined in chapter 16 of bursa malaysia’s listing guidelines. while a listed issuer through a request to the exchange initiates voluntary suspensions, it remains the exchange’s discretion to grant such a suspension. mandatory suspensions can be initiated either by the 2 such as the trading halt on the nyse following the september 11 attacks. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 127 exchange or the securities commission (sc). the sc can notify the exchange to suspend a stock from trading if it deems the issuer to be in breach or failure in compliance with the securities industry act 1983, the securities commission act 1993, or when the commission feels that “it is necessary or expedient in the public interest or where necessary for the protection of investors”. section 16.02 of the listing guideline provides the exchange the right to suspend at any time the trading of any listed security if the issuing company is undergoing a substantial corporate exercise, capital restructuring, stock conversion, or exercise in the event of any breach of the listing requirements. additionally; the exchange may also mandatorily suspend trading of a stock if in its opinion, “it is necessary or expedient in the interest of maintaining an orderly and fair market in securities traded on the exchange”. as is evident, both voluntary and mandatory suspensions can be the result of a wide range of reasons. while authorities obviously have broad powers to halt trading, companies as well can request the voluntary suspension of their stock for any number of reasons. in this paper, we examine a total of 291 firm-specific trading halts that occurred on bursa malaysia over the five-year period from 2000 to 2004. the paper is divided into four sections. section 2 provides a review of the literature, where as section 3 lays out our research questions, describing the data, research, design, and methodology implemented. section 4 presents the analysis of our results, while section 5 concludes the paper with our evaluation of the efficacy of trading halts in the malaysian context, along with providing implications for policy. 2. literature review much of the research interest in trading halts appears to have originated from the seminal work of hopewell and schwartz (1978). in that study of several hundred firm specific halts on the nyse, they had reported three key findings: 1. there exists a permanent price adjustment in response to new information released 2. there exists an anticipatory price behavior consistent with insider trading and information leakage. 3. post-suspension price behavior presents little, if any opportunity for systematic trading profits. additionally, they had also reported that suspensions of longer duration typically result in price adjustments of greater magnitude. most other u.s. based studies on trading halts have also examined their impact on trading volume and volatility. at least three such studies document significant increases in both volume and volatility, post-halt. lee, ready, & seguin (1994) had compared the impact of trading halts versus ‘pseudo-halts’, which are non-halt control periods on volume and volatility. they had concluded that rather than reducing both volume and volatility, the number of trading halts had increased. for the first, full day, post-halt, they had found trading volume to be 230% higher with volatility between 50 115% higher compared to pseudo-halts. further, the persistently high volume for days +2 and +3 does not seem to fit the “learning through trading model”. 128 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 these results are consistent with the earlier findings of ferris, kumar, and wolfe (1992), who had examined the impact of mandatory sec ordered trading halts. they had found that volume and volatility tended to be higher than normal in the pre-suspension period with the trend continuing in the immediate postsuspension period. they had also reported a permanent devaluation of stocks during the suspension with the extent of the devaluation dependent on the announced reason for the halt. examining trading halts on the nasdaq, christie, corwin, and harris (2002) had reported significantly higher volatility, volume, and bid-ask spreads in the period following halts. comparing their results with those based on the nyse and others, they had found trading halts to have important effects, independent of the market structure and the specific halt mechanism used. yet, other u.s. studies (while reporting similar findings with regards to price, volatility, and volume) have examined additional parameters. in addition to the argument that suspensions are almost always ‘bad news’,3 howe and schlarbaum (1986) had shown the existence of a correlation between the length of suspension (measured in trading days) and cumulative, abnormal returns. longer suspensions coincide with bigger, negative residuals. chen, chen, and valerios (2003) had examined the intraday data for 1992 of nyse stocks. their findings had showed that the type and significance of news determines the benefit from the trading halt. they had argued that halts can be beneficial in light of significant news coming to fruition. however, when a halt is called, pending a news release of little significance, the halt actually injects more noise into prices, undermining price discovery. a number of papers have examined trading halts in other countries. in analyzing stock specific suspensions on the london stock exchange, kabir (1994) had pointed towards anticipatory behavior, pre-halt, arguing that the presence of significantly positive abnormal returns up to a month following the reinstatement of trading had implied one of two things: • the complete impact of new information release takes place gradually • not all-relevant information is disclosed during the halt. wu (1998) had also examined the issue of information dissemination during halts for the hong kong market. he had found that mandatory suspension shows more effectiveness in disseminating information than voluntary suspensions. his findings about price reaction, volatility, and volume are largely in sync with that of u.s. studies. tan and yeo (2003) had studied the impact surrounding the type of news on voluntary suspensions in singapore. grouping firm-initiated suspensions into ‘favorable’ and ‘unfavorable’ news, they had found that while the first group showed significant, positive, abnormal returns around the event date, the latter group had suffered a prolonged decline. they had also pointed out that the much higher postsuspension volatility of returns implies the rationale behind voluntary suspensions to be a release of price sensitive info rather than to curb existing volatility. at least one study of the italian market (borsa italiana) and another of the portuguese market had produced results in conformity with findings elsewhere that 3 they had found that 80% of suspended securities have suffered a substantial devaluation. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 129 volume and volatility are higher, post-halt. generalizing across studies conducted in different markets, the broad conformity of results implies that market microstructures do not matter. what appears to matter is the type of information released, duration of the halt, and whether they are voluntarily or mandatorily imposed. 3. methodology and data 3.1 research design in line with previous works cited in section 2, we begin by examining the variables commonly impacted by trading halts (i.e., stock price reaction, volatility of returns, and trading volume). since the leakage of information has a direct impact on the efficacy of halts, we also explore for the evidence of such a leakage. in addition to these four factors, as the first study on trading halts in the malaysian context, we seek to add to the comprehensiveness of the study by including four, related dimensions: 1. type of halt (whether voluntary or mandatory) 2. category of news released (whether good, bad, or neutral) 3. duration of halt 4. frequency (whether the halt constitutes a single halt for the stock or is one of several). previous studies have included one or more of these four dimensions when examining price, volume, and volatility. results have shown that these dimensions could have a material effect on the overall effectiveness of trading halts. thus, we seek to examine each of the four standard variables of price reaction, volatility, volume, and information leakage within each of the four dimensions. for example, we seek to determine whether or not the price reaction, volatility, volume, and extent of information leakage are any different for voluntary versus mandatory halts, or when the news released is of a different type. given this objective, we frame the following 5 broad research questions: 1. what is the overall impact of trading halts on stock returns, volatility, and volume? 2. how different are these results when the halt is voluntary as opposed to mandatory? 3. what difference if any, does the type of news released make? 4. does the duration of the halt/suspension have any influence? 5. are the results of single halts any different from those of multiple halts? differentiating between voluntary and mandatory halts is fairly straightforward. where the issuing firm had requested a halt, it is the former; where either bursa malaysia or the sc imposed the halt, it is a mandatory halt. unlike ferris et. al., (1992) or tan and yeo (2003), who had used the positive/ negative daily abnormal returns prior to a halt to classify their sample in the ‘good’ or ‘bad’ news category, we examine the news released during the halt4. a judgment is then made on the category of news. most were fairly straightforward. 4 or news released just after the announcement of the halt. 130 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 announcements reporting increased earnings, profits, higher dividends, the winning of new contracts, etc., were categorized as ‘good’ news. the opposite would constitute ‘bad’ news. where the news was deemed to be neither, for example, unchanged earnings, etc., or where we were unsure, we classified it in the ‘neutral’ news category. during the duration of the halt, we had three classifications. a halt is classified to be: 1. ‘short’ if the halt is for one trading day5 2. ‘medium’ if the halt has a duration between 2 to 5 trading days. 3. ‘long’ if the halt has a duration longer than 5 trading days. finally, in determining whether a halt belongs to the single or multiple suspension category, we used a one-year cut-off. if a stock was suspended more than once within a one-year period, we classified it in the multiple suspension category. our dataset consists of 291 trading halts that occurred on bursa malaysia over the five-year period from 2000 to 2004. of the 291 total trading halts, the issuing firms had voluntarily requested 263 halts, whereas 28 trading halts were mandatory halts imposed by either bursa malaysia or the sc. the basis of our analysis is the daily closing price data of each sample for a 120-trading-day period around the announcement date of the halt. that is 60 days immediately prior to the announcement date and 60 days following the resumption of trading. (all daily price data were sourced from bloomberg.) others, such as information on announcements and firm specific information, were from bursa malaysia publications and newspapers. 3.2 methodology in line with almost all-previous work in this area, we used an event-study framework. thus, in examining the price reaction to an announcement of a trading halt, we had computed the abnormal returns (ar) and cumulative abnormal returns (car) for the two 60-day period windows for each firm. where necessary, we also examined the smaller windows within the 60-day windows. the daily car is computed as: (1) where, the daily abnormal return on day t for stock i is determined as: ar i,t = ar i,t r i,t (2) the abnormal return ar i,t is the difference of day t’s actual return of r i,t less the expected return of r i,t where: (3) rm equals the returns of the klse ci (kuala lumpur stock exchange composite index). 5 the shortest duration of halts in malaysia is one day. ˆ ˆ ∑ = == t t titi niiarcar 1 ,, .,......... tiiti rmbr ^^ , ^ += α the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 131 beta was estimated using daily stock and market returns for the 60-day period: -61 to -120 pre-halt announcement. next, we computed the daily mean abnormal return (that is, the average abnormal return across all samples on day t). the daily mean abnormal return (mar) is: (4) and variance (5) in addition to computing the daily car s for each of our sample companies, we computed the mean overall car s for all sample companies for each window period. the mean cumulative abnormal return (mcar) is determined as: (6) in determining the daily returns volatility for each window period, we first determined the volatility of returns across all sample firms by day (t). this is computed as: where τ it is the % return for firm i, on day t, cp it is the closing price for stock of firm i on day t. (7) for trading volume, we used the absolute number of stocks traded for each sample firm on day t. (8) where tv t is the mean trading volume across all sample firms on day t, tv i,t is the trading volume of stock for firm i on day t. tt n ar mar n ti ti t ...........,1 , == ∑ = tt n arvar marvar n ti ti t ...........,1 )( )( 2 , == ∑ = tt n car n car mcar n i ti n ti ti t ...........,1 )( 2 1 ,, == ∑∑ == ( )[ ] 100re 1,1,,, xcpcpcp titititi −−−=τ ( ) ( ) 2 ,re re n tvar tvar n ti ti t ∑ = = ntvtv n ti tit ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ = ∑ = , 132 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 while cars and abnormal returns are used to determine the impact of trading halts on price behavior, we examined the impact on returns volatility and trading volume. firstly, we employed the standard means test using the t-statistics and the non-parametric wilcoxon signed rank test (the latter test assumes the distribution is unknown or non-normal). additionally, the wilcoxon test uses the median to avoid test misspecifications that may arise from asymmetry in cross-sectional tabulations and non normal distribution. in comparing volatility, preand posthalt, we compared the variance between the samples using the f-test instead of the t-test. previous studies have shown that the results drawn from the event study can be sensitive to sample size and estimation period6. to avoid any bias that may arise from the selection of a preand post-event period, we examined the results by varying the window periods and by dividing the sample into those additional dimensions mentioned above (categorization by type of halt, type of news, duration, etc.). 3.3 testable hypotheses a number of testable hypotheses about the impact of halts on price, volatility, and volume may be inferred from previous literature. (i) price effect: a trading halt can either enhance price discovery by providing participants a “time-out” to assimilate impending new information as its proponents argue, or interfere with price discovery as its opponents point out. either way, more important than the halt is the information released during the halt. a company calls for a voluntary trading halt of its shares in order to release new information. in the case of a mandatorily required halt, even if no information is forthcoming immediately following the halt, the mere fact that authorities had stopped trading in the stock is a signal that there is a substantive problem. thus, in the absence of information leakage, the impact of trading halts on price behavior would be in one of the following three forms: 1. if information released is insignificant, there should be no abnormal returns on trading resumption. 2. if the information released is significant, a price reaction should occur (the kind of reaction being dependent on the type of news). 3. if the market had anticipated the information released during the halt, then we should expect ‘price continuation’; i.e. prices continue to move in the same direction. on the other hand, if the news released was unanticipated, then a price reversal could be the case. thus, with these three hypotheses, we test the following: 1. whether trading halts are significant events where price behavior is concerned 2. whether post-halt price behavior is dependent on the type of news 3. whether information released during halts merely reinforce anticipations or otherwise. 6 coutts, mills, and roberts (1994), cited in wu (1998) the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 133 (ii) effect on returns volatility: if trading halts require a time-out for participants to properly evaluate information during real or perceived market disequilibrium, then one should expect lower returns volatility, post-halt. alternatively, going by the logic of “learning through trading models”, the absence of transactions should exacerbate the uncertainty/disequilibria, thereby causing an increase in returns volatility, post-halt. (iii) effect on trading volume: there are two reasons why halts can be expected to have an impact on traded volume. firstly, the release of new information may require market participants to adjust their positions/exposure in the stock. volume increases, post-halt, as the stock changes hands. secondly, if one assumes trading in normal times to be evenly distributed over time, then it can be expected that when trading resumes following a halt, volume ought to be higher in order to compensate for the disruption. this also implies that the longer the trading halt, the greater the impact would be on volume. 4. results and analysis given the numerous permutations involved in our analysis, for ease of elucidation, we presented our results by the four variables of interest: • price • returns volatility • volume • information leakage. following an overall examination of the impact of trading halts on each of these variables, we then presented the results of our analysis examining each variable by the four dimensions: • type of halt • type of news released • duration of halt • frequency. the breakdown of sample size within each of these subcategories is shown in table 6. table 6: breakdown of voluntary halts sample by category note: the total number of voluntary halts in our sample was 263. the mandatory halts which had a sample size of 28 firms was not subdivided by category since these are irrelevant. mandatorily ordered halts are by definition ‘bad’ news, are of long duration and are only subjected to a single halt. type of news released duration of halt frequency of halt good news 120 short 162 single 106 neutral news 120 medium 92 multiple 157 bad news 23 long 9 total 263 total 263 total 263 134 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 4.1 price effect table 1 and figure 1 show the results of our analysis of daily cumulative abnormal returns (cars) for the 120-day period surrounding the halt announcement. table 1 shows the mean cars (mcars) for 3 pairs of window periods, 60 days pre/ post; 30 days pre/post; and 5 days pre/post announcement. the results of the paired samples t-test and the non-parametric wilcoxon signed ranks test for difference in means. the probability levels are also shown. overall, for the full sample of 291 firm-specific halts, mcars are higher for all 3 windows in the post-halt period. in other words, there is a positive price reaction once a halt is lifted. there is a big jump in the mcar for the +5 day window. this is followed by a steady rise in the two subsequent window periods +30 and +60. thus, prices are much higher 60 days after the resumption of trading relative to where they were 60 days before the announcement of halt. both statistical tests (t and wilcoxon) showed post-halt mcars to be significantly higher relative to their pre-halt windows at the 5% level. interestingly enough, when observing the mcars for the 3 pre-halt periods, we saw a steady increase as we approached announcement date. pre-halt mcar is at it’s maximum for the -5 day window. this steady buildup in prices is also clearly evident in figure 1 that plots daily cars. figure 1: daily cars for overall sample; voluntary and mandatory halts 4.2 price effect by type of halts the subsequent two columns show the results when the overall sample is separated into voluntary and mandatory halts. recall that our sample constituted of 263 voluntary and 28 mandatory halts. looking at the results for the voluntary halts, we see results very similar to the overall. all 3 post-halt mcars are higher and significant for the 60 and 30-day windows. we also see the marked increase in mcar just prior to the announcement. the mcar for the -5 day window is more than 5 times larger than that of the -60 day window. the results for voluntary halts are very similar to that of the overall sample. that should not be surprising given that the large majority (almost 90%) of our sample constituted of voluntary halts. -5 0 5 10 15 20 25 30 35 -80 -60 -40 -20 0 20 40 60 80 days c a r s overall voluntary mandatory t h e in tern a tio n a l jo u rn a l o f b a n kin g a n d f in a n ce, 2007/08 v ol. 5. n um ber 2: 2008: 125-148 135 table 1: cumulative abnormal returns by category window period price effect by news by duration frequency overall voluntary mandatory good neutral bad short med long single multiple (-60 to -1) 1.0770 .9846 1.9456 2.3463 2.6350 -1.12029 2.0517 .3981 -12.4045 .9547 1.1184 (+1 to +60) 8.8800 8.0651 16.5343 14.8738 2.7264 .4066 10.5503 4.2919 2.3364 11.4132 7.2065 (-30 to -1) 2.0115 1.9196 2.8749 3.7162 .6580 -.8714 3.6891 .5252 -15.7090 1.7727 2.2686 (+1 to +30) 7.3775 6.815 12.6598 12.8591 2.7632 -3.5775 9.3678 3.6907 -6.6638 10.5240 5.4042 (-5 to -1) 4.8120 5.143 1.6954 6.1787 4.8091 1.4904 7.6503 2.771 -15.4385 5.7903 4.5031 (+1 to +5) 6.7212 6.253 11.112 10.4326 3.5251 -1.3127 9.1165 3.6264 -17.3855 9.3882 5.7221 t-stat -62.680 (000) -50.949 (.000) -16.762 (.000) -81.080 (.000) -9.231 (.000) -2.544 (.014) -43.114 (.000) -13.947 (.000) -7.275 (.000) -50.665 (.000) -42.387 (.000) -25.200 (000) -18.766 (.000) -15.507 (.000) -54.777 (.000) -4.294 (.000) 4.708 (.000) -16.533 (.000) -7.130 (.000) -5.405 (.000) -25.960 (.000) -10.523 (.000) -4.187 (014) -2.656 (.057) -7.737 (.002) -8.516 (.001) 2.626 (.058) 6.647 (.003) -4.111 (.015) -1.256 (.277) .667 (.553) -6.308 (.003) -4.940 (.008) wilcoxon z-stat -6.736 (.000) -6.736 (.000) -6.736 (.000) -6.736 (.000) -5.904 (.000) -2.245 (.025) -6.736 (.000) -6.729 (.000) -5.382 (.000) -6.736 (.000) -6.736 (.000) -4.782 (.000) -4.782 (.000) -4.782 (.000) -4.782 (.000) -3.466 (.001) -3.651 (.000) -4.782 (.000) -4.391 (.000) -4.184 (.000) -4.782 (.000) -4.782 (.000) -2.023 (.043) -1.753 (.080) -2.023 (.043) -2.023 (.043) -1.753 (.080) -2.023 (.043) -2.023 (.043) -1.214 (.225) -.730 (.465) -2.023 (.043) -2.023 (.043) the table shows the mean cars by window period for the different categories. t-stat values shown are for paired sample test the prob. values are shown below in brackets. 136 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 analysis of mandatory halts produced some very interesting results. a first glance at figure 1 shows two things. firstly, relative to voluntary halts, daily cars in the post halt period is much higher. secondly, they are also much more volatile in both the preand post-halt period relative to the voluntary sample. going by mcar numbers in table 1, with the exception of the -5 day window, all mcars are higher for mandatory halts relative to voluntary ones. paired tests that compare the means across the two sample (as shown in table 2) confirm this. not only are mcars higher after the halt, but also, they are higher by several fold relative to pre-halt windows. daily cars exceed 30% as at day +60 (figure 1). these results are contrary to expectations. mandatory halts, being suspensions imposed by authorities, are a negative signal. it implies a breach/wrong doing or some other inadequacy on the part of the issuing firm. from this viewpoint, the huge positive cars, post-halt, are indeed a contradiction. however, when we consider the regulatory structure of mandatory halts and the survivor bias of our sample, the results are logical. malaysian authorities only initiate mandatory halts when there are serious inadequacies. in the post-1998 asian financial crisis environment, most of these inadequacies have to do with financial distress. upon having initiated a trading halt, the securities commission would require the company to come up with a restructuring plan. the restructuring plan is usually in the form of asset sales, debt restructuring, recapitalization, or some combination thereof. since trading in the stock will only be allowed to resume when the firm has come-up with a viable plan, inability to do so will mean continued suspension, followed by a possible delisting. a restructuring plan to be acceptable must obviously be one that will subsequently put the firm in a better financial position. thus, the 28 mandatory halts by definition are those that had successfully `restructured’ and were subsequently allowed to resume trading. this survivor bias and uniqueness of the malaysian regulatory requirement explains the hugely positive cars, post-halt. 4.3 price effect by type of news figure 2 shows the plot of daily cars for our sample of voluntary halts7, categorized by type of news. the difference in price reaction is obvious. the good news category shows a steady increase in prices all the way to day +60. the bad news category shows a sharp, initial decline for about 20 trading days (one month) before stabilizing and reversing the course. the neutral news category shows no distinct trend. the statistical tests in table 1, confirm this price behavior. for the good news sample, all 3 post-halt windows have significantly higher mcars. most of the positive cars happen in the first 5 days following announcement. for the neutral news category, though mcars for 60 days are marginally higher, both the t and 7 mandatory halts were not categorized by news, duration, or frequency of halt since these are irrelevant. mandatory halts are by definition ‘bad’ new; of long duration; and subject to a single halt. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 137 wilcoxon tests show price performance in the 5 days following announcement to be no different from the 5 days immediately prior. figure 2: daily cars by type of news thus, the price reaction in the neutral news category is largely muted. the mcars for bad news category is already in negative territory for the -60 and 30 day windows. however, the mcars show a positive build-up just before announcement as seen from the -5 days window. of course, all these positive buildups are erased following the announcement. the mcars are sharply negatively for the +5 and +30 day windows. the price reaction seen here is in line with expectations. what is interesting to note is the build-up of prices just before announcement. though table 1 shows this to be true for all 3 news categories, the build-up is most obvious in figure 2 for the good and neutral news categories. 4.4 price effect by duration of halts we next examined whether the duration of the halt matters for a price reaction. the results in table 1 confirmed the relevance. for short duration halts, mcars for all 3 windows are significantly higher, post-halt. we saw two obvious differences for medium term halts. the mcars, though higher, post-halt, are all lower relative to the ones we saw for short duration halts. secondly, both tests show mcars to be no different between the -5 and +5 day windows, implying that there is no significant initial price reaction when trading resumes. in sharp contrast to short and medium duration halts, long term halts show very different price behavior. mcars are already negative for all 3 windows even before the halt. when trading resumes, initially, there is a sharp, negative reaction. the stocks experience a mcars of -17% within the first week. this decline however abates with time. though the +5 and +30 day windows have negative mcars, it is marginally positive for +60 -10 -5 0 5 10 15 20 25 80 60 40 20 0 20 40 60 80 days c a r good neutral bad 138 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 days. interestingly enough, these results are similar to that of hopewell & schwartz (1978) who had found price adjustments of greater magnitude for halts of longer duration. table 2: cross comparison across sample by window period category mean t-stat wilcoxon zstatby type of halt voluntary vs. mandatory (-60 to -1) -.9610 -2.790 (007) -2.665 (.008) (-1 to -30) -.95533 -1.624 (.115) -1.697 (.090) (-1 to -5) 3.4483 12.128 (.000) -2.023 (.043) (1 +60) -8.4692 -12.638 (.000) -6.729 (.000) (1 + 30) -5.8446 -19.558 (.000) -4.782 (.000) (1 to 5) -4.8587 -5.427 (.006) -2.023 (.043) by news category good news vs. bad news (-60 to -1) 3.4666 15.110 (.000) -6.736 (.000) (-1 to -30) 4.5877 15.158 (.000) -4.782 (.000) (-1 to -5) 4.6882 9.025 (.001) -2.023 (.043) (1 +60) 14.4671 29.252 (.000) -6.736 (.000) (1 + 30) 16.4367 35.555 (.000) -4.782 (.000) (1 to 5) 11.7453 23.296 (.000) -2.023 (.043) by frequency of halt single vs multiple (-60 to -1) -.1636 -.997 (.323) -1.097 (.273) (-1 to -30) -.4958 -1.826 (.078) -2.005 (.045) (-1 to -5) 1.2871 1.917 (.128) -2.023 (.043) (1 +60) 4.2067 18.022 (.000) -6.736 (.000) (1 + 30) 5.1198 19.793 (.000) -4.782 (.000) (1 to 5) 3.6661 9.908 (.001) -2.023 (.043) by duration of halt short vs long (+1 to +60) 8.6287 6.352 (.000) -5.212 (.000) (+1 to +30) 16.5613 10.575 (.000) -4.782 (.000) (+1 to +5) 27.5355 11.484 (.000) -2.023 (.043) the table shows the mean cars by window period for the different categories. t-stat values shown are for paired sample test the prob. values are shown below in brackets. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 139 the above results, together with those in table 2 comparing mcars across categories, imply a link between duration of halt and price behavior, post-halt. generally, it appears from our results that shorter duration halts experience positive price reactions, post-halt, whereas long duration halts experience the opposite. these results are sensible when we consider the fact that all halts are disruptive, with longer term ones even more so. even if the halt was voluntarily requested, issuing companies will only want longer halts if they have more complex issues to solve. complicated problems will require a longer time-out. the negative mcars we saw for all three pre-halt windows would imply that may indeed be the case. companies needing long-duration halts already experience problems; thus they need time to sort these out. our final analysis with regards to price effect was to see if the frequency of halts had any different price behavior. the results in table 1 show similar price reactions, post-halt. when the mcars are compared between the two categories (table 2) for matched window periods, we see no difference, pre-halt, but when trading resumes, single halts outperform multiple halts. this out-performance is significant at the 5% level by both the parametric and non-parametric tests. 4.5 evidence of information leakage scrutinizing the mcars in table 1 shows an interesting feature. for almost all categories of analysis, we see a marked increase in mcars for the -5 day window. such a pattern is also clearly visible from the daily cars plotted in figures 1 & 2. this appears to be a tentative evidence of information leakage. the presence of such leakage has been documented for several markets in the previous studies cited in section 2. to seek confirmation of such leakages for bursa malaysia, we examined the mcars for any possible significant differences across two different window periods, 20 days before halt. the first is the 10-day pre-halt window (-20 to -11) and the final 10 days (-10 to -1). essentially, we want to see if there is a significant price change in the last 2 weeks leading to the halt, relative to the 2 weeks prior. the results are shown in table 3. at the 5% level, both the parametric and non-parametric tests show consistent results. for the overall sample of 291 companies, mcars for the last 10 days was indeed significantly higher than the 10 previous days. this initial evidence of leakage was reinforced when we examined the voluntary and mandatory subsegments. voluntary halts have mcars almost 3 times higher in the final 10 days than the previous 10-day window. mandatory halts on the other hand, have mcars more than 3 times lower in the last 10 days. the differences are statically significant for both cases. when the same 10-day windows were examined across the different type of news categories, mcars for the latter window period was significantly higher in all cases, even for the bad news category. what is interesting is that when we go from good to neutral to bad news categories, the mcars steadily reduce. finally, both the single and multiple suspension categories had significantly higher mcars for the last 10-day window8. 8 we did not evaluate the duration of the halt category since duration of halt cannot be known prior to or even at announcement. 140 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 table 3: test for evidence of information leakage by category of halt mean t-stat wilcoxon z-stat overall – voluntary & mandatory (-20 to -11) 1.6824 -6.333 (.000) -2.803 (.005) (-10 to -1) 3.8032 voluntary –suspension (-20 to -11) 1.4912 -8.446 (.000) -2.803 (.005) (-10 to -1) 4.0933 mandatory (-20 to -11) 3.477 3.565 (.006) -2.497 (.013) (-10 to -1) 1.0783 voluntary by news category good news (-20 to -11) 3.748 -4.179 (.002) -2.803 (.005) (-10 to -1) 5.259 neutral (-20 to -11) -.3488 -7.527 (.000) -2.803 (.005) (-10 to -1) 3.410 bad (-20 to -11) -.6825 -3.184 (-.011) -2.599 (.009) (-10 to -1) 1.574 by frequency of halt multiple (-20 to -11) 2.6763 -4.802 (.001) -2.803 (.005) (-10 to -1) 3.6217 single (-20 to -11) .7280 -8.891 (.000) -2.803 (.005) (-10 to -1) 4.636 the table shows the mean cars by window period for the different categories. t-stat values shown are for paired sample test the prob. values are shown below in brackets. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 141 summarizing the results, with the exception of mandatory halts that saw a marked price decline immediately prior to announcement, all voluntary halts (regardless of subcategory) had experienced a significant price build-up just prior to halt. there are two reasons why we believe insider trading is the cause of the sharp rise in prices. firstly, when we consider voluntary halts, even if the market can anticipate the release of good news, it is difficult for outsiders to know when a company will ask for a voluntary suspension. only those with inside information can tell the timing of a trading halt request. this can explain the very significant build-up in mcars/prices just prior to halt announcements that are then followed by the release of good news. secondly, going by the same logic, even if a firm’s financial distress is known, the timing of a mandatory halt is difficult to gauge for outsiders. yet, the fact that mcars are significantly negative just before an official announcement of halt, it appears to be the work of those trading on privileged information. while we believe our analysis provides sufficient evidence of information leakage, we are left questioning the unexplained, positive price build-up even for the bad news category9. 4.6 effect on trading volume lee et. al., (1994), ferris et. al. (1992), and christie et. al. (2002) had all shown similar findings with regards to volume and volatility. all three studies had shown a higher than normal volume and volatility in the pre-suspension period with the trend continuing in the immediate post-suspension period. where volume is concerned, our results appear to be very much inline with these u.s. based studies. figure 3 shows the mean daily volume for our overall sample and voluntary/mandatory categories for the 120-day period surrounding halt announcements. there is a clear build-up in daily volume in the period immediately before the halt announcement. the uptrend continues in the period immediately following trading resumption. in all three cases, the rise in traded volume is short lived. it peaks at about 5 days after resumption before sliding steadily back to normal levels. in fact, all the action appears concentrated between days -20 and +20. this is identical to ferris et. al. (1992) who had reported that volume levels returned to normal 20 days after suspension. results of our statistical tests for volume are shown in table 4. for the overall sample, mean volumes are higher for all 3 post-halt windows; however, the significance tests are mixed. for the sample of 263 voluntary halts, volume is significantly higher (post-halt) in the 5 and 60-day windows. this is in stark contrast to the sample of mandatory halts. volume, though higher for the +5 day window, is lower for both +30 and +60 day windows. both the t-test and wilcoxon show a significantly lower volume for post 60 days relative to 60 days, pre-halt. when we examined volume patterns by news category, though all three categories show higher volume for the +5 day window, in contrast to good and neutral news, the bad news category showed a significantly lower volume for the +60 day window. 9 mcars for the bad news category is negative for the first 10 day window, but turns marginally positive for the final 10 days. 142 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 figure 3: mean daily volume for overall, voluntary and mandatory halts when examining volume by duration of halt, short duration halts had no different volume in all 3 post-halt windows. halts of medium duration showed a significantly higher volume in the +5 and +30 windows. both tests confirmed this at the 5% level. this is interesting since it lends support to the argument that halts will lead to pent-up demand, and therefore, longer duration halts should see a bigger build-up in demand. indeed, this is true when we go from short to medium duration halts. however, our results for long duration halts are not consistent with this argument. volume is in fact significantly lower for the +5 and +60 day windows. we believe this could be due to the same argument we had made in explaining the negative cars for long duration halts. implying that a long duration halt was needed for a stock signifies more complicated problems; hence, the negative cars and significantly lower trading volumes. 4.7 impact on volatility our analysis of the impact of halts on returns volatility showed interesting results. recall that we measure volatility as the variance of daily returns. table 5 shows the results of our f-test and the non-parametric, wilcoxon test on the variance of daily returns. figures 4 and 5 show the plot of daily returns over the 120-day period. turning to table 5, variance of returns for the overall sample is significantly higher for all 3 post-halt windows. post-halt volatility increases as the window period is lengthened. this overall picture changes drastically when we decompose the sample by category. there is a marked contrast in the post-halt behavior of returns volatility between mandatory and voluntary halts. the mandatory sample has a much higher volatility relative to the voluntary sample, even before halts. this volatility increases very substantially after trading resumes for the mandatory sample. as is obvious from the table, post-halt volatility is sharply higher in the +5 day window than abates, 0 500 000 1000000 1500000 2000000 2500000 3000000 -80 -60 -40 -20 0 20 40 60 80 days v o lu m e overa ll voluntary mandatory t h e in tern a tio n a l jo u rn a l o f b a n kin g a n d f in a n ce, 2007/08 v ol. 5. n um ber 2: 2008: 125-148 143 table 4: mean daily volume window period price effect by news by duration frequency overall voluntary mandatory good neutral bad short med long single multiple (-60 to -1) 1286655.67 1356603.54 629645.35 822111.63 941522.34 1088635.79 854786.59 1023620.43 254520.92 689692.78 1187643.23 (+1 to +60) 2185269.80 2362928.92 516543.12 1997982.33 994148.79 770226.68 1644844.24 1141272.14 328407.22 774849.27 2399114.35 (-30 to -1) 1364379.15 1439878.43 655225.23 955564.17 900850.98 1082193.33 935325.93 985441.69 294677.40 684730.43 1289971.62 (+1 to +30) 2807778.98 3048612.53 545663.86 2685852.84 1115735.89 903458.55 2160076.39 1326152.05 273102.69 900419.21 3162613.77 (-5 to -1) 1492847.28 1557766.61 883069.28 953688.30 1236462.93 1250656.52 1038663.19 1292162.81 350597.77 1005471.20 1244040.27 (+1 to +5) 2141049.20 2232657.49 1280585.71 1376211.82 1770799.50 1219892.17 1376092.77 1966888.47 150451.11 1395489.80 1738650.14 t-stat -2.190 (.032) -1.830 (.078) -3.741 (.020) -2.216 (.031) -1.843 (.076) -4.142 (.014) 3.397 (.001) 1.218 (.233) -1.388 (.238) -1.609 (.113) -1.257 (.219) -3.440 (.026) -.939 (.351) -3.202 (.003) -3.587 (.023) 4.355 (.000) 1.758 (.089) 153 (.886) -1.468 (.147) -1.203 (.239) -1.823 (.142) -1.938 (.057) -4.383 (.000) -5.730 (.005) -3.102 (.003) .637 (.529) 3.392 (.027) -1.955 (.055) -3.222 (.003) -5.284 (.006) -1.453 (.151) -1.185 (.246) -2.063 (.108) wilcoxon z-stat -1.708 (.088) -2.396 (.017) -2.023 (.043) -1.855 (.064) -2.581 (.010) -2.023 (.043) -3.401 (.001) -2.252 (.024) -1.214 (.225) -4.667 (.000) -3.116 (.002) -2.023 (.043) -.199 (.842) -2.684 (.007) -2.023 (.043) -4.086 (.000) -1.944 (.052) -.135 (.813) -1.701 (.089) -2.417 (.016) -1.214 (.225) -1.362 (.173) -3.363 (.001) -2.023 (.043) -2.739 (.006) -.812 (.417) -.2.023 (.043) -2.010 (.044) -2.787 (.005) -2.023 (.043) -1.494 (.135) -2.910 (.004) -1.753 (.080) the table shows mean daily volume by window period for the different categories. t-stat values shown are for paired sample test the prob. values are shown below in brackets. 14 4 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 7/ 08 v ol . 5 . n um be r 2: 2 00 8: 1 25 -1 48 table 5: volatility of daily returns window period price effect by news by duration frequency overall voluntary mandatory good neutral bad short med long single multiple (-60 to -1) 2.554705 0.001254 26.5389 0.001138 0.001404 0.001767 0.00138 0.000976 0.001824 0.001354135 0.001106 (+1 to +60) 88.40587 0.001234 918.7779 0.000976 0.000917 0.001782 0.00102 0.000876 0.001212 0.001099934 0.000793 (-30 to -1) 3.472286 0.001357 36.07422 0.001295 0.00135 0.001897 0.00143 0.00106 0.003081 0.001350367 0.001368 (+1 to +30) 3.142628 0.000842 32.65298 0.000758 0.000844 0.001957 0.000958 0.000693 0.000256 0.00096044 0.000665 (-5 to -1) 4.707086 0.00212 48.90016 0.001632 0.002854 0.001365 0.002922 0.000976 0.000256 0.001648783 0.002818 (+1 to +5) 7.420658 0.00049 77.11724 0.000463 0.000455 0.000797 0.00052 0.000462 0.000243 0.000502735 0.000471 f-stat 9.08e-05 (0) 1.225924 (0.041689) 0.34929 (0) 0.23357 (0) 2.938413 (8.18e-18) 169.7847 (3e-216) 6.34e-05 (0) 1.227714 (0.298876) 0.345276 (0.003712) 1.440614 (0.023763) 2.258433 (6.07e-06) 36.14764 (1.96e-60) 2.844534 (1.29e-08) 3.175171 (4.51e-10) 477.5973 (7.5e-126) 0.62466 (0.138705) 0.69845 (0.203276) 2.03542 (0.051991) 1.713403 (0.000349 ) 1.938854 (1.61e-05) 217.5889 (2.9e-142) 2.628701 (3.2e-06) 11.01392 (1.77e-25) 5.811334 (1.29e-15) 1.849358 (0.201425) 1235.891 (1.49e-11) 0.92031 (0.454689) 1.941295716 (2.07403e05) 1.82645162 (9.64929e-05) 35.83203279 (2.45595e-78) 1.683927 (0.004039) 5.679062 (2.25e-17) 377.5843 (6.1e-106) wilcoxon z-stat -1.558 (.119) -3.650 (.000) -5.200 (.000) -2.307 (.021) -3380 (.001) -5.518 (.000) -1.070 (.285) -1.298 (1.94) -.319 (.750) -1.563 (.118) -2.208 (.027) -3.418 (.001) -2.092 (.036) -2.092 (.036) -4.254 (.000) -.274 (.784) -.791 (.429) -1.460 (.144) -3.262 (.001) -3.173 (.002) -4.264 (.000) -.343 (.732) -.857 (.392) -3.345 (.001) -.296 (.767) -2.073 (.038) -1.007 (.314) -1.049 (.294) -2.102 (.036) -4.501 (.000) -2.463 (.014) -2.703 (.007) -3.166 (.002) the tables shows average volatility of daily return by window period for the different categories. t-stat values shown are for paired sample test the prob. values are shown below in brackets. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 145 and is marginally lower for the +30 day window. for the 60-day window, post-halt volatility is several times higher. while one would expect higher volatility as the sample period is lengthened, the increase in variance is very high for the 60-day windows. 10 in stark contrast to these results, the sample of voluntary halts showed a significantly lower volatility, post-halt. volatility falls substantially in the +5 day window before rising steadily as the window period lengthens. in fact, as opposed to the mandatory sample, volatility is several fold lower for the 5 and 30-day windows, post-halt. figure 4, which shows a scaled plot of daily returns, captures this vast difference in volatility behavior between the two samples11. when we tried to explore the reasons for the very high volatility of the mandatory sample, we came up with 4 possible explanations. firstly, stocks of companies subjected to mandatory halts (being troubled companies to begin with), had been severely beaten down from their original ipo and par values. only about half of the sample was selling for above one ringgit, a typical par value. many selling below 50 sen were essentially “penny stocks”. so, one key reason for the high volatility is the very low traded price. small, absolute prices lead to large variance. figure 4: volatility of daily returns (scaled), voluntary vs mandatory secondly, mandatory halts experienced much longer suspensions. the average length of trading suspension for our mandatory sample was 42 days. howe and schlarboum (1986) had shown longer suspensions to coincide with bigger, negative residuals. thirdly, the low liquidity of these stocks might have been a factor, being that low liquidity tends to go hand-in-hand with high volatility. lastly, probably the 10 this appears to be the instance where the parametric and non-parametric tests produced inconsistent results. 11 daily returns of the mandatory sample were scaled by 10. -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 -80 -60 -40 -20 0 20 40 60 80 days v o la ti li ty voluntary mandatory 146 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 most important reason for the very high post-halt volatility is information released during suspension. as mentioned earlier, companies subjected to mandatory halts have to come up with viable restructuring plans before trading is allowed to be resumed. we believe the announcement of these plans, which are extensive by nature, leads to an increased uncertainty, especially when initially implemented. thus, the increased post-halt volatility. our results of higher volatility for mandatory halts in the post suspension period are in-line with those findings of ferris, et. al. (1992) who had examined sec ordered trading halts. there is also conformity with the results of wu (1998) who had examined the mandatory halts in hong kong. further, as reported in wu (1998), the largest change in value is typically on the 1st day of the resumption of trading. this too was consistent in our case. price variance on day 1 for our mandatory sample was 180%! in fact, price reaction was highest on day 1 for most of our categories. however, our findings of significantly lower volatility for voluntary halts are contradictory to wu (1998), who had found volatility to be higher postsuspension, even though they were lower than that of the mandatory sample. when the voluntary halts were categorized by type of news, the volatility results were in line with expectations. the good news category showed significantly lower volatility for all 3 windows, post-halt. neutral news showed similar results. the bad news category on the other hand had a volatility that was no different post/ pre-halt. both tests showed post-halt volatility to be no different at the 5% level. in fact, variance was marginally higher for the 30 and 60-day windows, post-halt. figure 5 plots changes in daily returns by news category. the bad news category obviously has higher fluctuations relative to the other two categories and higher volatility, post-halt, relative to its pre-halt volatility (a result consistent with the findings of tan and yeo [2003]). when volatility was examined by the duration of halt, both short and medium duration halts showed lower post-halt volatility. for long duration halts, both tests showed no change in volatility. single/multiple suspensions both displayed lower volatility, post-halt. figure 5: volatility of daily returns by type of news 0.03 0.02 0.01 0 0.01 0.02 0.03 0.04 80 60 40 20 0 20 40 60 80 days v o l a t il it y good neutral bad the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 147 5. conclusion most of our findings were broadly consistent and logical. trading halts are indeed significant events, with the type of halt (whether voluntary or mandatory) mattering tremendously. the type of news released during the halt is the critical determinant of how price, volume, and volatility would behave, post-halt. also, the duration of halt and frequency, whether single or multiple, appears to be largely inconsequential, while anticipatory behavior/information leakages appear to go hand-in-hand with trading halts. our results support the price efficiency hypothesis of trading halts. mandatory halts and the ‘bad news’ category had shown an increased volatility and reduced trading volume, thereby lending support to the “learning through trading argument”. with the exception of these two subsets, our overall results are consistent with the argument that trading halts help disseminate information, thus enhancing the price discovery process. trading halts had resulted in a positive price reaction, increased volume, and volatility. there was indeed a significant difference in the results of voluntary as opposed to mandatory halts. the type of news released during the halt had a huge impact on all three variables of price, volume, and volatility, post-halt. the duration of halt had an isolated impact and appears to be largely inconsequential. the frequency of halts, whether single or multiple, did not seem to matter either. comparing our results with that of previous studies, we found that for all three variables examined, there was broad conformity concerning our overall sample. however, when we refined the analysis by examining the sub categories, we found some interesting differences. we found significantly positive cars, post-halt, for our sample of mandatory halts, along with significantly lower volatility, post halt, for the voluntary sample. author statement: obiyathulla i. bacha, mohamed e. s. abdul rashid and roslily ramlee are staff members of the international islamic university malaysia. the authors acknowledge with thanks the financial support provided by the research centre at the iium. e-mail: obiya@iiu.edu.my references bhattacharya, u., and m. spiegel (1998). anatomy of a market failure; nyse trading suspensions (1974 – 1988), journal of business & economic statistics april, 216 – 226. chen, h., h. chen and n. valerios (2003) the effects of trading halts on price discovery for nyse stocks, applied economics 35: 91 – 97. christie, w.g., s.a. corwin, and j.h. harris, (2002). nasdaq trading halts: the impact of market mechanisms on prices, trading activity, and execution costs, the journal of finance lvii(3): 1443 – 1478. 148 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 125-148 duque, j., and a.r. fazenda (2003). evaluating market supervision through an overview of trading halts in the portuguese stock market, journal of financial regulation and compliance 11(4): 349 – 376. ferris, s.p., r. kumar and g.a. wolfe (1992) the effect of sec – ordered suspensions on returns, volatility and trading volume, the financial review 27(1): 1 – 34. hopewell, m.h., and a.l. schwartz (1978). temporary trading suspensions in individual nyse securities, the journal of finance 33(5): 1355 – 1373. howe, j. s., and g.g. schlarbaum (1986). sec trading suspensions: empirical evidence, journal of financial and quantitative analysis 21(3): 323 – 333. kabir, r., (1994). share price behaviour around trading suspensions on the london stock exchange, applied financial economics 24: 289 – 295. lee, m.c., m. j. ready and p.j. seguin (1994). volume, volatility, and new york stock exchange trading halts, the journal of finance xlix(1): 183 – 211. mcdonald, c.g., and d. michayluk (2003). suspicious trading halts, journal of multinational financial management 13: 251 – 263. tan, s.k., and w.y. yeo (2003). voluntary trading suspension in singapore, applied financial economics 13: 517 – 523. wu, l., (1998). market reactions to the hong kong trading suspensions: mandatory versus voluntary, journal of business finance & accounting 25(3 & 4): 419 – 437. international journal of banking and finance 3-1-2008 the efficiency of trading halts: emerging market evidence obiyathulla i. bacha mohamed eskandar s. a. rashid roslily ramlee recommended citation 1 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 https://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: charoenwong, c., ding, d. k., & pan, j. (2023). insider trading and corporate spinoffs. international journal of banking and finance, 18(2), 1-28. https://doi. org/10.32890/ ijbf2023.18.2.1 insider trading and corporate spinoffs 1charlie charoenwong, 2david k. ding & 3jing pan 1&3nanyang business school nanyang technological university, singapore 2lee kong chian school of business singapore management university, singapore 2corresponding author: davidding@smu.edu.sg received: 9/8/2021 revised: 13/10/2021 accepted: 13/10/2021 published: 25/6/2023 abstract the study reported here examines insider trading and the issue of undervaluation as a motive behind corporate spinoffs. the results show an unmistakable increase (decrease) in the number of insider purchases (sales) and net purchases (sales) in the four quarters prior to a spinoff announcement. in addition, relative to a benchmark period, insider selling is significantly lower, and their net purchases significantly higher, in the three quarters prior to a spinoff announcement, as compared to other periods. furthermore, announcement period excess returns for abnormal net insider purchases are significantly higher than excess returns for abnormal net insider sales. however, only firms with abnormal net insider purchases exhibit significant improvement in their long-run market and operating performance after a spinoff. the results seem to suggest that undervaluation is an important motive behind corporate spinoffs and that it is possible to 2 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 identify the quality of a spinoff firm based on insider trading behavior prior to its announcement. keywords: insider trading, undervaluation, corporate spinoffs. jel classification: g14, g34. introduction the information-based model of nanda and narayanan (1999) implies that undervaluation is an important motive for spinoffs. if a firm is undervalued due to information asymmetry between its management and the market, it may choose to do a spinoff to get correctly valued before approaching the external capital market for funds. empirical research is consistent with this argument. for example, ahn and denis (2004) and burch and nanda (2003) report a significant improvement in aggregate firm value following spinoffs. furthermore, krishnaswami and subramaniam (1999) find lower levels of information asymmetry and larger amounts of capital raised after spinoffs. choi (2020) reports a negative relation between the abnormal tone (optimistic or pessimistic) in a spin-off prospectus and the insider trading pattern (buy or sell) in the spun-off subsidiary within three months of the spin-off. his tests show that the negative relation holds only for transactions by insiders who had also been executives in the parent company prior to the spin-off. he finds that insider purchases result in substantial long-term excess returns, especially when they are accompanied by an abnormally negative tone. the present study provides complementary evidence on the link between undervaluation and corporate spinoffs. it analyzes personal trades made by corporate insiders in their firm’s stock prior to the announcement of a spinoff. since insiders have intimate knowledge of their firm’s operations, transactions that they make on their own behalf should reflect their belief regarding the prospects of the firm. if their firm is undervalued and a spinoff is undertaken to enhance market valuation, then insiders are expected to increase their share purchases, or decrease their sales, prior to the spinoff announcement. either action would result in a higher net share purchase measure for the corporate insider. in addition, a strong relation between insider trading prior to a spinoff announcement and improvements in the long-run performance of spinoff firms is expected. if only firms 3 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 with prior increases in net insider purchases exhibit a significant improvement in their long-run performance after a spinoff, but other firms do not, then insiders may tend to intentionally trade on superior information regarding the true value of their firm. otherwise, prior insider trading patterns may simply reflect a tendency for insiders to mechanically follow price increases or decreases. finally, if investors regard prior insider trading activity of a firm as a useful backdrop for evaluating its spinoff decisions, then the spinoff announcement period’s excess returns are expected to be significantly higher among firms with prior increases in net insider purchases. the purpose of the present study was to examine insider trading around corporate spinoffs. to test its hypotheses, the study first investigated insiders’ personal trading of their firm’s stock before the announcement of a spinoff. insider trading is defined as the open-market and private transactions by a firm’s top management. the sample consisted of 187 separate spinoff events by 172 parent firms from 1987 through 2006. the results showed a clear and near-monotonic increase (decrease) in the number of insider purchases (sales) and net purchases (sales) in the four quarters prior to the announcement. moreover, relative to a benchmark period, insider selling was significantly lower and the net purchases significantly higher in the three quarters prior to a spinoff announcement, as compared to other periods. as mentioned earlier, the study was aimed at investigating whether insider trading could predict spinoff announcements. the study sample was stratified based on insider trading activity one quarter prior to the announcement day. spinoffs with positive abnormal insider net purchases were classified as the “abnormal net purchases sample,” while those with zero or negative abnormal insider net purchases were classified as the “abnormal net sale sample.” consistent with the hypotheses proposed, it was found that the announcement period excess returns for the abnormal net purchases group were significantly higher than those for the abnormal net sales group, suggesting that the market viewed spinoff decisions by firms with prior abnormal net purchases as more compatible with shareholder interests. the present study also investigated the ties between insider trading activity and changes in firm performance around spinoffs. the results showed that firms with abnormal net purchases exhibited significant improvement in their long-run market and operating performance following spinoffs. the results seemed to imply that managers knew when their firms were undervalued and time their own trades accordingly. the rest of the paper is organized as follows. section 2 reviews the related research and develops the hypotheses to be tested. section 4 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 3 describes the sample selection procedure and research method. section 4 presents evidence on insider trading before the spinoff announcement. the results of the stock market performance and the operating performance around spinoffs are reported in section 5 and 6, respectively. section 7 concludes the paper. literature review and hypotheses development numerous studies, such as those by hite and owers (1983), miles and rosenfeld (1983), and schipper and smith (1983), among others, have documented a significant positive stock price reaction to the announcement of corporate spinoffs. there is considerable empirical evidence that spinoffs, on average, enhance long-run shareholder value (e.g., burch & nanda, 2003; ahn & denis, 2004). a variety of reasons has been presented in the literature to explain the value gains from spinoffs. among the most popular are improvements in corporate focus (daley et al., 1997; desai & jain, 1999), improvements in management incentives (schipper & smith, 1983), increase in investment efficiency (ahn & denis, 2004), relaxation of regulatory or tax constraints (schipper & smith, 1983), facilitation of a merger or takeover (cusatis et al., 1993), and elimination of cross subsidies (daley et al., 1997). however, as pointed out by krishnaswami and subramaniam (1999), if these motives were the only ones behind the separation of a parent from its subsidiary, then any other type of divestiture should work just as well as a spinoff. spinoffs differ from other forms of divestitures such as asset sells and equity carve-outs in that no capital is raised in spinoffs. therefore, a spinoff is an especially appropriate mode of separation when a firm is undervalued. this is because undervaluation does not affect the cash inflows to the firm since the “subsidiary is not being sold” (krishnaswami & subramaniam, 1999). nanda and narayanan (1999) develop an information-based model to explain the undervaluation of multi-divisional firms. they argue that diversified firms may be undervalued due to the information asymmetry between a firm’s management and the market. splitting the firm’s divisions into multiple business components will facilitate the market valuation of each component more accurately. it is clear then that the market value of the sum of the separated parts may be greater than the market value of the combined firm. one way to determine the undervaluation of firms that engage in spinoffs is to compare the excess value of firms prior to and following spinoffs. for example, ahn and denis (2004) 5 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 and burch and nanda (2003) report that pre-spinoff firms are valued at a discount. post-spinoff, they report a significant improvement in aggregate value and the diversification discount is eliminated. however, there is much debate about the accuracy of the compustat segment data used to measure excess value in most studies. the present study has provided complementary evidence to prior studies by measuring undervaluation with an approach that did not rely on the estimation of excess value. it analyzed personal trades made by corporate insiders in their firm’s stock prior to the announcement of a spinoff. since insiders have an intimate knowledge of their firm’s operations, transactions that insiders make on their own behalf are assumed to reflect their belief regarding the prospects of the firm. if a firm is undervalued and a spinoff is undertaken to enhance market valuation, then one can expect insiders to increase their share purchases or decrease their sales prior to a spinoff announcement. either action would result in an increase in their net share purchases. given the foregoing observations, hypothesis (1) was proposed. hypothesis 1: insider net share purchases increase prior to the announcement of a spinoff. studies dating back to the late 1960s and early 1970s, such as those by lorie and niederhoffer (1968), pratt and devere (1970), jaffe (1974), and finnerty (1976), have documented that corporate insiders earned significant abnormal profits by trading in stocks of their own firms. more recently, researchers have examined insider trading around corporate announcements of asset sales (hirschey & zaima, 1989), capital expenditures (john & mishra, 1990), corporate bankruptcy (seyhun & bradley, 1997), dividends (john & lang, 1991), earnings (penman, 1982; elliot et al., 1984), equity issues (clarke et al., 2001; kahle, 2000; lee, 1997), mergers and acquisitions (akbulut, 2005; boehmer & metter, 1997; seyhun, 1990b; song, 2005), and stock repurchases (lee et al., 1992). most studies have also reported significant changes in insider trading patterns before the public announcement. for example, lee et al. (1992) found evidence of increased buying by insiders prior to fixed price repurchase offers, and of their decreased selling prior to fixed price repurchases and repurchases that follow takeover-related events. kahle (2000) has shown that insider sales increase and purchases decrease prior to issues of equity and convertible debt by industrial firms. clarke et al. (2001) provide evidence that insider selling increases prior to both completed and canceled seasoned equity offerings. however, as lee (1997) points out, the abnormal insider trading patterns around 6 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 corporate announcements could simply reflect a tendency for insiders to sell (buy) and to mechanically follow price increases (decreases) rather than a tendency for them to take advantage of specific information unavailable to the public. one way to investigate this issue is to examine the relation between insider trading and the longterm performance of the firm. for example, lee (1997) shows that primary issuers significantly underperform their benchmarks in the long run, regardless of insiders’ prior trading pattern. for secondary issuers, only those with insiders selling their shares before the issuance significantly underperform their benchmarks in the long run, while the others do not. this indicates that primary and secondary issuers with insiders selling their shares before the issuance seem to be knowingly selling overvalued equity, while primary issuers with insiders purchasing shares before the issuance do not seem to be knowingly selling overvalued equity. in the current study, the researchers examined whether the insider trading pattern before a spinoff announcement was related to improvements in the long run stock market performance and operating performance of the firm around a spinoff. the study has assumed that insiders are aware of the correct valuation of their firm at the time of the spinoff decision and that not all spinoffs were motivated by undervaluation. whereas insiders of undervalued firms had an incentive to increase the net purchase of their firm’s stock, the motive disappeared in the absence of undervaluation. as such, firms in which undervaluation was a motive for the spinoff were distinguished from other firms by examining their insider trades. significant improvements in the long run stock market performance and operating performance following spinoffs were expected in firms with prior increases in insider net purchases. given these assumptions, hypothesis 2 was proposed. hypothesis 2: following a spinoff, firms with prior increases in net insider purchases will exhibit improvements in their long run stock market and operating performance. previous studies have also suggested that insiders’ personal trading of their firm’s stock prior to a corporate announcement was related to the market reaction to the announcement. for example, hirschey and zaima (1989) find that most positive market reactions to corporate sell-off announcements occur in closely held firms with insider netbuy activity six months before the sell-off announcement. john and lang (1991) present evidence that announcement day excess returns 7 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 are negative and significantly lower for firms with insider selling prior to a dividend initiation announcement than for the remaining firms with no insider trading, or with only insider buying. in the current study, the researchers examined whether investors similarly regarded the recent pattern of insider trading activity of the firm as a useful backdrop for evaluating their spinoff decisions. if the market views spinoff decisions by firms with recent increases in insider net purchases are compatible with shareholder interests, then the study will expect spinoff announcement period excess returns to be significantly higher among these firms than in others. this assumption led to hypothesis (3). hypothesis 3: spinoff announcement period excess returns are higher among firms with recent increases in insider net purchases. sample data and research methodology spinoff sample the sample was obtained from the sdc mergers and acquisitions database. the database identifies spinoffs from news articles. first a search of the database was conducted for all the spinoffs announced by firms listed on the nyse, amex, or nasdaq between 1987 and 2006. the reason for starting in 1987 was because the thomson financial insider filing data files, in which the insider trading data were accessed, began in 1986 and the nature of the present study necessitated that at least one year’s insider trading data prior to the announcement date of a spinoff be made available. spinoffs involving a real estate investment trust were excluded. an initial sample of 524 spinoffs was identified. the following data selection criteria were then applied to the initial sample: 1. to verify that each transaction in the data was indeed a spinoff by checking news articles from factiva and lexis-nexis. transactions involving tracking stock, equity carve-outs or distributions of common stock in other publicly traded firms that were not subsidiaries of the parent firm would not fall within the study’s definition of spinoffs and were excluded, resulting in a loss of 28 transactions; 2. a precise announcement date and ex-date for the spinoff had to be made available from factiva, lexis-nexis or the crsp database. a total of 58 transactions were lost due to the lack of an identifiable announcement date or ex-spinoff date or both; 8 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 3. a total of 30 spinoffs that were taxable were eliminated; 4. all together 8 spinoffs involving adrs (american depository receipts) and 9 involving firms with operations in a regulated industry (sic 4910-4949) were dropped; 5. another 28 spinoffs were excluded because they were undertaken to facilitate the parent’s or the subsidiary’s merger with some other firm; 6. a total of 11 spinoffs were also removed because their announcement occurred simultaneously with other corporate information disclosures (for example, the appointment of a new ceo); 7. all together 35 spinoffs were eliminated because the subsidiary’s stock was trading prior to the spinoff announcement date; 8. a total of 33 two-step spinoffs were dropped; 9. another 7 spinoffs were removed because they represented cases in which one firm engaged in multiple spinoffs and the interval between the announcement date of a spinoff and the ex-date of the immediate prior spinoff was less than 1 year; and 10. finally, 35 spinoffs were lost because of an absence of insider trade in the five years before the spinoff announcement date. the final sample consisted of 187 separate spinoff events by 172 parent firms. fifteen parent firms had two spinoffs separated by at least a period of one year. panel a of table 1 summarizes the sampling procedure used in this study. panel b reports the frequency of spinoffs corresponding to the year of announcement. description of sample panel a reports the sample selection procedure. spinoffs were identified from the sdc mergers and acquisitions database. the initial sample started with 524 spinoffs announced by non-reit firms listed on the nyse/amex/nasdaq between 1987 and 2006. the sample excluded the following cases: 1) the transaction was not a spinoff; 2) a precise announcement date or ex-spinoff date for the spinoff was unavailable; 3) the spinoff was taxable; 4) the spinoff involved adrs or firms with segments operating in the utilities industry (sic4910-4949); 5) the spinoff was undertaken to facilitate the parent’s or the subsidiary’s merger with some other firm; 6) the spinoff announcement occurred simultaneously with other corporate information disclosures; 7) the subsidiary was trading prior to the 9 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 spinoff announcement date; 8) one firm engaged in multiple spinoffs and the interval between the announcement date of the current spinoff and the ex-spinoff date of the previous spinoff was less than 1 year; 9) there was no insider trade in the 5 years before the spinoff announcement date. panel b reports the distribution corresponding to the year of announcement of the final sample of 187 spinoffs. table 1 panel a: sample selection procedure initial sample 524 reason for elimination not a spinoff transaction 28 unable to identify a precise announcement date or ex-spinoff date 58 taxable spinoff 30 adrs 8 utilities 9 spinoff undertaken to facilitate merger with some other firm 55 return data unavailable on the crsp database for the parent on the announcement date 28 contaminated simultaneous announcements 11 subsidiary publicly listed before the announcement date 35 two-step spinoffs 33 multiple spinoffs by the same parent 7 no reported insider trade in the 5 years before the announcement date 35 final sample 187 panel b: distribution of spinoffs by year of announcement year number year number 1987 5 1997 14 1988 9 1998 11 1989 7 1999 12 1990 9 2000 15 1991 5 2001 6 1992 9 2002 5 1993 10 2003 8 1994 14 2004 6 1995 19 2005 4 1996 18 2006 1 total 187 10 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 insider trading data insider trading data were obtained from the thomson financial insider filing data files, which included all insider activities reported on sec forms 3, 4, 5, and 144.1 these data were available from 1986. the sec has defined a corporate insider as a company’s officer, director or any beneficial owner of 10 percent or more, of any equity class of securities. previous research, however, indicated that trades by large shareholders who were not officers or directors did not convey much information (seyhun, 1986). consequently, this study has analyzed only top executives’ transactions, and it should be noted that top executives were defined as chairpersons of the board of directors, executive directors, controlling persons, presidents, and anyone holding the position of vice president and above (lee, 1997). both open market and private insider transactions were combined together in the present study because the thomson financial insider filing data files did not report these two types of transactions separately. following seyhun (1986), this study has excluded all duplicate, amended, inconsistent transactions, and any transaction involving less than 100 shares. in the present analysis, the focus was on several measures of abnormal insider trades. following clarke et al. (2001), abnormal insider trading was defined as the actual trading less expected trading. trading refers to sales, purchases, or net purchases (purchases minus sales) by insiders. expected trading is the mean insider trading of the firm in the 36-month period beginning 48 months prior to the spinoff announcement and ending 13 months prior to the announcement.2 trading can be defined in terms of the number of trades (e.g., clarke et al., 2001), the number of shares traded (e.g., kaestner & liu, 1996), or the dollar value of trades (e.g., pettit & venkatesh, 1995) by insiders. while all three measures were used in this study, the measure of abnormal insider trading using the number of trades would only be reported. results using other insider trading measures were found to be qualitatively similar. matching firms to compute long run abnormal returns, the procedure by desai and jain (1999) was used. this study selected four matching firms for each parent and each subsidiary as its sample. only crsp-listed nonspinoff firms were used as the pool of possible matching firms. from this list of possible matches, the firm with the same two-digit sic 11 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 code was selected as the sample firm. it was also the one closest to the sample firm in market capitalization in the month of the ex-spinoff date. the closest matching firm was then designated the first matching firm; the second closest, the second matching firm, and so on, until the fourth closest matching firm. the stock market return of the sample firm was then compared with the return of the first matching firm. if the first matching firm should disappear for some reason, the return of the second matching firm would be used from that point on. similarly, the third and fourth matching firms were used until they were exhausted, after which the return of the crsp value-weighted index was used as a matching proxy. following desai and jain (1999), this study created a pro-forma combined firm following the spinoff by weighting the return of the parent and that of its subsidiary, and by using their market values of equity at the end of the month of the ex-spinoff date. the combined matching firm’s return was created by weighting the return of the parent’s matching firm and the subsidiary’s matching firm in both the preand post-spinoff periods using similar weights. as a robustness check, two additional benchmarks were used for computing abnormal returns. the crsp value-weighted index return was the first benchmark and the second was the value-weighted industry return. industry here has been defined as all the non-spinoff firms that have the same 2-digit sic code as the sample firm. however, the results from the sensitivity analysis would not be reported as they were found to be qualitatively similar to those reported in this paper. insider trading before spinoff announcement table 2 presents the analysis of insider trading in the four quarters prior to the spinoff announcement. the first three columns report the average quarterly insider sales, purchases, and net purchases, which are measured as insider purchases minus insider sales. the last three columns report insiders’ abnormal sales, purchases, and net purchases. these have been defined as actual sales, purchases, and net purchases minus expected sales, purchases, and net purchases, respectively. expected insider sales, purchases, and net purchases are the mean sales, purchases, and net purchases in the 36-month period, beginning 48 months and ending 13 months prior to the announcement. if undervaluation is an important motive for spinoffs, then insiders have an incentive to increase buying or decrease selling before spinoff announcements. consequently, one should be able to 12 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 observe significant abnormal insider net purchasing prior to spinoff announcements. table 2 quarterly insider trading before the announcement of spinoffs average abnormal purchases sales net purchases purchases sales net purchases q-4 0.08 1.60 -1.51 -0.13*** 0.02 -0.08 q-3 0.26 1.09 -0.83 0.07 -0.34** 0.61*** q-2 0.17 1.07 -0.95 -0.07** -0.40*** 0.38** q-1 0.18 0.76 -0.63 -0.03 -0.61*** 0.64*** the first three columns of table 2 indicate a clear, almost monotonic, increase in purchasing and net purchasing and a steady decline in selling in the four quarters prior to the spinoff announcement. the last three columns show that sales were abnormally low relative to historical levels in the three quarters prior to the announcement. in contrast, there was no evidence of abnormally high purchases. this may be due to the impact of insider trading regulations which deter insiders from increasing purchases when they have favorable private information about the firm. however, insiders do not act illegally if, instead of increasing purchases, they simply decrease sales. finally, significant abnormal net purchases were evident for the three quarters before the announcement, primarily due to less insider selling. overall, the findings are consistent with the assumption that undervaluation is an important motive for spinoffs. table 2 reports the average number of insider purchases, sales, and net purchases per firm per quarter in the four quarters before the announcement of spinoffs for a sample of 187 spinoffs. net purchases are measured as purchases minus sales. abnormal sales/purchases/net purchases are defined as actual sales/purchases/net purchases minus expected sales/purchases/net purchases in the given quarter. expected sales/purchases/net purchases are the mean quarterly sales/purchases/ net purchases of that firm in the 36-month period, beginning 48 months prior to the announcement of spinoffs and ending 13 months prior to the announcement. all variables were winsorized at the 5th and the 95th percentiles. the symbols *, ** and *** indicate significance of the test statistics at the 0.1, 0.05 and 0.01 levels, respectively. 13 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 firms’ market performance around spinoffs in this section, the spinoff announcement excess returns as well as the firms’ long run stock market performance around spinoffs will be examined. the sample used was stratified based on the abnormal insider net purchases during the quarter prior to the spinoff announcement day. spinoffs with positive abnormal insider net purchases are classified as “abnormal net purchases sample,” while those with zero or negative abnormal insider net purchases are classified as “abnormal net sales sample.” the announcement period excess returns as well as the long-term abnormal returns between the two sub-samples were then compared. announcement period abnormal returns panel a of table 3 reports the abnormal returns over different time intervals around the announcement of spinoffs. abnormal returns are estimated using the value-weighted nyse/amex/nasdaq index return and the market model with parameters estimated over days -250 through -50 relative to the announcement date. for the entire sample, the average (median) two-day cumulative abnormal return was 3.24% (2.38%) in the window [-1, 0]. both the mean and the median were significant at the 1% level. significant mean (median) returns of 2.70% (1.89%) and 4.36% (3.94%) were also found on day 0 and in the window [-1, +1], respectively. what is remarkable about the announcement period excess returns was the difference between the abnormal net purchases and abnormal net sales sub-samples. the average two-day cumulative abnormal return in the abnormal net purchases sample was 4.12%, which was more than twice that of the abnormal net sales sample at 2.01%. a t-test of the difference in the means between the two sub-samples and a wilcoxon’s signed rank test for the difference in the medians indicate that the announcement period excess returns of the two sub-samples were significantly different from each other at the 1% level and the 5% level, respectively. these results also persisted through the other two time-intervals investigated, namely, day 0 and the period [-1, +1]. panel b of table 3 examines the previous results using univariate and multivariate regression analyses. models 1 through 3 used cumulative abnormal returns over three different time intervals, day -1 to 0, day 0, and day -1 to 1, respectively, as dependent variables. the independent variable, anp, is an indicator variable that is set to one for firms in the abnormal net purchase sub-sample, and zero otherwise. all three 14 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 regressions showed a significant and positive relationship between the anp and abnormal returns, suggesting that firms in the abnormal net purchase sub-sample experienced significantly higher announcement period abnormal returns. in models 4 to 6, the study has included the arpr, focus, and spinsize as control variables. the arpr is the parent firm’s prior annual return minus the industry-and-size matched control firm’s prior annual return. focus is a dummy variable that is equal to one if the two-digit sic code of the subsidiary is different from the two-digit sic code of the parent, and zero otherwise. spinsize is the spun-off firm’s market value of equity at the end of the month of the ex-spinoff date, divided by the sum of the parent’s and the spun-off firm’s market value of equity. the results of multivariate regressions were consistent with those in the univariate analysis. in all three models, the anp was significantly and positively related to the abnormal returns. furthermore, the results show that firms with a larger spinsize experienced higher announcement abnormal returns. altogether, the results are consistent with the expectation that insider trading activity provides a useful indicator for the market to evaluate corporate spinoff decisions. moreover, spinoff decisions by firms with recent increases in insider net purchases are viewed by the market as being more compatible with shareholder interests. table 3 announcement period abnormal returns panel a: announcement period abnormal returns variables all firms (1) abnormal net purchases (2) abnormal net sales (3) t/z statistics for tests of differences (2)-(3) car (day -1 to day 0) mean 3.24%*** 4.12%*** 2.01%*** 2.60*** median 2.38%*** 2.91%*** 1.82%*** 2.22** no. of obs. 187 109 78 ar (day 0) ev mean 2.70%*** 3.72%*** 1.28%*** 3.68*** median 1.89%*** 3.00%*** 1.37%*** 2.65*** no. of obs. 187 109 78 car (day -1 to day 1) mean 4.36%*** 5.28%*** 3.07%*** 2.15** median 3.94%*** 5.12%*** 3.18%*** 2.37** no. of obs. 187 109 78 (continued) 15 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 panel b: regressions of announcement period abnormal returns dependent variables coefficient estimates car(-1, 0) ar(0) car(-1, 1) car(-1, 0) ar(0) car(-1, 1) constant 0.02 0.01 0.03 0.00 0.00 0.01 (3.30)*** (2.83)*** (3.86)*** (0.44) (0.36) (0.60) anp 0.02 0.02 0.02 0.03 0.03 0.03 (2.61)*** (3.66)*** (2.14)** (3.33)*** (3.60)*** (2.22)** arpr 0.00 0.00 -0.01 (0.58) (0.57) (-0.60) focus 0.00 0.01 0.01 (0.05) (1.10) (0.87) spinsize 0.04 0.04 0.05 (1.62) (2.02)** (1.70)* adjusted r2 0.03 0.06 0.02 0.07 0.10 0.04 sample size 187 187 187 169 169 169 panel a of table 3 shows the mean and median announcement period abnormal returns for 187 spinoff announcements during the period 1987-2006. abnormal returns are estimated using the valueweighted nyse/amex/nasdaq index return and the market model with parameters estimated over days -250 to -50 relative to the announcement date. the abnormal net purchases sample comprises firms with positive abnormal net insider purchases in the quarter prior to the announcement day. the remaining sample is classified as the abnormal net sales sample. the t-test and the wilcoxon signed ranks test were used to test the significance of the mean and median, respectively. the symbols *, ** and *** indicate significance of the test statistics at the 0.1, 0.05 and 0.01 levels, respectively. panel b of table 3 shows regressions of announcement period abnormal returns on the anp and other control variables. the anp is an indicator variable that is set to one for firms in the abnormal net purchases sub-sample and zero otherwise. arpr is the parent firm’s prior annual return minus the industry-and-size matched control firm’s prior annual return. focus is a dummy variable that is equal to one if the two-digit sic code of the subsidiary is different from the two-digit sic code of the parent and zero otherwise. spinsize is the spun-off firm’s market value of equity at the end of the month of the ex-spinoff date, divided by the sum of the parent and the spinoff 16 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 market value of equity. heteroskedasticity-consistent estimates of t-values are reported in parentheses. sample sizes varied due to the missing data. the symbols *, ** and *** indicate significance of the test statistics at the 0.1, 0.05 and 0.01 levels, respectively. performance of pro forma combined firms before spinoff table 4 reports the results for the pro-forma combined firms. the results in panel a show that in the three-year pre-spinoff period (prior to the month of announcement), firms in the entire sample significantly underperformed their benchmarks. the average (median) abnormal return one year before the announcement (month -12 to -1) was -15.19% (-8.26%). the average (median) abnormal returns two and three years prior to the announcement were -21.51% (-8.97%) and -21.41% (-16.07%), respectively. these returns were all large in magnitude and were statistically significant at the 1% level. more interestingly, as can be seen in panel b and c, the performance of the abnormal net purchases sub-sample was substantially different from that of the abnormal net sales sub-sample during the pre-spinoff period. table 4 stock market performance of the pro-forma combined firm time period n raws rawm ar panel a: all firms am -36 to am -1 158 mean 45.43% 66.84% -21.41%*** median [43.47%] [55.51%] [-16.07%***] am -24 to am -1 163 mean 24.75% 46.26% -21.51%*** median [24.62%] [40.46%] [-8.97%***] am -12 to am -1 165 mean 8.05% 23.24% -15.19%*** median [9.01%] [16.12%] [-8.26%***] am 169 mean 5.67% 1.23% 4.43%*** median [4.81%] 0.49% [3.54%***] ex 169 mean 1.94% 1.43% 0.51% median [0.44%] 0.93% [0.81%] ex +1 to ex +12 157 mean 18.60% 9.29% 9.31%** median [16.77%] [11.29%] [2.72%*] ex +1 to ex +24 131 mean 29.68% 21.90% 7.78% median [24.40%] [27.06%] [1.64%] ex +1 to ex +36 109 mean 51.61% 39.18% 12.43% median [43.55%] [40.25%] [6.14%] panel b: abnormal net purchases sample (continued) 17 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 time period n raws rawm ar am -36 to am -1 97 mean 41.61% 67.59% -25.98%*** median [38.96%] [57.47%] [-16.36%***] am -24 to am -1 100 mean 20.22% 49.51% -29.28%*** median [23.75%] [41.17%] [-12.11%***] am -12 to am -1 101 mean 3.34% 24.06% -20.72%*** median [5.71%] [13.94%] [-12.31%***] am 104 mean 7.12% 1.78% 5.34%*** median [6.25%] 0.27% [4.48%***] ex 104 mean 1.95% 2.19% -0.24% median [0.75%] 0.93% [0.82%] ex +1 to ex +12 92 mean 13.69% 6.34% 7.35% median [11.67%] [8.96%] [4.18%] ex +1 to ex +24 79 mean 24.78% 18.83% 5.95% median [21.49%] [24.07%] [1.55%] ex +1 to ex +36 66 mean 42.48% 34.06% 8.43% median [41.00%] [38.51%] [3.45%] panel c: abnormal net sales sample am -36 to am -1 61 mean 51.50% 65.64% -14.13% median [46.16%] [50.00%] [-15.78%] am -24 to am -1 63 mean 31.95% 41.11% -9.17% median [27.88%] [40.32%] [-2.01%] am -12 to am -1 64 mean 15.49% 21.95% -6.46% median [17.01%] [17.70%] [0.15%] am 65 mean 3.33% 0.36% 2.97%** median [1.58%] [0.66%] [1.79%*] ex 65 mean 1.91% 0.21% 1.71% median [-0.41%] [0.54%] [0.57%] ex +1 to ex +12 65 mean 25.54% 13.45% 12.08% median [21.67%] [15.41%] [1.92%] ex +1 to ex +24 52 mean 37.13% 26.56% 10.56% median [36.41%] [30.77%] [2.89%] ex +1 to ex +36 43 mean 65.61% 47.04% 18.58% median [48.58%] [42.27%] [10.59%] panel b of table 4 presents the results for the abnormal net purchases sub-sample. the average (median) abnormal returns for one to three years prior to the announcement was -20.72% (-12.31%), -29.28% (-12.11%), -25.98% (-16.36%), respectively. all these returns were statistically significant at the 1% level. panel c of table 4 reports the results for the abnormal net sales sub-sample. in contrast with the results for abnormal net purchases, none of the abnormal returns in the abnormal net sales sub-sample during the pre-spinoff period was significant. hence, the significant negative abnormal returns in the 18 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 full sample appeared to be driven solely by the results in the abnormal net purchases sub-sample. a comparison of the performance of the abnormal net purchases sub-sample to that of the abnormal net sales sub-sample indicates that, on average, the abnormal net purchases subsample underperformed the abnormal net sales sub-sample by 14.27% (p-value=0.08), 20.12% (p-value=0.08), and 11.85% (p-value=0.34), respectively, for holding periods of one, two, and three years prior to the spinoff announcement. the table 4 shows raw buy-and-hold returns of the sample firms (raws), raw buy-and-hold returns of the matching firms (rawm), and abnormal returns (ar) over several periods for the full sample of pro-forma combined firms, as well as for the abnormal net purchases and the abnormal net sales sub-samples. the abnormal net purchases sample comprised firms with positive abnormal net insider purchases in the quarter prior the announcement day. the remaining sample was classified as the abnormal net sales sample. a pro-forma combined firm following the spinoff was created by weighting the return of the parent and that of its subsidiary using their market values of equity at the end of the month of the ex-spinoff date. the matching firm’s return was created by weighting the return of the parent’s matching firm and the subsidiary’s matching firm in the pre-spinoff period, as well as the post-spinoff period using the above weights. for each parent and each subsidiary in the study sample, four matching firms were selected. the matching firms selected were the ones with the same two-digit sic code as that of the sample firm and were closest to the sample firm in market capitalization in the month of the exspinoff date. the closest matching firm was designated as the first matching firm; the second closest matching firm was designated as the second matching firm, and this designation process was carried out on the third and until the fourth matching firm. the stock market return on the sample firm was then compared with the return on the first matching firm. if the first matching firm should disappear for some reason, the return on the second matching firm was used from that point on, and then the third and so on until all the four matching firms were exhausted. after that, the return of the crsp value-weighted index was used. am is the month of the announcement date. ex is the month of the ex-spinoff date. sample sizes varied due to the missing return data. median values are shown in brackets. the symbols *, ** and *** indicate significance of the test statistics at the 0.1, 0.05 and 0.01 levels, respectively. 19 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 performance of the pro-forma combined firms after spinoff after the completion of a spinoff (following the month of the ex-spinoff date), the pro-forma combined firms in the full sample significantly outperformed their matching firms for a holding period of up to one year (see table 4, panel a). the average (median) abnormal return in the first year following the spinoff (month +1 to +12) was 9.31% (2.72%), which was significant at the 5% (10%) level. for holding periods of two years (month +1 to +24) and three years (month +1 to +36), the average (median) abnormal returns were 7.78% (1.64%) and 12.43% (6.14%), respectively, and they were not statistically significant. in the abnormal net purchases sub-sample, the post-spinoff average (median) abnormal returns were 7.35% (4.18%), 5.95% (1.55%), and 8.43% (3.45%), respectively, for holding periods of one to three years (see table 4, panel b). in the abnormal net sales sub-sample, the corresponding values were 12.08% (1.92%), 10.56% (2.89%), and 18.58% (10.59%), respectively (see table 4, panel c). however, none of these returns was statistically significant. furthermore, a comparison of the performance of the two sub-samples indicates no significant difference between them. finally, when the performance of the proforma combined firms before and after the spinoff was compared, it was found that the abnormal net purchases sub-sample exhibited a significant improvement in market performance around the spinoff, whereas the abnormal net sales sub-sample did not. in sum, the results that have been presented are consistent with the prediction that insider trades reflect superior information on the prospects of the firm. performance of parents after spinoff the post-spinoff performance of parent firms and their subsidiaries were examined separately to test whether there would be any difference between abnormal net purchases and abnormal net sales. table 5 reports the results for parent firms following a spinoff. panel a of table 5 presents results for all parents. the results show that parent firms had earned significant positive average (median) abnormal returns of 10.57% (3.35%), and 16.36% (8.96%) over a holding period of one and three years, respectively, and insignificant average (median) abnormal returns of 7.70% (-2.18%) over a holding period of two years. panel b and c report the findings for parents in the abnormal net purchases and sales sub-samples, respectively. parent firms with abnormal net purchases had earned average (median) abnormal returns of 9.34% 20 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 (4.54%), 6.25% (-2.18%), and 12.31% (5.19%) for holding periods of one, two, and three years, respectively. those with abnormal net sales had earned abnormal returns of 12.31% (2.54%), 9.90% (-0.89%), and 22.56% (19.04%) over the same holding periods. however, none of these returns was statistically significant. table 5 stock market performance of the parents following spinoffs time period n raws rawm ar panel a: all parents ex +1 to ex +12 157 mean 18.38% 7.81% 10.57%** median [17.21%] [11.59%] [3.35%*] ex +1 to ex +24 131 mean 26.90% 19.20% 7.70% median [22.65%] [22.71%] [-2.18%] ex +1 to ex +36 109 mean 51.42% 35.06% 16.36%* median [42.30%] [41.33%] [8.96%*] panel b: abnormal net purchases sample ex +1 to ex +12 92 mean 11.51% 2.17% 9.34% median [14.24%] [9.06%] [4.54%] ex +1 to ex +24 79 mean 21.48% 15.23% 6.25% median [19.42%] [22.08%] [-2.18%] ex +1 to ex +36 66 mean 41.79% 29.48% 12.31% median [41.17%] [41.16%] [5.19%] panel c: abnormal net sales sample ex +1 to ex +12 65 mean 28.11% 15.79% 12.31% median [23.23%] [20.91%] [2.54%] ex +1 to ex +24 52 mean 35.13% 25.23% 9.90% median [37.03%] [31.44%] [-0.89%] ex +1 to ex +36 43 mean 66.20% 43.64% 22.56% median [48.97%] [41.33%] [19.04%] table 5 shows raw buy-and-hold returns of the sample firms (raws), raw buy-and-hold returns of the matching firms (rawm), and abnormal returns (ar) over several periods for the full sample of parent firms, as well as for the abnormal net purchase and the abnormal net sales sub-samples. the abnormal net purchases sample comprised parent firms with positive abnormal net insider purchases in the quarter prior to the announcement day. the remaining sample was classified as the abnormal net sales sample. for each parent in the study sample, four matching firms were selected. the matching firms selected were the ones with the same two-digit sic code as that of the 21 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 sample firm and were closest to the sample firm in market capitalization in the month of the ex-spinoff date. the closest matching firm was designated as the first matching firm; the second closest matching firm was designated as the second matching firm, and this designation process was carried out on the third and until the fourth matching firm. the stock market return on the sample firm was then compared with the return on the first matching firm. if the first matching firm should disappear for some reason, the return on the second matching firm was used from that point on, and then the third and so on until all the four matching firms were exhausted. after that, the return of the crsp value-weighted index was used. am is the month of the announcement date. ex is the month of the ex-spinoff date. sample sizes varied due to the missing return data. median values are shown in brackets. the symbols *, ** and *** indicate significance of the test statistics at the 0.1, 0.05 and 0.01 levels, respectively. performance of subsidiaries after spinoff table 6 reports the results for subsidiary firms. panel a of table 6 shows that the sample of all subsidiaries exhibited a strong positive performance following a spinoff. the average (median) abnormal returns for the entire sample of subsidiaries for holding periods of one, two, and three years were 22.34% (12.24%), 19.56% (17.63%), and 19.69% (13.64%), respectively. all of these returns were significant at the 10% level or better. panel b shows that subsidiaries in the abnormal net purchases sub-sample had earned significant average (median) abnormal returns of 25.93% (12.21%), 22.88% (27.53%), and 22.50% (16.44%) over holding periods of one, two, and three years, respectively. the corresponding abnormal returns for subsidiaries in the abnormal net sales sub-sample were 17.26% (12.24%), 14.52% (-0.11%), and 15.39% (6.57%), respectively. however, only abnormal returns for a one year holding period were significant. relative to parent firms, the market performance of subsidiaries was much stronger. overall, the results indicate that there was an improvement in the market performance of pro-forma combined firms following a spinoff, and these were due primarily to the superior performance of their subsidiaries. the table 6 shows raw buy-and-hold returns of the sample firms (raws), raw buy-and-hold returns of the matching firms (rawm), and abnormal returns (ar) over several periods for the full sample of subsidiaries, as well as for the abnormal net purchase and the abnormal net sale sub-samples. the abnormal net purchase sample comprised firms with positive abnormal net insider purchases in the 22 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 quarter prior to the announcement day. the remaining sample was classified as the abnormal net sale sample. table 6 stock market performance of the subsidiaries following spinoffs time period n raws rawm ar panel a: all subsidiaries ex +1 to ex +12 157 mean 26.88% 4.55% 22.34%*** median [21.82%] [7.58%] [12.24%***] ex +1 to ex +24 131 mean 46.81% 27.24% 19.56%** median [44.36%] [22.58%] [17.63%**] ex +1 to ex +36 109 mean 69.75% 50.06% 19.69%* median [70.71%] [49.43%] [13.64%**] panel b: abnormal net purchases sample ex +1 to ex +12 92 mean 30.07% 4.14% 25.93%*** median [23.09%] [4.65%] [12.21%***] ex +1 to ex +24 79 mean 46.92% 24.04% 22.88%** median [47.02%] [21.68%] [27.53%**] ex +1 to ex +36 66 mean 65.05% 42.55% 22.50%* median [68.34%] [46.40%] [16.44%**] panel c: abnormal net sales sample ex +1 to ex +12 65 mean 22.37% 5.11% 17.26%** median [20.18%] [9.17%] [12.24%*] ex +1 to ex +24 52 mean 46.62% 32.11% 14.52% median [41.56%] [25.54%] [-0.11%] ex +1 to ex +36 43 mean 76.98% 61.58% 15.39% median [70.71%] [53.52%] [6.57%] for each subsidiary in the study sample, four matching firms were selected. the matching firms selected were the ones with the same two-digit sic code as that of the sample firm and were closest to the sample firm in market capitalization in the month of the ex-spinoff date. the closest matching firm was designated as the first matching firm; the second closest matching firm was designated as the second matching firm, and this designation process was carried out on the third and until the fourth matching firm. the stock market return on the sample firm was then compared with the return on the first matching firm. if the first matching firm should disappear for some reason, the return on the second matching firm was used from that point on, and then the third and so on until all the four matching firms were exhausted. after that, the return of the crsp value-weighted 23 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 index was used. am is the month of the announcement date. ex is the month of the ex-spinoff date. sample sizes varied due to the missing return data. median values are shown in brackets. the symbols *, ** and *** indicate significance of the test statistics at the 0.1, 0.05 and 0.01 levels, respectively. operating performance changes around spinoffs in this section, changes in operating performance around spinoffs are examined. following daley et al. (1997), the ratio of operating cash flow (compustat annual data item #13) to total assets (compustat data item #6) as a measure of operating performance was used and was referred to as the return on assets (roa). table 7 reports changes in the roa from year -1 (the year before the ex-spinoff year) to year +1 (the year after the ex-spinoff year) for pro-forma combined firms.3 similar to the analysis of firms’ stock market performance carried out in the present study, the focus was on the adjusted roa of matching-firms. for each parent and subsidiary in the study sample, a matching firm with available data on the compustat annual database was selected. the matching firm selected was the one closest to the sample firm in terms of market capitalization, and the same two-digit sic code in the month of the ex-spinoff date. this study created a pro-forma roa measure for each combined firm in the post-spinoff period by weighting the roa of the parent and of its subsidiary with their respective market value of equity at the end of the month of the ex-spinoff date. similarly, the matching firm’s combined roa was created by weighting the roa of the parent’s and subsidiary’s matching firm in the preand post-spinoff periods, using the same weights. a firm was included in the sample as long as it had at least one year’s operating performance data before and after the spinoff. table 7 reports the mean and median change in operating performance for the pro-forma combined firms from year -1 (the year before the ex-spinoff year) to year +1 (the year after the ex-spinoff year). the abnormal net purchase sample comprised firms with positive abnormal net insider purchases in the quarter prior to the announcement day. the remaining sample was classified as the abnormal net sale sample. operating performance was measured as the ratio of operating cash flow (compustat annual data item #13) to total assets (compustat data item #6). for each parent and each subsidiary in the study sample, one matching firm with data available on the compustat annual database was selected. the matching firm selected was the one closest to the sample firm in market capitalization and had the same two-digit sic 24 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 as that of the sample firm in the month of the ex-spinoff date. the pro-forma combined firm’s roa measure in the post-spinoff period was computed by weighting the roa of the parent and that of its subsidiary, and by using their market values of equity at the end of the month of the ex-spinoff date. the matching firm’s roa was created by weighting the roa of the parent’s matching firm and the subsidiary’s matching firm in the pre-spinoff period, as well as the post-spinoff period using the above weights. a firm would be included in the sample as long as it had at least one year’s operating performance data before and after the spinoff. median values are shown in brackets. the symbols *, ** and *** indicate significance of the test statistics at the 0.1, 0.05 and 0.01 levels, respectively. table 7 change in operating performance around spinoffs for the pro-forma combined firms year relative to spinoff n unadjusted match-firm-adjusted panel a: abnormal net purchases sample -1 73 mean 13.10% -1.66% median [12.90%] [-0.42%] 1 73 mean 13.36% 0.97% median [15.03%] [1.20%] 73 mean 0.26% 2.63%** median [0.50%] [2.20%***] panel b: abnormal net sales sample -1 41 mean 14.88% 7.57% median [14.48%] [0.58%] 1 41 mean 12.88% 3.26% median [10.97%] [1.66%] 41 mean -2.00% -4.31% median [0.80%] [-0.95%] the study has stratified the sample and classified them as abnormal net purchases and sales, and has reported the mean and median change in operating performance (roa) for the pro-forma combined firms from year -1 (the year before the ex-spinoff year) to year +1 (the year after the ex-spinoff year). the abnormal net purchases sample comprised firms with positive abnormal net insider purchases in the quarter prior to the announcement day. the remaining firms were then classified as 25 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 the abnormal net sales sample. the findings are as reported in table 7. in the abnormal net purchases sub-sample, the mean (median) change in matching-firm-adjusted roa was 2.63% (2.20%) around the spinoff year. both the mean and the median changes were significant at least at the 5% level. in the abnormal net sales sub-sample, the corresponding mean and median changes were -4.31% and -0.95%, which were not significant at the 10% level. the wilcoxon two-sample median test rejected the equality of median roa changes across both the abnormal net purchases and the abnormal net sales sub-samples at the 10% significance level. overall, the results in table 7 seem to suggest that only the abnormal net purchases sub-sample exhibited a significant improvement in operating performance following a spinoff. these findings support the study hypothesis that firms with prior increases in insider net purchases will have an improved postspinoff long run operating performance. conclusion the present study has used insider trading to examine undervaluation as a motive behind corporate spinoffs. the sample consisted of 187 separate spinoff events by 172 parent firms during the period 19872006. if undervaluation is a motive for spinoffs, as has been predicted by the model of nanda and narayanan (1999), self-interested managers have an incentive to increase their share purchases or to decrease their sales prior to a spinoff announcement. either action leads to an increase in net insider share purchases. consistent with the study hypothesis, it was found that there was a clear and almost monotonic increase (decrease) in the number of insider purchases (sales) and net purchases (sales) in the four quarters prior to a spinoff announcement. in addition, relative to the benchmark period, insider sales (net purchases) were significantly lower (higher) in the three quarters prior to a spinoff announcement. the study sample has been stratified based on the insider trading activity one quarter prior to the announcement day and then classified into the following two groups: abnormal net purchases and abnormal net sales. it was found that the announcement period excess returns for abnormal net purchases were significantly higher than those for abnormal net sales, suggesting that the market viewed the spinoff decisions by firms with prior abnormal net purchases as more compatible with shareholder interests. moreover, only firms in the abnormal net purchases group exhibited significant improvement in their long-run stock market and 26 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 operating performance following a spinoff, suggesting that managers seemed to be aware that their firms were undervalued and hence, time their own trades accordingly. overall, the results seem to imply that undervaluation is an important motive behind corporate spinoffs and that it is possible to identify the quality of a spinoff firm based on the insider trading behavior prior to its announcement. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. endnotes 1 the sec has recently modified the electronic data gathering, analysis, and retrieval (“edgar”) system to allow persons to file electronically, securities ownership and transaction reports pursuant to section 16(a) of the securities exchange act of 1934 (forms 3, 4 and 5). persons required to file notifications of the proposed sale of securities pursuant to rule 144 under the securities act of 1933 (form 144) may now also file electronically, wherein the issuer of the securities is a public company. 2 for spinoffs announced before jan 1, 1990, the benchmark period begins on jan 1, 1986. this would be less than 36 months since the insider trading data from the thomson financial insider filing data files started in 1986. the study results are robust by excluding these spinoffs. 3 the study focused on the roa change from year -1 to year +1. there would be a large loss of observations due to the missing compustat data if the investigation was on two or more years around the year of spinoff. references ahn, s., & denis, d. j. (2004). internal capital markets and investment policy: evidence from corporate spinoffs. journal of financial economics, 71, 489-516. akbulut, m. e. (2005). market misevaluation and merger activity: what do managers’ insider trades tell us? working paper, university of southern california. 27 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 allen, j. w. (2001). private information and spin-off performance. the journal of business, 74, 291-306. boehmer, e., & netter, j. m. (1997). management optimism and corporate acquisitions: evidence from insider trading. managerial and decision economics, 18(7/8), 693-708. burch, t. r., & nanda, v. (2003). divisional diversity and the conglomerate discount: evidence from spinoffs. journal of financial economics, 70, 69-98. choi, w. (2020). disclosure tone of the spin-off prospectus and insider trading. journal of accounting and public policy, 39(1), 106692. clarke, j., dunbar, c., & kahle, k. m. (2001). long-run performance and insider trading in completed and canceled seasoned equity offerings. journal of financial and quantitative analysis, 36, 415-430. cusatis, p. j., miles, j. a., & woolridge, j. r. (1994). some new evidence that spinoffs create value. journal of applied corporate finance, 7, 100-107. daley, l., mehrotra, v., & sivakumar, r. (1997). corporate focus and value creation: evidence from spinoffs. journal of financial economics, 45, 257-281. desai, h., & jain, p. c. (1999). firm performance and focus: longrun stock market performance following spinoffs. journal of financial economics, 54, 75-101. elliot, j., morse, d., & richardson, g. (1984). the association between insider trading and information announcements. rand journal of economics, 15, 521-536. finnerty, j. e. (1976). insiders and market efficiency. journal of finance, 31, 1141-1148. hirschey, m., & zaima, j. k. (1989). insider trading, ownership structure, and the market assessment of corporate sell-offs. the journal of finance, 44, 971-980. hite, g. l., & owers, j. e. (1983). security price reactions around corporate spin-off announcements. journal of financial economics, 12, 409-416. jaffe, j. f. (1974). special information and insider trading. journal of business, 47, 410-428. john, k., & lang, l. h. p. (1991). insider trading around dividend announcements: theory and evidence. the journal of finance, 46, 1361-1389. john, k., & mishra, b. (1990). information content of insider trading around corporate announcements: the case of capital expenditures. the journal of finance, 45, 835-855. kaestner, r., & liu, f. y. (1996). going private restructuring: the role of insider trading. journal of business finance and accounting, 23, 779-806. 28 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 1–28 kahle, k. m. (2000). insider trading and the long-run performance of new security issues. journal of corporate finance, 6, 25-53. krishnaswami, s., & subramaniam, v. (1999). information asymmetry, valuation, and the corporate spin-off decision. journal of financial economics, 53, 73-112. lee, d. s., mikkelson, w. h., & partch, m. m. (1992). managers’ trading around stock repurchases. the journal of finance, 47, 1947-1961. lee, i. (1997). do firms knowingly sell overvalued equity? the journal of finance, 52, 1439-1466. lorie, j. h., & niederhoffer, v. (1968). predictive and statistical properties of insider trading. journal of law and economics, 11, 35-51. miles, j. a., & rosenfeld, j .d. (1983). the effect of voluntary spinoff announcements on shareholder wealth. the journal of finance, 38, 1597-1606. nanda, v., & narayanan, m. p. (1999). disentangling value: financing needs, firm scope, and divestitures. journal of financial intermediation, 8, 174-204. penman, s. h. (1982). insider trading and the dissemination of firm’s forecast information. journal of business, 55, 479-503. pettit, r. r., & venkatesh, p. c. (1995). insider trading and long-run return performance. financial management, 24, 88-103. pratt, s. p., & devere, c. w. (1970). relationship between insider trading and rates of return for nyse common stocks, 19601966. in lorie, j., & brealey, r. (eds), modern developments in investment management. praeger. schipper, k., & smith, a. (1983). effects of reconstructing on shareholder wealth the case of voluntary spin-offs. journal of financial economics, 12, 437-467. seyhun, h. n. (1986). insiders’ profits, costs of trading, and market efficiency. journal of financial economics, 16, 189-212. seyhun, h. n. (1990a). overreaction or fundamentals: some lessons from insiders’ response to the market crash of 1987. the journal of finance, 45, 1363-1388. seyhun, h. n. (1990b). do bidder managers knowingly pay too much for target firms? journal of business, 63, 439-464. seyhun, h. n., & bradley, m. (1997). corporate bankruptcy and insider trading. journal of business, 70, 189-216. song, w. h. (2005). does overvaluation lead to bad mergers? evidence from insider trading and long-term performance. working paper, boston college. the information conveyed by debt and equity announcements in australia the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 61 the information conveyed by debt and equity announcements in australia oraluck arsiraphongphisit deakin university, australia abstract this paper aims to report the findings associated with the effects of share prices around the disclosure dates of four different debt-and-equity fund-raising events over a 12-year period from 1991 to 2003 in australia. by applying the well-known event study approach, along with the data-trimming procedures, a new idea is to remove all known confounding events and make corrections for thin-trading bias. the observed statistically significant price effects are consistent with theories: a positive price effect is observed for straight-debt and private placements whereas negative price effects occur when convertible debt and rights issues are announced. the results pertaining to the private placement effect is reported for the first time on this market. these findings are consistent with leverage, agency, and asymmetric information theories. it is believed that this study contributes new evidence on private placements and other events adding to existing literature surrounding the matter at hand. keywords: equity private placement, equity rights issue, straight debt, convertible debt, cumulative abnormal returns jel classification: d92, g32 1. introduction a great deal of empirical analysis, particularly during the 1980s and 1990s, has been devoted to the observation of share price behavior surrounding the announcements of debt and equity financing, along with the way in which the issues are marketed. however, the majority of these studies have been limited to the u.s., as well as some european and asian countries. the researchers had also examined data between 1965 and 1997; therefore, the findings have become obsolete since market structure has changed substantially over the last decade. to the author’s knowledge, no study of this kind on straight debt, convertible debt, equity, and private placements exists in australia. there are only two rights-issue studies to date.1 1 only two rights studies have been done in australia. this includes a published paper by ball, brown, and finn (1977), as well as a phd thesis by dehnert (1993). ijbf 62 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 thus, the primary objective of this paper is to provide robust evidence on share price responses to these issues for a given period, ultimately adding new findings to the australian finance literature. in addition, this study seeks to examine and compare the effects of the four debt and equity capital events announced and issued over a simultaneous period, a phenomenon not seen elsewhere. to the author’s knowledge, the analysis of the four financing types has not yet been reported. this study should also contribute fresh evidence to the literature while at the same time, resolving the debate over any ambiguous results on the rights issue price effects in the australian market. corporations raise external capital by borrowing first and then offering different types of securities through various methods of capital financing. aside from bank lending, equity financing has long been popular as a primary source of funds for firms. the methods used in raising equity externally were predominantly private placements and rights issues, accounting for 69 percent of the overall equity funding in 2006.2 current equity funding trends would suggest that private placements and rights issues must have dominant price effects in the equity market. although the debt market is less developed than the equity market, it has been growing significantly. in fact, private sector debt, including straight and convertible debts, has grown threefold in the past decade. it accounts for around 75 percent of the total debt securities outstanding (about a$810 billion or us$705 billion) in 2006.3 although secondary issues of both debt and equity are considered as new forms of financing for firms, they convey opposite information to the market about value. based on theories such as the tax hypothesis (modigliani and miller, 1963), agency theory (jensen and meckling, 1976), and asymmetric information model (myers and majluf, 1984), the announcement of publicly traded straight debt (cheaper funds as long as debt is not beyond optimal) is associated with positive share price reactions. the sale of equity securities is generally associated with negative share price responses (overvalued shares, thus lower cost of funding). being that convertible debt is a hybrid security, its debt and equity-related features have implications on changes in shareholders’ wealth, thus affecting share price negatively. however, the literature further suggests that the market not only negotiates a firm’s value through the types of securities issued, but also through methods of raising capital. equity private placements involve the sale of shares to institutions and/or high net worth investors. they generally convey favourable news through the so-called “white knights effect”. the market in contrast reacts negatively to publicly traded equity securities like that of rights issues, which are pro-rata offerings of new shares to existing shareholders in proportion to their current share ownership of a firm in this market. management is prone to resort to this type of funding when share prices are overvalued, thus a lower cost of capital, hence signalling an overvaluation to those purchasing rights. the effect is also due to the discount offered to the subscribers. evidence on share price reactions to different security types and their issuing methods are somewhat inconclusive. studies on straight debt issues in u.s. markets have reported mixed results. for example, de jong and veld (2001) had reported 2 source: australian stock exchange. 3 ibid. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 63 positive market responses to straight debt, whereas howton, howton, and perfect (1998) had detected negative price effects. however, market responses observed in european and asian countries are consistently positive (e.g. in europe by brounen and eichholtz (2001) and in japan by kang and stulz (1996)). extensive studies, particularly those done in the u.s., concentrated on share price reactions to rights issues and to a lesser extent, private placements. as found in studies conducted by goh, gombola, lee and liu (1999) and cronqvist and nilsson (2003), market participants generally perceive new information regarding private placements as a signal of good news, hence the allegory of the white knight coming to the rescue of troubled firms, suggesting a positive price reaction. in contrast, announcements regarding rights issues are perceived as a signal of bad news, probably eliciting a negative share price reaction (e.g. in miles and perterson (2002) and martin-ugedo (2003)). however, unlike private placements where positive share price reactions are consistently observed over different countries, this is yet to be investigated in australia. the literature on market reactions to rights issue announcements has yielded mixed results as well.4 for instance, while studies in canada, new zealand, the u.s., and the u.k. reveal negative share price reactions to rights issue announcements, those in european and asian countries had reported mixed results. the findings in this study are similar to that evidence found in the u.s. for all types of issues, except for straight debt. however, in terms of rights issues, the results are in contrast with studies in japan, korea, and singapore. in this study, significant positive price effects to public sales of straight debt issues and equity private placements are revealed, whereas a significant negative effect is evident for publicly traded convertible debt securities and equity rights issues. these effects are consistent with theories, such as the tax hypothesis (under a classical tax regime), agency theory, and asymmetric information model. also, the negative effects of announcements involving equity rights issues resolves the conflicting findings reported hitherto, perhaps due to the robust results from the absence of confounding effects in our research method, where the use of high frequency data and corrections for data errors were implemented. section 2 briefly reviews the relevant literature on theoretical issues, where as section 3 discusses the data, methodology, and statistical procedures for data analysis. lastly, section 4 contains the results of the analyses, while a conclusion is provided in section 5. 2. literature on financing effect a number of hypotheses have been advanced to explain share price behaviours resulting from an announcement of financing decisions. this section relates the theories that explain how markets react to different security types and their issuance methods. these theories will be articulated here in terms of their relevance to the study. 4 mixed results are also evident in australian studies. that is while ball et al. (1977) report a positive price effect to rights issues, dehnert (1993) detects a negative price effect. 64 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 tax hypothesis: the classical work of modigliani and miller (1958) perhaps best outlines the rudiments of this theory: in a perfect market, where there is no tax, financing choices are irrelevant to firm value. in a subsequent study, modigliani and miller (1963) relaxed some of the assumptions in their original model to include the effects of corporate tax in order to properly reflect the benefits of a tax shield. given the tax-effect of borrowings, firm value can be increased through the use of debt. miller (1977) argued that with the differences in personal tax rates across investors, debt should have no impact on firm value while deangelo and masulis (1980) suggest that at some point the personal tax disadvantage of retaining debt is compensated by the corporate tax advantage of debt. consequently, the net tax benefit is often positive, making debt still attractive. the results appear to support this interpretation. thus, the underlying theme in this hypothesis is that the issuance of debt should lead to positive share price responses, which in turn should increase firm value. new equity effectively reduces a shareholders’ tax shield per share while new debt will generally increase their tax shield benefits. moreover, the magnitude of these effects should be related to the amount of tax shield benefits obtained from debt financing (brounen and eichholtz, 2001). therefore, low non-debt tax shield firms may face a larger announcement effect on share prices than high non-debt tax shield firms (deangelo and masulis, 1980). conversely, high non-debt tax shield firms may face a larger announcement effect than low non-debt tax-shield firms (suchard, 1997). notwithstanding the above themes, the tax hypothesis generally has no implications for the methods of issuing securities (i.e. public or placements etc.), and thus it is unlikely to explain the effect that these methods may have on market expectations unless the specific conditions of the firms’ capital structure are known. also, the above arguments are mainly applied to countries operating within a classical tax system (i.e. no imputation, a fact in this market). under dividend imputation tax regimes in which australian firms operate, tax should not play an important role in debt or equity financing choices as shareholders are indifferent between debt and equity financing, given the role of imputation credits or single taxation at the corporate level, offsetting the tax on dividends. the tax hypothesis therefore provides no explanation under the imputation tax system. agency model: jensen and meckling (1976) are among the first to relate the economic theory of the firm as did others (berle & means, (1932); coase, (1937); & smith, (1776)) to modernize finance theory. in essence, they had argued that several parties have competing interests in the firm, including managers who act as agents for the principals, i.e., shareand debt-holders. they had further suggested that managers are expected to act in the best interest of shareholders. however, as utility maximizers, they will not always do so, which will invariably result in the incurrence of agency costs. they had observed that conflicts of interests between shareholders and managers are generally associated with the separation of ownership and control. they had predicted that the lower the managers’ ownership, the higher the conflict of interest between shareholders and managers, and the higher the agency costs of equity. therefore, the issue of equity, i.e. a decrease in managers’ ownership, conveys unfavourable news, which results in a downward share price movement. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 65 however, the conflict of interest between shareholders and managers can be mitigated through debt financing, as debt reduces the amount of free cash flow available to a manager (jensen, 1986; stulz, 1990), thus increasing the monitoring of the managers by lenders. this is also known as the free cash flow hypothesis. debt financing increases the managers’ stake in a firm, binding the managers to perform and utilise assets more efficiently in order to meet debt obligations. consequently, debt issuance should convey good news to the market, creating the potential for positive effects on share prices relative to equity issuance. this line of thought can also be viewed with leland and pyle’s (1977) signalling model. a manager’s willingness to increase his/her stake in a firm is positively related to share prices and in turn firm value. thus, investors will see the raising of equity as a signal of overpriced shares. jensen and meckling (1976) also identify another type of a conflict of interest between shareholders and debt-holders. this has been the subject of many studies, e.g. galai and masulis, (1976); jensen and meckling (1976); myers (1977); and smith and warner (1979), all of which generally suggest that shareholders have incentives to maximise their own benefits, and not necessarily maximise firm value, at the expense of debt-holders (masulis, 1988). this is the basis for a further theory, typically referred to as the wealth transfer hypothesis. given that contractual restrictions of a firm’s ability to issue debt (i.e. debt covenants) cannot completely prevent a manager’s decisions to alter the capital structure of a firm, debt-holders risk and wealth can be still affected by changes in the capital structure (fama and miller, 1972). the option pricing theory also predicts this. indeed, galai and masulis (1976) had demonstrated that wealth can be transferred from debt-holders to shareholders when there is an unexpected increase in the risk associated with the underlying assets of the firm’s assets. as a result, the firm’s value may decrease (masulis, 1988). new debt can increase both the risk associated with asset degradation and risk of current debt-holders. firstly, the higher the amount of debt, the greater the financial distress costs, and secondly, debt-holders bear the loss in the value of debt through the cost of failure or poor investment returns given the risks associated with underlying assets. note also that, while shareholders bear the loss in the value of equity, this loss can be offset by the gain of wealth transferred from debt-holders if investments are value-decreasing, and earn unlimited returns from profitable investments (harris and raviv, 1991). this implies that debt financing expropriates wealth from debtholders to shareholders. as a result, share market participants interpret debt issues as favourable news and react positively to debt announcements. conversely, new equity financing reduces the risk of existing debt, and therefore transfers wealth from shareholders to debt-holders. as a result, investors view the issue of new equity as unfavourable news and react negatively to equity announcements. however, it should also be noted that debt-holders are able to protect their interests from activities that decrease shareholder value (e.g., by debt covenants and discounts in securities), in which case, shareholders will bear the cost of investing in value-decreasing projects. this 66 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 effect is commonly termed the asset substitution effect, which is also known as the agency cost of debt financing (harris and raviv, 1991). the issue of new debt affects both the debt-holders’ claim on the value of the firm’s assets and the potential expropriation of wealth from debt-holders to shareholders. this implies that the specific characteristics of securities have implications for the wealth of stakeholders. given this theme, smith and warner (1979) and green (1984) had suggested the issue of a convertible debt (or warrants) as a compromised device to reduce conflicts of interest between debt-holders and shareholders. they argued that the conversion feature of convertibles makes it less attractive for shareholders to invest in non-profitable projects. this is because convertible debt-holders can convert their debt claims into shares, enjoying wealth increasing activities from the above expropriation process (harris and raviv, 1991). asymmetric information model: the information asymmetry theory suggests that managers are privy to inside information and are thus at a considerable informational advantage compared to outsiders regarding a firm’s activities, value, and investment opportunities. given this informational advantage, ross (1977) suggested that managers have an incentive to convey information about their firm’s prospect to the market through their financing activities. as managers’ compensation packages and career paths are more often linked to a firm’s level of performance, they are unlikely to commit to additional debt unless they are optimistic about future earnings and cash flows of the firm. thus, issuing debt would be perceived by the market as a signal of quality and optimism regarding the future prospects of the firm. this would concomitantly lead to positive movements in share price. similar price behaviour is predicted by leland and pyle (1977). they had suggested that the level of share ownership retained by managers is a signal that managers are ‘willing to invest in their own project’. given that they have superior knowledge regarding future cash flows and prospects of the firm, the signal is positive (particularly given that they are potentially increasing their diversification risk by colloquially speaking ‘retaining all their eggs in one basket’). rational investors would thus perceive the level of managers’ share ownership as a credible signal regarding a firm’s quality (masulis and korwar, 1986). negative share price responses are therefore predicted should managers issue new shares, which in turn reduce their portion of firm ownership. myers and majluf (1984) had extended ross’s initial model to account for investment decisions. their model is known as the adverse selection model. it is motivated by the asset value of a firm. they argue that if a firm requires funding to finance a profitable investment and management believes that the firms prevailing share price exceeds its intrinsic value, then the firm will issue equity (or to a lesser extent, convertible debt). as a result, a firm will use equity to finance the project only if the value of the firm increases and its existing shareholders benefit more than new shareholders. this follows the asymmetric information hypothesis that managers have superior information and act in the best interest of existing shareholders. market participants perceive this as a wealth transfer mechanism and react negatively to equity financing, thus resulting in negative returns. note also that the magnitude the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 67 of the share price change is related to the extent to which a firm’s share price is over/undervalued. the greater the overvaluation of a firm’s share price, the more negative the market responds. brennan and kraus (1987) and stein (1992) however suggest that such an adverse selection problem associated with equity financing can be mitigated through the use of convertible debt. an alternative information asymmetry explanation is advanced by miller and rock (1985), namely implied cash flow change hypothesis. as in ross (1977) and leland and pyle (1977), this hypothesis suggests that the quality of the firm is associated with a firm’s performance. their model assumes an accounting principle where sources and uses of funds must be matched. it predicts that a firm tends to raise new funds when its cash flow is weakening. thus, an unexpected, new source of outside financing through either debt or equity alerts the market of some potentially dismal news, suggesting that the firm faces a higher risk of weak cash flow performance relative to the inability of the firm in achieving future obligations. unlike the previous signalling hypothesis, miller and rock (1985) had predicted a negative association between share price and either debt or equity offerings. in summary, asymmetric information theory generally predicts debt announcements are favourable and equity announcements are unfavourable. in addition, negative market responses can be reduced if there is little information asymmetry between market participants (dierkens, (1991) and korajczyk, lucas and macdonald, (1991)). rights issues offers no change in the existing fractional ownership (as hypothesised by leland and pyle (1977). there are also no apparent wealth transfers from existing to new shareholders, as assumed by myers and majluf (1984). thus, only ross’s (1977) and miller and rock’s (1985) hypotheses may be of relevance when explaining share price patterns surrounding rights issue announcements. following these themes, since rights issues are an equity raising activity, we can expect a negative share price reaction once the news of rights offerings is available. kabir and roosenboom (2003) had suggested that although asymmetric information hypotheses can explain share price patterns around rights issues (i.e., if existing shareholders take up the rights issues and sell them to new investors), they do not apply if existing shareholders take up and hold the rights issues. hertzel and smith (1993) had extended myers and majluf’s (1984) study to include private placements. they posit that undervalued firms with profitable opportunities can mitigate the information asymmetry effect by using private placements as an alternative mechanism for equity financing. they explain that, when firms issue equity privately, a small group of private investors are able to evaluate value more closely than would public investors in the usual public issue process (goh et al., 1999). thus, if a firm does not have good future prospects or a stable flow of expected earnings, it cannot afford to carry out such an action, and accordingly, would not engage in such scrutiny. this is consistent with the findings in szewczyk and varma (1991). they base their analysis on the approach used by myers and majluf (1984), where private placements are considered as close substitutes for financial slack, suggesting that markets should react non-negatively to private financing. in this sense, the signalling hypothesis predicts that private 68 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 placements of both equity and debt convey positive news to the market and should improve a firm’s share price. this is also consistent with evidence found in wruck (1989), hertzel and smith (1993), and lee and kocher (2001). 3. data and methodology 3.1 data firm specific data: the initial sample consists of 137 straight debt issues, 80 convertible debt issues, 806 private placements, and 358 rights issues. they were announced and issued by companies listed on the australian stock exchange (asx) over the thirteen-year period from 1991, the timeframe in which the economy was relatively stable despite the asian crisis that affected the australian economy the least among those respective countries in the asia-pacific region. to be included in the initial sample, the announcements of interest must have a clear public disclosure date. also, events that are announced concurrently with any potential confounding events are excluded from the study. these potential, confounding events that of which are announced within a five-day period before or after the event date are those that have the propensity to impact share prices. additionally, announcements made by firms during the first year of listing were excluded. for each observation, the event date, daily share price, as well as any and all relevant market and financial data were collected. the primary source of data for event dates and market data were datanalysis, sirca, and bloomberg databases. the financial data was collected from various sources, including connect4, datanalysis, and aspect financial analysis databases. to ensure the reliability of results, this study further takes into account the problem of infrequent and thin trading in data the screening process because many securities (particularly small companies) are infrequently or thinly traded (bartholody and riding, 1994). this problem often prevails in the data collection process where there is a lack of share price data for securities with no trading (most studies do not reveal this information). this screening process yielded 60 straight debt, 43 convertible debt, 377 private placement, and 158 rights observations. summary statistics for all security offerings used in this study are presented in table 1. table 1: descriptive statistics table 1 reports the descriptive statistics for 60 straight debt, 43 convertible debt, 377 private placement, and 158 rights observations. mean median std. deviation minimum maximum panel a: straight debt issues total assets ($) 10,124,910,295 2,334,736,000 30,162,132,835 3,102,000 251,714,000,000 total liabilities ($) 8,893,626,045 1,918,707,000 28,351,428,868 56,000 235,950,000,000 interest bearing debt ($) 7,677,315,987 1,063,745,000 23,096,331,375 0 188,008,000,000 total equity ($) 1,344,513,469 821,210,000 2,275,036,775 12,242,868 15,761,000,000 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 69 outstanding shares 394,074,218 358,103,552 304,094,768 21,000,000 1,447,649,492 tax shield from issues ($) 90,298,551 61,649,114 95,422,265 2,700,000 720,000,000 market capitalisation ($) 1,919,555,344 940,678,285 3,622,466,176 13,440,000 29,459,667,162 issue sizes ($) 257,198,883 176,429,076 266,072,544 7,500,000 2,000,000,000 issue sizes/market cap. 13.40% 18.76% 7.35% 55.80% 6.79% panel b: convertible debt issues total assets ($) 1,054,359,992 175,248,636 1,840,866,059 2,716,752 4,955,700,000 total liabilities ($) 702,210,163 40,314,000 1,202,725,141 601,516 3,098,700,000 interest bearing debt ($) 434,569,767 11,790,000 720,040,856 0 1,645,200,000 total equity ($) 352,149,829 121,365,000 670,190,819 2,115,236 1,857,000,000 outstanding shares 327,885,066 301,530,000 273,906,965 32,968,005 699,922,121 tax shield from issues ($) 52,147,769 13,680,000 59,635,283 340,000 133,307,195 market capitalisation ($) 868,892,404 321,964,176 1,318,521,882 3,626,481 3,645,225,166 issue sizes ($) 148,117,452 38,000,000 170,578,584 1,000,000 392,079,984 issue sizes/market cap. 17.05% 11.80% 12.94% 27.57% 10.76% panel c: private placement issues total assets ($) 32,802,403 11,180,762 64,421,760 572,175 342,515,578 total liabilities ($) 16,165,084 873,561 43,964,997 10,476 235,366,118 interest bearing debt ($) 6,686,697 21,028 13,959,072 0 57,845,602 total equity ($) 16,637,320 9,058,056 22,375,731 561,699 107,149,459 outstanding shares 107,691,381 96,856,178 85,691,044 12,417,764 419,206,323 market capitalisation ($) 44,866,576 25,921,896 63,717,898 1,494,935 324,512,090 issue sizes ($) 5,852,752 3,464,800 8,164,678 104,125 40,000,000 issue sizes/market cap. 13.04% 13.37% 12.81% 6.97% 12.33% panel d: rights issues total assets ($) 150,252,117 10,299,000 283,557,471 4,024,867 853,240,000 total liabilities ($) 59,215,302 6,094,513 129,956,012 19,000 399,196,000 interest bearing debt ($) 34,289,941 4,200,000 74,100,821 0 228,432,000 total equity ($) 91,036,815 10,124,000 165,623,888 3,910,232 454,044,000 outstanding shares 176,948,617 64,167,271 210,767,173 21,291,190 604,360,999 market capitalisation ($) 76,681,078 13,227,512 130,620,247 3,755,870 355,281,744 issue sizes ($) 14,915,486 2,055,707 22,376,723 850,956 56,146,969 issue sizes/market cap. 19.45% 15.54% 17.13% 22.66% 15.80% (continued) 70 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 the first four rows of each panel show the summary measures in dollar size the total assets, total liabilities, interest bearing debt, and total equity of firms issuing securities in question. firms issuing straight debt have the highest median amount of all interested characteristics. the dollar size of the offerings is presented in the second last row of each panel. the median amount ranges are from $2.06 million for rights issues to $176.5 for straight debt offerings. the median market capitalisation of firms issuing straight debt is also much larger than that of firms issuing other types of securities. however, based on comparable dollar amounts measured relative to the market capitalisation of the issuing firms, the percentage of the median amount issued relative to the market capitalisation is 18.76%, 11.8%, 13.37%, and 15.54% for straight debt, convertible debt, private placement, and rights issues, respectively. these figures imply that firms rely more heavily on straight debt than other types of issues. nevertheless, should they need equity financing, they will prefer rights issues to private placements. perhaps this is so being that private placements are subject to a limited number of shares or amount that can be funded. the median issue size from straight debt issues is about $176.5 million with a minimum of $7.5 million and a maximum of $200 million. this represents 18.76% of the firm’s market capitalisation before the straight debt issues. this amount is larger than that of the u.s. market, in which the percentage of straight debt issues represents 13.6% of the firm’s market capitalisation over the period over 1977-1984 (mikkelson and partch, 1986). as hybrid securities can be structured to mitigate informational effects, firms in australia appear to undertake some convertible debts over the past decade. the median issue size from the convertible debt issues is about $38 million, representing 11.8% of the firm’s market capitalisation. this percentage is smaller than the u.s. market’s figure, amounting to 13.2% over the period of 19721982 (mikkelson and partch, 1986) and 20% over the period of 1978-1992 (lewis, rogalski, and seward, 2003). unlike debt financing, the percentage of the issue size to the market capitalisation of firms issuing private placement in australia is higher than in the u.s. although the median issue size from the private placement issues is small (about $3.5 million), it represents 13.37% of the firm’s market capitalisation before the issues. this figure is slightly higher than that of the u.s. (11.75%) over the period of 1980-1987 (hertzel and smith, 1993). lastly, australian firms also raised capital through rights issues, representing 15.54% of the firm’s market capitalisation with the median amount of $2 million and maximum mount of $56 million. however, when comparing descriptive figures between countries/studies, one should be aware of the analysis periods, which in most cases are different. share prices and market index: closing share prices were used and the all ordinaries accumulation index (xao_a) was used as a proxy for the market index. for each event, the event date is defined as time 0. daily share prices and xao_a are collected for the period surrounding the respective events. these data were collected from the datastream and sirca databases. similar to earlier studies, the data used in this study are adjusted for stock dividends and changes in capital, such as stock splits. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 71 methodology the analysis of share price reactions was conducted within a standard event study framework as described in brown & warner (1980; 1985). in this study, the market adjusted returns method is employed for abnormal return computations for both private placements and rights issues. the event periods are various in order to ensure that the period is long enough to capture the significant effect of the event, yet short enough to exclude confounding effects (mcwilliams & siegel, 1997). market adjusted model: the market adjusted return model describes ex-post abnormal returns as follows: ar i,t = r i,t -r m,t (3) where : t : days measured relative to the event date ar i,t : excess or abnormal returns r i,t : the rate of return on security i in period t r m,t : the rate of return on market index in period t table 2 on the following page represents the predicted signs of abnormal returns following each type of financing event based on theories discussed in theoretical section. in general, a positive share price reaction to debt securities is expected, whereas a negative share price effect is expected for equity securities. however, when placing the securities privately, the market appears to act positively to both debt and equity issues. this model estimates expected or predicted returns as follows: r i,t = r m,t (4) consistent with strong (1992), for each announcement, the return on each day being studied is computed as follows: (5) where p i,t : the price of security i in period t, adjusted for capital changes d i.t : the dividend received from security i in period t. the abnormal returns across firms on the same event date are cumulated across firms to obtain the average abnormal return over time. then, the cumulative average residuals (cars, or the sum of average abnormal returns during time t within the event window) are calculated based on fama, fisher, jensen and roll’s (1969) procedure. the following arithmetic method is applied: ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ + = −1, ,, , ln ti titi ti p dp r 72 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 table 2: predicted signs of abnormal returns following four financing events debt security equity security private placements public offerings public offerings straight debt convertible debt shares rights of shares tax hypothesis# + +/-^ agency theory free cash flow hypothesis wealth transfer hypothesis + + + + + signalling hypothesis firm quality hypothesis (ross, 1977and leland & pyle, 1977) adverse selection model (myers & majluf, 1984) implied cash flow change hypothesis (miller & rock, 1985) + + + + + # under the classical tax regime, the tax hypothesis generally predicts a positive (negative) share price reaction to debt (equity) announcements. the imputation tax system aims to eliminate the double taxation inherent in the classical system and to remove any tax advantages of debt. therefore, while firms under the classical system are biased toward debt, those under the imputation system are biased toward equity. this subsequently gives rise to a neutral effect between debt and equity. the table however, presents the effects under the classical tax system. ^ a convertible debt can be structured as equity-liked or debt-liked security. if it is designed to primarily possess straight debt features, a positive share price reaction is expected. however, a negative effect is expected if the issue predominantly possesses equity features. ( 6 )∑= = n i tit ar n aar 1 , 1 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 73 ( 7 ) where aar t : the average abnormal returns for n securities in period t, n : the number of securities in the portfolio and car (j,k) : cumulative abnormal returns between time j and k. the statistical procedure for data analysis in this study employs the conventional t-test statistics. the null hypothesis is tested using test statistics discussed in brown and warner (1985), corrado (1989), and corrado and zivney (1992). this test depends upon the cross-sectional independence of the securities’ abnormal returns for the correct specifications (corrado, 1989). according to brown and warner (1985), if the assumption that the cross-sectional independence is valid, the t-statistics used in this study is better than the one that equals to aar t /σ aar . the t-statistics of the average abnormal return for any event day t used in this study is: t statistics of aar = (8) = (9) = (10) = (11) where t statistics of aar : t-test of mean standardized abnormal returns, see aar : the standard error of estimates of the average abnormal returns of observations within the estimation period, σ aar : the standard deviation of the average abnormal return, aar int : the average of the average abnormal returns during the interested period, m : the beginning of the interested period, n : the end of the interested period, and p : number of days in the interested period or m-n +1. the t-statistics of cumulative abnormal return for any specific interval is: ∑ = = k jt tkj aarcar ),( aar t see aar n aarσ ( ) )1( 2 int − ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ −∑ = = p aaraar nt mt t ∑ = = nt mt taar n 1 74 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 t statistics of car = (12) = (13) where see car : the standard error of estimates of cars of observations within the event period, k : the beginning of the event period, l : the end of the event period, and t : number of days in the event period or k-l +1. 4. results 4.1 share price reactions to public straight debt announcements aars and cars for the sample of straight debt are presented in table 3. the results show that straight debt issues have a slight effect on share prices. in fact, the aar for the day prior to the event date (day -1), the event date (day 0), and the day after the event date (day -1) are 0.29 percent, 0.22 percent, and 0.09 percent, respectively. they are positive but are not statistically significant. in addition, the cars for the whole test period (-10, +10) and the pre-event period (-10, -1) are positive at 1.33 and 0.71 percent, respectively, and statistically significant at the 0.05 level. the car for the on-event period (-1, +1) is also positive at 0.60 percent, but it is more statistically significant at the 0.01 level. although the car for the post-event period (+2, +10) is positive, it is not statistically significant. these results are consistent with a market that is efficient in the semi-strong form in which share prices adjust rapidly to new information. it is also noted that the positive, average, three-day market responses to straight debt announcements observed in this study are consistent with those results obtained from studies done in spain by arrondo & gomez-anson (2003) and in the u.s. by howe and shilling (1988). the majority of u.s. studies report negative results. through the observance of industrial firms, arrondo and gomez-anson (2003) revealed a positive return of 0.68 percent, which is statistically significant at the 0.01 level. similarly, while observing real estate investment trusts (reits), howe and shilling (1988), had reported a significantly positive return of 1.94 percent. positive results are also found in other studies; however, they are insignificant (e.g. in the u.s. by johnson (1995) and lewis et al. (1999), in japan by kang and stulz (1996), and in european countries by brounen and eichholtz (2001)5. the positive share price reactions resulting from straight debt issue announcements in this study are consistent with the agency hypothesis, including the free cash flow and wealth transfer hypotheses, and the asymmetric information model, as discussed in the previous sections. however, the results are inconsistent with the tax hypothesis in 5 those six european countries include france, netherlands, norway, portugal, spain, and the u.k. car lk see car ),( aart σ the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 75 ∑ = = n i tit ar n aar 1 , 1 ∑ = = k jt tkj aarcar ),( table 3: daily average abnormal returns and cumulative abnormal returns surrounding public straight debt announcements aar (%) car (%) std dev. see t-statistics panel a: aar and car -10 0.2246 0.2246 0.0151 0.0020 1.1449 -9 0.0257 0.2503 0.0163 0.0021 0.1201 -8 -0.0353 0.2150 0.0127 0.0016 -0.2146 -7 -0.0050 0.2100 0.0149 0.0020 -0.0256 -6 0.1494 0.3594 0.0175 0.0023 0.6561 -5 0.0086 0.3680 0.0119 0.0016 0.0549 -4 -0.0492 0.3188 0.0210 0.0027 -0.1801 -3 0.0421 0.3609 0.0171 0.0022 0.1887 -2 0.0595 0.4204 0.0162 0.0021 0.2852 -1 0.2931 0.7135 0.0184 0.0025 1.1923 0 0.2220 0.9356 0.0184 0.0025 0.9041 1 0.0866 1.0222 0.0145 0.0019 0.4498 2 -0.1049 0.9172 0.0118 0.0015 -0.6909 3 -0.0722 0.8450 0.0158 0.0020 -0.3541 4 0.0798 0.9249 0.0152 0.0020 0.4065 5 0.0105 0.9354 0.0121 0.0016 0.0663 6 0.0348 0.9701 0.0131 0.0017 0.2037 7 -0.1263 0.8439 0.0090 0.0012 -1.0689 8 0.1291 0.9730 0.0130 0.0017 0.7674 9 -0.0973 0.8756 0.0129 0.0017 -0.5848 10 0.4628 1.3385 0.0132 0.0018 2.6182** panel b: cumulative windows -10,+10 1.3385 0.1449 0.6642 2.0151** -10,-1 0.7135 0.1143 0.3614 1.9745** -1,+1 0.6018 0.1049 0.1817 3.3121*** +2,+10 0.3163 0.1834 0.5502 0.5748 *** indicates statistical significance at the 0.01 level. ** indicates statistical significance at the 0.05 level. * indicates statistical significance at the 0.10 level. a b 76 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 the imputation context where the market should be indifferent between issuing debt or equity securities. 4.2 share price reactions to public convertible debt announcements aars and cars for the sample of convertible debt announcements are presented below in table 4. although the results show that convertible debt issues have a significant negative effect on share prices during the event period, the issues did not significantly affect share price movements on particular trading days during the observed period. while the aar for day -1 is positive, the aars for day 0 and day -1 are negative. in addition, all of these aars are not statistically significant. both cars for the entire test period, i.e. days -10 to +10, and the pre-event period, i.e. days -10 to -1, are not statistically significant. while the former window demonstrates a negative effect to the convertible debt announcements, the latter window shows a positive effect. however, the analysis of a small event window proves the phenomenon to be a significant event, as it is apparent from the table that cars for both of the on-event periods, i.e. days -1 to 0 and -1 to +1, are negative at -0.61 percent and -1.19 percent, and statistically significant at the 0.05 and 0.01 level, respectively. the insignificance of cars during the post-event period, i.e. days +2 and +10, are consistent with market efficiency in which share prices adjust rapidly to new information and no abnormal returns should be earned as share prices already fully reflect all available information. this study reports a negative, average, two-day market reaction to convertible debt announcements. the abnormal return and the significant t-test findings obtained from this study are consistent with results obtained in many u.s. studies, like that of lewis et al. (1997) & burlacu (2000). the share price responses observed in these studies were all statistically significant at the 0.01 level. while most hypotheses such as the tax hypothesis (under a classical tax regime), free cash flow hypothesis, firm quality hypothesis (leland and pyle, 1977; ross, 1977), and adverse selection model (myers and majluf, 1984), predict that leverage-increasing events convey favourable news to the market and leveragedecreasing events convey unfavourable news, the results of this study contrast with this theoretical prediction. the significant negative results obtained from this study may be a result of the expected conversion time of the issue. the market will price the convertible issue based upon the expected conversion period. if the time to expected conversion is short, the market will react to convertible issues in a similar fashion to when it responds to equity issues (davidson, glascock, and schwartz, 1995). in other words, the market may perceive dilution effects that result from the conversion of the issue into equity of a firm. such uncertainty of wealth per share is therefore immediately impounded to the price at the time when the convertible security is issued. another plausible explanation is in regards to the characteristics of convertible issues. the dichotomised characteristics (debt and equity claims) of convertible issues make it difficult, or if not impossible, for investors to precisely estimate an issuing firm’s financial leverage and/or analyse the real purpose of convertible debt. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 77 ∑ = = n i tit ar n aar 1 , 1 ∑ = = k jt tkj aarcar ),( a b table 4: daily average abnormal returns and cumulative abnormal returns surrounding convertible debt aar (%) car (%) std dev. see t-statistics panel a: aar and car -10 0.5019 0.5019 0.0354 0.0054 0.9733 -9 -0.5243 -0.0223 0.0338 0.0052 -0.8909 -8 -0.0344 -0.0568 0.0386 0.0059 -0.0589 -7 1.1167 1.0599 0.0374 0.0058 1.6350 -6 -0.7929 0.2670 0.0432 0.0068 -0.7835 -5 0.4906 0.7576 0.0664 0.0101 0.9697 -4 0.2525 1.0101 0.0328 0.0051 0.3751 -3 -0.2939 0.7162 0.0441 0.0067 -0.4163 -2 0.0774 0.7936 0.0463 0.0071 0.1017 -1 -0.4508 0.3428 0.0499 0.0076 -1.1420 0 -0.1589 0.1839 0.0259 0.0039 -0.2534 1 -0.5797 -0.3959 0.0411 0.0063 -0.6185 2 -2.0247 -2.4205 0.0615 0.0094 -1.4268 3 0.6060 -1.8145 0.0931 0.0142 1.0992 4 0.1594 -1.6551 0.0357 0.0055 0.2380 5 -0.2020 -1.8571 0.0439 0.0067 -0.3570 6 0.0168 -1.8403 0.0371 0.0057 0.0261 7 -0.0674 -1.9077 0.0423 0.0064 -0.1238 8 0.1314 -1.7763 0.0357 0.0054 0.1517 9 0.3323 -1.4440 0.0568 0.0087 0.5084 10 -0.1322 -1.5762 0.0429 0.0065 -0.2128 panel b: cumulative windows -10,+10 -1.5762 0.6297 2.8855 -0.5463 -10,-1 0.3428 0.5766 1.8233 0.1880 -1,+1 -1.1895 0.2156 0.3734 -3.1851*** +2,+10 -1.1804 0.7523 2.2570 -0.5230 *** indicates statistical significance at the 0.01 level. ** indicates statistical significance at the 0.05 level. *indicates statistical significance at the 0.10 level. 78 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 consequently, this may distort the positive effect of the first claim component of the convertible issue. the findings of this study are however consistent with the model developed by miller and rock (1985), which they had predicted a negative correlation associated with either debt or equity offerings. miller and rock’s (1985) model relied on the decision to obtain new financing rather than types of financing. they had suggested that the decision to raise new funds reveals the weakening cash flow of an issuing firm. thus, they had predicted the same share price response for both debt and equity announcements. 4.3 share price reactions to private placement announcements aars and cars for the sample of private placements are presented on the following page in table 5. the results show that the aars for the day prior to the event date (day -1), the event date itself (day 0), and the day after the event date (day -1) are positively and statistically significant at the 0.01 level, except for day 0, which is statistically significant at the 0.05 level. in addition, the cars for the whole test period (-15, +10), the pre-event period (-15, -1), and the on-event period (-1, +1) are positively and statistically significant at the 0.01 level. however, the car for the post-event period (+2, +10) is negative and not statistically significant. similar to the previous results, these findings are consistent with a market that is efficient in the semi-strong form in which share prices adjust rapidly to new information. it is noted also that the positive, average market responses regarding private placement announcements are consistent with those results obtained from u.s. studies done by hertzel and rees (1998), and goh, et al. (1999); in the u.k. by slovin, sushka, and lai (2000); in sweden by cronqvist and nilsson (2003); in japan by kang and stulz (1994); and in hong kong by wu and wang (2002). the significantly positive aars prior to the announcement date and the significantly positive cars during the pre-announcement period are consistent with the adverse selection model, as proposed by myers and majluf (1984), in which a firm will issue equity if it believes that its prevailing share price exceeds its intrinsic value. also, information leakage that a firm will privately place equity can explain the nature of positive aars prior to the announcement day. the positive share price reactions resulting from the announcement of privately placed equity in this study are also consistent with the previous evidence. although many hypotheses (such as the tax, agency, free cash flow, and wealth transfer hypotheses, as well as the asymmetric information model) generally predict that the market will react negatively to equity issue announcements, evidence shows that the issuing method is likely to influence share price responses to new security issues. the argument for positive reactions to placement is that private investors are likely to enhance the monitoring system, as well as improving the expertise of an issuing firm, which in turn, mitigates the agency costs. however, public offerings do not offer monitoring mechanisms, nor do they increase the level of cash in hands of the manager, which results in increased agency costs. thus, market participants react positively for they view a private placement as favourable news. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 79 ∑ = = n i tit ar n aar 1 , 1 ∑ = = k jt tkj aarcar ),( table 5: daily average abnormal returns and cumulative abnormal returns surrounding private placement announcements aar (%) car (%) std dev. see t-statistics panel a: aar and car -15 0.6488 0.6488 0.0552 0.0040 1.6158 -14 0.6646 1.3134 0.0586 0.0043 1.5595 -13 0.3426 1.6560 0.0598 0.0044 0.7873 -12 1.0077 2.6637 0.0558 0.0041 2.4834 ** -11 0.2435 2.9072 0.0609 0.0044 0.5497 -10 0.3012 3.2084 0.0536 0.0039 0.7721 -9 0.8124 4.0209 0.0899 0.0065 1.2428 -8 1.1253 5.1462 0.0688 0.0050 2.2496** -7 0.9331 6.0793 0.0483 0.0035 2.6575*** -6 0.9723 7.0516 0.0570 0.0041 2.3454** -5 0.9523 8.0039 0.0634 0.0046 2.0650** -4 0.9622 8.9661 0.0614 0.0045 2.1532** -3 0.0318 8.9979 0.0496 0.0036 0.0881 -2 0.3352 8.6627 0.0655 0.0048 0.7038 -1 1.9161 10.5787 0.0628 0.0046 4.1969*** 0 1.0756 11.6543 0.0628 0.0046 2.3539** 1 1.3317 12.9860 0.0602 0.0044 3.0411*** 2 0.1988 13.1848 0.0584 0.0042 0.4680 3 0.0275 13.2123 0.0590 0.0043 0.0641 4 0.0578 13.2701 0.0554 0.0040 0.1434 5 0.2724 13.5424 0.0543 0.0039 0.6896 6 0.1011 13.4414 0.0597 0.0043 0.2328 7 0.2315 13.6728 0.0464 0.0034 0.6851 8 0.6020 13.0708 0.0530 0.0039 1.5620 9 0.5012 12.5695 0.0530 0.0039 1.2998 10 0.4222 12.1474 0.0594 0.0043 0.9776 panel b: cumulative windows -15,+10 12.1474 0.6240 3.1817 3.8179*** -15,-1 10.5787 0.5375 2.0817 5.0817*** -1,+1 4.3233 0.4308 0.7462 5.7940*** +2,+10 0.8386 0.3345 1.0036 0.8356 *** indicates statistical significance at the 0.01 level. ** indicates statistical significance at the 0.05 level. * indicates statistical significance at the 0.10 level. a b 80 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 ∑ = = n i tit ar n aar 1 , 1 ∑ = = k jt tkj aarcar ),( table 6: daily average abnormal returns and cumulative abnormal returns surrounding right issues announcements aar (%) car (%) std dev. see t-statistics panel a: aar and car -15 0.3610 0.3610 0.0448 0.0037 0.9827 -14 0.2859 0.0751 0.0421 0.0034 0.8398 -13 0.0366 0.0385 0.0451 0.0036 0.1009 -12 0.0271 0.0656 0.0501 0.0041 0.0667 -11 0.3930 0.4586 0.0448 0.0036 1.0891 -10 0.5038 0.9624 0.0441 0.0036 1.4125 -9 0.5275 1.4898 0.0411 0.0033 1.5806 -8 0.0713 1.5611 0.0388 0.0032 0.2243 -7 0.1765 1.7376 0.0427 0.0035 0.5040 -6 0.6471 2.3847 0.0462 0.0038 -1.6984* -5 0.2395 2.1452 0.0442 0.0036 0.6742 -4 0.5592 2.7044 0.0501 0.0041 1.3795 -3 0.6115 3.3159 0.0501 0.0041 1.5059 -2 0.0662 3.3821 0.0408 0.0033 0.1988 -1 0.3283 3.7105 0.0473 0.0038 0.8562 0 1.2652 4.9757 0.0399 0.0033 3.8230*** 1 1.4059 6.3816 0.0473 0.0039 3.6196*** 2 0.2705 6.6521 0.0503 0.0041 0.6565 3 0.3043 6.3478 0.0547 0.0044 0.6861 4 0.1469 6.2009 0.0424 0.0034 0.4283 5 0.0243 6.2253 0.0419 0.0035 0.0702 6 0.3909 5.8343 0.0571 0.0047 0.8390 7 0.3344 6.1688 0.0561 0.0046 0.7306 8 0.6739 6.8427 0.0420 0.0034 1.9670** 9 0.3934 7.2361 0.0479 0.0039 1.0196 10 0.0838 7.3199 0.0413 0.0033 0.2503 panel b: cumulative windows -15,+10 7.3199 0.4380 2.2336 3.2772*** -15,-1 3.7105 0.3040 1.1775 3.1513*** -1,+1 2.9995 0.5858 1.0146 2.9564*** +2,+10 0.9383 0.3483 1.0450 0.8979 a b *** indicates statistical significance at the 0.01 level. ** indicates statistical significance at the 0.05 level. * indicates statistical significance at the 0.10 level. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 81 4.4 share price responses to rights issue announcements the daily aars and cars for the sample of rights issues are presented in table 6. the aars for the event date (day 0) and the day after the event date (day -1) are negative and statistically significant at the 0.01 level. in addition, the cars for the whole test period (-15, +10), the pre-event period (-15, -1), and the on-event period (-1, +1) are negatively and statistically significant at the 0.01 level. as is expected, the car for the post-event period (+2, +10) is also negative, yet, not statistically significant. the average, negative abnormal returns attributable to rights issue announcements are consistent with previous australian studies conducted by the likes of dehnert (1993). dehnert (1993) had employed both mean adjusted return and risk-adjusted return (market model) methods in her study. she had also reported a two day car of -1 percent for the former method and -1.2 percent for the latter. these results are statistically significant at the 0.01 level. the result is similar to this study in which a negative, three-day car of -2.99 percent exists with statistical significance at 0.01. in contrast to these findings, by employing monthly returns, the early australian study by ball et al. (1977) had reported a positive event day return of 2.6 percent. however, unlike this and dehnert’s (1993) study, ball et al. (1977) had examined rights issues with bonus elements in their study. the market reactions to rights issues announcements in international studies are mixed. for instance, in the u.s., miles and perterson (2002) and martin-ugedo (date?), had generally reported a statistically significant and negative abnormal return. similarly, in the u.k., levis (1995), slovin, et al. (2000), and korteweg and reneboog (2003) had observed a negative 1 percent statistically significant abnormal return. however, these results are contrary to the findings obtained in asia (e.g. in japan by cooney, kato and schallheim (2003); in korea by dhatt, kim and mukherji (1996); and in singapore by tan, chng, and tong (2002)). the negative share price reactions resulting from equity rights issue announcements in this study are consistent with the agency hypothesis, including free cash flow and wealth transfer hypotheses, as well as the asymmetric information theory, as discussed in previous sections. 4.5 analysis of debt and equity event over the same period the pattern of returns from all announcements of debt and equity offerings is consistent with the previous studies. that is, positively significant price effects to straight debt and equity private placement announcements are revealed, whereas significant negative effects are evident for convertible debt and rights issues. these results are shown graphically in figure 1. the time series plots also indicate that share prices tend to plateau without substantially changing after the announcements. interestingly enough, the three-day price effects of all types of issues around the announcement date are statistically significant at the 0.01 level. while the market reacts most strongly to private placement issues, the price effect of straight debt offerings is least pronounced. hertzel and smith (1993) had explained that when 82 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 issuing equity privately, a firm allows institutional investors and/or a small group of private investors to evaluate its quality closely. these investors generally possess facilities and expertise about the market. they in fact are the hardest profit-driven investors in the market and will generally invest should a firm be undervalued. thus, their decision to take on a private offer indicates a positive signal about an issuing firm’s quality, consequently enhancing the share price of a firm. the magnitude of a positive abnormal return for private placements is correlated with the extent to which the asset is undervalued. in effect, the private placement issues in this study attract the three-day car of 4.32 percent, followed by the 3 percent obtained from the rights issues. the three-day cars of convertible debt and straight debt issues are -1.19 percent and 0.6 percent, respectively. a comparison of price effects to these issues is shown graphically in figure 1. note that the aars on the announcement date of straight and convertible debt are positive and negative, respectively. these values are however statistically insignificant due to a weak price effect on the event day. figure 1: comparison of daily cumulative abnormal returns surrounding straight debt, convertible debt, private placement and rights issue announcements 5. conclusion in the past two decades, empirical studies and literature on the valuation effects of debt and equity issue announcements, as well as its methods of issue have progressed substantially. however, this kind of research is scarce in the australian market. in australia, debt has remained a dominant external funding source for non-financial firms, with a steady increase in debt securities outstanding from $270 billion in 1993 -10.00 -5.00 0.00 5.00 10.00 15.00 -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 days relative to announcement car (%) st ra ig h t d e b t c onv e rti b le de b t p riva te p lac e m e n ts ri g h ts iss u es the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 83 to $810 billion in 2006.6 australian firms nonetheless rely heavily on the equity market as a secondary source of funds. the volume of equity issues has grown substantially from $10 million in 1993 to $61.5 million in 2006, especially through private placement and rights issues, representing 49.8 percent of total equity raised in the australian equity market in 2006. as there is a continuing trend towards popular funding through straight debt, convertible debt, private placement, and rights issues, as well as a rare study of price effects in the australian market, it is appropriate to investigate the information conveyed by these securities. this study analysed share price reactions to announcements of various types of debt and equity funding over a current 13-year period. on average, the findings indicated a positive, statistically significant share price response to the announcement of public straight debt and equity private placement issues. in contrast, a negative, statistically significant price effect to the announcement of public convertible debt and rights issues are observed. despite the statistical significance of cars over the three-day period surrounding the announcement of straight debt and convertible debt issues, the on-event aar of these securities are statistically insignificant. the results however, indicate that market participants appear to use the financing information of firms at the time of announcements, reacting quickly upon the information. over the same period, the price effect to all debt and equity issues examined in this study is consistent with the theoretical explanations from that of the agency theory to the asymmetric information model. author statement: oraluck arsiraphongphisit, school of business, deakin university, australia. email: oraluck@deakin.edu.au. references arrondo, r., and gomez-anson, s., (2003). a study of spanish firms’ security issue decision under asymmetric information and agency costs, applied financial economics, 13, 10, 771-82. ball, r., brown, p., and finn, f., (1977). share capitalisation, changes, information and australian equity markets, australian journal of management, 2, 2, 105-25. bartholody, j., and riding, a., (1994). thin trading and the estimation of betas: the efficacy of alternative techniques, the journal of financial research, 17, 2, 241-54. berle, a. a., and means, g. c., (1932). the modern corporation and private property, macmillan, ny. brennan, m.j., and kraus, a., (1987). efficient financing under asymmetric information”, journal of finance, 42 (december), 5, 1225-43. brounen, d., and eichholtz, m.a., (2001). capital structure theory: evidence from european property companeis’ capital offerings, real estate economics, 29, 4, 615-32. 6 source: australian stock exchange. 84 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 brown, s. j., and warner, j. b., (1980). measuring security price performance”, journal of financial economics, 8, 205-58. brown, s. j., and warner, j. b., (1985). using daily stock returns: the case of event studies, journal of financial economics, 14, 3, 3-31. burlacu, r., (2000). new evidence on the pecking order hypothesis: the case of french convertible bonds, journal of multinational financial management, 10, 3-4, 439-59. coase, r., (1937). the nature of the firm”, economica, 4, 386-405. cooney, j.w. jr., kato, h.k., and schallheim, j.s., (2003). underwriter certification and japanese seasoned equity issues”, review of financial studies, 16, 3, 949-82. corrado, c. j., (1989). a nonparametric test for abnormal security-price performance in event studies, journal of financial economics, 23, 2, 385-95. corrado, c. j., and zivney, t. l., (1992). the specification and power of the sign test in event study hypothesis tests using daily stock returns, journal of financial and quantitative analysis, 27, 3, 465-78. cronqvist, h., and nilsson, m., (2003). agency costs of controlling minority shareholders, financial and quantitative analysis, 38, 4, 695-719. davidson iii, w.n., glascock, j.l., and schwartz, t.v., (1995). signalling with convertible debt, journal of financial and quantitative analysis 30 (september), 3, 425-40. de jong, a., and veld, c., (2001). an empirical analysis of incremental capital structure decisions under managerial entrenchment, journal of banking and finance, 25 (october), 10, 1857-95. deangelo, h., and masulis, r.w., (1980). optimal capital structure under corporate and personal taxation, journal of financial economics, 8, 3-27. dehnert, j., (1993). a study of rights issues of equity: a theoretical and empirical analysis, phd thesis, the australian graduate school of management, university of new south wales, sydney, new south wales. dhatt, m. s., kim, y. h., and mukherji, s., (1996). seasoned equity issues: the korean experience”, pacific-basin finance journal, 4, 31-43. dierkens, n., (1991). information asymmetry and equity issues, journal of financial and quantitative analysis, 26 (june), 2, 181-99. fama, e. f., and miller, m. h., (1972). the theory of finance, dryden press, hinsdale, ill. galai, d., and masulis, r. w., (1976). the option pricing model and the risk factor of stock, journal of financial economics 3, 1/2, 53-81. goh, j., gombola, m. j., lee, h. w. l., and liu, f.y., (1999). private placement of common equity and earnings expectation”, the financial review, 34, 3, 1-19. green, r.c., (1984). investment incentives, debt, and warrants, journal of financial economics 13, 1, 115-36. harris, m., and raviv, a., (1991). the theory of capital structure, journal of finance 46, 1, 394-355. hertzel, m., and rees, l., (1998). earnings and risk changes around private placements of equity, auditing and finance 13, 1, 21-35. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 85 hertzel, m. g., and smith, r. l., (1993). market discounts and shareholder gains for placing equity privately”, journal of finance 48, 2, 459-85. howe, j.s., and shilling, j.d., (1988). capital structure theory and reit security offerings”, the journal of finance 43 (september), 4, 983-93. howton, s.d., howton, s.w., and perfect, s.b., (1998). the market reaction to straight debt issues: the effects of free cash flow, journal of financial research, 21, 2, 219-28. jensen, m. c., (1986). agency costs of free cash flow, corporate finance, and takeovers, american economic review 76, 2, 323-29. jensen, m. c., and meckling, w. h., (1976). theory of the firm: managerial behavior, agency costs and ownership structure, journal of financial economics 3, 4, 305-60. johnson, s.a., (1995). dividend payout and the valuation effects of bond announcements”, journal of financial and quantitative analysis 30 (september), 3, 407-23. kabir, r., and roosenboom, p., (2003). can the stock market anticipate future operating performance? evidence from equity rights issues, journal of corporate finance 9, 93-113. kang, j.k., and stulz, r.m., (1996). how different is japanese corporate finance? an investigation of the information content of new security issues, review of financial studies 9, 1, 109-39. korajczyk, r.a., lucas, d., and macdonald, r.l., (1991). the effect of information releases on the prices and timing of equity issues, review of financial studies 14, 685-708. korteweg, a. g., and renneboog, l., (2003). the choice between rights-preserving issue methods. regulatory and financial aspects of issuing seasoned equity in the uk, working paper series, graduate school of business and tilburg university department of finance, university of chicago, chicago. lee, h. w., and kocher, c., (2001). firm characteristics and seasoned equity issuance method: private placement versus public offering, journal of applied business research, 17, 3, 23-36. leland, h. e., and pyle, d. h., (1977). informational asymmetries, financial structure, and financial intermediation, journal of finance, 32, 2, 371-87. lewis, c.m., rogalski, r.j., and seward, j.k., (1997). the information content of value line convertible bond rankings, journal of portfolio management 24 (fall), 1, 42-52. ---------(1999). is convertible debt a substitute for straight debt or for common equity? financial management 28 (autumn), 3, 5-27. ---------(2003). industry conditions, growth opportunities and market reactions to convertible debt financing decisions, journal of banking and finance 27 (january), 1, 153-81. martin-ugedo, j.f., (2003). equity rights issues in spain: floation costs and wealth effects, journal of business finance and accounting 30, 9/10, 1277-305. masulis, r., (1988). the debt/equity choice, ballinger publishing company, u.s.a. masulis, r., and korwar, a. n., (1986). seasoned equity offerings: an empirical investigation, journal of financial economics 15, 91-118. 86 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 61-86 mcwilliams, a., and siegel, d., (1997). event studies in management research: theoretical and empirical issues, academy of management journal 40, 3, 623-57. mikkelson, w. h., and partch, m. m., (1986). valuation effects of security offerings and the issuance process, journal of financial economics 15, 31-60. miles, j.a., and peterson, m.a., (2002). an analysis of nonunderwritten rights offers: the case of closed-end funds, the journal of financial research, the journal of financial research 25, 2, 187-206. miller, m. h., (1977). debt and taxes, journal of finance 32, 2, 261-75. miller, m. h., and rock, k., (1985). dividend policy under asymmetric information”, journal of finance 40, 4, 1031-51. modigliani, f., and miller, m. h., (1958). the cost of capital, corporation finance and the theory of investment, the american economic review 48, 3, 261-97. modigliani, f., and miller, m. h., (1963). corporate income taxes and the cost of capital: a correction, the american economic review 53, 433-43. myers, s. c., and majluf, n. s., (1984). corporate financing and investment decisions when firms have information that investors do not have, journal of financial economics, 18, 187-221. ross, s., (1977). the determination of financial structure: the incentive signaling approach, bell journal of economics 8, 1, 23-40. slovin, m. b., sushka, m. e., and lai, k. w. l., (2000). alternative flotation methods, adverse selection, and ownership structure: evidence from seasoned equity issuance in the u.k, journal of financial economics 57, 2, 157-90. smith, a., (1776). an inquiry into the nature and causes of the wealth of nations, dublin, u.s. smith, c. w. j., and warner, j. b., (1979). on financial contracting: an analysis of bond covenants, journal of financial economics 7, 2, 117-61. stein, j.c., (1992). convertible bonds as backdoor equity financing, journal of financial economics 32 (august), 32, 1. strong, n., (1992). modelling abnormal returns: a review article, journal of business finance and accounting 19, 4, 533-53. stulz, r., (1990). managerial discretion and optimal financing policies, journal of financial economics 26, 3-27. suchard, j.a., (1997). valuation effects of rights issues of convertible debt, working paper, school of banking and finance, university of new south wales. szewczyk, s. h., and varma, r., (1991). raising capital with private placements of debt”, journal of financial research,14, 1, 1-13. tan, r. s. k., chng, p. l., and tong, y. h., (2002). private placements and rights issues in singapore, pacific-basin finance journal 10, 29-54. wruck, k. h., (1989). equity ownership concentration and firm value: evidence from private equity financings, journal of financial economics 23, 1, 328. wu, x., and wang, z., (2002) vol. 2003 social science research network electronic paper collection. international journal of banking and finance 3-1-2008 the information conveyed by debt and equity announcements in australia oraluck arsiraphongphisit recommended citation contractual structures and payoff patterns of sukå«k securities contractual structures and payoff patterns of sukūk securities meysam safari segi university, malaysia _______________________________________________________________ abstract this paper is about the six different sukūk securities, which originated in 1990s, and are now traded in some 11 markets as the new debt-like securities classed under islamic finance. the outstanding value of these contracts is estimated to be us $850 billion. this paper proposes a classification for sukūk contracts as pure debt, equity-based, and asset-backed based on the intrinsic nature and purpose of fund-raising. this classification has more practical use compared to existing classifications. further, the contract peculiarities of the six instruments (mudārabah, mushārakah, murābahah, ijārah, salam, and istisnāh) are carefully specified for the first time. to start a discussion on how the economic behavior may be modeled for theory building, the potential cash flow pattern of each type of sukūk contracts is specified. the paper aims to contribute to advanced studies by specifying the basic behavioral characteristics. key words: bond-like socially responsible funding; sukūk certificates; people; planet before profits; islamic finance; special purpose company; asset-backed debt contract jel classification: g12, z12 ________________________________________________________________ 1. introduction sukūk is the plural form of sakk, which in arabic means legal instrument, deed, or check.1 it was used in pre-islamic era as a withdrawal certificate (a form of check) on deposits in financial firms or authorized financial merchants, much like the modern day cheques. later, these certificates became instrument for trading as bills; and then, it became debt instruments traded among willing holders. although it is reported that the turkish empire used sukūk to finance re-development of infrastructures after the devastating wars during 11th-13th centuries, the modern version of sukūk came into market only in 1990 in malaysia 1 during the 3rd century ad, financial firms in persia (currently known as iran) and other territories in the persian sassanid dynasty issued letters of credit known as “chak” (kharazmi, 1895). in post-islamic arabic documents this word has been transformed into “ṣakk” (floor, 1990). ijbf safari: contractual structures and payoff patterns of suk?k securities 68 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 with a rm125 million (equivalent to usd33 million) issued by a private firm.2 publicly traded issues saw its debut in 2000: the market has now six different sukūk, each designed for a different funding need of private and public issuers. the market is estimated to have an outstanding value of about us$850 billion as of 2011, and is growing at about 15-20 percent a year, mostly in islamic countries of which malaysia has two-thirds of the value (see ariff, iqbal, & shamsher, 2012). the rest of the paper is organized into six sections. in the next section, readers will find a description of the economic-cum-financial behavior resulting from the use of dramatically different principles in originating, issuing, trading and payoff patterns of these new market participants. section 3 discusses about various types of sukūk and provides a classification in this regard. sections 4 reviews each type of ṣukūk in some detail. the paper ends with a conclusion on section 5. 2. sukūk, new instruments resulting in new behavior currently, sukūk are considered as islamic funding certificates, although similar instruments existed prior to seventh century. the contract allows businesses to borrow funds in a manner compliant with islamic principles (i.e., sharī’ah). compliance with important ethical and societal restrictions are required for approval by a special ethics body at the regulatory institutions: sukūk instruments promise to pay out of profits of the funding at the business, and not as a guaranteed interest payment; funds are prohibited to be used for a number of purposes considered anti-social (gambling, prostitution, drugs, cigarettes, alcohol etc.); assets of borrower are transferred to a special purpose company which is created and controlled by the lenders. hence, sukūk is somewhat like socially responsible investment firms that put people and planet before profits.3 sukūk structures are designed based on the purpose and conditions of financing needs, and are not (as in conventional borrowing) general purpose borrowing. therefore, sukūk is a financial instrument with complex characteristics. it may even have some features of common equity (in which case it is termed as mushārakah, which is share-like funding effort but with a finite period over which the sukūk holders would receive their money back with a share of profits. another more common sukūk is mudārabah, a form of 2 for further reading refer to (adam & thomas, 2004; jalil, 2005; dusuki, 2009; vishwanath & azmi, 2009; shaikh, 2010; wan abdullah, roudaki, & clark, 2010) 3 over the last 45 years there has been slow growth of mutual and private funds that subscribe to socially responsible investments and they put people and planet before profit as their motto. there are about 470 of such funds managing about us$400 billion. the sukūk contracts likewise prohibit investments in certain goods/areas. the difference between the two is that the use of funds from sukūk is priced using profit-sharing or some variations of it, and are not based on paying a pre-agreed fixed interest irrespective of the outcome of the use of funds. in this manner, a single behavior is encouraged, namely that the investors and the firm agree to share in the risk of the investments (hence a profit share) while the firm has to give ownership of part of the assets to the lender, which restricts unlimited borrowing by the businesses. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 69 borrowing closer to common debt but with periodic profit shares specified ahead of contracts. another form is asset-backed loans (ijārah, murābahah, or others). for instance, one may issue an ijārah sukūk, which is a lease agreement while another business may issue istisnāh sukūk as project finance. therefore, there are various forms of sukūk contracts with peculiar risk factors and payback schemes all tied to the purpose for which the funding is raised. in europe, high net worth individuals are brought together by financial institutions to organize private lending via sukūk contracts for long term investments in energy sector for example, on the basis of profit shares, which yield higher returns to investors. similar contracts are drawn up in several financial centers which have set up regulatory framework to tap into the high net worth individuals especially in the persian gulf region to raise much needed capital for private sector. the outstanding value of private sector non-traded sukūk is unknown, although press reports indicate that the size of this market is as big as the public-traded market in 11 countries. accounting and auditing organization for islamic financial institutions (aaoifi, 2004) defined sukūk as “certificates of equal value representing, after closing subscription, receipt of the value of the certificates and putting it to use as planned, common title to shares and rights in tangible assets, usufructs and services, or equity of a given project or equity of a special investment activity”. international islamic financial market (iifm) (2010) defined sukūk as a commercial paper that provides an investor with ownership in an underlying asset. it is asset-backed trust certificates evidencing ownership of an asset or its usufruct. it has a stable income and complies with the principle of sharī”ah. unlike conventional bonds, sukūk needs to have an underlying tangible asset transaction either in ownership or in a master lease agreement. islamic financial services board (ifsb) (2009) definition of sukūk is “sukūk (plural of sakk), frequently referred to as “islamic bonds”, are certificates with each sakk representing a proportional undivided ownership right in tangible assets, or a pool of predominantly tangible assets, or a business venture (such as a murārabah). these assets may be in a specific project or investment activity in accordance with sharī’ah rules and principles”. ifsb definition of sukūk differs from conventional interest-based securities (i.e., bonds) in a number of ways, including: • the funds raised through the issuance of sukūk should be applied to investment in specified assets rather than for general unspecified purposes. this implies that identifiable assets should provide the basis for sukūk. • since the sukūk are backed by real underlying assets transferred with lenders owning pro-rata shares, income generated by the assets must be related to the purpose for which the funding is used. • the sukūk certificate represents a proportionate ownership right over the assets in which the funds are being invested. the ownership rights are transferred, for a fixed period ending with the maturity date of the sukūk, from the original owner (the originator) to the sukūk holders. safari: contractual structures and payoff patterns of suk?k securities 70 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 local authorities of islamic countries such as qatar financial center (qfc)4 require the authorized firms to comply with international definition, while some others define sukūk in their own way, as in malaysia. securities commission of malaysia,5 which issued the ‘guidelines on the offering of islamic securities’, defines sukūk as “a document or certificate which represents the value of an asset”. liquidity management center (lmc) in bahrain defines sukūk as “a certificate of equal value representing undivided shares in ownership of tangible assets, usufruct, and services or (in the ownership of) the asset of particular projects or investment activity”.6 finally, some authors such as ariff, safari, and shamsher (2012) define sukūk contracts as “a funding (debt) arrangement agreed to between a party providing the funds (investor) and the counterparty (a government or a firm or an individual) borrowing the funds for the purposes of using the funds to engage only in permissible economic production/services.” as these definitions imply, there is no time-horizon specification for sukūk. in other words, sukūk may be issued for long term financing as well as short term financing. short term sukūk with maturities as short as one month are issued by various issuers. some of the earliest short term sukūk are the 30and 91-days salam sukūk issued by bahrain monetary agency (bma)7 in 2001. bank negara malaysia (bnm)8 also issues one-year short-term ijārah sukūk. since 1994, bnm has introduced islamic interbank money market (iimm) sukūk as a short term intermediary to provide a ready source of short-term investment outlets based on islamic principles. although iimm works under the concept of debt-like murārabah financing, they have not issued sukūk for this specific purpose. 3. classification of different types of sukūk in the course of conducting their businesses, companies require funds to initiate, operate, promote, and expand their businesses (same applies to conventional firms). this need is addressed either from internal or external sources. external financing could be obtained by issuing new equity-share certificates (infinite-period mushārakah) as a financial instrument in the same manner as conventional counterparts do through issuing common stocks. this also can be achieved by issuing funding certificates (sukūk) as an islamic finite-period “loan” for productive uses of funds. the distinguishing factor between sukūk and conventional financing lies in the borrowing process. sukūk should not deal 4 qatar financial center (qfc) in the islamic finance rulebook (ifsi) under section ifsi 6.2.2 requires the authorized firm to comply with the aaoifi standards. 5 securities commission malaysia website: http://www.sc.com.my/main.asp?pageid=448, accessed on 17/07/2012 6 obtained from lmc documents at http://www.lmcbahrain.com/pdf/about-ṣukūk.pdf, accessed on 17/07/2012 7 in 2006, renamed as central bank of bahrain (cbb). 8 central bank of malaysia. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 71 in payment or receipt of ribā. instead they use profit shares. the debt contract specifies the sharing of risk of the business with a promise of reward that is not fixed. it is variable and depends on the amount of profits. sukūk offer many ways to acquire funds via debt. the most important requirement is that the pure debt certificates are not tradable at any price but the face value of the debt while the other two forms of debt to be described later could be traded. mudārabah and mushārakah are called the primary modes of sukūk (chapra, 1998) because they are based on the profit-loss sharing (pls). (however, due to severe asymmetric information problem of these contracts, these two forms of islamic financial instruments funds less than 10 per cent of all funds.) in pls methods, the outcome of investment is completely based on the performance of the project and hence, is not predetermined. the predetermined factors are the duration of investment and the ownership in special purpose firm or distribution ratio of profit of the project. in contrast to pls methods, in secondary methods of financing, the outcome of investment for investors is to some extent predetermined and is not fully tied to the performance of the investment project. for instance, in an ijārah lease contract, the investor will benefit from a predetermined rental (i.e., lease) fees for a certain period of time. however in the secondary modes of finance, the principal investment should not be guaranteed by the issuer. present literature provides two types of classification for sukūk. first classification is based on issuer type: sovereign sukūk (government, quasigovernment) and corporate sukūk. second classification is based on contract forms, including but not limited to the six types already mentioned in this paper. the first two (i.e., murārabah and mushārakah) are sometimes considered as major or primary modes of finance because they are based on pls method whereby the profit or loss arising from the project financed by the sukūk issue is shared among the parties according to a pre-agreed ratio. another classification for sukūk is based on the underlying nature of the financing contract and its characteristics i.e., pure debt, equity-based, or assetbacked sukūk. as explained later, there are various contractual frameworks for financing in the jurisprudence applying to these. these contracts include mudārabah (silent partnership or capital trust) for pure-debt sukūk, mushārakah (full partnership or joint venture) for equity-based sukūk, and murābahah (procurement financing), ijārah (leasing), salam (advance payment), and istisnāh (financing manufacturing projects) for asset-backed sukūk. figure 1 depicts these modes of financing. each of the main categories may then, if necessary, be subcategorized further to accommodate various types of contracts. to accommodate various types of asset-backed sukūk contracts, and based on their intrinsic purpose, three more subcategories are suggested as property-backed (semi-collateralized) sukūk, advance or deferred payment sukūk, and project financing sukūk. these classifications are summarized later. safari: contractual structures and payoff patterns of suk?k securities 72 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 figure 1: sukūk based on their underlying contractual structure 4. sukūk types 4.1 mudārabah mudārabah sukūk are based on the murārabah contract. mudārabah, or “partnership in profit”, contract is one of the premier financing methods dating back to pre-islamic era. abdul-gafoor (2006) documented that muhammad (pbuh) 9 used murārabah with khadijah, a rich woman from mekkah,10 about fifteen years prior to the dawn of islam. mudārabah is a contract between at least two parties: rabb al-māl (pl. arbāb al-māl) meaning capital owner(s), who provides funds and mudārib or entrepreneur who brings the entrepreneurship and management, in order to perform a specific activity or venture. the rabb al-māl or capital owner could be of any type ranging from individual investors, investment companies, or banks. at the maturity of the contract or at some certain predetermined time, generated profits from the venture are shared between contracting parties according to a pre-agreed ratio. in case of loss, each party should bear the loss of his contribution to the venture. hence, the capital owner should bear all financial losses and the entrepreneur should bear the operating losses such as time and effort. however, in case of negligence or misconduct by the entrepreneur, then, as ibn qudama states “the loss is a result of a misuse or violation of the conditions of the contract on the part of the working partner, then he alone will be liable to cover it” (quoted in daryanani, 2008). capital owner is not allowed to have a management role in the mudārabah venture contract. mudārib is considered a trustee as well as an agent of the rabb al-māl. mudārib, as a trustee, 9 short form for “may allah’s peace be upon him”. 10 khadijah bint khuwaylid (555-619 ad) later became the first wife of prophet muhammad (pbuh). pure debt  mudārabah equity-based •mushārakah asset-backed •murābahah •ijārah •salam •istisnāh international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 73 figure 2: classification of various sukūk contracts based on their financial characteristics is responsible for possible losses due to the willful negligence. moreover, as an agent, mudārib is required to use and manage the capital in a way that generates the maximum profit for the venture (chapra, 1998). furthermore, similar to equity holders, murārabah investors’ profit is in line and proportionate to the performance of the firm. ebrahim (1999) illustrates murārabah contract as a combination which has features of both equity and debt. ijārah moṣūfah bi dhimmah diminishing mushārakah ijārah (leasing) manfa’ah ijārah istiṣnāh (projek financing) manfa’ah ijārah moṣūfah bi dhimmah murābahah (procurement financing) salam (advance payment) muzara’ah (agreculture) musāqah (fruit orchid) mushārakah muqārasah (farm land with trees/crops) wākalah (agent) mushārakah (fullpartnership, joint venture) property backed advance or differed payment project financing muḍarabah (silent partnership/capital trust) equity based asset baked pure debt ṣukūk milkiyat al khadamāt safari: contractual structures and payoff patterns of suk?k securities 74 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 however, they do not benefit from all aspects of shareholders, like capital gains, and do not have some of their rights such as attending or voting at annual general meeting. on the other hand, in case of bankruptcy, sukūk holders are in a higher position (have preferred claims) to the equity of shareholders (wilson, 2004). although mudārabah may be applied in various economic activities, the majority of islamic jurists and scholars hold the view that mudārabah contracts are most suitable for trading activities. in practice, however, the implication of mudārabah contracts is limited due to the operational difficulties and business ethics constraints. the inefficient tax system, high rate of illiteracy, inadequate accounting standards and the practice of keeping a double set of accounts on the part of the majority of business people are major constraints on the practical implementation of the pls system in muslim countries (khan, 2003). in order to issue a mudārabah sukūk a special purpose vehicle (spv) company is set up. this spv will act as rabb al-mal, or capital owner, and the originator will act as mudārib. alvi et al. (2010) described the process of payback of mudārabah sukūk and highlighted that the proceeds of the issue collected by the spv from the sukūk investors are applied as the capital of the mudārabah which the mudārib will manage for a share in the profits, the profit sharing ratio being specified at the outset. while there should not be a predetermined rate of return in a mudārabah contract, the sukūk issued until early 2008 have been designed in a way to ensure that sukūk holders receive the so-called indicative rate of return announced at the inception of the issue. alvi et al. (2010) highlighted that they achieved this by including clauses in the mudārabah agreement that specified a ‘maximum’ rate of return. any profit to be generated above that rate of return would be directed to a reserve account, which could be used to cover any shortfall in future years. in case of insufficient profits as well as insufficient funds in the reserve account, the issuing company would be required to provide sharī’ah-compliant funding to meet the shortfall and make it up to the indicative rate of return, in effect guaranteeing the rate of return independent of the actual profit generated. usmani (2007), chairman of the aaoifi sharī’ah council, has ruled against the practice of guaranteeing the indicative rate of return. as mentioned before, mudārabah sukūk are pure debt. in other words, by nature, it is mere case of borrowing money, thus, the mudārabah sukūk securities are not tradable. however, alvi et al. (2010) highlighted that the mudārabah sukūk, in practice, are structured in a way to be tradable. mudārabah sukūk are tradable and negotiable if the mudārabah assets do not comprise entirely of the sukūk proceeds (in which case it will be all liquid assets and cannot be traded). in most mudārabah sukūk, there is a combination of tangible assets and sukūk proceeds, plus the mudārib is allowed to mingle his own assets with those of the mudārabah, hence mostly meeting the sharī’ah-compliance requirement of having more than 51% of the assets in tangible form for tradability and negotiability. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 75 the simplest form of mudārabah sukūk is form of a zeroreward11mudārabah, which does not provide any form of reward before maturity. the only payback that sukūk holders receive is the undetermined maturity payback. the pay-off structure of zero-reward mudārabah sukūk is depicted in figure 3. in this diagram, the maturity payment (i`) is not predetermined and is based on the performance of the venture. figure 3: cash flows pattern of zero-reward mudārabah sukūk other possible payoff structure of the mudārabah sukūk contract is when issuer pays periodical rewards to sukūk holders. however, based on the characteristics of mudārabah contract, amount of reward cash flows should not be predetermined and hence, must be based on the performance of the venture in that period. the sharing ratio is, however predetermined. similar to zero-reward mudārabah sukūk, the maturity payment is also undetermined and based on the venture’s performance. the pay-off structure of such general mudārabah sukūk is depicted in figure 4. figure 4: cash flows pattern of reward paying mudārabah sukūk 4.2 mushārakah 11 we used the term “reward” as a substitution for interest-based term “coupon”. reward is the periodical payment that the sukūk holder is entitled to receive. safari: contractual structures and payoff patterns of suk?k securities 76 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 mushārakah is the next type. iqbal and molyneux (2005) defined mushārakah as “an arrangement where two or more parties establish a joint commercial enterprise and all contribute capital as well as labor and management as a general rule”. in contrast to mudārabah contract, mushārakah investors have the right to participate in management of the business partnership, however, this right is entrusted to each investor (shinsuke, 2007). it could be derived that mushārakah contract may require establishment of a new partnership or company, where mushārakah contract parties are the participants and owners (wilson, 2004). mushārakah sukūk could be issued based on such financing concept. mushārakah type of equity finance demands that both profit ratio and length of the joint venture agreement is decided in advance. similar to mudārabah, loss is shared in proportion to the capital contribution unless the loss is proven to be due to negligence of one party (daryanani, 2008). therefore, all profits and losses generated from the mushārakah are shared among the parties on the basis of the pre-agreed ratio. as a result, mushārakah is basically suitable for financing private or public companies and projects and is also practiced by islamic banks, where it is typically performed through joint ventures between banks and business firms for a certain operation (gait & worthington, 2007). mushārakah, due to its nature and advantages in providing equal (but proportionate) benefits for all parties, has support of all islamic scholars and is valid under sharī’ah principles. however, el-gamal (2000) suggests that parties to mushārakah usually need the help of legal expert to ensure that any potential ribā or gharar is carefully avoided. on the discussion about mushārakah contracts, chapra (1998, p. 7) concluded that “the only requirement of the sharī’ah would be justice, which would imply that the proportional shares of partners in profit must reflect the contribution made to the business by their capital, skill, time, management ability, goodwill and contacts. anything otherwise would not only shatter one of the most important pillars of the islamic value system, but also lead to dissatisfaction and conflict among the partners and destabilize the partnership. the losses must, however, be shared in proportion to capital contribution and the stipulation of any other proportion would be ultra vires and unenforceable.” lewis and algaoud (2001) suggested two ways to perform a mushārakah contact. however, both types are based on the same general concept of mushārakah, where its parties (capital owner and entrepreneur) are ensured an equitable share in the profit or loss on pre-agreed terms. the difference lies in the pre-agreed sharing ratio. in the first method, this pre-agreed ratio is fixed and non-changeable for the whole period of the contract while in the second type, the ratio is declining. the diminishing, or declining mushārakah contract is preferred by some financiers since it allows them to release their capital from the investment by reducing its equity share each year and receiving periodic profits based on the remaining share balance. on the other hand, the equity share of the entrepreneur increases over time to the extent that he or she becomes the sole owner of the firm. there are varieties of sukūk based on the mushārakah contracts. sukūk based on diminishing mushārakah are gaining momentum since they enable international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 77 islamic banks or sharī’ah-compliant investment companies to provide up-front investment funding to the issuer. in this regard, both parties establish a special purpose vehicle (spv) to administer the sukūk. in order to issue a diminishing mushārakah sukūk, the issuer transfers the ownership of an asset to the spv to enter the partnership agreement. on the other hand, investors enter the agreement by paying cash. therefore, both the investors and the issuer are equity partners in the spv. however, the investors share in the spv diminishes over time as the issuer pays installments to investors to repurchase their respective share in the asset. these installment payments plus the issuer’s rental payments for use of asset (i.e., asset’s generated income) constitute the cash flow stream for sukūkholders. in fixed-ratio mushārakah sukūk, the cash flow stream for the sukūkholder is only constituted from the income generated from the asset and not the installment part. the structure of diminishing mushārakah sukūk is depicted in the figure 5. figure 5: diminishing mushārakah sukūk structure flexibility in payments schedule and amounts has made diminishing mushārakah sukūk more convenient to use. however, it should be highlighted that all arrangements should be agreed upon ex-ante by all parties to the spv. the payments are usually monthly or quarterly, but not necessarily in equal amounts (wilson, 2008). smaller installments could be made during the initial period of the sukūk, with most of the asset value or spv capital remaining with the investors, but the amount of the installment payments could increase in a linear fashion, or according to some predetermined formula. as the issuer’s share in the asset increases through the buy-back process, the periodical rental safari: contractual structures and payoff patterns of suk?k securities 78 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 might be expected to decrease due to the decline in remaining share. however, this does not necessarily have to be the case, especially if there is capital appreciation in the value of the asset. in other words when installment and rental payments are aggregated, they might be constant, diminishing or increasing over time, provided both parties agree to the formula used and the documentation is transparent. the simplest payoff structure of mushārakah sukūk contract is the form of zero-reward fixed mushārakah sukūk which does not provide any form of reward during its tenure. the only cash flow that sukūk holders receive is the undetermined maturity payback. its cash flow structure is the same as the zeroreward mudārabah sukūk and is depicted in the figure 6. the amount of maturity payment (i`) is undetermined and should be based on the venture’s performance. figure 6: cash flows pattern of zero-reward mushārakah sukūk another possible payoff structure of the fixed mushārakah sukūk contract is when issuer pays periodical rewards to sukūk holders. however, based on the characteristics of mushārakah contract, amount of cash flows should not be predetermined and hence, must be based on the performance of the venture. similar to zero-reward mushārakah sukūk, the maturity payment (i.e., i’ in the diagram) is also undetermined and based on the venture’s performance. the pay-off structure of such fixed mushārakah sukūk is depicted in figure 7. in this figure, r i is the periodical reward payment distributed among sukūk holders, which is based on the actual performance of the venture in each period. figure 7: cash flows pattern of fixed mushārakah sukūk with floating-reward payments international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 79 diminishing mushārakah ṣukūk contract governs the mushārakah contract where the profit ratio of the capital owner declines over time, eventually reaching zero at the maturity time. the periodical payments, then, constitute from two parts; reward and portion of original equity capital. the periodical payment, in this case, could be engineered in various ways. first possible cash flow pattern may be the case where the periodical payment is fixed at a certain amount. the pay-off structure of such diminishing mushārakah sukūk is depicted in figure 8. in this figure, r i is the periodical payment distributed among ṣukūk holders, which is sum of reward amount (based on the actual performance of the venture in that period) and the portion of original capital paid back to the capital owners. figure 8: cash flows pattern of diminishing mushārakah ṣukūk with fixedamount periodical payments periodical payments in diminishing mushārakah sukūk contract could be designed in a way that the periodical payment amounts follow a constant growth model over its tenure. the pay-off structure of such diminishing mushārakah sukūk is depicted in figure 9. in this figure, r i is the periodical payment distributed among sukūk holders, which is sum of reward amount (based on the actual performance of the venture in that period) and the portion of original capital paid back to the capital owners. figure 9: cash flows pattern of diminishing mushārakah sukūk with growingamount periodical payments safari: contractual structures and payoff patterns of suk?k securities 80 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 4.3 murābahah murābabah is the third type used for sukūk. murābahah contracts govern the process of buying, purchasing, or importing an item by one party, mainly an islamic bank, and then reselling it to other party. the markup is the bank’s profit for funding this transaction (gait & worthington, 2007). islamic banks’ benefit is generated from the markup on the cost of purchase of the goods which is agreed upon in advance (metwally, 2006; shinsuke, 2007). murābahah contracts, which are based on a cost-plus basis, are especially used for foreign trade and working capital financing for circumstances in which banks purchase raw materials, goods or equipment and sell them to the customer (lewis, 2007). daryanani (2008) highlighted the fact that the ownership of the murābahah asset remains with the islamic bank until all payments are settled, in contrast to the conventional system where the ownership of the assets is immediately transferred to the buyer. therefore, from modern finance point of view, it is equivalent to an asset-backed risky loan and is a popular substitute for interestbased conventional trade financing in islamic banks. to initiate the murābahah, the customer provides the detailed specification and prices of the required goods to be purchased or imported to the bank. having received the application documents, the islamic bank, analyzes and collects the required information from the vendors, especially on the price and payment conditions, then, the bank and murābahah-applier agree on the terms of the deal and finally, the bank purchases the goods or commodities and resell them to the customer. in order to conduct a murābahah contract, following requirements should be met (obaidullah, 2005): i) goods and commodities mentioned in the contract must be classified, clearly identified according to commonly accepted standards and must exist at the time of sale; ii) islamic financier must hold the ownership of the goods at the time of sale to the buyer; iii) the cost in terms of net purchase price and the markup must be known at the time of sale and be declared to the customer. in other words, if the bank succeeds in price reduction by obtaining a discount from the vendor; this discount should be shared with the customer; iv) schedule for delivery of goods, as well as payments must be specified in the contract and cannot be changed in the life of the contract. v) murābahah contract must be based on sale of commodity or tangible goods and cannot be based on sale of money. khan and bhatti (2008) claimed that the murābahah contracts constitute the majority (54 percent) of the total financing and investment portfolios of ten largest islamic banks during the 2004-2006 period. however, murābahah was more popular a decade before, 1994-1996, by 65 percent (iqbal, 1998). international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 81 murābahah contracts typically have short term maturities. hence, islamic banks basically use murābahah for short-term investment and liquidity management. however, murābahah contracts have low returns, leading to an inefficient use of funds and lower rate of return for islamic banks (abdul majid, 2003). the way murābahah contracts conducted in practice contradicts with sharī’ah principles because islamic banks transfer all costs of insurance of murābahah goods against possible risks of damage, destruction, and theft to the murābahah customer (bashir, 1999). islamic banks also perform some other actions that are not sharī’ah compliant such as benchmarking interest rate to fix the returns on murābahah, assigning higher markups for murābahah contracts with longer periods, charging fines to customers who delay installment payments, and recovering losses from customers who breach their promises to buy the murābahah goods (homoud, 1994). murābahah contract, by itself, is only a contract of appointing bank as an agent, on behalf of customer, in the process of purchase of goods and do not necessary involve in financing the purchase (obaidullah, 2005). in other words, in the basic murābahah contract, customer should pay the cost of goods and the profit margin immediately after the delivery of goods. however, the customer can pay by deferred installments (bay bi thaman al-ājal) or a deferred lump sum without an increase over the original value (bay bi thaman al-ājal). with a murābahah sukūk an islamic bank securitizes its trading transactions with a proportion of the fixed markup providing the return to the sukūk investor, and the bank using the repayment from its trading client to repay the sukūk holder on termination of the contract (wilson, 2008). bay mu’ajjalmurābahah sukūk is the case where issuer pays back the total amount borrowed at a certain time in a predetermined lump sum payment. payback structure of bay mu’ajjal murābahah sukūk would only have a maturity payment (i`) that constitutes the original amount plus the predetermined markup as shown in figure 10. figure 10: cash flows pattern of bay mu’ajjalmurābahah sukūk bay bi thaman al-ājal sukūk is the case where issuer pays back the total amount borrowed in deferred installments. payback structure of bay bi thaman al-ājal sukūk can be in two types. in first type, periodical payments are equal and fixed in amount and maturity payment (i`) is predetermined. cash flows of this form of bay bi thaman al-ājal sukūk is depicted in figure 11. safari: contractual structures and payoff patterns of suk?k securities 82 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 another type for payback structure of bay bi thaman al-ājal sukūk is where the periodical payments follow a growth model over its tenure. the growth pattern, as well as the initial periodical payment amount, should be predetermined. figure 11: cash flows pattern of bay bi thaman al-ājal sukūk with fixed-amount reward as the general rule of murābahah sukūk, the maturity payment is also known in advance. the cash flows diagram of this payback structure is depicted in figure 12. figure 12: cash flows pattern of bay bi thaman al-ājal sukūk with growingreward 4.4 ijārah ijārah or lease contract is another form based on rent-like payments. ijārah, which means “to give something on rent” (lewis & algaoud, 2001), is the reward or recompense that proceeds from a rental contract between two parties, where the lessor (the owner of the asset) leases capital asset to the lessee (the user of the asset) (gait & worthington, 2007). there is a tendency toward lease financing (ijārah) in islamic banking sector, since it promises higher yields than trade finance (murābahah) and also has longer financing horizons, which is an important feature for business investments (daryanani, 2008). international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 83 to be a compliant contract, the ijārah contract should satisfy some conditions. the primary requirement is that the lessor must be a real owner and in possession of the asset to be leased under contract. as a result, the lessor should solely bear all risks and uncertainties associated to the asset and be responsible for all damage, repair, insurance, and depreciation of the asset (khan and bhatti, 2008). it could be inferred that charging rental payment is not allowed until the lessee actually receives the possession of the asset and shall pay the rental only as long as it is in usable condition. moreover, in case of manufacturing defects which are beyond the lessee’s control, the lessor is responsible. however, the lessee is responsible for the proper upkeep and maintenance of the leased asset. the intention of posing such restriction in ijārah contract by sharī’ah is to protect both parties to the contract by reducing the uncertainty and ambiguity from the agreement (wilson, 2004). in addition to that, both lessor and lessee should be clear on purpose of ijārah and the usage of assets, moreover, the ijārah purpose must comply with sharī’ah (al-omar & abdel-haq, 1996). there are two forms of leasing contracts, or ijārah, in islamic finance. ijārah, or direct leasing contract, is the case where the lessee uses the capital asset owned by the lessor, with his/her permission, for a specific period of time for a monthly or annually rental fee. the owner bears the ownership title for the whole contract period and should perform the ownership responsibilities such as insurance and maintenance (zaher & hassan, 2001). in ijārah contract possession in asset should be transferred back to the owner after the contract matures. in other words, in pure ijārah contracts, there is no option to transfer the ownership of the asset at maturity. ijārah wa iqtinā’, or hire purchase, is the case of contract where the basic intention is transferring the ownership after completing the leasing period. ijārah wa iqtinā’ is popularly practiced when islamic bank purchases equipment or some other capital asset based on the request of an individual or institutional customer and then rents it to the customer for a certain fixed rent. on the other hand, the customer promises to purchase the equipment or asset within a specified period to transfer the ownership from the islamic bank to the customer (al-jarhi & iqbal, 2001). however, it should be noted that the lease contract is completely separate and independent from the contract of purchase of residuals, which has to be valued on a market-basis and cannot be fixed in advance. the purchase contract should be an optional, non-binding contract because the quality and the market price of the asset at the end of the lease period are unclear (chapra, 1998). one other approach is the case where the ownership is gradually transferred to the customer. in this case, and in addition to the regular rental payment, the customer shall pay installments of the value of the asset in order to reduce the ownership share of the lessor in the asset until the ownership is fully transferred to the lessee (metwally, 2006). ijārah wa iqtinā’, having a strong support from sharī’ah scholars, is widely used in the real estate, retail, industry, and manufacturing sectors (iqbal, 1998). ijārah sukūk is based on the ijārah contract. in order to issue ijārah sukūk, the originator, who primarily owns the assets, sells the assets to a special safari: contractual structures and payoff patterns of suk?k securities 84 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 purpose vehicle (spv), which is typically a company in an offshore tax-free site. then the spv leases back the assets to the issuer at a specific predetermined rental fee and then the spv securitize the ownership in the assets by issuing ṣukūk certificates to the public investors (lewis, 2007). these sukūk certificates represent an undividable share in the ownership of the assets which entitle the sukūk-holders to distribution of the rental payments on the underlying assets. however, the rental payment could be fixed or floating for the whole period, dependent on the leasing contract between the spv and originator. since these sukūk certificates represent ownership in real assets, they could be traded in a secondary market. the role of the spv in conducting ijārah sukūk is the management of the ṣukūk cash flows, particularly receiving periodical rentals and installments from the originator and disbursing them to the ṣukūk-holders (aseambankers, 2005). thus, after maturation of sukūk, the spv no longer has a role and consequently will be ceased from existence. however, the ijārah ṣukūk is typically issued for periods longer than five years and could be considered as long term debt certificates. this may raise the issue of spv’s default risk, so, the investors typically receive a direct guarantee from the issuer of the spv obligations (wilson, 2008). this guarantee also includes the obligation by the issuer to repurchase the asset from the spv at the end of the ijārah contract at the original sale price. source: adapted from wilson (2008), abdul majid (2003), el-gamal (2007), bose and mcgee (2008) figure 13: ijārah sukūk structure international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 85 wilson (2008) suggested that due to its nature, spv does not have any of risks associated with banks, in other words, spv is bankruptcy remote. in other words, if the issuer faces the bankruptcy, the creditors to the issuer cannot claim the assets held by the spv or otherwise interfere with the rights of the sukūkholders with respect to the underlying assets (gurgey & keki, 2008). as a result, spv would be attractive to both issuers and investors, and this may justify the relatively high legal establishment costs. kamali (2007) claimed that due to the fixed and predetermined nature of rental cash flow, the ijārah sukūk-holders receive steady income that is even more risk averse than common stocks. however, he mentioned general market conditions, price movements of real assets, ability of the lessee to pay the rental or installments, maintenance, and insurance cost as sources of risks to the ijārah sukūk. he concluded that because of these risk factors, the expected return on some of ijārah sukūk may not be precisely predetermined and fixed. thus, the fixed rental may only represent a maximum that is subject to some possible deductions. the major criticism of ijārah sukūk is that the return is variable or floating in most cases. moreover, this variable rate, sometimes for simplification reasons, is mostly benchmarked or “pegged” to an interest-based index such as the london interbank offered rate (libor) for usd based sukūk and local rates for other currencies. usmani (2002) criticized this practice by associating ribā to this form of ijārah sukūk practice. sharī’ah scholars suggested the usage of other non-interest benchmarks for pricing and evaluation purposes. in order to overcome the ribā issue, government or sovereign sukūk could be assessed by macroeconomic indicators and corporate sukūk could be assessed based on the company performance indicators. ijārah sukūk can have various types of payback structure. in the simplest form, ijārah sukūk payback could be as fixed reward payments and an undetermined maturity payment. the formal ijārah contract does not have the option for parties to transfer the ownership of the asset at the end of the period. thus, at the end of an ijārah contract, the asset should be returned to the owner (i.e. capital owner or the spv). in order to transfer the ownership back to the issuer at the maturity, one should use ijārah wa iqtina’ contract. ijārah wa iqtinā’ sukūk is form of ijārah contract where the ownership of the asset will be transferred to lessee (i.e. issuer) at the maturity of the sukūk. however, the maturity payment is not determined at the issuance time of sukūk. the valuation of the asset in ijārah wa iqtinā’ sukūk should be conducted at the maturity time, when the market value of the asset is recognized and maturity payment (i’) is set to be equal to the market value at that point in time. such cash flow pattern is depicted in figure 14. in this diagram, r i is the periodical reward payments that is fixed and predetermined and i’ is the undetermined maturity payment. rewards in ijārah sukūk contracts may follow a growth model over its tenure to compensate the actual increase of the rental fees in the market during this period. this form of payback structure might be more practical in long term ijārah sukūk securities. the payback structure of this form of sukūk contract safari: contractual structures and payoff patterns of suk?k securities 86 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 figure 14: cash flows pattern of ijārah sukūk with fixed-amount periodical reward payment and undetermined maturity payment is illustrated in the figure 15. in this form of contract, the amount of maturity payment (i`) is undetermined and will be determined only at the maturity time based on the actual market value of the asset. the amount of reward payments (r i ) are predetermined and set to follow a growing pattern. figure 15: cash flows pattern of ijārah sukūk with growing-reward payments and undetermined maturity payment 4.5 salam salam contract refers to a sale where a seller promises to supply a specific commodity to a buyer at a future date in return for an advanced price, paid in full on the spot. sale of goods that are not currently under ownership and possession of the seller or do not exist at the contract time is generally prohibited because of extreme uncertainty (gharar). however, salam and istisnāh (manufacturing project finance) contracts are exceptions to this general ruling which facilitate financing process of agricultural and industrial projects under certain conditions. iqbal and molyneux (2005) defined salam as: “salam is a sale contract in which the price is paid in advance at the time of contracting against delivery of the purchased goods/services at a specified future date”. in order to be sharī’ah international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 87 compliant, a salam contract should comply with these following requirements (gait & worthington, 2007). i) the commodities and goods have not came into existence at the time of the contract. ii) the quality and quantity of goods must be known at the time of contract. iii) the delivery schedule and venue must be determined at the contracting time. iv) buyer must pay the entire price of goods in advance to the seller at the time of contract. therefore, salam contract is used when buyer pays to order a specific product to be arranged by seller and be delivered on a certain point in time. salam is an old type of contract which dates back to the pre-islamic era when it was widely used especially for agricultural produce (khan & bhatti, 2008). salam contract is approved by sunnah. ibn abbas narrated: “the messenger of allah came to madinah and found its inhabitants entering salam contracts (with the price paid in advance) in fruits for one, two, and three years. he said: whoever enters into a salam contract let him specify a known volume or weight, and a known term of deferment” (el-gamal, 2000). moreover, in another narration the holy prophet (pbuh) has said: “whoever enters into salaf12, should stipulate a determined weight and measurement, and a determined date of delivery” (almasri, 2003). salam contract has benefitted both buyer and seller. seller could finance working capital needs and buyer could benefit from the difference between purchase price and the commodity price at the delivery time, which tends to be higher. in the meantime, the purchaser can finance the advance payments by issuing certificates against the salam contract’s goods, at purchase price; and, since these certificates represent real assets, they could be sold to the public or be traded in the secondary market. the return for the buyers of these certificates is the difference in the commodity prices at the maturity time and the discounted price they have paid. in order to issue a salam sukūk, a not-for-profit special purpose vehicle (spv) should be created as a separate legal entity for the duration of the sukūk to administrate the flow of payments between issuer and investors as well as holding the title of the underlying asset (dommisse & kazi, 2005). to issue ṣukūk, the issuer should transfer the title to the assets to the spv, which in turn issues certificates of participation to the public investors. salam sukūk represent an undivided right to an interest in the asset. in order to obtain these certificates of participation the investors should make an advance payment which entitles them to a future payback of the investment plus a fixed pre-agreed mark-up (wilson, 2008). one of the famous examples of salam sukūk is the “bahrain sukūk alsalam” which is primarily designed to broaden the depth and liquidity of the bahrain’s market. this short-term treasury-bill-type sukūk has aluminum as 12 salaf has the same meaning and interpretation as of salam. safari: contractual structures and payoff patterns of suk?k securities 88 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 its underlying asset (abdul majid, 2003). at maturity time, the issuer which is “bahrain monetary agency” acts as a sukūk-holders’ agent and sells the commodity at a pre-agreed price that guarantees promised return rate to the sukūk-holder (el-gamal, 2007). as mentioned above, in salam contracts, capital owner pays in advance for a commodity to be delivered in future. the capital owner (i.e., sukūk holder) would benefit from the difference between amount paid in advance and the market price of the commodity at maturity. the maturity reward (i’), thus, is undetermined at the issuance. the payback structure of salam sukūk is depicted in figure 17. source: wilson (2008) figure 16: salam sukūk structure figure 17: cash flows pattern of salam sukūk 4.6 istisnāh istisnāh means manufacturing. commission to manufacture is the proper method of financing working capital in the manufacturing and construction sectors (elgamal, 2000). gait and worthington (2007) defined istisnāh as “a manufacturing contract which allows one party to obtain industrial goods with either an upfront international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 89 cash payment and deferred delivery or deferred payment and delivery”. istisnāh has the cost reduction benefit for the issuer in the term that all or part of the working capital is obtained from external resources and consequently the final product is manufactured at a lower cost (lewis & algaoud, 2001). as mentioned before, sales of goods that are not currently under ownership and possession of the seller is generally prohibited. however, like salam, istisnāh, is an exception to this general ruling, which refers to a contract whereby manufacturer-seller agrees to produce and deliver a specific predetermined good in specified quantity on a given date in the future at a fixed price (chapra, 1998). in contrast to salam, there is no obligation for lump sum advance payment under istisnāh. the price may be paid in accordance to the production progress or partly in advance and the rest at the time of delivery. another distinction point of istisnāh contract is that the delivery time could be unknown or unspecific at the time of contract. finally, the subject matter of an istisnāh contract is usually a commodity or item which involves manufacturing (iqbal & molyneux, 2005). wilson (2004) explained the process of issuance of istisnāh sukūk and highlighted that istisnāh sukūk has also become the contractual form for financing construction projects. in order to issue istisnāh sukūk, a parallel istisnāh contract is used between financier and the actual subcontractor of the project. project commissioner provides details of the technical, financial, and project management specifications of the project to the financier. then, financier sets up a tender to find the best subcontractor for the project. among all information required in the bid, they should also specify their proposal for selling the completed parts of the project over time and the amount of expected in installment payments. this stream of revenues which is based on the expected installments income for a specific period of time can be used for issuing the debt certificates. however, since the istisnāh certificates are not based on any real asset and are solely representing debt obligation, they are not tradable in secondary market at discount. figure 18: cash flows pattern of istisnāh sukūk istisnah sukūk could only be exchanged at face value or be used to purchase goods or services whose price is equal to the face value of the certificate. however, transferring the debt contract from the financier to a supplier of goods safari: contractual structures and payoff patterns of suk?k securities 90 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 or service needs the permission of the original debtor commissioning the project (wilson, 2004). as mentioned above, in istisnāh contracts, capital owner pays in advance for a project to be constructed in future. the capital owner (i.e., sukūk holder) would benefit from the difference between amount paid in advance and the market price of the project at maturity. the maturity reward (i’), thus, is undetermined at the issuance. the payback structure of istisnāh sukūk is depicted in figure 18. 5. conclusion this paper compares various definitions of sukūk contracts and summarizes these definitions as a funding (debt) arrangement agreed between a party providing funds (investor) and the counterparty (a government or a firm or an individual) borrowing the funds for the purposes of using the funds to engage only in permissible economic production/ services (ariff, safari & shamsher, 2012). this definition satisfies the conceptual requirements of a sukūk contracts. this paper also categorizes various sukūk contracts based on their intrinsic nature. sukūk securities are issued for different purposes, however, they can be classified into three main types as: pure debt, equity-based, and asset-backed. pure debt sukūk securities, which include the mudārabah contracts, are merely silent partnership: process of raising funds through loan-like contracts. equitybased contracts include mushārakah and diminishing mushārakah contracts and are similar to finite-term equity ownership for investors. asset-backed sukūk contracts include the murābahah, ijārah, salam, and istisnāh. these are contracts that ownership of an asset is transferred to investors as a part of the fund raising process. paper reviews major types of sukūk contracts (mudārabah, mushārakah, murābahah, salam, ijārah, and istisnāh ) by reviewing and summarizing the literature available on them and providing the best definition and highlighting their unique features and their structure. it also draws possible cash flow patterns pertaining to each form of contracts. to commence the more arduous task of building mathematical models of how the six sukūk instruments should be valued, we provide a very useful starting point, especially by charting the payoffs of each type of the general class of sukūk. this paper therefore is hoped to provide a clear understanding of the economic behavior that results from a different way of constructing the financial arrangement for working capital and long term capital under a more advanced socially responsible funding arrangement that limits too much borrowing, makes borrowing conditional on purpose, sharing risk while avoiding extreme uncertainty (which is the hallmark of 21st century finance) and clearly showing the payoff patterns to be different from those of conventional bond market behavior. author information: the author wishes to thank professor mervyn lewis (university of south australia) for his comments on a thesis and relevant to international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 91 this paper. the author also wishes to extend his sincere thanks to professors mohamed ariff (bond university) and shamsher mohamad (inceif) for their valuable contributions to this research. the author was sponsored by khazanah nasional malaysia (a sovereign wealth fund) as a scholar while conducting this research. an earlier version of this paper was accepted for presentation at islamic finance conference at the aston university. the author is solely responsible for any errors: e-mail: meysam.safari@gmail.com. references aaoifi. (2004). sharīʿah standards for financial institutions. bahrain: accounting and auditing organization for islamic financial institutions (aaoifi). abdul-gafoor, a. l. m. (2006). mudārabah-based investment and finance. journal of islamic banking and finance, 23(4), 78-98. abdul majid, a. r. (2003). development of liquidity management instruments: challenges and opportunities. paper presented at the international conference on islamic banking: risk management, regulation and supervision. adam, n. j., & thomas, a. (2004). islamic bonds: your guide to issuing, structuring and investing in ṣukūk. london: euromoney books. al-jarhi, m. a., & iqbal, m. (2001). islamic banking: answers to some frequently asked questions. unpublished occasional papers. jeddah: islamic research and training institute. al-masri, r. y. (2003). market price of salam on the date of delivery: is it permissible? journal of king abdulaziz university: islamic econ., 16(2), 29-32. al-omar, f., & abdel-haq, m. (1996). islamic banking: theory, practice & challenges. karachi: oxford university press. alvi, i. a., mohammed, a. r., khan, g. z., nasser, u. m., naseer, b., & khan, m. s. (2010). ṣukūk report: a comprehensive study of the international sukūk market. manama, bahrain: international islamic financial markets. ariff, m., iqbal, m., & shamsher, m. (2012). the islamic debt market for ṣukūk securities: the theory and practice of profit sharing investment. cheltenham, uk: edward elgar. ariff, m., safari, m., & shamsher, m. (2012). ṣukūk securities, their definitions, classification and pricing issues. in m. ariff, m. iqbal and m. shamsher (eds.), the islamic debt market for sukūk securities: the theory and practice of profit sharing investment. cheltenham, uk: edward elgar. aseambankers. (2005). capitalising on opportunities in the sukuk industry. aseambankers malaysia berhad. kuala lumpur: aseambankers. bashir, a. h. m. (1999). risk and profitability measures in islamic banks: the case of two sudanese banks. islamic economic studies, 6(2), 1-26. bose, s., sr., & mcgee, r. w. (2008). islamic investment funds: an analysis of risks and returns. working paper 1310449 ssrn elibrary. safari: contractual structures and payoff patterns of suk?k securities 92 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 capital adequacy requirements for ṣukūk, securitizations and real estate investment, ifsb-7 c.f.r. § 1.1 (2009). chapra, m. u. (1998). the major modes of islamic finance. paper presented at the 6th intensive orientation course on “islamic economics, banking and finance”. daryanani, n. (2008). a deeper understanding on the prohibition of ribā. unpublished masters’ thesis. university of nottingham. dommisse, a., & kazi, w. (2005). securitization and sharīʿah law. banker middle east, 1 july, 7-8. dusuki, a. w. (2009). challenges of realizing maqāsid al-sharīʿah (objectives of sharīʿah) in islamic capital market: special focus on equity-based ṣukūk. paper presented at the 3rd usm-isdev international islamic management conference on islamic capital market. ebrahim, m. s. (1999). integrating islamic and conventional project finance. thunderbird international business review, 41(4-5), 583-609. el-gamal, m. a. (2000). a basic guide to contemporary islamic banking and finance. islamic society of north america. el-gamal, m. a. (2007). mutuality as an antidote to rent-seeking sharīʿah arbitrage in islamic finance. thunderbird international business review, 49(2), 187-202. floor, w. (ed.) (1990). encyclopaedia iranica online edition. new york: columbia university. gait, a. h., & worthington, a. c. (2007). a primer on islamic finance: definitions, sources, principles and methods. university of wollongong. gurgey, u., & keki, e. (2008). ṣukūk in turkey. international financial law review, 27(11), 111-111. homoud, s. h. (1994). progress of islamic banking the aspiration and realities. islamic economic studies, 2(1), 71-80. iqbal, m. (1998). islamic banking. in m. kahf (ed.), lessons in islamic economics (pp. 493-524). jeddah: islamic research and training institute. iqbal, m., & molyneux, p. (2005). thirty years of islamic banking: history, performance and prospects. houndmills, basingstoke, hampshire; new york: palgrave macmillan. jalil, a. (2005). islamic bonds issues: the malaysian experience. paper presented at the 6th asian academy of management conference, 9-11 december 2005, casuarina ipoh, perak, malaysia. kamali, m. h. (2007). sharīʿah analysis of issues in islamic leasing. journal of king abdulaziz university: islamic econ., 20(1), 3-22. khan, m. m. (2003). interest-free finance: the islamic banking and finance movement in pakistan (1980-2002). unpublished phd thesis, swinburne university of technology, melbourne. khan, m. m., and bhatti, m. i. (2008), development in islamic banking: a financial risk-allocation approach. the journal of risk finance, 9(1), 40 51 al-khawarizmi. (1896). mafatiḥ al-ulum. in g. v. vloten (ed.), 38 moshe gill. leiden. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 7 contractual structures and payoff patterns of sukuk secutities: 67-93 93 lewis, m. k. (2007). islamic banking in theory and practice. monash business review, 3(1), 1-8. lewis, m. k., & algaoud, l. m. (2001). islamic banking. cheltenham, uk: edward elgar. metwally, m. m. (2006). economic consequences of applying islamic principles in muslim societies. journal of islamic banking and finance, 23(2), 11-33. obaidullah, m. (2005). islamic financial services. jeddah: islamic economics research center. shaikh, s. a. (2010). a brief review and introduction to practiced islamic banking and finance. unpublished working paper, university of east, pakistan. shinsuke, n. (2007). beyond the theoretical dichotomy in islamic finance: analytical reflections on murābahah contracts and islamic debt securities. kyoto bulletin of islamic area studies, 1(2), 72-91. usmani, m. t. (2002). an introduction to islamic finance. the hague: kluwer law international. usmani, m. t. (2007). ṣukūk and their contemporary applications. bahrain: accounting and auditing organization for islamic financial institutions (aaoifi), 14. vishwanath, s. r., & azmi, s. (2009). an overview of islamic ṣukūk bonds. the journal of structured finance, 14(4), 58-67. wan abdullah, w. r., roudaki, j., & clark, m. (2010). the evolution of the islamic capital market in malaysia. paper presented at the sixth accounting history international conference “accounting and the state”. warde, i. (2000). islamic finance in the global economy. edinburgh: edinburgh university press. wilson, r. (2004). overview of the ṣukūk market. in n. j. adam & a. thomas (eds.), islamic bonds: your guide to issuing, structuring and investing in sukūk. united kingdom: euromoney institutional investor.wilson, r. (2008). innovation in the structuring of islamic sukūk securities. humanomics, 24(3), 170 181. zaher, t. s., & hassan, m. k. (2001). a comparative literature survey of islamic finance and banking. financial markets, institutions & instruments, 10(4), 155-199. safari: contractual structures and payoff patterns of suk?k securities 1 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 https://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: ma’aji, m. m., shrubsall, r., & anderson, e. o. (2023). determinants of sme success or failure in frontier markets. international journal of banking and finance, 18(1), 1-30. https://doi. org/10.32890/ ijbf2023.18.1.1 determinants of sme success or failure in frontier markets 1muhammad muhammad ma’aji, 2robert shrubsall & 3ediri okoloba anderson camed business school, phnom pehn, cambodia 1corresponding author: muhammad@cam-ed.com received: 9/8/2021 revised: 13/10/2021 accepted: 13/10/2021 published: 5/1/2023 abstract sme competitiveness, financial stability, and and its continuity as a going concern are important considerations for company stakeholders. this research is utilizing logistic regression to examine the determinants of sme business success or failure in cambodia. a total of 314 successful and 78 failing smes were examined. information was gathered through a survey questionnaire that had been previously validated in various countries based on the financial and non-financial information of smes. the study found that four factors, namely the owner’s education level, owner’s marketing skills, customer complaints, and the age of the business were important in determining the success or failure of smes. the predictive accuracy of the model was 95.5 percent, with an r-square estimation of 64 percent. policymakers can use the results of this study to prepare and train aspirant entrepreneurs and current sme proprietors to keep proper records and maintain financial control, and develop marketing 2 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 and other skills to assist their businesses to succeed. the findings can provide insights for entrepreneurs, especially with respect to efficient resource allocations that may help avert or lessen the rate of failure among smes. the findings can also help strengthen the sme sector in cambodia, by providing more employment opportunities, assisting in better income distribution, and eventually leading to long-term prosperity and competitiveness. keywords: business failure, business success, logistic regression, resource-based theory, smes. jel classification: g30, g39. introduction small businesses or smes are viewed as one of the essential supports to economic development in terms of creating job opportunities, enhancing indigenous skills and innovations, building market intensity, and improving the well-being of the overall population (jahur & quadir, 2012). ayyagari et al. (2011) reported that 95 percent of the total businesses globally are smes, accounting for 60 percent of the global employment. more precisely, in the us, germany, the uk, and france, smes accounts for approximately 51 to 56 percent of these countries gdp (mcgowan, 2019). smes in the southeast asian region accounts for 96 percent of total businesses, with a 95 percent contribution to domestic employment and up to 53 percent contribution to the gdp (sme corp malaysia, 2013). for example, in indonesia and malaysia, the sme contribution to the gdp was 58.7 and 35.9 percent, respectively. therefore, policymakers consider the competitiveness and success of smes to be of paramount importance in maintaining high employment and income generation, and achieve sustainable economic growth. consequently, these contributions of smes have triggered state and regional initiatives to further strengthen the growth of smes around the globe (european commission, 2013). in this regard, cambodia is no exemption. smes in the country are seen as playing an important role in creating employment opportunities, creating income for low-income households, increasing social stability, as well as economic growth and creating a competitive and a vibrant private sector. between 2000 and 2012, the number of smes 3 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 has expanded by 268 percent, representing in excess of 75 percent of all established enterprises and generating 50 to 90 percent of the total employment (national institute of statistics (nis), 2011; forsinetti, 2012). there are more than 156,890 smes employing more than one million workers (savi, 2019). moreover, cambodia’s average annual gdp growth stood at around 7 percent for the past 10 years, signaling how both the economy and businesses are thriving at a steady rate. the cambodian government through its key rectangular strategy has come up with and implemented a number of initiatives to further develop the sme sector and make it more competitive and vibrant in the country’s continued economic development. for example, the government in 2018, introduced a decentralized sme registration platform, as well as tax exemptions, provided technical training and financial assistance to further enhance the competitive positioning of smes (sokhorng, 2018) . smes that successfully completed a tax registration (or updated their tax registration for existing smes) of their enterprise would receive a tax exemption for three to five years (sokhorng, 2018). despite such initiatives, most of the smes around the world go bankrupt (neshamba, 2006) at a early stage of their life cycle (pena, 2004), or at the accelerated growth stage (hayward et al., 2006). for example, an approximately 60 percent failure rate among smes has been reported in malaysia (ahmad & seet, 2009). studies in the uk showed an estimated 35 percent of smes went bankrupt after the first three years of initial start-up, and 55 percent failed after five years (gray et al., 2012). similarly, australia also recorded a high failure rate of about 62 percent of the smes failing in the third year in business, while an estimated 74 percent of smes failed in the fifth year of business (chancharat, 2011). in nigeria, an estimated 60 to 70 percent of smes went bankrupt in the first three years of business (akingbolu, 2010). similarly, in south africa, 63 percent of small businesses failed to pass the second year of operations (cant &wiid, 2013). likewise, in cambodia, there have been a number of smes that failed (at an early stage or otherwise) due to the challenges they faced in the course of their business operation. therefore, there is a need to further investigate the causes of the poor performance and failure of smes in cambodia. the competitiveness, financial stability and going concern of the smes are important considerations for the company’s stakeholders, including the 4 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 shareholders, financial institutions, suppliers, employees, government, customers, and society in general. furthermore, the costs and spillover effect of business failure have a broad impact on all the stakeholders. in light of these concerns, the continuous tracking of sme performance, successful or otherwise, should be an important issue for the corporate sector. therefore, the purpose of this research is to investigate the causes or predictors of sme success or failure in cambodia. to date there has been no such study carried out in cambodia, as such the present study will be the first ever attempt to determine the success versus failure factors of smes in the country. this study aims to make a practical contribution, as the model put forward can be used in cambodia to improve the performance of smes, as well as increasing their chances of survival. the cambodian government is currently making efforts to enhance and develop the sme sector so that it may become a more vibrant contributor to the country’s economic development. in addition, the paper aims to provide more insights for entrepreneurs, policymakers, and different stakeholders, so as to assist them through a redistribution of resources and assets that may reasonably help to eschew poor performance or failure among smes. the findings of the study can help to strengthen the sme sector in cambodia. the ultimate goal is to enhance job opportunities, provide better income distribution and allocation, improve social cohesion, and in the end contribute to an increased economic growth and development. literature review there is lack of general theory about business failure that can help to distinguish between successful and failing firms (dimitras et al., 1996). the selection of business failure predictors is subject to the popularity and predictive power of each predictor in previous studies. therefore, this section provides a review of the related underlying theory used in selecting the predictors, existing empirical literature on sme business failure, and the commonly used statistical methods to predict business failure. resource-based theory the resource-based theory (rbt) was used in this study to highlight factors that could lead to the success or failure of smes. wernerfelt 5 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 (1984) introduced the rbt and used the theoretical framework to identify the internal resource capabilities that could lead to the survival of firms and the achievement of sustainable competitive advantage, especially for smes (ong & ismail, 2008). these resources can either be tangible, such as assets, access to funds and location among others; or intangible, specifically in the areas of human resources which constitute knowledge and awareness, skills and expertise, customer service, reputation and status, entrepreneurial orientation, network dispositions and inspiration that allows the development of specific capabilities (pratono & mahmood, 2015; wernerfelt, 1984). the concept has been broadly utilized as a hypothetical base for understanding how a firm’s resources and capabilities can drive organizational performance (crook et al., 2008). numerous researchers have argued that it is a company’s capacity to organize its resources that clarifies its performance differentials (newbert, 2007). the theory suggests that business owners need to have explicit assets that will assist them to observe new opportunities and amass different resources so that the organizations can succeed (alvarez & busenitz, 2001). lussier (1995) utilized the rbt in his 15 success versus failure elements to learn how resource-based variables add to the success or failure of a typical firm in the us. likewise, the rbt was utilized by thornhill and amit (2003) to separate firms that have encountered financial distress during the early life cycle stages because of the deficiency of financial assets or lack of knowledge among the executives. moreover, o’cass and sok (2013) employed the same theory to look at the degree to which the mix of marketing skills, intellectual assets and product innovation impacted the performance of smes. empirical studies on sme failure prediction models over the last four decades, the incidence of bankruptcy cases has led to a growing interest from researchers in business failure prediction, to the extent that it has developed into a key research area in finance. previous studies have revolved around the search for the best failure prediction model for smes, focusing on financial data from both listed corporations and small businesses since the pioneering model of altman in 1968. altman’s (1968) model used multiple discriminant analysis (mda) to predict bankruptcy among us-listed companies using financial ratios. since then, a number of prediction models for listed companies have been developed in the literature (altman, 1968; beaver, 1967; blum, 1974). 6 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 similarly, the increasing number of business failures among small businesses in the us has motivated edmister (1972) to adopt altman’s (1968), beaver’s (1967) and blum’s (1974) bankruptcy prediction models of listed companies. these models were based on financial data and looked for evidence among small businesses in the same market. using 19 selected financial ratios from their studies, edmister (1972) employed the mda method to differentiate between failed and non-failed smes. he found that for the year prior to failure, seven financial variables were found to be important in forecasting business failure among small businesses. they included current liabilities to equity, equity to sales ratio, annual fund flows to current liabilities, inventory to sales ratio, working capital to sales ratio, and quick ratio to robert morris associates (rma) up-trend. the findings further implied that financial ratios are applicable to predicting the business failure of small firms. following edmister (1972), a number of studies have developed prediction models using financial data (see abdullah, ma’aji & khaw, 2016; altman & sabato, 2007; altman et al., 2020; blum, 1974; edmister, 1972; keasey & watson, 1987; lussier, 1995; ma’aji et al, 2019; ma’aji et al., 2018; ma’aji et al., 2020; ohlson, 1980; platt & platt, 1990; and zmijewski, 1983). among the most predictable variables in these models were the leverage ratios, efficiency ratios, profitability ratios, and liquidity ratios. moreover, the research field has been extended by developing prediction models using qualitative data. researchers have claimed that these models also provided a complete or integrated clarification for the failure of smes. lussier (1995), for example, used nonfinancial data to predict failure among us smes, and this attempt was regarded as among the first models to utilized such variables. the model comprised 15 significant variables recognized in over 20 previous studies. the studies used non-financial variables which were examined utilizing the rbt. this has helped to broaden our understanding of the use of resources and capabilities in new firms by concentrating on the recognition and gaining of capitals and assets that are vital for a firm’s long-term achievement (lichtenstein & brush, 2001). the findings showed that successful firms featured less troubles in getting and holding qualified staff than the unsuccessful ones. contrary to expectations, the findings additionally showed that the unsuccessful firms had staff with more industry and work experience with board members that are well educated. gimeno et al. (1997), flerackers (1998), and reynolds and miller (1989) found 7 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 similar results regarding these variables. a possible explanation is that there is a need for a young mind and fresh ideas to be a successful entrepreneur. furthermore, the studies revealed that failed companies had prepared more detailed financial and other business plans than successful companies (gimeno et al., 1997; flerackers, 1998; reynolds & miller, 1989). these findings emphasized that there is a need for companies to think out-of-the-box (houben et al., 2005; bono & mcnamar, 2011). additionally, studies have shown that having a business partner does not increase the chance of business success and is not an indicator for business success (cooper et al. 1991). their model accurately estimated 81 percent of the successes and failures in the sample chosen, which was consistent with the model of lussier (1995). moreover, the studies of teng et al. (2011) and lussier and pfeifer (2001) found that recruitment is an important predicting factor, while lussier and pfeifer (2001) also found that knowledge of the owner and the decision-making expertise of the board members were found to be important variables in explaining success or failure among small businesses. the majority of academic scholars have been developing and testing failure prediction models based on smes in developed countries, rather than the ones in developing countries. furthermore, fewer studies have used non-financial variables in their failure prediction studies.this is the result of the problem faced in retrieving the financial and non-financial data of small businesses (ma’aji et al., 2019; lussier & corman, 1995). furthermore, numerous small businesses are informal and cannot be found in the formal sector of the economy in frontier markets such as cambodia. the need for non-financial indicators rests on the foundation that the use of financial factors are the sole indicators of organizational achievement and performance (behr & guttler, 2007). the use of financial factors such as ratios in failure prediction research has gained lots of deliberation within the finance literature as the financial indicators are mostly highlights of firms’ past performance. consequently, the prediction models based on financial variables might not be appropriate for the future (lussier & corman, 1995; keasey & watson, 1987). therefore, the use of historical accounting principles may impact on the reliability of the failure prediction models since there is a propensity for managers to manipulate the information, more so in the case of small businesses, 8 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 where there is often a lack of comprehensive and efficient internal control mechanisms (agarwal & taffler, 2007). the limited evidence available would seem to suggest that the fundamental causes of business failure among small businesses in cambodia are numerous, and any concerned stakeholder attempting to predict which companies are more susceptible will have to depend on a broad set of information on which to support his/her forecast. in sum, this will consist of governance, industry specific information, macroeconomic lead indicators, and quality of management, as well as ratios relating to the specific company, all of which are expected to enhance understanding of factors determining small business success or failure. methodology financial and non-financial information about small businesses in developing countries is difficult to access (lussier & corman, 1995; ma’aji et al., 2019) and furthermore, a number of these firms are not registered as in the formal sector. therefore, data was collected as part of a comprehensive survey questionnaire, used previously and validated in different regions and countries, to gather financial and non-financial information (lussier, bandara, & marom 2016; lussier & pfeifer, 2001; lussier, 1995; lussier, 2005; lussier & halabi, 2010; gyimah et al., 2019). the selection of businesses used in this research followed the definition provided by the sub-committee of the sme secretariat of the royal government of cambodia (rgc). the subcommittee of the sme secretariat has classified cambodian smes into four categories (see table 1), namely micro, small, medium and large enterprises. these smes are categorized based on total assets, excluding land and building and the number of staff in the business. this study has employed a random sampling method in order to choose the sample of smes, it is a method that is regularly utilized in firm-level research (carrero-morales, 2015; ma’aji & barnett, 2019). furthermore, a total of 565 questionnaires were administered to entrepreneurs and managers of businesses. a cover letter was attached to each questionnaire sent as an overview of the objective of the study. it also guaranteed the confidentiality of the evidence provided by each respondent. it must be noted that the cambodian business community is not used to this type of survey due to the culture of secrecy embedded 9 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 among businesses. the questionnaires were presented in person and appointments were scheduled in advance in order to improve the response rate and to guarantee business owners understood all of the questions raised. a total of 427 questionnaires were completed, resulting in a response rate of 76 percent. however, 35 of the responses were incomplete. thus, only 392 of the responses were usable for the research, of which 314 (80%) were from firms classified successful and 78 (20%) classified as failing firms. moreover, the combined sample of successful and failing smes were taken from phnom pehn, the capital city of cambodia. more specifically, 20 percent (78 firms) was from the micro category, 40 percent (157 firms) from the small category, 25 percent (98 firms) from the medium segment and 15 percent (59 firms) from the large category as per the definition of the rgc’s sub-committee of the sme. table 1 definition of smes categories no of employees assets excluding land (usd) micro less than 10 less than 50,000 small 11 – 50 50,000 – 250,000 medium 51 – 100 250,000 – 500,000 large over 100 over 500,000 note. sourced from the rgc’s sub-committee on sme secretariat, 2005 and 2007. analysis model and variables logistic regression was employed to assess the model based on a dummy dependent variable. logistic regression is best utilized in situations where the observed result for a dependent variable can have only two likely outcomes, and in the case of this study a successful or failing firm. the prediction model used in this study is as follows in equation (1): zi = β’xi + ui (1) where: zi = failing if zi> 0; successful otherwise xi = capital; financial record and financial control; working capital; management experience; professional advice; educational level; business planning; leadership skills; product/service timing; customer complaint; marketing skills; age of firm; under-skilled 10 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 labour; government support; infrastructure condition (see definitions in appendix 1) ui = error term zi ranges from –α to +α the probability and likelihood function for failing can be defined as follows in equation (2): (2) a logistic distribution function is represented by eqn.(2). if pi represents the probability of failing as given in eqn. (2), then (1-pi) would be the probability of success, as is shown in equation (3): (3) the company was classified as failing if the calculated probability from the logistic model was more than 0.5; otherwise, it was considered successful. variable measures dependent variable the study used dichotomous variables of success or failing as dependent variables. it followed the method of using profitability as a measure of success or failure for the reason that at the time of executing this study, cambodia do not have any bankruptcy law for determining bankruptcy, or any government agency to document bankrupt businesses. the use of profitability as a determinant of firm success or failure is used by many researchers. researchers have categorized an sme as “successful” if the level of the current profit is the same as the industry average profit or above the industry average profit, and a business is categorized as “failing” if the current level of profit is below the industry average profits, or the business is currently making losses from operations (lussier & pfeifer, 2001; gyimah et al., 2019; guzman & lussier, 2005; halabi & lussier, 2010; hyder & lussier, 2016). on a scale of 1 to 4, the owner of a firm, or the respondents were asked to select their level of profit, 1 – positive profit for more than 10 years, 2 – positive profits for 2 to 5 consecutive years, 3 – unpredictable profits/losses, and 4 – currently not making profits. a dummy variable was created, and respondents infrastructure condition (see definitions in appendix 1) ui = error term zi ranges from –α to +α the probability and likelihood function for failing can be defined as follows in equation (2): pi = e(y = 1/ xi) = 1 1+𝑒𝑒 −(𝛽𝛽’𝑥𝑥𝑖𝑖 + 𝑢𝑢𝑖𝑖) (2) a logistic distribution function is represented by eqn.(2). if pi represents the probability of failing as given in eqn. (2), then (1-pi) would be the probability of success, as is shown in equation (3): 1 – pi = 1 1+ 𝑒𝑒 𝑍𝑍𝑖𝑖 (3) the company was classified as failing if the calculated probability from the logistic model was more than 0.5; otherwise, it was considered successful. variable measures dependent variable the study used dichotomous variables of success or failing as dependent variables. it followed the method of using profitability as a measure of success or failure for the reason that at the time of executing this study, cambodia do not have any bankruptcy law for determining bankruptcy, or any government agency to document bankrupt businesses. the use of profitability as a determinant of firm success or failure is used by many researchers. researchers have categorized an sme as “successful” if the level of the current profit is the same as the industry average profit or above the industry average profit, and a business is categorized as “failing” if the current level of profit is below the industry average profits, or the business is currently making losses from operations (lussier & pfeifer, 2001; gyimah et al., 2019; guzman & lussier, 2005; halabi & lussier, 2010; hyder & lussier, 2016). on a scale of 1 to 4, the owner of a firm, or the respondents were asked to select their level of profit, 1 – positive profit for more than 10 years, 2 – positive profits for 2 to 5 consecutive years, 3 – unpredictable profits/losses, and 4 – currently not making profits. a dummy variable was created, and respondents that chose 1 and 2 were coded as 0, representing “successful smes," and respondents that chose 3 and 4 were coded as 1, representing “failing smes”. independent variables for the 15 independent variables, the age of the firm and length of management experience were considered ratio measurement, while capital, financial records and financial control, business planning, working capital, use of professional advice, leadership skills, educational level, product/service timing, marketing skills, customer complaints, under-skilled labour, government supports and infrastructure conditions were all measured on a 5-point likert scale. statistical analysis the study used three levels of statistical analysis. firstly, a descriptive examination was carried out to understand the mean differences of the variables used in the samples. then, diagnostic tests were carried out, such as the multicollinearity test (i.e., the pearson correlation examine the association that exists between the independent variables and the strength of their linear association). lastly, the model was run using logistic regression model, employing success or failing as the dependent variable in relation to the 15 independent variables. infrastructure condition (see definitions in appendix 1) ui = error term zi ranges from –α to +α the probability and likelihood function for failing can be defined as follows in equation (2): pi = e(y = 1/ xi) = 1 1+𝑒𝑒 −(𝛽𝛽’𝑥𝑥𝑖𝑖 + 𝑢𝑢𝑖𝑖) (2) a logistic distribution function is represented by eqn.(2). if pi represents the probability of failing as given in eqn. (2), then (1-pi) would be the probability of success, as is shown in equation (3): 1 – pi = 1 1+ 𝑒𝑒 𝑍𝑍𝑖𝑖 (3) the company was classified as failing if the calculated probability from the logistic model was more than 0.5; otherwise, it was considered successful. variable measures dependent variable the study used dichotomous variables of success or failing as dependent variables. it followed the method of using profitability as a measure of success or failure for the reason that at the time of executing this study, cambodia do not have any bankruptcy law for determining bankruptcy, or any government agency to document bankrupt businesses. the use of profitability as a determinant of firm success or failure is used by many researchers. researchers have categorized an sme as “successful” if the level of the current profit is the same as the industry average profit or above the industry average profit, and a business is categorized as “failing” if the current level of profit is below the industry average profits, or the business is currently making losses from operations (lussier & pfeifer, 2001; gyimah et al., 2019; guzman & lussier, 2005; halabi & lussier, 2010; hyder & lussier, 2016). on a scale of 1 to 4, the owner of a firm, or the respondents were asked to select their level of profit, 1 – positive profit for more than 10 years, 2 – positive profits for 2 to 5 consecutive years, 3 – unpredictable profits/losses, and 4 – currently not making profits. a dummy variable was created, and respondents that chose 1 and 2 were coded as 0, representing “successful smes," and respondents that chose 3 and 4 were coded as 1, representing “failing smes”. independent variables for the 15 independent variables, the age of the firm and length of management experience were considered ratio measurement, while capital, financial records and financial control, business planning, working capital, use of professional advice, leadership skills, educational level, product/service timing, marketing skills, customer complaints, under-skilled labour, government supports and infrastructure conditions were all measured on a 5-point likert scale. statistical analysis the study used three levels of statistical analysis. firstly, a descriptive examination was carried out to understand the mean differences of the variables used in the samples. then, diagnostic tests were carried out, such as the multicollinearity test (i.e., the pearson correlation examine the association that exists between the independent variables and the strength of their linear association). lastly, the model was run using logistic regression model, employing success or failing as the dependent variable in relation to the 15 independent variables. 11 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 that chose 1 and 2 were coded as 0, representing “successful smes,” and respondents that chose 3 and 4 were coded as 1, representing “failing smes”. independent variables for the 15 independent variables, the age of the firm and length of management experience were considered ratio measurement, while capital, financial records and financial control, business planning, working capital, use of professional advice, leadership skills, educational level, product/service timing, marketing skills, customer complaints, under-skilled labour, government supports and infrastructure conditions were all measured on a 5-point likert scale. statistical analysis the study used three levels of statistical analysis. firstly, a descriptive examination was carried out to understand the mean differences of the variables used in the samples. then, diagnostic tests were carried out, such as the multicollinearity test (i.e., the pearson correlation examine the association that exists between the independent variables and the strength of their linear association). lastly, the model was run using logistic regression model, employing success or failing as the dependent variable in relation to the 15 independent variables. analysis of results descriptive statistics table 2 summarizes the statistics of all the independent variables for both successful and failing smes. a mean difference test was conducted between both the successful and failing smes, and the findings were found to be in line with the expectations of the study. there were, however, significant differences between the characteristics of both the successful and failing smes. the results showed that wcap, mgex, prad, educ, mrkt, cucm, bage and gvst were significantly different at the 1 percent level. furthermore, the variable uskl was significantly different at the 5 percent level, while the ldsk was significantly different at the 10 percent level. however, 12 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 the difference between the frfc, capt, bpln, psti and infc variables were insignificant. table 2 descriptive statistics healthy firm unhealthy firm variables mean std. dev mean std. dev sig. frac 4.4222 0.6530 4.2143 0.6820 0.611 wcap 3.7222 0.7499 4.0000 1.3969 0.002*** capt 3.9111 1.0238 3.5238 1.0178 0.461 mgex 25.244 11.188 7.9762 2.9918 0.000*** bpln 3.6699 1.1548 3.4676 1.1227 0.233 prad 3.6111 0.9081 3.9286 0.7775 0.006*** educ 2.7889 0.9656 2.1905 0.3974 0.000*** ldsk 3.8111 0.9349 3.9048 0.7262 0.063* mrkt 3.6333 1.0648 2.8095 0.6713 0.000*** cucm 3.2925 1.3648 2.0948 0.7861 0.000*** psti 3.2468 1.2111 2.3242 1.2219 0.570 bage 15.289 8.1364 6.0952 2.5164 0.000*** uskl 3.5556 0.9494 3.4524 0.7715 0.047** gvst 4.3111 0.7289 4.1429 0.3542 0.000*** infc 1.8333 0.6910 1.7143 0.7083 0.435 n 314 78 note. *, **, *** significant at the 10 percent, 5 percent and 1 percent levels, respectively. capital (capt), financial record (frac), working capital (wcap), management experience (mgex), business plan (bpln), professional advisors (prad), education level (educ), leadership skills (ldsk), product/service timing (psti), marketing skills (mrkt), customer complaint (cucm), business age (bage), under-skilled labour (uskl), government support (gvst), infrastructure condition (infc). number of observation (n). overall, the mean comparison between the two groups suggests that smes with tight working capital control, greater management experience, the presence of a professional adviser, whose owners had marketing skills prior to starting their business, have effective and efficient communication with customers, with more than seven years of operation and adequate government support have a better chance of success. moreover, smes with a skilled workforce and a demonstration of strong leadership skills from the top management have a greater chance of success. 13 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 ta bl e 3 p ea rs on c or re la tio n a na ly si s v if fr fc w c ap c ap t m ge x b pl n pr a d e du c l ds k m rk t c uc m ps ti b a ge u sk l g vs t in fc fr fc 1. 74 8 1 w c ap 1. 49 6 -0 .4 45 0 1 c ap t 1. 34 8 -0 .0 22 5 -0 .0 39 0 1 m ge x 2. 51 9 0. 27 2* * -0 .1 65 4 0. 11 16 1 b pl n 1. 16 0 -0 .0 22 9 -0 .0 30 5 0. 13 07 0. 07 93 1 pr a d 1. 16 3 -0 .1 49 5 0. 09 34 -0 .0 42 5 -0 .2 06 0. 04 98 1 e du c 1. 31 0 0. 06 39 0. 11 29 0. 10 79 0. 20 0* 0. 06 85 -0 .1 61 9 1 l ds k 1. 44 2 -0 .3 48 0 0. 05 25 0. 03 00 -0 .0 58 2 0. 00 20 -0 .0 40 3 0. 09 60 1 m rk t 1. 82 1 0. 24 0* * -0 .0 27 5 0. 34 7* * 0. 38 1* * 0. 06 80 -0 .0 58 2 -0 .0 96 3 -0 .3 08 ** 1 c uc m 1. 35 1 0. 27 0* * -0 .1 43 6 -0 .0 65 5 0. 31 9* * 0. 01 67 -0 .0 54 8 0. 07 91 -0 .0 06 8 0. 30 1* * 1 ps ti 1. 27 1 0. 20 9* -0 .1 67 8 0. 06 25 0. 26 2* * 0. 21 1* -0 .1 47 5 0. 13 88 -0 .1 87 0. 19 4* 0. 17 2* 1 b a ge 3. 01 6 0. 13 59 -0 .0 27 2 0. 01 96 0. 48 6* * 0. 05 86 -0 .1 11 9 0. 25 7* * -0 .0 70 3 0. 25 9* * 0. 28 4* * 0. 23 5* * 1 u sk l 1. 23 3 0. 00 55 -0 .1 52 2 0. 10 42 0. 05 40 0. 08 31 0. 05 69 -0 .0 61 5 0. 08 78 0. 17 7* 0. 20 3* 0. 09 33 -0 .0 30 5 1 g vs t 1. 46 2 0. 33 9* * -0 .2 93 0. 12 98 0. 14 57 0. 24 6* * -0 .2 07 0. 07 76 -0 .1 73 0. 20 2* 0. 14 06 0. 31 6* * 0. 01 75 0. 13 68 1 in fc 1. 25 6 -0 .0 39 2 -0 .0 66 7 0. 20 4* 0. 07 29 0. 12 42 0. 01 53 0. 19 1* 0. 19 7* 0. 01 09 0. 04 46 0. 01 57 0. 03 63 -0 .1 57 9 -0 .1 04 0 1 n ot e. * , * *, * ** s ig ni fic an t a t t he 1 0 pe rc en t, 5 pe rc en t a nd 1 p er ce nt le ve ls , r es pe ct iv el y. c ap ita l ( c ap t) , f in an ci al r ec or d (f ra c) , w or ki ng c ap ita l (w c ap ), m an ag em en t e xp er ie nc e (m ge x) , b us in es s pl an ( b pl n) , p ro fe ss io na l a dv is or s (p ra d) , e du ca tio n le ve l ( e du c) , l ea de rs hi p sk ill s (l ds k) , pr od uc t/s er vi ce t im in g (p st i) , m ar ke tin g sk ill s ( m rk t) , c us to m er c om pl ai nt (c uc m ), b us in es s a ge (b a ge ), u nd er -s ki lle d la bo ur (u sk l) , g ov er nm en t su pp or t ( g vs t) , i nf ra st ru ct ur e co nd iti on (i nf c ). n um be r o f o bs er va tio n (n ). 14 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 pearson correlation test was utilized to examine the connection between the independent variables and the outcomes of the test are as shown in table 3. the results reveal that the association between the variables is moderately low, ranging from 0.0020 to 0. 0.486. to additionally confirm that multicollinearity is not an issue in this research, variance inflating factor (vif) was accounted for in table 4. in the event that a variable is having a vif of more than 10, or the tolerance result is lower than 0.10, then there is the problem of multicollinearity in the study (gujarati & porter, 2003). however, since in this study all the independent variables had vif values ranging from 1.160 to 3.016, as is shown in table 3, the results suggest that there was no multicollinearity problem in the study. validity and reliability of the model table 4 presents the model parameter estimates from the logistic regression. the hosmer and lemeshow (h&l) test is a statistical test showing how well the logistic regression models fit. the h&l test is commonly employed to answer questions about how well and sound the model fit the sample data. based on this study, the model did fit the sample data, since the observed and expected event rates in the sub-groups were similar, which indicated that the model is reliable. furthermore, the h&l chi-square of 0.2924 and p-value 0.743 implied that the models fitted the data. correspondingly, the logistic regression results examining the model chi-square was 109.749, with the model’s significance level at less than 0.01 (p-value = 0.000). consequently, the results supported the model’s sound validity as it can classify a group of smes as “successful” or “failing” more precisely than random guessing by 99 percent. the cox & snell r square was 0.66, signaling that the model is able to predict 66 percent of the variability in the scores. the model classified correctly 307 out of the 314 successful smes, accounting for 97.8 percent, and 74 out of the 78 failing smes correctly, representing 95.2 percent, with an overall classification rate of 97 percent. therefore, the results support the model’s ability to predict successful or failing smes in cambodia. due to the model’s high predictive value, it will be able to correctly predict an smes as successful or failing by 97 percent most of the time, compared to 50 percent of the time for a random model. 15 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 table 4 logistic regression model for cambodia model parameter estimates variable name model coefficient odds ratio wald sig. constant 16.8099 1976% 2.1398 0.144 frac -0.2760 -24.1% 1.748 0.283 wcap 0.6690 95.2% 1.3308 0.080* capt -0.2213 -19.9% 0.1323 0.716 mgex -0.1698 -15.6% 0.4184 0.518 bpln -0.6742 -49.0% 1.1655 0.280 prad 1.0412 183.3% 1.1866 0.276 educ -1.9394 -85.6% 2.6924 0.004*** ldsk 0.2802 32.3% 0.0685 0.794 mrkt -2.2710 -89.7% 3.6619 0.050** cucm -1.0314 -64.3% 3.3588 0.067* posts -0.0188 -1.9% 0.0010 0.975 bage -0.7757 -54.0% 6.9903 0.008*** uskl 0.7641 114.7% 0.5633 0.453 gvst -0.1407 -13.1% 0.0149 0.903 infc -0.2149 -19.3% 0.0347 0.852 model test results -2 log likelihood 23.755 model chi-square 109.749 model significance 0.000 cox & snell r square 0.66 hosmer and lemeshow test 0.2924 classification result correctly classified cases healthy 97.8% unhealthy 95.2% overall 97.0% note. *, **, *** significant at the 10 percent, 5 percent and 1 percent levels, respectively. capital (capt), financial record (frac), working capital (wcap), management experience (mgex), business plan (bpln), professional advisors (prad), education level (educ), leadership skills (ldsk), product/service timing (psti), marketing skills (mrkt), customer complaint (cucm), business age (bage), under-skilled labour (uskl), government support (gvst), infrastructure condition (infc). number of observation (n). reduced model for cambodia the most significant variables that the study was able to distinguish between successful smes from failing ones was at the 90 and 95 percent 16 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 confidence interval from the logistic regression model (see table 5). these variables were working capital (t-value, p = 0.080), owner’s education level (t-value, p = 0.004), owner’s marketing skills (t-value, p = 0.050), customer complaints (t-value, p = 0.067) and the age of the business (t-value, p = 0.008). however, based on the significant variables in table 4, the study ran another logistic regression analysis using only those variables. the results shown in table 6 indicates that only four variables are significant determinants, namely educ, mrkt, cucm, and bage. thus, the reduced model’s function for cambodia can be expressed as: zi = β (educ, mrkt, cucm, and bage) as shown in table 6, the h&l chi-square is 5.518 and the p-value is 0.701 for the reduced model, which still implied that the models fitted the data. additionally, the logistic regression model chi-square is 131.43 with the model’s significance level less than 0.01 (p-value = 0.000). consequently, the results supported the reduced model’s sound validity, as it can classify a group of smes as “successful” or “failing” more precisely than random guessing by 99 percent. the cox & snell r square was slightly reduced to 0.64, signaling that the model is able to predict 64 percent of the variability in the scores. the model classified correctly 300 out of the 314 successful smes, accounting for 95.6 percent, and 74 out of the 78 failing smes correctly, representing 95.2 percent, with an overall classification rate of 95.5 percent. overall, the reduced model too has the ability to predict successful or failing smes in cambodia. the model is able to correctly predict an smes as successful or failing by 95.5 percent most of the time, compared to 50 percent of the time for a random model. table 5 presents the odds ratio for the significant variables. the odds ratio measures the relationship between an exposure and an outcome. the odds ratio describes the chances that a result will happen given a particular exposure, contrasted with the odds of the result occurring without that exposure. at the point when logistic regression is estimated, the regression coefficient (b1) is the estimated increase in the log odds of the result per unit increase in the value of the exposure. all in all, the exponential function of the regression coefficient (eb1) is the odds ratio associated with a one-unit increase in the exposure (szumilas, 2010). 17 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 table 5 logistic regression reduced model for cambodia model parameter estimates variable name model coefficient odds ratio wald sig. constant 16.164 1046% 12.832 0.000*** wcap 0.585 79.5% 2.175 0.140 educ -1.568 -79.2% 4.443 0.035*** mrkt -2.001 -86.5% 7.389 0.007** cucm -0.898 -59.3% 3.530 0.060* bage -0.683 -49.5% 8.565 0.008*** model test results -2 log likelihood 30.697 model chi-square 134.43 model significance 0.000 cox & snell r square 0.639 hosmer and lemeshow test p-value 0.701 classification result correctly classified cases healthy 95.6% unhealthy 95.2% overall 95.5% note. *, **, *** significant at the 10 percent, 5 percent and 1 percent levels, respectively. capital (capt), financial record (frac), working capital (wcap), education level (educ), marketing skills (mrkt), customer complaint (cucm), business age (bage). in terms of the owner’s education (educ), the odds ratio showed that businesses that start with an owner having a higher level of education has an increased chance of success by 79.2 percent. in other words, there is a 79.2 percent likelihood that a business will be healthy or successful if the owner has a higher education when he/she established the business. for marketing skills (mrkt), the clarification could be that for each additional score towards marketing skill, the odds of “successful” (which is coded 0) will increase by 86.5 percent. meaning that there is an 86.5 percent chance that businesses will be healthy or successful when the owners have marketing skills. the odds of customer complaints (cucm) is 59.3 percent, which implies that there is a 59.3 percent likelihood that businesses that receive 18 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 customer complaints most of the time have a greater chance of being healthy or successful. the result for business age (bage) shows that there is a 49.5 percent likelihood that younger businesses are more prone to not being successful. individual variables in the model the findings show that education (educ) is negatively associated with business failure. the findings also show that business owners with higher education when they setup their business have a more likelihood of being successful than owners without higher education. individuals that have a higher education have acquired a certain amount of knowledge and set of skills, and have acquired capabilities that help personal development in certain professions which might help them to better manage their businesses, hence increasing their chance of being successful. a high level of education can allow for the necessary intellectual skills that can empower an individual to better appraise business opportunities during the course of business operations, which could lead to higher productivity and efficiency (davidsson & honig, 2003; gauthier, 2006; jiménez et al., 2015). furthermore, business owners with higher education tend to be more confident, thus helping their exploration of entrepreneurial activity, while reducing risks which are the traits of being a successful entrepreneur (jiménez et al., 2015). this finding is consistent with the tenets of human capital theory that have proposed that human capital contributes to business success (becker & rosen, 1992). firms are inclined to expand investments in human capital development in order to enhance performance outcomes of employees, which in return can lead to the firm’s growth (unger et al., 2011). furthermore, the marketing skills (mrkt) of a business owner were negatively and significantly related to business failure at the 5 percent level. the results revealed that entrepreneurs with marketing skills are more likely to be successful compared to business owners without them. marketing skills and capabilities can help develop a business sense that reacts to market developments and changes, such as changes in consumer taste and preferences, competitors’ moves, and technological changes. these, in turn, can enable businesses to leverage on their capabilities and resources for value creation and successfully anticipate customer’s potential needs. according to habtoor and alharbi (2020), this will help companies avoid surprises 19 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 from new waves of market rivalry based on new technological trends, change in consumer preferences and value propositions. this finding was also consistent with that in previous studies (habtoor & alharbi 2020; bharadwaj et al., 1993; narver & slater 1990), which found that marketing skills gives organizations the capacity to generate satisfactory information about rival activities and responses, and can draw on sufficient intelligence to counter the action of competitors. this allows businesses to deliver superior customer value, which in turn helps develop the basis for a competitive advantage (porter, 1985). additionally, customer complaints (cucm) were negatively related to businesses failure and was significant at the 1 percent level. the findings show that businesses that received customer complaints most of the time have more chances of being successful than firms that received fewer customer complaints. when a customer’s complaint highlights a problem, whether about the firm’s product, employees or internal processes, the company can investigate and solve the problem and thus, prevent further complaints in the future. frequently debated anecdotally is the fact that for every one complaining customer, there are possibly many more other customers with the similar issue, but stay quiet (plymire, 1991). accordingly, studies have found that customers whose complaints were handled quickly can often turn into loyal customers and even brand advocates. customers who have a complaint handled quickly go on to spend more on future purchases, thereby increasing the probability of success for the company (habtoor & alharbi 2020; bell & luddington, 2006; bell & mengüç 2002; brodie et al., 2011; groth et al., 2001; lockrey, 2015). through taking care of customer complaints productively, businesses can grow and develop in fulfilling and satisfying customers beyond their expectations, which could lead to repeat sales and subsequently can prompt the flourishing of business operations and overall performance. the age of a business (bage) has a negative coefficient and significance at the 1 percent level. the finding indicates that younger businesses are more likely to be unhealthy compared with older ones, thus, the longer the existence of an sme, the higher the chances are for it to be successful, probably because older smes have more experience and management capabilities (expertise). older firms are more likely to tap into the appropriate customer segment and deliver differentiated 20 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 products and services that meet the expectations of customers, and this will subsequently help them gain further customer loyalty and build up a better rapport with suppliers (majumdar, 1997). long established companies are more likely to have a competitive advantage and resource capabilities in their development stages as they grow, while young companies are expected to suffer a lack of sufficient resources and capabilities (thornhill & amit, 2003). additionally, success can be as a result of a firm’s activities over the years. for example, mergers, absorptions, evolution from being micro, small, medium to large may happen over a number of years. these situations can be quite common given the nature of the sme business. previous studies also find that the business age is negatively associated with bankruptcy (abdullah et al., 2016; altman et al., 2010; ma’aji et al., 2018; ma’aji et al., 2020; shane, 1996). conclusion issues relating to smes and failure prediction have been the focus of many scholarly and regulatory debates around the world. this is the result of the significant role of smes in overall economic development and sustainability. consequently, early recognition and understanding of risks leading to business failure is imperative for instituting, sustaining, and growing a business, as well as for the economy in general. business owners need to recognize the risks associated with failure and attain the resources and capabilities identified in this study in order to enhance their chance of being successful. in summary, the results from this study indicate that long established smes that handle customer complaints in a timely manner, have owners who possess higher education and strong marketing skills, and have a tight control over the working capital, have a better chance of success in cambodia. these findings are significant, for the reason that they can help entrepreneurs, management, policymakers and regulators, financial institutions, business consultants and advisers and investors to better understand how a business in cambodia can succeed and evade failure. the research limitations in this paper include the fact that cambodia does not have a database on failed or bankrupt businesses. therefore, the level of profitability of firms was utilized to order to categorize 21 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 firms as either successful or failing. despite the fact that it was incredibly hard to find failed business owners and get them to honestly complete the survey, future study would be more robust if information from firms that have failed or gone bankrupt can be gathered, and then matched them with successful firms. moreover, although the present study included various industries, it did not discuss their different characteristics separately. future studies ought to consider each sector or industry independently when predicting business success or failure, in order to find out sector specific determinants and which industries are more prone to failure in cambodia. acknowledgment this research is fully supported by the camed business school, phnom pehn, cambodia. the authors would therefore, to express their gratitude to camed business school for the financial support. references abdullah, a., ma’aji, m., & khaw, l. (2016). the value of governance variables in predicting financial distress among small and medium-sized enterprises in malaysia. asian academy of management journal of accounting & finance, 12, 77–91. http://dx.doi.org/10.21 315/aamjaf2016.12.s1.4 agarwal, v., & taffler, r. j. (2007). twenty-five years of the taffler z-score model: does it really have predictive ability?. accounting and business research, 37(4), 285-300. https://doi. org/10.1080/00014788. 2007.9663313 ahmad, n., & seet, p. (2009). dissecting behaviours associated with business failure: a qualitative study of sme owners in malaysia and australia. asian social science, 5(9), 98–104. https://doi.org/ 10.5539/ass.v5n9p 98 akingbolu, r. (2010). why 70 percent of smes fail in nigeria? this day live. http://www.thisdaylive.com/articles/why-70-ofsmesfailinnigeria%2 0experts/719 11/ alvarez, s. a., & busenitz, l. w. (2001). the entrepreneurship of resource-based theory. journal of management, 27(6), 755– 775. https://doi.org/10.1177/014920630102700609 22 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 altman, e. (1968). financial ratios, discriminant analysis and the prediction of corporate bankruptcy. journal of finance, 23, 589-609. https://doi.org/10.2307/2978933 altman, e., & sabato, g. (2007). modeling credit risk for smes: evidence from us market. a journal of accounting, finance and business studies (abacus), 43(3), 332-357. https://doi. org/10.2139/ssrn.8723 36 altman, e. i., esentato, m., & sabato, g. (2020). assessing the credit worthiness of italian smes and mini-bond issuers. global finance journal, 43, 100450. https://doi.org/10.1016/j. gfj.2018.09.003 ayyagari, m., demirgüç-kunt, a., & maksimovic, v. (2011). small vs. young firms across the world – contribution to employment, job creation, and growth. policy research working paper 5631. the world bank development research group. beaver, w. (1967). financial ratios as predictor of failure. journal of accounting research, 4, 71-111. https://doi. org/10.2307/2490171 becker, w. e., & rosen, s. (1992). the learning effect of assessment and evaluation in high school. economics of education review, 11(2),107-118. https://doi.org/10.1016/0272-7757(92)90002-k behr, p., & guttler, a. (2007). credit risk assessment and relationship lending: an empirical analysis of german small and mediumsized enterprises. journal of small business management, 45(2), 194-213. https://doi.org/10.1111/j.1540-627x.2007.00209.x bell, s., & luddington, j. (2006). coping with customer complaints. journal of service research, 8(3), 221-233. https://doi. org/10.1177 /1094670505283785 bell, s., & menguc, b. (2002). the employee-organization relationship, organizational citizenship behaviors, and superior service quality. journal of retailing, 78(2), 131-146. https:// doi.org/10.1016/ s0022-4359(02)00069-6 bharadwaj, s., varadarajan, p., & fahy, j. (1993). sustainable competitive advantage in service industries: a conceptual model and research propositions. the journal of marketing, 83-99. https://doi.org/10.1177/002224299305700407 blum, m. (1974). failing company discriminant analysis. journal of accounting research, 12(1), 1-25. https://doi. org/10.2307/2490525 23 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 bono, j., & mcnamar, g. (2011). from the editors. academy of management journal, 47(1), 657–660. https://doi.org/10.5465/ amj.2 011.64869103 brodie, r. j., hollebeek, l. d., jurić, b., & ilić, a. (2011). customer engagement: conceptual domain, fundamental propositions, and implications for research. journal of service research, 14(3), 252-271. https://doi.org/10.1177/1094670511411703 cant, c., & wiid, a. (2013). establishing the challenges affecting south african smes. international business & economics research journal, 12(6), 707–716. https://doi.org/10.19030/ iber.v12i6.7869 carrero-morales, g. i. (2015). factors affecting the success and failure of small and medium enterprises in puerto rico. 60th annual icsb world conference, dubai: icsb, 111–11 chancharat, n. (2011). smes failure prediction: literature review. abac journal, 31(3), 70-81. http://www.assumptionjournal. au.edu/ index.php/abacjournal/article/view/145 cooper, a., gascon, j., & woo, c. (1991). a resource-based prediction of new venture survival and growth. academy of management proceedings, 1, 68–72. https://doi.org/10.5465/ ambpp.1991.4976561 crook, t. r., ketchen, d. j., combs, j. g., & todd, s. y. (2008). strategic resources and performance: a meta-analysis. strategic management journal, 29(11), 1141–1154. https:// doi.org/10.1002 /smj.703 davidsson, p., & honig, b. (2003). the role of social and human capital among nascent entrepreneurs. journal of business venture, 8(3), 301-331. https://doi.org/10.1016/s0883-9026(02)00097-6 dimitras a., zanakis s., & zopudinis c. (1996). a survey of business failures with an emphasis on failure prediction methods and industrial applications. european journal of operational research, 90(3), 487-513. https://doi.org/10.1016/03772217(95)00070-4 edmister, r. o. (1972). an empirical test of financial ratio analysis for small business failure prediction. journal of financial and quantitative analysis, 7(2), 1477-1493. https://doi. org/10.2307/232 9929 european commision. (2013). a recovery on the horizon? annual report on european smes 2012/2013. http://aei.pitt.edu/58189/1/ annual-report-smes-2013_en.pdf 24 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 flerackers, p. (1998). het success van nieuwe bedrijven: slaag en faalfactoren van nieuwe bedrijven in nederland (unpublished master’s thesis). faculty of social sciences, utrecht. forsinetti, c. (2012). financing smes in asean, the case of cambodia. the phnom penh post. https://www.phnompenhpost. com/post-plus/financing-smes-asean-case-cambodia. gauthier, j. (2006). the content of secondary education around the world: present position and strategic choices. united nations educational, scientific and cultural organization, geneva. gimeno, j., folta, t. b., cooper, a. c., & woo, c. y. (1997). survival of the fittest? entrepreneurial human capital and the persistence of under performing firms. administrative science quarterly, 42, 750-783. https://doi.org/10.2307/2393656 gray, d., saunders, m., & goregaokar, h. (2012). success in challenging times: key lessons for uk smes. working paper, university of surrey. groth, m., gutek, b. a., & douma, b. (2001). effects of service mechanisms and modes on customers’ attributions about service delivery. journal of quality management, 6(2), 331348. https://doi.org/10.1016/s1084-8568(01)00043-8 gyimah, p., appiah, k., & lussier, r. (2019) success versus failure prediction model for small businesses in ghana, journal of african business. https://doi.org/10.1080/15228916.2019.162 5017 gujarati, d. n., & porter, d. c. (2003). basic econometrics (ed.). mcgrew hill book. guzman, j. b., & lussier, r. n. (2015). success factors for small businesses in guanajuato, mexico. international journal of business and social science, 6(11), 1–7. https://doi. org/10.1.1.1069.3340&rep habtoor, n., & alharbi, a. (2020). importance of human factors to organizational performance. international journal of management and human science, 4(1), 11-22. https://ejournal. lucp.net/index.php/ ijmhs/article/view/965 hayward, m. l., shepherd, d. a., & griffin, d. (2006). a hubris theory of entrepreneurship. management science, 52(2), 160– 172. https://doi.org/10.1287/mnsc.1050.0483 houben, g., bakker, w., & vergauwen, p. (2005). assessing the non-financial predictors of the success and failure of young firms in the netherlands. journal of economics and applied informatics, 1, 5-14. https://doi.org/10.1287/mnsc.1050.0483 25 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 hyder, s., & lussier, r. n. (2016). why businesses succeed or fail: a study on small businesses in pakistan. journal of entrepreneurship in emerging economies, 8(1), 82–100. https://doi.org/10.1108/jeee-03-2015-0020 jahur, m., & quadar, s. (2012). financial distress in small and medium enterprises (smes) of bangladesh: determinants and remedial measures. economia series management, 15(1), 432-444. https://ide as.repec.org/a/rom/econmn/v15y2012i1p46-61.html jiménez, a., palmero-cámara, c., gonzález-santos, m. j., gonzález-bernal, j., & jiménez-eguizábal, j. a. (2015). the impact of educational levels on formal and informal entrepreneurship. brq business research quarterly, 18(3), 204-212. https://doi.org/10.10 16/j.brq.2015.02.002 keasey, k., & watson r. (1987). non-financial symptoms and the prediction of small company failure: a test of argenti’s hypotheses. journal of business finance and accounting, 14(3), 335-354. https://doi.org/10.1111/j.1468-5957.1987.tb00099.x lichtenstein, b., & brush, c. (2001). how do “resource bundles” develop and change in new ventures? a dynamic model and longitudinal exploration entrepreneurship. theory and practice, 25, 37-58. https://doi.org/10.1177/104225870102500303 lockrey, s. (2015). a review of life cycle based ecological marketing strategy for new product development in the organizational environment. journal of cleaner production, 95, 1-15. https:// doi. org/10.1016/j.jclepro.2015.02.022 lussier, r. n. (1995). a nonfinancial business success versus failure prediction model. journal of small business management, 33 (1), 8–20.https://www.proquest.com/docview/221008471?pqorigsite=gsch olar&fromopenview=true lussier, r. n., bandara, c., & marom, s. (2016). entrepreneurship success factors: an empirical investigation in sri lanka. world journal of entrepreneurship, management and sustainable development, 12(2), 102-112. https://doi.org/10.1108/ wjemsd-10-2015-0047 lussier, r. n. (2005). a success versus failure prediction model for the real estate industry. american journal of business, 20(1), 47–53. https://doi.org/10.1108/19355181200500005 lussier, r. n., & corman, j. (1995). a business success versus failure prediction model for entrepreneurs with 0–10 employees. 26 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 journal of small business strategy, 7(1), 21–36. https:// libjournals.mtsu.edu/ind ex.php/jsbs/article/view/327 lussier, r. n., & halabi, c. e. (2010). a three-country comparison of the business success versus failure prediction model. journal of small business management, 48(3), 360–377. https://doi. org/10.1111/j.1540-627x.2010.00298.x lussier, r. n., & pfeifer, s. (2001). a cross national prediction model for business success. journal of small business management, 39(3), 228–239. https://doi.org/10.1111/0447-2778.00021 ma’aji, m., abdullah, a., & khaw, l. (2019). financial distress among smes in malaysia: an early warning signal. international journal of business & society, 20(2), 775-792. http://www. ijbs.unimas.my/ images/repository/pdf/vol20-no2-paper21.pdf ma’aji, m. (2019). modelling business failure among small businesses in nigeria. journal of economics finance and accounting, 6(2), 72-81. https://doi.org/10.17261/pressacademia.2019.1046 ma’aji, m., abdullah, a., & khaw, l. (2018). predicting financial distress among smes in malaysia. european scientific journal, 14(7), 91-102. http://dx.doi.org/10.19044/esj.2018.v14n7p91 ma’aji, m. m., & barnett, c. (2019). determinants of capital budgeting practices and risks adjustment among cambodian companies. archives of business research, 7(3), 171-182. https://doi. org/10.14738/abr.73.6351 ma’aji, m., sok, p., & long, c. (2020). working capital financing preference among small businesses in cambodia. journal of economics and sustainability, 2(1), 48-61. https://doi. org/10.32890/jes majumdar, s. (1997). the impact of size and age on firmlevel performance: some evidence from india. review of industrial organization, 12(2), 231-241. https://doi. org/10.1023/a:1007766324 749 mcgowan, l. (2019). smes and their importance to the uk economy. bcrs business loans. https://bcrs.org.uk/smes-andtheirimportance-to-the-uk-economy/ narver, j. c., & slater, s. f. (1990). the effect of a market orientation on business profitability. journal of marketing, 54(4), 20-35. https://doi.org/10.1177/002224299005400403 national institute of statistics (nis). (2011). economic census of cambodia. national institute of statistics. https://www.nis. 27 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 gov.kh/index.php/en/12-publications/16-economic-census2011-final-results neshamba, f. (2006). why do some small businesses grow faster and become “successful” while others do not get beyond the ‘foothills’? journal of african business, 7(1-2), 9-30. https:// doi.org/10.130 0/j156v07n01_02 newbert, s. l. (2007). empirical research on the resource-based view of the firm: an assessment and suggestions for future research. strategic management journal, 28(2), 121–146. https://doi. org/1 0.1002/smj.573 o’cass, a., & sok, p. (2013). exploring innovation driven value creation in b2b service firms: the roles of the manager, employees, and customers in value creation. journal of business research, 66(8), 1074–1084. https://doi.org/10.1016/j. jbusres.2012.03.004 ohlson, j. (1980). financial ratios and the probabilistic prediction of bankruptcy. journal of accounting research, 18(1), 109-131. https://doi.org/10.2307/2490395 ong, j. w., & ismail, h. b. (2008). sustainable competitive advantage through information technology competence: resource-based view on small and medium enterprises. communications of the ibima, 1(7), 62-70. https://ibimapublishing.com/uploads/ articles/cibima/2008/ 320097/320097.pdf pena, i. (2004). business incubation centers and new firm growth in the basque country. small business economics, 22(3), 223– 236. https://doi.org/10.1023/b:sbej.0000022221.03667.82 platt, h., & platt, m. (1990). a note on the use of industry-relative ratios in bankruptcy prediction. journal of banking and finance, 15(6), 1183-1194. https://doi.org/10.1016/03784266(91)90057-s plymire, j. (1991). complaints as opportunities. journal of consumer marketing, 8(2), 39-43. https://doi. org/10.1108/07363769110035009 porter, m. e. (1985). competitive strategy: creating and sustaining superior performance. the free, new york. pratono, a., & mahmood, r. (2015). mediating effect of marketing capability and reward philosophy in the relationship between 28 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 entrepreneurial orientation and firm performance. j. glob. entrepreneurship res., 5(5), 1-12. https://doi.org/10.1186/ s40497-015-0023-x reynolds, p., & miller, b. (1989). new firm survival: analysis of a panel’s fourth year (pp. 159–172). frontiers of entrepreneurship research, babson college savi, k. (2019, 10 april). pm highlights achievements in khmer new year message. the phnom pehn post. https://www. phnompenhpost.com/national/pm-highlights-achievementskhmer-new-year-message shane, s. (1996). hybrid organizational arrangements and their implications for firm growth and survival: a study of new franchisors. academy of management journal, 39(1), 216-234. https://doi.org/10.5465/256637 smecorps. (2013). annual report 2012/2013. smecorp malaysia. sokhorng, c. (2018). smes set to boost the kingdom’s economy. https://m.phnompenhpost.com/business/smes-set-boostkingdoms-economy szumilas m. (2010). explaining odds ratios. journal of the canadian academy of child and adolescent psychiatry, 19(3), 227–229. https://www.ncbi.nlm.nih.gov/pmc/articles/pmc2938757/ teng, h., bhatia, s., & anwar, s. (2011). a success versus failure prediction model for small businesses in singapore. american journal of business, 26(1), 50–64. https://doi. org/10.1108/193551 81111124106 thornhill, s., & amit, r. (2003). learning about failure: bankruptcy, firm age, and the resource-based view. organization science, 14(5), 497-509. https://doi.org/10.1287/orsc.14.5.497.16761 unger, j. m., rauch, a., frese, m., & rosenbusch, n. (2011). human capital and entrepreneurial success: a meta-analytical review. journal of business venturing, 26(3), 341-358. https://doi. org/10.1016/j.jbusve nt.2009.09.004 wernerfelt, b. (1984). a resource-based view of the firm. strategic management journal, 5(2), 171–180. https://doi.org/10.1002/ smj. 4250050207 zmijewski, m. (1984). methodological issues related to the estimation of financial distress prediction models. journal of accounting research, 22, 59-86. https://doi.org/10.2307/2490859 29 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 appendix 1: lussier (1995) model of 15 variables variables type questions capt financial indicators businesses that start undercapitalized have a greater chance of failure than firms that start with adequate capital. frac financial indicators businesses that do not keep updated and accurate records have a greater chance of failure than firms that do. wcap financial indicators businesses with insufficient business resources (finance, raw material or finished product, equipment) have a greater chance of failure than firm’s sufficient resource. mgex management indicators businesses managed by people without prior management experience have a greater chance of failure than firms managed by people with prior management experience. bpln management indicators businesses that do not develop specific business plans have a greater chance of failure than firms that do. prad management indicators businesses that do not use professional advisors have a greater chance of failure than firms using professional advisors. a more recent source of professional advisors is venture capitalists educ management indicators owners without higher education who start a business have a greater chance of failing than owners with higher education ldsk management indicators businesses with insufficient domain of leadership skills from top management have a greater chance of failure than firms with. psti market competition indicators businesses that select products/services that are too new or too old have a greater chance of failure than firms that select products/services that are in the growth stage. mrkt market competition indicators business owners without marketing skills have a greater chance of failure than owners with marketing skills. cucm market competition indicators businesses that received customer complaints most of the time have a greater chance of failure than firms that received less customer complaint. bage market competition indicators younger business has a greater chance of failing than older businesses. (continue) 30 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 1–30 variables type questions usk environment indicators shortage of skilled people in the area made it difficult to obtain a qualified employee and contributed to the failure of business. gvst environment indicators lack of financial support in the area made it difficult to gather sufficient capital and contributed to the failure of businesses. infc environment indicators inadequate infrastructure conditions where the business was located made it difficult to transport for consumer to buy goods. noted: capital (capt), financial record (frac), working capital (wcap), management experience (mgex), business plan (bpln), professional advisors (prad), education level (educ), leadership skills (ldsk), product/service timing (psti), marketing skills (mrkt), customer complaint (cucm), business age (bage), under-skilled labour (uskl), government support (gvst), infrastructure condition (infc) insider trading and large chapter 11 bankruptcies in usa the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 1 insider trading and large chapter 11 bankruptcies in usa tareque nasser and benton e. gup the university of alabama abstract the enron and worldcom debacles raised questions about the state of corporate governance in the united states. insider trading is one aspect of corporate governance highlighted in these cases. in this paper, we explore insider trading of large chapter 11 bankruptcy filing firms during the twelve-year period of 1995-2006. we find that insiders in these firms, on average, do not use private information for gain or loss avoidance. keywords: insider trading, bankruptcy, and corporate governance jel classification: g32, g34 1. introduction several large firms that filed for bankruptcy in recent years used unscrupulous accounting and business practices including, but not limited to, insider trading. for instance, several executives of worldcom were either convicted or had confessed to fraud and illegal insider trading. bernard ebbers, chief executive of worldcom, was sentenced to 25 years in prison for orchestrating his $11 billion fraud of the bankrupted telecommunication giant.1 similarly, after confessing to fraud and illegal insider trading, some enron executives too were convicted. as a result, lawmakers passed the sarbanes-oxley act in order to enforce corporate governance. insider trading is one aspect of corporate governance that needs to be examined. in this paper we explore insider trading in large american publicly traded firms with assets of $1 billion or more that filed for chapter 11 bankruptcy protection in the test period. both the securities and exchange commission (sec) and academic researchers refer to a firm’s chairman, directors, officers, and principal shareholders with 10 1 dionne searcey, shawn young & kara scannell, “ebbers is sentenced to 25 years for $11 billion worldcom fraud”, wall street journal (eastern edition), new york, n.y., jul 14, 2005. pg. a.1. ijbf 2 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 percent or more of any equity class of securities as the insiders.2 the term insider trading includes both legal and illegal conduct. the sec considers insider trading as legal when corporate insiders buy and sell stock in their own companies without violating any rules laid out to protect the general investors’ interest. similarly, the sec characterizes illegal “insider trading” as the buying or selling of securities by any insider in breach of a fiduciary duty or other relationship of trust and confidence, while in possession of material, nonpublic information about the security. insider trading violations may also include “tipping” such information to an individual and securities trading by those who misappropriate such information. however, these tip-initiated trades are not documented and filed with the sec as insider trading. the phrase insider trading in finance literature is often used both in the general sense and as the illegal use of private information for personal gain. in most cases it is the latter. it is generally assumed that readers can distinguish between the two from the context of the discussion. when there are no prohibitive insider trading rules, insiders may have ample incentive to take advantage of their private information. so, the actual magnitude and pervasiveness of insider trading depends upon the trade-offs between the benefits and costs that insiders observe and perceive. the securities exchange act of 1934 (hereafter, 1934 act) was enacted to insure a “fair and honest” market. the abusive use of private information by insiders is dealt in three ways (bettis, duncan, and harmon, 1998): 1. insiders are obligated to disgorge any profit from the buying and subsequent selling of securities within a six-month period, even if the trade is not based on private information. this is known as the short-swing rule. 2. there are numerous reporting requirements by insiders such that any trading undertaken by insiders is transparent. insiders of the publicly traded companies need to report their trades and changes in ownership to the sec using three different forms.3 these are forms 3, 4, and 5. a. form 3 is the initial statement of beneficial ownership for all officers. b. form 4 reports change in an insider’s ownership position. c. form 5 is the annual statement of change in beneficial ownership. in addition to the above three forms, form 144 is the insider’s declaration of their intention to sell restricted stock. 2 the legal definition of a corporate insider is any person who is obligated to report and file his/her securities trading with the sec. they include a company’s officers and directors, as well as any beneficial owners of more than ten percent of a class of the company’s equity securities registered under section 12 of the securities exchange act of 1934. 3 previously, insiders were to report any changes in their holding by the 10th day of the following month. however, after the sarbanes-oxley act of 2002, insiders are required to report their trade within two business day after the trade has occurred. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 3 3. it is unlawful to use any private nonpublic information to make profit through trading. in fact, the insider trading sanction act of 1984 (itsa) and insider trading and securities fraud enforcement act of 1988 (itsfea) require more severe civil and criminal penalty for insider trading. ista stipulates that the penalty that may be imposed is determined by the court in light of the facts and circumstances, but should not exceed three times the profit gained or loss avoided as a result of such unlawful purchase, sale, or communication. the itsfea allows criminal fines up to $1,000,000 from individuals, and maximum jail up to 10 years. how effective the sec has been with discouraging insider trading is a question explored in many studies with mixed evidence (seyhun, 1992; bettis et al. 1998; bris, 2005). here, we attempt to add to the literature concerning the state of insider trading in large u.s. corporate firms facing bankruptcy. we seek to examine a sample of 129 large firms that filed for bankruptcy over the period of 1995-2006. appendix a lists out the names of these 129 firms, their bankruptcy filing date, and the asset size prior to bankruptcy. all these firms had assets over a billion dollars two years prior to their bankruptcy filing. although a total of 137 firms met the criteria of our sample selection, we had to drop 8 firms because of data limitations. table 1 shows some characteristics relating to the bankruptcy for all of the 129 firms. in this table, “fraud” implies bankruptcies caused by fraud claims (including securities fraud claims) against the company. these cases often began with financial difficulties from other causes, which were concealed from the investors until they were severe enough to cause the bankruptcy. “disposition” is the outcome of the bankruptcy case. “confirmed” means that the court confirmed a plan of reorganization. the term “§ 363 sale” means that the debtor sold all or nearly all of its assets during the chapter 11 case.4 “pending” means that the case remains pending in the bankruptcy court. “emerged” means that the company has either emerged as it was before the bankruptcy or as some other form (i.e., taken over or merged with some other firms), but did not die out. as shown in table 1, 123 out of the 129 firms sampled had their chapter 11 bankruptcy case filed by the debtor, 5 by the creditors, and 1 by both the debtor and creditors. note that 10 out of the 129 firms shown in table 1 had committed fraud, which ultimately led to bankruptcy.5 out of these 10 firms, 8 firms emerged from bankruptcy. one firm did not survive, while another’s case is still pending. 4 usually, if a chapter 11 debtor wanted to sell all or nearly all of its assets, it would do so through a plan of reorganization or a plan of liquidation. lately however, bankruptcy courts have permitted such a sale based on section 363 of the bankruptcy code prior to the plan solicitation and confirmation process, so long as certain requirements are satisfied. this is commonly known as the “§ 363 sale”. 5 the 10 firms that filed for chapter 11 because of fraud are indicated in appendix a with the symbol *. source: lynn m. lopucki’s bankruptcy research database (webbrd). 4 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 table 1: characteristics of 129 sample firms filing for chapter 11 during 19952006 filing parties: creditor debtor both 5 123 1 cause of bankruptcy: fraud other than fraud 10 119 disposition: confirmed § 363 sale pending 92 16 21 emerged did not emerge 67 25 source: lynn m. lopucki bankruptcy research database (webbrd). the bankruptcy courts have confirmed reorganization plans for 92 firms, while 21 cases are pending trial. the remaining 16 firms of the 129 firms went though the “§ 363 sale.” out of 92 firms whose reorganization/liquidation plans have been confirmed by the bankruptcy court, 67 emerged as either the same firm as before or in some other form, while 25 firms did not survive. interestingly, none of the firms whose insiders were charged with fraud went through a “§ 363 sale”. this implies that at least some of the insiders of the 50 firms were at risk of losing their jobs because of either fraud or dissolution.6 on the other hand, it is usually the firms that file for the chapter 11 bankruptcy, not the creditor. these facts suggest that there were several reasons why insiders were unlikely to do trading based on insider information. firstly, illegal insider trading is punishable by incarceration; secondly, losing their job because of illegal insider trading may impair their future employability; and thirdly, the chapter 11 bankruptcy process may enable managers to save their job at the time of financial distress. in this paper, we compare the level of insider trading of large firms filing for bankruptcy to a set of control firms. we find that insiders of firms filing for bankruptcy do not significantly sell more or buy less than the control firms in similar 6 (50 = 10 + 25 + 16 – 1). the number of bankruptcy filings caused by fraud = 10; number of firms whose reorganization/liquidation plan was confirmed, but did not survive = 28; number of firms that went through “363” sale = 15; and number of firms that committed fraud and did not survive = 2. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 5 industries. both the number of trading insiders and the dollar volume of trading in each quarter, for 8 quarters, before the bankruptcy filing date, measure insider trading. section 2 aims to discuss the relevant literature for our study, while section 3 covers the data and methodological issues measuring insider trading. section 4 describes our own study and its results, with section 5 naturally providing the conclusion. 2. review of literature 2.1 insider trading in the event of bankruptcy although not focused on illegal insider trading, numerous academic studies report that insiders earn abnormal returns.7 alternatively, many studies document the presence or absence of insider trading around various corporate events or public announcements. for instance, elliot, morse and richardson (1984) had tested whether insiders trade profitably before the public release of earnings, dividends, bond ratings, mergers, and bankruptcies. they had found that most insiders’ trading was not related to these events. penman (1982) on the other hand, found that insiders do time their trade according to the annual earning forecast disclosure date, earning abnormal returns. karpoff and lee (1991) had found that insiders sell shares prior to new equity issues, while harlow and howe (1993) had found that insiders accumulate shares abnormally prior to management buyout. similarly, the question of whether insiders take advantage of material, nonpublic information of the impending bankruptcy has yielded mixed results. loderer and sheehan (1989) do not find insiders “bailing out” of the firm. in contrast, gosnell, keown, and pinkerton (1992) had found that insider selling increases significantly as the bankruptcy announcement approaches, particularly for over the counter (otc) traded firms. seyhun and bradley (1997) had found that insiders systematically sell and buy stocks around the bankruptcy event to make abnormal profits. ma (2001) however, found that there is not much selling by insiders before bankruptcy. but there is the significant lack of purchases by the insiders before the bankruptcy. iqbal and shetty (2002) had viewed the problem a little differently. they had found that there exists significant insider trading around the month when the market anticipates the bankruptcy, not around the month when the bankruptcy filing is announced. 2.2 motivations for insider trading the extent of insider trading depends upon the expected benefits and costs, whether they can foresee the future bankruptcy. in an experimental study using college students, beams, brown, and killough (2003) had found that the subjects are more likely to trade based on insider information to avoid loss than to achieve abnormal 7 see, for instance, jaffe (1974); seyhun (1986); rozeff and zaman (1988); lin and howe (1990); meulbroek (1992); and lakonishok and lee (2001). 6 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 gain. it is often difficult to pinpoint whether a particular trade is motivated by selfish personal gains, or if it is carried out to implement the implicit compensation contract for the insiders. recent, notorious bankruptcy cases are examples where insiders resorted to illegal measures for personal gains at the expense of the shareholders. it is often not clear whether insider trading (in a general sense) is a result of an agency problem or whether it reduces agency problems. it can be argued both ways. carlton and fischel (1983) had argued that the ability of an agent to exploit his/her informational advantage could be part of the implicit or explicit contract between the shareholder and the agent. in fact, trading allows an agent to renegotiate the contracts when underlying conditions change (dye (1984). however, fischer (1992) had argued that when there is an agency problem, trading by the agents might in turn aggravate the agency problem. fischer (echoing easterbrook, 1981), argued that the uncertain compensation through trading results in suboptimal risk sharing. also, the agents may have added an incentive to cook the books and take actions that may create profitable trading opportunities for agents at the expense of shareholders. when bankruptcy occurs, there is high probability the insiders can lose their jobs and any other associated quasi-rents. this may prompt them to take actions consistent with what fischer (1992) had argued, particularly in absence of significant, legal deterrence mechanisms. moreover, if the insiders bail out by selling their stock, the subsequent restructuring process may not be in the best interest of the other shareholders. stockholders’ gains are directly related to the insiders’ holding because the insiders’ wealth is at stake betker (1995). 2.3 chapter 11 bankruptcy and insider trading under u.s. bankruptcy laws, firms have a choice between filing for chapter 7 and chapter 11, where they go through liquidation and reorganization processes, respectively. under chapter 7, a firm ceases its business immediately, and goes through the liquidation process by an appointed trustee to pay off its debts. whereas under chapter 11, the incumbent managers still remain in control as the firm continues to operate for the duration of the reorganization process. at the same time, a reorganization plan is adopted through which all pre-bankruptcy claims are settled or renegotiated. however, management with strong bargaining power can pressure the creditors to accept their terms [white (1989), and managers bargaining power is the greatest when firm is close to solvency (betker (1995)]. it is also interesting to note that during the chapter 11 bankruptcy process, the only way the stockholder’s interest is represented is by the management.8 betker (1995) had reported how courts have held that shareholders cannot hold a meeting to replace the directors of insolvent firms. all these facts point out that the invoking chapter 11 bankruptcy process is a wonderful opportunity for the managers to save their jobs, while at the same time, retain significant bargaining power when the firm is on borderline solvency. bradley and rosenzweig (1992) had stated that “…chapter 11 does not require that a debtor be insolvent in order to qualify for reorganization, and it includes a strong presumption favoring retention of management throughout the reorganization 8 this obviously involves deviations from absolute priority in chapter 11 bankruptcy. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 7 process. thus, in the ordinary case, a chapter 11 filing transforms a corporate debtor into the “debtor-in-possession” and leaves existing management in control of the firm’s resources.” if management is relatively certain that the bankruptcy restructuring process is just a temporary adjustment to solidify their grip on the reign of the firm, we can expect that there would be no insider trading. this is even more pertinent to big firms as it is often believed that they are “too-big-to-fail.” 2.4 impediments to insider trading many public companies have policies governing the trading of their securities by officers, directors, and employees, with the goal of preventing trades at times when insiders may be in possession of material nonpublic information (bettis, coles and lemmon, 2000). such policies are designed to ensure compliance with securities trading prohibitions. they generally contain periodic “window periods” or “blackout periods” tied to the company’s announcements of financial results or other corporate developments. during window and blackout periods, trades by insiders are conditionally permitted or prohibited, respectively. bettis et al. (2000) had reported that although firms are not required by law to implement periodic insider trade restricting policies. the absence of such policies may be deemed as a reckless conduct by the firms from the regulator’s point of view. insider trading policies often require insiders to obtain an advance trading clearance with a designated member of the management. public companies normally include trading windows or trading blackout periods in insider trading policies to prevent insiders from trading at or around the time of earnings announcements or the dissemination of other sensitive corporate information. in the event of job loss due to bankruptcy, managers want to preserve their future employment opportunities. lopucki and whitford (1993) had argued that managers could not breach their fiduciary duty at the expense of the firm because they may be stigmatized, which in turn, may adversely affect their marketability. however, lopucki and whitford did not discount the possibility of some unscrupulous insiders who resort to illegal means for immediate gain at the expense of shareholders and their own future human capital loss. to a certain extent, the incidence of informed insider trading also depends upon the corporate governance system in place. for instance, becht, bolton, and roell (2003) had asserted that executive stock options (eso) “…are at best an inefficient financial incentive and at worst create new incentive or conflict-of-interest problems of their own.” they had also criticized other mechanisms of corporate governance for their inability to exert corporate control. holmstrom and kaplan (2003) however, had responded by stating that this is an exaggeration. they had argued that although there were instances of stock price manipulation because of the eso carried out by firms like worldcom, these are by no means representative. holmstrom and kaplan had contended that the u.s. corporate governance system with the recent regulatory overhauling like sarbanes-oxley act of 2002 (sox) is “…likely to make a good u.s. system [a] better one.” bettis et al. (2000) had investigated the corporate policies restricting insider trading, reporting that by november 1996, 92 percent of their sample firms have instituted policies to restrict 8 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 and manage insiders’ trading, with 80 percent of the sample firms having explicit blackout periods prohibiting insider trading. thus, the extent to which the corporate control mechanisms deter insider trading is an empirical question. 2.5 laws and insider trading the deterrence of informed insider trading requires both good corporate control and effective enforcement of insider trading regulations. effective enforcement does not necessarily imply a total prohibition on inside trading. for instance, shin (1996), while modeling an optimal regulation of insider trading, had concluded that in the presence of research-informed market professionals, some insider trading is better than a total ban on insider trading to help decrease the loss of liquidity traders. similarly, demarzo, fishman and hagerty (1998) had recognized that effective insider trading regulation requires monitoring and enforcement in a market setting. however, in doing so, regulators must use market data that is very costly to decipher. based on their model, they had concluded that it is optimal for the regulators to investigate if and only if trading volume exceeds some threshold, where the threshold depends on the information released concerning share value. demarzo et al. had also proposed that if the insider is caught trading more than above a certain critical level, then he/she must pay the maximum, feasible penalty. a study conducted by bris (2005), based on data from 52 countries, had found that the incidence and profitability of insider trading increases with the enforcement of insider trading laws. he had measured insider trading before the tender offer announcement and found that after some enforcement of insider trading laws, insiders had appropriated a larger portion of the takeover gains. however, he had concluded that harsher laws work better in reducing insider trading. bris had found that the u.s., among all 52 countries, has the toughest insider trading regulation and the lowest profit for the insiders. most academic studies on insider trading deterrence concluded that harsher penalties discourage insider trading. it seems that the u.s. legal system is going towards a harsher penalty over major insider trading and frauds as evidenced at the beginning of this paper. agrawal and jaffe (1995) had reported that even though the sec may not actively prosecute short-swing trades, lawyers generally challenge them if they perceive the prosecution to be profitable. agrawal and jaffe had explained that all these lawyers have to do is to buy a share of the company where they find any violations of the short-swing rule and file the suit as a shareholder. when successfully prosecuted, they received the legal fees from the trading profits that the insider returns to the firm. based on the previously cited studies, we infer that the two most likely types of illegal insider trading involves: 1. small trades that fall into the category of short-swing trades and those that the regulatory authority may overlook; 2. large trades that result of gross breach, fiduciary duties. insider-trades of intermediate size are not worth the risk of getting caught at the expense of future economic and human capital loss. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 9 3. methods and data 3.1 measuring insider trading 3.1.1 use of purchases and sales information to measure insider trading one of the earliest measurements of insider trading conducted by lorie and niederhoffer (1968) is based on identifying an intensive selling month or buying month, and then measuring the market price movement in the six months subsequent to the event. they define an intensive selling (buying) month as a month with at least two more sellers (buyers) than buyers (sellers) among the insider of a company. however, lorie and niederhoffer do not measure the presence of insider trading around any particular corporate event or information released; rather, they simply attempt to establish its existence. jaffe (1974) had used a modified method to measure insider trading. he computed the number of net buyers9 and sellers for each firm during a randomly picked month. he then defined a month as a “month of net purchasers” if the number of net buyers is more than the number of net sellers, similarly defining the “month of net sellers.” he had regarded these months as information based trading events. jaffe had then employed a technique similar to the event study methodology developed earlier by ball & brown (1968) and fama, fisher, jensen, and roll (1969) to measure the abnormal return. 3.1.2 using car to assess insider trading the use of abnormal returns and cumulative abnormal returns (car) is widely recognized in academic studies. seyhun and bradley (1997) had based their whole insider trading argument using event study methodology in several ways. one obvious way they had implemented the event study method is to measure the abnormal returns and car. however, seyhun and bradley had altered the standard event study method to take account of some unique features of the data during the event of bankruptcy. for example, it is quite normal that before bankruptcy is announced, several firms either cease trading of their shares temporarily or they are delisted altogether. also, when the prices fall to single digit levels, standard car estimates may grossly misrepresent the true abnormal return over the period. for these reasons, instead of using a standard market model, seyhun and bradley had used a bootstrapping method to calculate the mean abnormal holding period return over different periods.10 another way they use the event study method is to compute abnormal insider trading, where instead of returns, they measure insider trading in trading volume. 9 a net buyer here is if an insider buys more days than sells in a month. a net seller is defined analogously. 10 seyhun and bradley (1997) had measured the abnormal returns, the difference between the return of the investing portfolio of firms that filed for bankruptcy, and the average return of the bootstrap distribution for the set of control firms’ bankrupt portfolios. 10 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 3.1.3 setting the benchmark level of insider trading while measuring insider trading, it is crucial to establish a normal insider-trading level that accounts for portfolio rebalancing, liquidity needs, or any other reasons except private information based trading. most “abnormal” insider trading is evident by documenting any significant excess trading, lack of trading, or deviations from the normal level in the direction indicated by the nature of the information. therefore, how the researchers account for the normal level of trading is reflected in their methods they had implemented. in many cases, the normal level of trading is established using a set of control firms. typically, it involves matching the subject firms’ two digit sic code, while controlling for size in terms of either asset size (see loderer and sheehan, 1989) or market capitalization (gosnell et al., 1992 and ma, 2001). the matching is generally done based on data related to the above; that is, two to five years prior to the event. seyhun and bradley (1997) had criticized this method. firstly, they had picked a set of control firms from the same industry with similar problems (e.g. financial distress). secondly, with bankruptcy, the subject firms may shrink in size over the period, creating a size disparity between the subject and the control firms. however, seyhun and bradley (1997), with their modified use of event study to measure abnormal insider trading, did not bypass the use of control firms. in their study, seyhun and bradley had centered the time series of the insider trading data on the event date, and then measured the average trading per firm for each month over the event window. here, they had designated sales as a negative number and purchase as a positive number. they had measured the abnormal insider trading (both in number of shares traded and in dollars) for each period (in months) as the difference between the actual trading activity for a given firm and the amount of insider trading activity of a “control portfolio” of similar sized firms over the same period. they had constructed these control portfolios by first calculating the mean annual equity value of all firms reporting any insider trading activity over their study period. then, they had ranked these firms, excluding the subject sample firms, according to market capitalization, categorizing them into deciles. using this method, they had created the “expected” insider trading data for portfolios of firms of each size decile and insider category for each month over the study period. however, we believe that judiciously choosing the time of matching (i.e., not too far away from the event time) may alleviate the size disparity problem significantly, neglecting those similar methods as adopted by seyhun and bradley. furthermore, agrawal and jaffe (1995) had argued that matching based on two-digit sic classification is broad enough to have little or no effect on the control firm’s event. 3.1.4 bypassing the control firms to assess insider trading many researchers forgo the use of control firms altogether by trying to establish the normal level of insider trading by using time series data. karpoff and lee (1991) had developed such a technique, while those by the likes of harlow & howe (1993), iqbal & shetty (2002), and irani (2003) have used it. karpoff and lee the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 11 (1991) had derived the mean abnormal and the cumulative number of net sellers and associated statistics to assess the abnormal number of net sellers in the event of new issues of common stock. harlow and howe (1993) on the other hand, had used the same method to find the abnormal number of net buyers for leverage buyout announcements. when researchers do not use control firms, they use other approaches. one general approach used by finnerty (1976), penman (1985), john and lang (1991), and lakonishok and lee (2001) had involved the development of indices or ratios. the typical way of constructing these would be computing ratios as or or some other complicated variations. here, p may represent the number of insiders purchasing stock, the number of purchasing transactions, the number of shares purchased, the dollar value of transactions, or any other construct that researchers feel would convey information related to the study. similarly, s would represent any of the above corresponding constructs related to sales. these ratios were used in different regression equations to explain returns or abnormal returns, thereby, evaluating insider trading. 3.1.5 method used in this paper many researchers believe in the efficacy of control firms for benchmarking purposes. of course, the form of analysis may differ considerably. for instance, while assessing insider purchases around merger announcement, agrawal and jaffe (1995) had used matched control firms in addition to time-series controls. they had used matchedpair t-statistics for measuring the cross-sectional difference between the means of the target and control firms. however, ma (2001) had argued that insider-trading activity is quite infrequent, non-normal, and highly skewed in magnitude, thus calling for statistical analysis that does not require distributional assumptions. he argued that non-parametric methods are suitable for this purpose. hence, he used the wilcoxon signed-rank test (which is a non-parametric technique) to compare insider-trading measures of bankruptcy firms with those of the control firms. we use this method in our study along with other descriptive statistics. the next obvious issue is the appropriate measures of insider trading. agrawal and jaffe (1995) had used several measures for insider trading. these are the number of insiders who buy in a given period, shares purchased in a period, dollar value of purchase, and percentage of equity bought; whereas, ma (2001) uses the number of transactions, number of insiders, and dollar amount of trade. other measures like percentage of outstanding equity purchased, percentage of total share purchase volume, and percentage of total dollar purchase volume are also used in the literature. against this background, we concentrate on two constructs: 1. the number of insiders buying or selling each quarter; and 2. the purchases or sales in dollars (trading volume) for each quarter. the number of insiders helps to explain how widespread insider information utilization is present within the firm. generally, we do not expect many insiders trading at the same time unless there is some motivating reason, such as the awareness of price changing information. hence, if we see that the number of insiders who p-s p+s p-s (p+s)/2 12 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 are trading in a particular period is significantly higher, it may be because of inside information. for instance, if we find the number of insiders that are selling in a particular quarter for the firm filing for bankruptcy is abnormally high, then it is evidence of insider trading. finally, trading volume captures the magnitude of use of insider information. we will use these two constructs at the same time. if only one of them is significant, while the other is not, it is not a strong case of insider trading. however, if both are significant at the same time, it is strong evidence of insider trading. 3.1.6 justification of the method used in this paper although there is no clear cut evidence that any one method described above is superior to others, we selected the method used in this paper for several reasons. firstly, we were in search of a simple proven test to measure the extent of insider trading. the method that we had adopted meets this criterion. secondly, we avoid the usual criticism of the use of control firms to set normal insider trading levels. later, we will show how the criticism of seyhun andd bradley (1997), in using control firms from the same industry may produce firms with similar problems like financial distress, is absent in the control sample. moreover, seyhun and bradley’s other objection in using control firms was that the subject firms might shrink in size over the period, creating a size disparity between the subject and the control firms. they had avoided this by using shorter time periods (i.e., using two years prior data instead of five years prior data for the matching process). thirdly, insider-trading activity is infrequent, non-normal, and highly skewed in magnitude. hence, using a non-parametric technique like the wilcoxon signedrank test, we can do away with the distributional assumption that is needed for other statistical techniques. 3.2 data 3.2.1 sources of sample firms professor lynn m. lopucki’s bankruptcy research database (webbrd)11 provides an extensive listing of firms that filed for bankruptcy in the u.s. bankruptcy courts. it also has a data query, web interface, where researchers can get lists of bankruptcy filing firms with a variety of specifications. we used this web query to get a list of firms that had an asset size greater than or equal to a billion dollars prior to filing for bankruptcy during the 1995-2006 period. the asset size of the firms filing for bankruptcy in the webbrd is recorded in current dollars. however, we also further screened these firms based on the criterion that firms should have asset size, as reported in compustat, over a billion dollars two years prior to filing for bankruptcy. this criterion is imposed to ensure that there is no mistake in the asset size reported in the webbrd,12 as well as the fact that we had decided 11 researchers who want to know more about webbrd’s data sources and quality may browse (http://lopucki.law.ucla.edu/frequently_asked_questions.htm). 12 we found some discrepancies between the asset size reported in webbrd and compustat, after accounting for current dollars conversion. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 13 to use matched control firms based on asset size and industry for our analysis.13 it yielded 137 firms. unfortunately, we were not able to get the insider trading data for 8 firms. hence, our final sample includes 129 firms that had assets of more than a billion dollars that filed for bankruptcy during the 1995-2006 period. 3.2.2 getting the insider trading data most of the previous studies surrounding insider-trading used data from the ownership reporting system (ors) or the official summary of securities transactions and holding, which is published by the sec. currently, thomson financial provides the same information through tfn insider filling data (tfn insiders). we used the thomson financial insiders data (tfn insiders) to get the insider trading data. note that it is possible that some insiders buy and sell their securities without reporting their trade to the sec (meulbroek, 1992). however, we have no information on either accepting or refuting such activity. hence, the data that we are using is based on only the reported trades. to extract the appropriate data, we needed each firm’s corresponding cusip issuer code (cusip6), and the date when they filed for bankruptcy. webbrd only provides the bankruptcy-filing year without the exact filing date. therefore, we obtained the bankruptcy filing date by searching the 10-k wizard sec filings. when these were not available in the 10-k wizard, we obtained the chapter 11 filing dates through web browsing different news sources and “bankruptydata. com.” the cusip6 was derived from matching each firm’s name within compustat and crsp. we obtained insider-trading data for the 129 firms in our sample covering a two-year period prior to the bankruptcy from tfn insiders. we only use the open market purchases or sales data in tfn insiders, excluding purchases related to options exercises. insider trading activity equals zero for any firm that has a registered insider documented by tfn insiders, but has no open market trades reported by tfn insiders during the sample period. we exclude trades that buy or sell fewer than 100 shares or with a total value exceeding $100. trades lacking transaction prices in tfn insiders are excluded from our sample. we use trading data for two years prior to the filing for bankruptcy. although we believe that insiders cannot foresee the actual bankruptcy as far as two years down the road, we still use the two-year time span, since it is the shortest period used in the previous literature. 3.2.3 constructing the set of control firms like previous studies, we used a set of control firms to assess the prevalence of insider trading. given the choice between matching control firms based on market capitalization and asset size, we opted for the latter for two reasons: 1. we picked our sample firms based on asset size; 13 as it will be explained later in the paper, we matched asset sizes for two years prior to bankruptcy using compustat industrial annual and a 2-digit sic code to get the control firms. 14 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 2. rapidly growing firms can have a market capitalization that is equal to or greater than that of some mature firms that are growing slowly or shrinking (gup and agrawal, 1996). thus, market capitalization, per se, can be misleading when comparing across the two samples. consequently, we matched asset sizes for two years prior to the bankruptcy using compustat industrial annual and a 2-digit sic code to get the control firms. although, we could obtain a match for most of the firms, if the firm’s asset size was missing in the two-year priors to bankruptcy, we used one-year prior data. also, if the closest match was an adr of some foreign firm, another bankruptcy filing firm, or a firm not in the tfn insiders, then we found the next closest match in asset size. table 2: comparison of bankrupt firms and matched control firms panel a: summary statistics of asset size (in million dollars) bankrupt firms asset matched firms difference n 129 129 129 mean 6,614 6,375 554 t-test 1.2363 median 2,579 2,399 79 wilcoxon sign-rank test 0.858 std. dev. 12,167 12,018 2,131 panel b: number of firms with zero trading bankrupt firms matched firms purchase 29 31 sales 28 21 both 16 13 none of these statistics are acceptable at the normal confidence levels. the descriptive statistics of the sample bankrupt firms and their control counterparts are shown in panel a of table 2. the mean and the median of the bankruptcy sample firms’ asset size are $6,614 and $2,579 million, respectively. for their matched control firms, they are $6,375 and $2,399 million dollars, respectively. the standard deviation of the asset size for the subject firms is $12,167 million and for the control firms is $12,018 million. the t-test for the mean difference and the wilcoxon sign-rank test for the median difference are not statistically significant in any of the commonly used significant levels. hence, the asset size distributions of both the samples are similar. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 15 as shown in panel b of table 2, 16 firms had no insider trades for two years prior to their bankruptcy. for the matched firms, 13 firms had no insider trading during the same period. based on panel b of table 2, we find that 41 (29 + 28 – 16) of the bankrupt firms have no insider purchase and/or sales during the period under examination. the corresponding number for the matched control firms is 39 (31 + 21 – 13). this implies that more than 30% of the bankrupt firms and their corresponding control firms had no insider trades for the two years prior to the filing date. stated otherwise, insider trading is an infrequent activity. 4. results and analysis we cannot talk about the existence of insider trading before we look into the abnormal returns. it is obvious that significant, negative abnormal returns prior to bankruptcy filings provide the motive for insider trading. panel a of table 3 shows average cumulative abnormal returns (caar) of the bankrupt firms over different periods.14 these are calculated based on the ols market model using the crsp equally weighted index. for the 58 trading days, the period starting 60 days before the date the firm filed for bankruptcy and ending 2 days before the filing date relative to the filing date (-60, -2), the average caar is -38.18%. the patell z-value is 12.201 and is significant at the 0.1 percent level. similarly, for the period (-30, -2) and (-1, 0), the average caars are -39.86 percent and -20.35 percent, respectively, while both are significant at the 0.1 percent level. this test determines whether the abnormal stock returns equal zero, assuming cross-sectional independence. from this, we can conclude that there is ample loss of returns at stake, thus creating a strong motivation for insider trading. table 3: caar of bankrupt firms and their matched control firms panel a: bankrupt firms days n caar patell z (-60,-2) 45 -38.18% -12.201*** (-30,-2) 45 -39.86% -14.613*** (-1,0) 43 -20.35% -23.317*** panel b: matched control firms days n caar patell z (-60,-2) 94 -2.80% -0.997 (-30,-2) 94 -1.70% -1.075 (-1,0) 94 -1.24% -2.143* the patell z-test examines whether abnormal stock return equals zero assuming cross sectional independence. the symbol *and *** denotes statistical significance at the 0.10 and 0.001 levels respectively using a 1-tail test. 14 note that the number of firms used in estimating the car is smaller than 129 sample bankrupt firms. this is so due to the fact that many firms stopped trading their shares some time prior to the bankruptcy filing, hence, they lack data to calculate the car. 16 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 panel b of table 3 shows the average caar of the control firms over the same period used for their matching bankrupt firms. we are interested in the average caar values to establish that the control firms are not going through similar financial distress or some other events that will cause abnormal returns. we find that for the three time periods (-60, -2 days), (-30, -2 days), the average caars are -2.80 percent, -1.70 percent, and -1.24 percent, respectively, and only the period (-1, 0 days) caar is significant at the 5 percent level of significance. figure 1 shows the caar over the 60 days prior to the filing date for both the samples. the matched control firms’ caar does not show any significant movement compared to that of the bankrupt firms. this dispels the possibility raised by seyhun and bradley (1997) that using control firms from the same industry may produce firms with similar problems like financial distress. figure 1: bankrupt firms and their matched control firms’ caar for a period of two months prior to filing for bankruptcy 7 0 . 0 0 % 6 0 . 0 0 % 5 0 . 0 0 % 4 0 . 0 0 % 3 0 . 0 0 % 2 0 . 0 0 % 1 0 . 0 0 % 0 . 0 0 % 1 0 . 0 0 % 2 0 . 0 0 % 3 0 . 0 0 % 6 0 5 0 4 0 3 0 2 0 1 0 0 days matched firms' caar bankrupt firms' caar c a a r as previously mentioned, we compared the bankrupt firms to matched control firms in order to evaluate the extent of insider trading. table 4 provides the descriptive statistics on trading volume in millions of dollars. we categorized firms into three categories: 1. positive net purchase (i.e., firm purchased more dollars worth of shares than they have sold); 2. negative net purchase; 3. zero net purchase (essentially, these are firms with no trading activity). data on panel a is based on two years of insider trading data of both the bankruptcy and control firms. the two-year time span starts from two years prior to the date of the bankruptcy, and the same time span is used for the corresponding control firm. similarly, data on panel b and c is based on quarterly and monthly the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 17 data, respectively, prior to the bankruptcy. if the insiders use their private information about future bankruptcy, they would sell more than they would purchase, and we would expect more firms categorized as the negative net purchase firms. however, in panel a, we observed that in the bankruptcy sample, only 58 of the 129 firms were negative net purchase firms, which is roughly equal to the number of positive net purchase firms. in the control sample, negative net purchase firms are almost thrice in number compared to the positive net purchase firms. over a shorter period of time, as shown in panels b and c, the number of firms with negative net purchase declined sharply. table 4: descriptive statistics of trading volume over different time horizons prior to the filing date for bankruptcy panel a: net purchases of firms for two years prior to bankruptcy filing date (in million dollars) positive net purchase negative net purchase zero mean median sd mean median sd b an kr up t f ir m s purchase 8.81 1.21 20.69 2.34 0.18 7.33 sale 1.73 0.09 6.31 77.99 8.12 188.86 net purchase 7.08 0.95 16.51 -75.65 -6.47 189.21 n 55 58 16 m at ch ed f ir m s purchase 31.78 2.88 106.99 3.89 0.20 27.83 sale 4.39 0.18 12.08 80.51 11.67 224.76 net purchase 27.39 1.95 96.59 -76.62 -11.09 215.06 n 33 83 13 panel b: net purchases of firms for one quarter prior to bankruptcy filing date (in million dollars) positive net purchase negative net purchase zero mean median sd mean median sd b an kr up t f ir m s purchase 2.32 0.11 4.42 0.05 0.00 0.22 sale 0.97 0.00 2.84 1.72 0.23 3.34 net purchase 1.35 0.11 2.41 -1.67 -0.23 3.33 n 15 24 90 m at ch ed f ir m s purchase 1.77 0.16 3.81 0.01 0.00 0.03 sale 0.05 0.00 0.16 15.93 1.45 63.34 net purchase 1.72 0.15 3.69 -15.92 -1.45 63.34 n 22 51 56 panel c: net purchases of firms for one month prior to bankruptcy filing date (in million dollars) positive net purchase negative net purchase zero mean median sd mean median sd b an kr up t f ir m s purchase 1.06 0.56 1.57 0.00 0.00 0.00 sale 0.05 0.00 0.13 1.32 0.13 3.13 net purchase 1.01 0.43 1.57 -1.32 -0.13 3.13 n 6 10 113 18 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 m at ch ed f ir m s purchase 0.48 0.06 0.78 0.01 0.00 0.02 sale 0.00 0.00 0.01 5.34 0.76 12.50 net purchase 0.48 0.05 0.79 -5.33 -0.76 12.50 n 10 29 90 a closer inspection of the two samples had revealed that all statistics in the negative net purchase category are larger in size for the control firms. similar results were found for the positive net purchase firm. hence, the statistics reported in table 4 suggest that the insiders of the bankrupt firms as a whole do not sell or buy more than other firms. also, insiders of a larger number of bankrupt firms stopped trading as the period for bankruptcy filing approaches. one may argue that some insiders of the bankrupt firms do sell more shares, but the dollar value is lower because of lower stock prices. however, the use of private information would imply that insiders should have sold the shares before the price declined. for instance, the ceo of warnaco group inc., one of the bankrupt firms in our sample, sold approximately 1 million shares for $12.4 million 18 months before the filing date, buying back these shares for $3 million, 12 months later (6 months before bankruptcy), ultimately netting $9.4 million. we do not know the real reason of these trades, yet, it raises questions about the use of private information. table 5: trading volume this table presents the summary statistics of the trading volume along with t-test for mean equality and wilcoxon signed rank test for the equality of the median. if the test statistics take a negative value that implies the bankrupt firms’ trading volume is lower than that of their matched control. the last two columns present the percentage of banks that have zero insiders trading (in panel a only purchase, and in panel b only sales). note that since the median data is always zero, it is not presented here. panel a: purchase volume (in million dollars) bankrupt firms matched samples statistics % of zeros quarter mean s.d. mean s.d. t-test wilcoxon z target matched -1 0.28 1.64 0.31 1.68 -0.127 -2.025** 86.05 75.19 -2 0.64 4.98 0.66 4.20 -0.030 -1.135 79.84 70.54 -3 0.95 6.20 0.91 6.19 0.057 1.489 69.77 76.74 -4 0.34 1.65 0.26 1.42 0.390 0.227 68.99 72.87 -5 0.80 5.23 0.89 5.34 -0.143 0.267 65.89 70.54 -6 1.09 5.73 6.87 58.40 -1.119 -0.273 70.54 69.77 -7 0.54 2.95 0.60 3.02 -0.151 1.211 53.49 68.99 -8 0.17 0.54 0.13 0.62 0.488 0.614 68.22 68.22 (continued) the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 19 panel b: sales volume (in million dollars) bankrupt firms matched samples statistics % of zeros quarter mean s.d. mean s.d. t-test wilcoxon z target matched -1 0.43 1.83 6.31 40.40 -1.651* -3.563*** 76.74 57.36 -2 1.04 5.23 6.16 32.70 -1.752* -3.484*** 80.62 62.02 -3 2.42 14.90 10.90 86.60 -1.093 -3.668*** 81.4 59.69 -4 1.79 10.80 13.10 83.90 -1.512 -3.703*** 72.09 51.94 -5 3.65 17.20 4.22 16.10 -0.288 -1.485 72.87 62.02 -6 3.19 14.60 3.69 19.00 -0.414 -2.240** 72.09 58.91 -7 14.00 86.90 2.99 9.59 1.431 -0.780 63.57 60.47 -8 9.34 46.70 5.59 30.80 0.783 -0.467 62.79 55.81 panel c: net purchase volume (in million dollars) bankrupt firms matched samples statistics % of zeros quarter mean s.d. mean s.d. t-test wilcoxon z target matched -1 -0.15 1.83 -6.00 40.40 1.641 2.520** 69.77 43.41 -2 -0.40 7.28 -5.51 33.10 1.710* 2.592*** 64.34 44.19 -3 -1.46 14.80 -9.99 86.90 1.095 3.562*** 58.91 45.74 -4 -1.45 10.90 -12.80 83.90 1.522 3.365*** 50.39 41.86 -5 -2.85 18.10 -3.33 17.20 0.230 1.310 50.39 44.19 -6 -2.10 15.80 3.18 57.90 -1.054 1.703* 52.71 41.86 -7 -13.40 87.00 -2.39 10.10 -1.436 1.162 37.21 40.31 -8 -9.17 46.70 -5.45 30.80 -0.776 0.859 41.86 40.31 the symbols *,**, and *** denote statistical significance at the 0.10, 0.05, and 0.01 levels, respectively, using a 2-tail test. we performed the wilcoxon signed-rank test as well as t-test on two insidertrading measures: trading volume in dollars and the number of insiders. the results are presented in tables 5 and 6. each of the insider trading measures is documented for purchases (panel a), sales (panel b), and net purchases (panel c) for the 8 quarters prior to the bankruptcy. if we are to find profitable or loss avoiding insider trading by the bankrupt firm, the test statistics should have negative signs for purchases and net purchases, and positive signs for sales of both insider trading measures.15 however, only few statistics show the sign consistent with insider trading, and they are not even significant at any accepted level. on the contrary, any significant statistics that the table presents points to no insider trading. note that for both insider-trading measures (i.e., tables 5 and 6), insider sales (i.e., panel b) and z-statistics are significant and negative for the quarters -4 to -1. this implies that insiders of control firms are significantly selling more than that of the bankrupt 15 the test statistics are computed as (bankruptcy firms quarter i – control firms quarter i , where i = -8, -7, …, -1). 20 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 table 6: number of insiders buying or selling this table presents the summary statistics of the number of insiders buying or selling along with t-test for mean equality and wilcoxon signed rank test of the equality for the median. if the test statistics take a negative value that implies the bankrupt firms’ number of insiders is lower than that of their matched control. the last two columns present the percentage of banks that have more than one insiders trading (in panel a only purchase, and in panel b only sales). note that since the median data is always zero, it is not presented here. panel a: number of insiders buying bankrupt firms matched samples statistics % of > 2 insiders quarter mean s.d. mean s.d. t-test wilcoxon z target matched -1 0.32 1.04 0.60 1.61 -1.718* -2.125** 6.20 11.63 -2 0.40 1.11 0.70 2.28 -1.556 -1.820* 6.20 10.85 -3 0.70 1.39 0.78 3.33 -0.240 1.172 16.28 12.40 -4 0.81 1.57 0.67 3.25 0.410 1.127 19.38 9.30 -5 0.95 1.92 0.69 1.51 1.163 1.017 18.60 16.28 -6 0.72 1.52 0.64 1.54 0.409 0.734 16.28 10.85 -7 1.08 1.71 0.71 1.44 1.920* 2.363** 22.48 16.28 -8 0.98 2.01 0.66 1.66 1.464 1.035 20.93 13.95 panel b: number of insiders selling bankrupt firms matched samples statistics % of > 2 insiders quarter mean s.d. mean s.d. t-test wilcoxon z target matched -1 0.33 0.70 1.14 1.88 -4.632*** -3.976*** 6.98 27.13 -2 0.38 1.09 0.92 1.60 -3.339*** -3.561*** 6.98 20.16 -3 0.35 1.24 1.36 2.71 -4.001*** -4.691*** 5.43 27.13 -4 0.53 1.36 1.22 1.98 -3.032*** -3.397*** 10.08 24.03 -5 0.71 1.76 1.03 2.11 -1.269 -1.344 14.73 20.93 -6 0.71 1.58 1.25 2.14 -2.591** -2.484*** 14.73 27.13 -7 1.01 2.44 1.07 2.35 -0.209 -0.361 16.28 21.71 -8 1.30 2.99 1.15 1.90 0.526 -0.583 19.38 24.81 panel c: net number of insiders buying bankrupt firms matched samples statistics % of > 2 insiders quarter mean s.d. mean s.d. t-test wilcoxon z target matched -1 -0.02 1.12 -0.54 2.60 2.052** 2.109** 11.63 37.21 -2 0.02 1.56 -0.22 2.89 0.938 1.281 12.40 29.46 -3 0.35 1.85 -0.59 3.93 2.412** 4.208*** 20.93 35.66 -4 0.27 2.09 -0.54 3.51 2.217** 3.986*** 27.13 31.78 -5 0.23 2.66 -0.34 2.61 1.724* 2.305** 31.01 34.11 -6 0.02 2.27 -0.60 2.50 2.195** 2.371** 28.68 34.88 -7 0.07 3.10 -0.36 2.85 1.181 1.956* 35.66 35.66 -8 -0.32 3.74 -0.49 2.54 0.458 1.485 36.43 33.33 the symbols *,**, and *** denote statistical significance at the 0.10, 0.05, and 0.01 levels, respectively, using a 2-tail test. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 21 firms. similarly, z-statistics reported in panel c of table 5 and 6 confirms that insiders of control firms are either selling more or buying less than that of bankrupt firms. table 5 shows that in the quarter prior to bankruptcy, less than 25 percent (15 percent) of the bankrupt firms have insider selling (purchasing). similarly, table 6 shows that in the same quarter, less than 7 percent of the bankrupt firms have more than two insiders selling, whereas for the corresponding control firms, this figure is more than 25%. hence, contrary to popular belief, insiders of large public firms do not engage in trading based upon private information for private gain or loss avoidance. it does not mean that there exists no insider trading, but rather, insider trading is more of an exception than a rule. table 7: wilcoxon signed-rank test base on monthly data this table presents statistics for wilcoxon signed rank test of the equality for the median. if the test statistics take a negative value that implies the bankrupt firms’ trading volume or number of insiders, whichever is appropriate, is lower than that of their matched control. z-statistics for trading volume in dollars z-statistics for number of insiders month purchase sale net purchase purchase sale net purchase -1 -1.417 -3.391*** 2.489** -1.411 -3.478*** 2.353** -2 -1.297 -2.695*** 1.679* -1.232 -2.553** 1.152 -3 -0.469 -2.730*** 2.502** -0.793 -2.634*** 2.060** -4 -0.704 -2.350** 1.538 -1.240 -1.896* 0.510 -5 0.451 -2.695*** 2.793* 0.225 -2.636*** 2.374** -6 -0.695 -2.738*** 1.537 -1.008 -2.329** 1.092 -7 0.487 -1.496 1.448 0.725 -1.627 1.570 -8 1.739 -2.479** 2.568** 1.323 -2.802*** 3.159*** -9 1.136 -3.995*** 3.612*** 1.076 -4.035*** 3.558*** -10 -0.177 -2.715*** 1.807* 0.305 -2.827*** 2.933*** -11 0.709 -2.560** 2.601*** 1.041 -2.471** 2.696*** -12 1.206 -2.895*** 3.288** 1.361 -2.647*** 3.191*** the symbols *,**, and *** denote statistical significance at the 0.10, 0.05, and 0.01 levels, respectively, using a 2-tail test. since insider-trading activity is very infrequent, we used quarterly data for the previous analysis. for a robustness check, we reported test statistics in table 7 with similar results based on monthly data. in fact, we conducted a battery of robustness checks. for instance, we increased the sample size by lowering the asset size cut off point, as well as using an alternative control group that matched firms based on market capitalization. all cases were essentially the same. we also checked for insider trading in subgroups. for example, we examined for any discernible differences between the 50 firms that “either filed for bankruptcy because of fraud or 22 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 did not survive the bankruptcy process”. however, we did not find any differences in the insider trading behavior between these two groups. our results are similar to those of loderer and sheehan (1989). seyhun and bradley (1997) had criticized loderer and sheehan’s (1989) sample size and data, arguing that they were unable to capture all insiders trading. seyhun and bradley (1997) had argued that loderer and sheehan (1989) used proxy statements to collect the insider trading data, and this data selection technique yields a sample of mostly large exchange listed firms. we, however, intentionally picked large firms in our sample to observe the behavior of the insiders during the last twelve years as there are many cases of large firms filing for bankruptcy amid a torrent of bad corporate governance practice allegations. ma’s (2001) results are similar to ours. he had found that there is significantly less insider purchase, especially prior to the bankruptcy, yet, no significant insider selling was present. ma had used a relatively small sample of 89 during the period of 1982 through 1990 for firms that filed for chapter 11 bankruptcy. the obvious question is whether the results presented in tables 5 and 6 are expected or not. one of the arguments based on chapter 11 bankruptcy process is the favoring of insiders, while the other is based on a good corporate governance and legal system. we realize that the majority of the chapter 11 filings are self-initiated (i.e., debtor initiated). also, filing for chapter 11 helps managers to restructure their firm while keeping their job. therefore, insiders may perceive that either saving their job or future employability outweighs any potential benefit from insider trading. however, there exists a considerable amount of controversy surrounding the state of corporate governance in the u.s. holmstrom and kaplan (2003) had provided evidence that u.s. corporate governance is in much better shape than the rest of the world. also, as evidenced by bettis et al. (2000), most u.s. firms have instituted mechanisms restricting insider trading. moreover, current developments toward more transparent and speedy reporting instituted by the sarbanes-oxley act of 2002 and harsher penalty imposed by the legal system on corporate fraud and insider trading, severely deters any large insider trading. 5. conclusion in this paper, we examined the extent of insider trading for 129 large chapter 11 bankruptcy filings firms for the twelve-year period of 1995-2006. although, these firms suffered a significant reduction of their stock prices over the period as documented by significant negative cars, the insiders of these firms are found not to engage in trading that is significantly different from firms of similar size and of similar industry in most cases. in some cases where the trading is significantly different, it shows that insiders of similar non-bankrupt firms are either selling more or buying less than in the bankrupt firms. there are, however, incidences of insider trading as reported in the media (these are the exception, rather than common practice). there are several possible explanations as to why this is so. one reason is that managers of these the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 23 large firms find it profitable to file for chapter 11 bankruptcy to save their jobs. any insider trading activity would put their current job or future potential jobs at risk. other reasons are that there is a good corporate governance and legal system in place that actively deters insider trading. author statement: tareque nasser is the corresponding author at the university of alabama, 200 alston hall, box 870224,al 35487, united states of america. (1)205-348-7592; fax: (1)-205-348-0590; email: tnasser@cba.ua.edu. benton gup’s affiliation is with the birmingham campus of the same university. references agrawal, anup, and jeffery f. jaffe, (1995). does section 16b deter insider trading by target mangers? journal of financial economics 39, 295-319. ball, ray and phillip brown, (1968). an empirical evaluation of accounting income numbers, journal of accounting research 6, 159-178. beams, joseph d., robert m. brown and larry n. killoug, (2003). an experiment testing the determinants of non-compliance with insider trading laws, journal of business ethics 45, 309-323. becht, marco, patrick bolton, and alisa roell, (2003). corporate governance and control, in g. m. constantinides, m. harris and r. stulz, ed.: handbook of economics of finance, elsevier b.v. betker, b. l, (1995). management’s incentives, equity’s bargaining power, and deviation from absolute priority in chapter 11 bankruptcies, journal of business 68, 161-183. bettis, j.c., j.l coles, and m.l. lemmon, (2000). corporate policies restricting trading by insiders, journal of financial economics 57, 191-220. bettis, j.c., w. a. duncan, and w. k. harmon, (1998). the effectiveness of insider trading regulations, journal of applied business research 14, 53-70. bradley, michael and michael rosenzweig, (1992). the untenable case for chapter 11, yale law journal 101, 1043-1095. bris, arturo, (2005). do insider trading laws work? european financial management 11, 267-312. carlton, denis w. and daniel r. fischel, (1983). the regulation of insider trading, stanford law review 35, 857-895. demarzo, peter m., michael j. fishman, and kathleen m. hagerty, (1998). the optimal enforcement of insider trading regulations, journal of political economy 106, 602-632. dye, ronald a, (1984). insider trading and incentives, journal of business 57, 295313. easterbrook, frank, 1981, insider trading, secret agents, evidentiary privileges, and the production of information, supreme court review, 309-365. elliot, john, dale morse, and gordon richardson, (1984). the association between insider trading and information announcements, rand journal of economics 15, 521-536. 24 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 fama, eugene, lawrence fisher, michael jensen and richard roll, (1969). the adjustment of stock prices of new information, international economic review 10, 1-21. finnerty, joseph e., (1976). insiders’ activity and inside information: a multivariate analysis, journal of financial and quantitative analysis 11, 205-215. fischer, paul e., (1992). optimal contracting and insider trading restrictions, journal of finance 47, 673-694. gosnell, t., a. keown, and j. pinkerton, (1992). bankruptcy and insider trading: differences between exchange-listed and otc firms, journal of finance 47, 349-362. gup, b. e. and p. agrrawal, (1996). the product life cycle: a paradigm for understanding financial management, financial practice and education 6, 41-48. harlow, w. v. and john s. howe, (1993). leverage buyouts and insider nontrading, financial management 22, 109-118. holmstrom, bengt, and steven n. kaplan, (2003). the state of u.s. corporate governance: what’s right and what’s wrong? working paper, national bureau of economic research. iqbal, z. and s. shetty, 2002, insider trading and stock market perception of bankruptcy, journal of economics and business 54, 525-535. irani, afshad j., (2003). management earning forecast bias and insider trading: comparison of distressed and non-distressed firms, journal of business & economic studies 9, 12-25. jaffe, j. f., (1974). special information and insider trading, journal of business 47, 410-428. karpoff, jonathan m., and daniel lee, (1991). insider trading before new issue announcements, financial management 20, 18-26. lakonishok, j. and i. lee, (2001). are insider trades informative? review of financial studies 14, 79-111. lin, j., and j. howe, (1990). insider trading in the otc market, journal of finance 45, 1273-1284. loderer, c. f. and d. p. sheehan, (1989). corporate bankruptcy and manager’s selfserving behavior, journal of finance 44, 1059-1075. lopucki, lynn m., and william c. whitford, (1993). corporate governance in the bankruptcy reorganization of large, publicly held companies, university of pennsylvania law review 141, 669-800. lynn m. lopucki’s bankruptcy research database (webbrd), http://lopucki.law. ucla.edu/index.htm. lorie, james h. and victor niederhoffer, (1968). predictive and statistical properties of insider trading, journal of law and economics 11, 35-51. ma, y., (2001). insider trading behavior prior to chapter 11 bankruptcy announcements, journal of business research 54, 63-70. penman, s., (1982). insider trading and the dissemination of firm’s forecast information, journal of business 55, 479-503. penman, s., (1985). a comparison of the information content of insider trading and management earning forecasts, journal of financial and quantitative analysis 20, 1-17. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 25 rozeff, m. s. and m. a. zaman, (1988). marker efficiency and insider trading: new evidence, journal of business 61, 25-44. seyhun, n., (1986). insider’s profits, costs of trading, and market efficiency, journal of financial economics 16, 189-212. seyhun, n., (1992)., the effectiveness of insider trading sanctions, journal of law and economics 35, 149-182. seyhun, n. and m. bradley, (1997). corporate bankruptcy and insider trading, journal of business 70, 189-216. shin, jhinyoun, (1996). the optimal regulation of insider trading, journal of financial intermediation 5, 49-73. white, michelle j., (1989). the corporate bankruptcy decision, journal of economic perspective 3, 129-151. 26 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 appendix a: list of n=129 sample bankrupt firms the asset size in millions of dollars as reported in the compustat two years prior to the bankruptcy filing year. name bankruptcy filing date asset size (in million dollars) worldcom, inc. july 21, 2002 98,903 conseco, inc. december 18, 2002 58,589 refco finance inc. october 17, 2005 48,765 enron corp. december 2, 2001 33,381 global crossing ltd. january 28, 2002 30,185 pacific gas & electric co. april 6, 2001 29,715 calpine corp. december 20, 2005 27,304 delta air lines, inc. september 14, 2005 26,356 ual corporation (united airlines) december 9, 2002 24,355 mirant corp. july 14, 2003 22,754 adelphia communications corp. june 25, 2002 21,499 delphi corporation october 8, 2005 20,904 kmart corp. january 22, 2002 14,630 reliance group holdings, inc. june 12, 2001 14,616 northwest airlines corporation september 14, 2005 14,154 finova group, inc. march 7, 2001 14,050 ntl, inc. may 8, 2002 13,026 nrg energy, inc. may 14, 2003 12,895 pg&e national energy group july 8, 2003 10,329 federal-mogul corporation october 1, 2001 9,945 us airways, inc. august 11, 2002 9,127 at home corp september 28, 2001 9,104 xo communications, inc. june 17, 2002 9,085 dana corporation march 3, 2006 9,019 comdisco, inc. july 16, 2001 7,807 home holdings, inc. january 15, 1998 7,593 mcleodusa, inc. january 30, 2002 7,366 arm financial group, inc. december 20, 1999 7,138 anc rental corp november 13, 2001 6,350 bethlehem steel corp october 15, 2001 5,536 integrated health services, inc. february 2, 2000 5,393 ltv corp. december 29, 2000 5,324 owens corning october 5, 2000 5,101 trenwick group ltd. august 20, 2003 4,929 genuity inc. november 27, 2002 4,899 budget group inc. july 29, 2002 4,520 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 27 psinet may 31, 2001 4,492 loral space & communications ltd. july 15, 2003 4,390 armstrong world industries, inc. december 6, 2000 4,273 global telesystems, inc. november 14, 2001 4,002 amerco june 20, 2003 3,773 fleming companies, inc. april 1, 2003 3,655 iridium llc august 13, 1999 3,646 asia global crossing, ltd. ^ november 18, 2002 3,633 chs electronics, inc. ^ april 4, 2000 3,572 solutia, inc. december 17, 2003 3,408 kaiser aluminum corp. february 12, 2002 3,343 covanta energy corp. april 1, 2002 3,295 collins & aikman may 17, 2005 3,191 flag telecom holdings, ltd april 12, 2002 3,079 winstar communications, inc. ^ april 18, 2001 3,065 touch america holdings inc ^ june 19, 2003 3,059 spectrasite holdings, inc. november 15, 2002 3,054 mariner post-acute network, inc. january 18, 2000 3,037 tower automotive, inc. february 2, 2005 2,846 philip services corp. * june 25, 1999 2,823 winn-dixie stores, inc. february 21, 2005 2,790 hayes lemmerz international, inc. december 5, 2001 2,777 usg corp. june 25, 2001 2,773 warnaco group inc. june 11, 2001 2,763 viatel inc. may 2, 2001 2,704 encompass services corporation november 19, 2002 2,700 northwestern corp. september 14, 2003 2,617 chiquita brands international, inc. november 28, 2001 2,596 sun healthcare group, inc. october 14, 1999 2,579 national steel corp. ^ march 6, 2002 2,565 w.r. grace & company april 2, 2001 2,493 fruit of the loom, inc. december 29, 1999 2,483 leap wireless international inc. april 13, 2003 2,451 contifinancial corp. ^ may 17, 2000 2,355 exide technologies april 14, 2002 2,299 trump hotels & casino resorts inc. november 21, 2004 2,196 trans world airlines, inc. ^ january 10, 2001 2,137 pegasus satellite communications, inc. ^ june 2, 2004 2,111 polaroid corp ^ october 12, 2001 2,040 peregrine systems, inc. * september 22, 2002 2,004 highlands insurance group, inc. ^ october 31, 2002 2,001 28 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 rcn corporation may 27, 2004 1,990 ames department stores, inc. august 20, 2001 1,975 service merchandise company, inc.^ march 27, 1999 1,951 heilig-meyers company august 16, 2000 1,948 loews cineplex entertainment corp february 15, 2001 1,907 spiegel inc. march 17, 2003 1,890 adelphia business solutions, inc. * march 27, 2002 1,889 burlington industries, inc. * ^ november 15, 2001 1,876 amf bowling, inc. july 3, 2001 1,827 rsl communications, ltd. ^ march 19, 2001 1,803 allegiance telecom inc. ^ may 14, 2003 1,775 exodus communications, inc. ^ september 26, 2001 1,743 aurora foods inc. * ^ december 8, 2003 1,723 unicapital corporation ^ december 11, 2000 1,670 hechinger company ^ june 11, 1999 1,668 viasystems group inc. march 11, 2003 1,667 pillowtex corp. (2000) november 14, 2000 1,654 interstate bakeries corporation september 22, 2004 1,646 world access, inc. ^ april 24, 2001 1,630 icg communications, inc. november 14, 2000 1,625 magellan health services inc. october 1, 2002 1,611 paging network, inc. july 24, 2000 1,581 penn traffic co march 1, 1999 1,564 nationsrent, inc. december 17, 2001 1,559 boston chicken, inc. * october 5, 1998 1,544 polymer group, inc. may 11, 2002 1,508 dvi inc. ^ august 25, 2003 1,478 pinnacle holdings, inc. may 21, 2002 1,470 lodgian, inc. december 20, 2001 1,424 atlas air worldwide holdings inc. january 30, 2004 1,401 westpoint stevens inc. june 1, 2003 1,369 criimi mae inc. october 5, 1998 1,367 gentek, inc. october 11, 2002 1,351 payless cashways, inc. july 21, 1997 1,344 united companies financial corporation ^ march 1, 1999 1,337 it group, inc. ^ january 16, 2002 1,323 imperial sugar company january 16, 2001 1,281 morrison knudsen corp. june 25, 1996 1,273 anchor glass container corp. september 13, 1996 1,264 apw ltd. may 16, 2002 1,214 washington group international, inc. * may 14, 2001 1,196 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 1-29 29 american business financial services, inc. january 21, 2005 1,159 gst telecommunications, inc. ^ may 17, 2000 1,151 oakwood homes corp. ^ november 15, 2002 1,149 covad communications august 15, 2001 1,148 mobilemedia communications, inc. ^ january 30, 1997 1,143 teligent inc may 21, 2001 1,132 inacom corp. ^ june 16, 2000 1,104 grand union company june 24, 1998 1,061 metals usa, inc. november 14, 2001 1,049 sunterra corp may 31, 2000 1,021 caldor corporation ^ september 18, 1995 1,006 the symbols * and ^ represent firms filing for bankruptcy because of fraud and firms that did not survive the bankruptcy process, respectively. international journal of banking and finance 3-1-2008 insider trading and large chapter 11 bankruptcies in usa tareque nasser benton e. gup recommended citation mushä†rakah and mudä†rabah instuments: theory development to assess their characteristics ijbf mushārakah and mudārabah instruments: theory development to assess their characteristics naeem chowdhury bangladesh institute of development studies ________________________________________________________________ abstract until very recently, equity financing is more in line with the spirit of islam as compared with debt financing. as such, mushārakah and mudārabah are the preferred modes for islamic investors. it is therefore a matter of a concern that even after four decades after islamic finance captured the attention of investors around the globe, these two purest genres account only for a small percentage of islamic financial industry. the paper examines more searchingly for the reasons why mushārakah and mudārabah occupy such insignificant place among islamic financial products. this paper offers an analytical insight. keywords: islamic finance; mushārakah; mudārabah; rothchild-stiglitz framework; heterogeneity jel classification: p5 ________________________________________________________________ 1. introduction the central question addressed in this paper is about one major challenge leaping off the page of virtually any narrative concerning the evolution of islamic finance industry in the world. as any serious student would anticipate that challenge is this: why do the two of the purest islamic capital-versus-noncapital instruments have remained such a small percentage of islamic financial institutions’ portfolio all over the world? those two most authentic, pristine, channels are mushārakah and muḍārabah contracts. many analysts and commentators have commented upon this, and this paper attempts to answer this industry experience. while the specific characteristics of mushārakah and mudārabah will form the bulk of the rest of the paper, suffice to say that sharing profits as well as losses based on a pre-agreed profit-sharing ratios rather than pre-agreed profits in the contract is central to these two financing types. the paper describes the characteristics of mushārakah and mudārabah, especially the latter. certain advantages could ensue in probing that question while emphasizing incentivecompatibility across transactors with economic diversity. conventional neochowdhury: mush?rakah and mud?rabah instuments 26 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 classical economics have worked out a widely-used framework specifying prerequisites to choices among risky outcomes. central to that framework are the use of indifference curves, and the use of incentive-compatibility to provide the logic to the conclusions reached. we propose to use that framework with necessary changes to provide an analytical answer to this dilemma of the industry. the original framework is from rothschild and stiglitz (1976). the concepts of ‘fair-odds line’, marginal rate of substitution, zero and non-zero rates of profits, and ‘budget lines’ are used with appropriate changes in this paper to stay in line with the institutional and incentives assumptions made in that work. equilibria are established when neither the rabb-al-māl (capital provider) nor the mudārib (entrepreneur) has any material incentive to disturb any given status quo, earning ‘zero-profits’. when the risk-pool is homogeneous, the informational regime is by assumption complete and an equilibrium is achieved such that all capital to be provided by the rabb-al-māl will be mobilized and absorbed by the mudāribs. when mudāribs differ in their opinion on loss probability as well as on when information is asymmetric, the loss probability difference is unknown to the rabb-al-māl but is known to the mudāribs, so that less than complete absorption of the capital in the economy will likely result. the paper invokes agent heterogeneity and information frictions (adverse selection and moral hazards), the former is introduced in terms of differential loss probability. it establishes that in islamic murārabah markets, the interposition of mudārib heterogeneity and information advantage will necessarily lead to separating ‘equilibria’ as the most likely competitive solution: pooling ‘equilibria’ will tend to phase themselves out. that said, separating ‘equilibria’ will require the establishment of both voluntary signaling and enforced screening to be well-articulated institutional features of the market, so that the problems posed by hidden information (adverse selection) and hidden actions (moral hazard) are substantively neutralized. next, the paper examines in depth the reasons why mudārabah is accorded such insignificant share in league tables of islamic financial products. we do that by invoking agent heterogeneity and information frictions (adverse selection and moral hazards). heterogeneity is introduced in terms of different loss probability. the paper establishes that, as in the conventional insurance markets, in islamic murārabah markets too, the interposition of mudārib (entrepreneur) heterogeneity and information advantage will necessarily lead to separating ‘equilibria’ as the most likely competitive solution. pooling ‘equilibria’ will tend to phase themselves out. that said, separating ‘equilibria’ will require the establishment of both voluntary signaling and enforced screening to be wellarticulated in the institutional features of the market, so that the problems posed by hidden information (adverse selection) and hidden actions (moral hazard) are substantively neutralized, if not altogether eliminated. the rest of the paper is divided into three sections. the next section is about concept development within the rothchild-stiglitz framework. the reader will find in the third section the logical development of the framework extended to islamic finance. relevant conclusions are drawn in the concluding section 4. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 27 2. concept development the objective of this section is to assess the trends in islamic finance industry in terms of the quantitatively small take-up of the various sharī’ah-compliant finance as well as other miscellaneous islamic financial products. the section provides brief definitions of the various lines of financial products from secondary sources. this would support our claim that these two modes of financing occupy a small share of the capital market. 2.1 definitions mudārabah instrument is a trustee partnership agreement. in it, two or more parties combine human and financial capital to share the risk of running an enterprise together, with the purpose of earning halal (permissible) profits. one of the two parties is called mudārib (the entrepreneur) who has the necessary skills, expertise and diligence but lacks capital. the second party called rabb-almāl provides the required financing to the entrepreneur. this is congruent with all economic practices of all human societies. such a capital provider may be an islamic bank, a finance company or a private investor, as for example as happens in private placements by entrepreneurs, which has become very popular since the 1990s as a fund raiser. the management of the enterprise and the production work are the exclusive responsibility of entrepreneur. profits generated are shared between the two parties according to a pre-agreed ratio, not fixed returns on capital provided. to create a balanced contract, if a loss were to occur (due to no fault of the entrepreneur), the capital provider will lose his capital, and the entrepreneur will lose the value of time and effort she invested in nurturing the business. under a court’s finding of any willful neglect by entrepreneur, the capital provider could sue for damages on grounds of neglect. mushārakah is a partnership or joint venture agreement whereby two or more parties contribute capital and management expertise to pursue earning halal (permissible) profits to be shared among them. all providers of capital are entitled to participate in management but are not necessarily required to do so, if one or more of them wish to opt out of full participation, he/she can do so with the agreement of rest of the members on terms to be mutually agreed. profit is distributed among the partners in pre-agreed ratio while the loss is borne by each partner strictly in proportion to capital contributions. the capital of the company can be money or valuables such as merchandise or capital goods or inventories. 2.2 why have mushārakah and mudārabah not become major modes of finance in practice? if personal circumstances or preferences prohibit going into entrepreneurship as a solo effort, partnership as embedded in the two instruments is allowed under the islamic legal framework. that is one reason, as also the reason given after the landmark decisions of sovereigns in about five centuries ago, to chowdhury: mush?rakah and mud?rabah instuments 28 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 permit their citizens to engage legally in sole proprietorship, joint ventures and also as joint stock companies (starting from the days of the columbus’s joint venture with queen isabella of spain, in the case of conventional partnership). islamic teachings laud the principle of pursuing profits while exhorting parties to contract splitting exposure to risk-taking equitably. in the more than twocenturies old history of modern banking, the idea of sharing risk has been pushed back through financing arrangements, where the entrepreneur solely bears the risk and the financiers developed contracts that do without any risk-sharing, requiring payments for funding. the qurãn states, “allah has made buying-and-selling permissible, but ribā impermissible”: [2:276]. buying and selling legally-permitted goods is all about pursuing legitimate returns while embracing risks equitably is enshrined in contracts. mushārakah and mudārabah contracts are the most authentic financing products in islam, because returns qualified by risks provide their bedrock. one of the first tests of genuineness in islamic financing is about the proportion it accords to mushārakah/mudārabah. the available evidence emphatically suggests that about ninety percent of sharī’ah-compliant financing as of late 1990s used to be on murābaḥah (sale on deferred payment basis) or joint-venture (iqbal et al., 1998). more recently, the percentage share of these contracts mushārakah/ muḍārabah in the total funding provided by banks during the six-year period ending in 2011 averaged about 7 per cent whereas the corresponding percentage owing to murābaḥah (a third form) is as high as 85 per cent (goud, 2012). a long-term evaluation of the operation of islamic bank of bangladesh (ibbl) over about a quarter century shows the following. over the first three years, mushārakah/mudārabah had accounted for some 15 per cent of banking finance whereas during the latest two decades that percentage shrank practically to almost single digits (belal, abdelsalam & nizamee, 2010). in general, mushārakah and mudārabah accounts account for less than a tenth of the financing provided by islamic banks worldwide (iqbal & molyneux, 2005). the sub-total is found to range anywhere between 9.7 and 7.8 per cent (goud, 2012). for those in the know, such isolation of variable-returns products in islamic finance is readily understandable, though regrettable. the underlying argument is all about incentive-incompatibility. for a financing structure to work economically and sustain interest over time, sufficient incentives of participant interest-groups need to be dovetailed. otherwise, the contract will soon slacken off. a simple sale contract between a buyer and a seller takes place only if there is synchronization of needs and an agreement on price.1 in the event of a default of a conventional debt contract, the accumulation of arrear interest is itself a deterrent to default. in a murābahahbased sale contract for deferred payment, such a deterrent is itself less effective in the absence of a penalty clause. however, it provides less protection to the 1 the remainder of this paragraph borrows heavily from dar (2007). international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 29 financier as they cannot benefit from the proceeds of the penalty, which are paid out to designated charities. this less favorable treatment of the financier may result in higher pricing of islamic financing products based on murābahah. murābahah makes the incentives of the providers and the users of capital substantively compatible. this is largely the message of table 1. 2.3 comparative incentive-compatibility among different islamic financial products table 1 contains summary statistics on a variety of sharī’ah-compliant financial products. whereas for murābahah, the deficit relative to ribā-based regime is passed off as being only marginal, for mushārakah and mudārabah, the deficits are said to be significantly less. for three other financial products, the incentivecompatibility are said to be only moderately less compared with (roughly) equivalent conventional banking systems. table 1: incentive features of some of the islamic financing products modes residual rights control rights incentive compatibility remedial measures murābaḥah fund user fund user marginally less penalty clause, buyer undertaking ijāra wa iqtina ʿ fund user financier moderately less undertaking from financier salam financier fund user moderately less parallel salam istiṣnā ʿ shared shared moderately less parallel istiṣnā ʿ muḍārabah shared fund user significantly less strong monitoring and supervision mushārakah shared shared significantly less strong monitoring and supervision source: hassan and lewis (2007). several explanations have been offered as to why murābahah is so ubiquitous, pervasive and durable. the motives include, (i) capital protection while generating sharī’ah-compliant profits through making financing available; (ii) risk transfer through an internal market; (iii) hedging against the undermining of the environment of financing integrity arising from informational frictions; and (iv) other means. also, in any modern economy, partnership tends to be the most common form of business form because of the ease with which partnership can provide capital. in addition, the laws provide easy exit from partnership to sole ownership, if the parties desire to go that way. chowdhury: mush?rakah and mud?rabah instuments 30 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 2.4 why should we care if proportion of mushārakah/mudārabah is small? the small size of mushārakah/mudārabah as a percentage of sharī’ah-compliant financing raises questions about the genuineness of islamic finance in economic life. this arises mainly on two grounds. first, by restricting to a small percentage share, the most original and authentic mode of finance in islam presents islam’s these two financing models as an apologist. second, compared with murābahah, which is the fastest growing segment in the islamic finance industry, mudārabah/mushārakah have the potential of rendering careers in business and entrepreneurship more broadly-based, and thus leveling the playing field of investment. murābahah can of course be readily tailored to scale up with regard to a whole range of physical, income-generating assets such as diagnostic laboratories, private hospitals, dentists’ clinics, beauty-parlors, all manner of testing (soil, water, etc.), cyber-cafes, desktop-printing shops, engineering consulting firms, fire-advisors’ businesses, etc. these are well outside and beyond the oft-quoted space of real estate and vehicle purchases. that contract can cast a wide net in terms of how many categories of businesses it could serve well. that said, these murābahah-favorite business categories mostly feature relatively high entry barriers due to specialization and the associated humancapital requirement. they are heavily skewed towards specialized services and imply a narrow demographic base. they will of necessity ride the crest of an expensive kind of human capital accumulation, which in any case will be limited in number. no matter how exuberant, the problem surrounding such growth will still leave islam’s implementation record removed from the bonafide maqāsid al-sharī’ah. the latter are much more likely to prefer financing of a larger, direly under-capitalized mass of small and medium enterprises in production, manufacturing, and trade. in many economies, with only one percent of businesses being outside the fold of tiny, small or medium enterprises, these statistics actually represent the very heart of a financing problem that needs to be pointed out. most entrepreneurial units typically struggle on account of a mismatch between the availability of business skills and capital. murābahah does not reach a large number of needy people. all this will undoubtedly amount to a strategy of obliging the top of the ladder investors. given the dominance of murābahah type contracts, the typical spectacle of sharī’ah-compliant financing appears to be incongruous with transforming the entrepreneurship in the economy that could advance socio-economic objectives and mandates of societies. provisioning of capital is among the most strategically important action driving economic growth in any economy. the provisioning of capital via the rabb-al-māl mode is akin to supplying blood in a body. without effective coupling of capital, and technical, organizational, managerial and human-resources skills (of entrepreneurs) no income-generating project can bear fruit. even so, the centrality of capital in the development process is inescapable as is well known and acknowledged by all economists, including the islamic economists. modern international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 31 economic analysis suggests that almost two-thirds of a dollar of gdp in an average economy is the result of capital injection to production process: labor input accounts for almost the rest of the productivity. of course, the compact between the two partners would need to be framed by the goals set and constraints to be faced by a business as agreed to by both the financier and the entrepreneur. the savings that the financiers bring aboard will have involved conscious postponement of consumption by today’s owners over many past periods. the opportunity cost of today’s capital in terms of past sacrifice can be very significant. today’s capital is the prospective mainspring of all future streams of income for the rabb-al-māl. while one always expects one’s endowment of capital to perform financially, the opposite, however, may no doubt could happen, thus introducing asymmetric expectations. consider in contrast the situation of the mudārib. the entrepreneur’s claim to the prospective profits stems from his business skills in general. for a manufacturing concern, such skills will run across the whole gamut of production process from ideas through the prospective product positioning to picking a path towards operational success from an intelligent analysis of the competitive scenarios and finally to efficient transformation of all inputs into a profitable proposition. the entrepreneur is potentially irreplaceable in any productive economic activity as he/she brings to the table a repertoire of specialist skills, capabilities and expertise, which at times could be unique. there can be no denying the unique contribution that a good mudārib can potentially make to the success of a project. recent evidence of world economic recovery after the global financial crisis attest to the fact that jobs are created by the entrepreneurial activities of innovative persons. 2.5 rabb-al-māl deserves priority in policy formulation having stated that, it is still imperative to consider at least three dimensions of the comparative situation of the entrepreneur and the financier before one is able to absorb the incentives/disincentives that characterize them. first, capital owned by a financier is like a general-purpose vehicle. it is a malleable resource. as long as capital is in the form of unencumbered money, it represents a liquid form of buying power ready to be deployed in any productive sector for procuring goods or services or for paying for their transformation into any pre-set blueprint that an entrepreneur has prepared for socially useful products or services. like all general purpose vehicles, capital faces the prospect of potentially profitable deployment across real sectors of an economy without the encumbrances of any significant adjustment or menu costs. in contrast, the same is in general not true of the muḍārib; if not in its entirety, then certainly to a degree. unless the economy offers the services within well-defined markets of the full array of technological, managerial, logistical, information-technology skills, an entrepreneur has to agree to supply all or a part of what is required from the diverse array of competencies in entrepreneurship. it is impracticable for any one person or well-knit team to wrap their heads chowdhury: mush?rakah and mud?rabah instuments 32 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 around such diverse skills and capabilities for anything but a given production sector of interest. this suggests strongly a state of relative immobility: the entrepreneur’s skills are not intrinsically mobile across sectors. at a further remove, this suggests that the mudārib is exposed to non-negligible adjustment or menu costs when he/she has to change the sector of involvement. the main point is that a general purpose resource has an advantage that is denied to a more sector captive, specialized, resource of a mudārib. second, owners of capital would likely enjoy greater negotiating leverage compared with the owners of sector-specific resources and skills. whereas capital is likely to be universally in demand across several sectors or industries, the mudārib’s services are unlikely to face equally brisk demand regardless of the sector of interest. once again, one sees capital’s generalized advantage to be contrasted with a disadvantage associated with relatively anemic demand across sectors and industries for services of mudāribs. third is the existence of information asymmetry. asymmetric information comes in two varieties. there is adverse selection, something what debraj ray has called hidden information (ray, 1998). hidden information is about an economic agent actively seeking to create more positive, favorable, congenial perceptions about the prospective value of contribution(s) to an enterprise. there is no doubt that relative to the rabb-al-māl, the informational balance is very clearly with the muḍārib. we make a fuller presentation about the various aspects of informational asymmetry in the next section. what the rabb-al-māl brings to the equation suffers from not even a small measure of adverse selection (hidden information) or moral hazard (hidden action). the rabb-al-māl must simply deliver the amount of money committed to investment. only when financier honors this commitment, one can monitor this fully and in a costless manner. that commitment, whose fulfillment takes places at the ‘input’ stage, will need to be honored to the last dime. in contrast, the payoff in any transaction is often bound up with the output. output is typically far less easily amenable to contractual formulation to everyone’s satisfaction than inputs (entrepreneurship and finance). the entrepreneur’s true contribution will be to the output although that output is not very clearly contractible, which becomes a decisive factor in the present context. given the role of hidden information and hidden action in it, the output-producing process becomes an important source of uncertainty in the life of an enterprise. this introduces the important idea that the mudārib is potentially susceptible to engage in an adverse selection activity when a murārabah contract is signed. before going any further, one further relatively minor point needs to be covered. one should be aware that the following analysis is more relevant to pure types of murārabah than to mixed mudārabah. the assumption in this paper is that many of the individual mudāribs could all be relatively young people with professional qualification or unique ideas, but without much cash resources to start an enterprise. their ability to put up collateral in cash or cash-equivalents may be too trivial or narrow. it may become absolutely impracticable for such mudāribs to front-load the deal by pledging something of economic value to financiers. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 33 2.6 indifference curve analysis in modern economics, the first use of indifference curves under asymmetric information was in analyzing welfare effects of insurance contracts. there are important differences between using indifference curves in discussing insurance markets and markets for islamic partnership contracts. the transactions are monetary and two-way in nature in insurance: premium is paid by the insured with the expectation that someday they will be traded off for an indemnity to be paid by the insurance company. the equilibrium in insurance markets is said to take place when premiums paid into the plan is equal to the indemnity paid out. premiums paid appear as debit entries on customers’ books of accounts but as credit entries on insurers’ books. indemnities appear as credit entries on customers’ books of accounts but as debit entries on insurers’ books. this brings about incentive-compatibility of the insured versus the insurer. prima facie, it could be said that analogous conditions to premiums and indemnity do not exist in the islamic risk-sharing partnership contracts. at least, such a characterization would be entirely inconsistent with the profiles and attributes that mushārakah and muḍārabah possess. secondly, then how can one conduct indifference curve analysis when discussing the choice of a method for trading between different states of outcome across a range of uncertainty? 3. model development consider a mudārabah market where each potential rabb-al-māl faces two states of the world. a profit result, in which case the wealth is w+p; and a loss result, in which case the wealth is w-l. a state diagram is often used in the analysis implied here of risky bundles of economic outcomes. before joining the mudārabah, we assume that the rabbal-māl’s wealth endowment is equal to w. after a mudārabah contract is agreed, the rabb-al-māl’s wealth endowment is changed as follows: (1) the probability of a loss occurring in any given year is assumed to be p. therefore, the foregoing assignment essentially states that the wealth endowment in the event of a profit is w+p, where p is the profit that accrues to the rabb-al-māl. in the event of a loss, the endowment is changed to w-l, where l is the loss that accrues to the rabb-al-māl. here, the parameters merely describe the murārabah pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ chowdhury: mush?rakah and mud?rabah instuments 34 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 contract, with both profit and loss being shared between the rabb-al-māl and mudārib based on the profit and loss sharing ratio agreed beforehand. we choose to look at the economics of dealing in risky bundles through the filter of the rabb-al-māl. we also assume that all arbāb-al-māl (plural of rabb-al-māl) in an economy with islamic finance being available are identical in all essential respects. this then suffices for the use of the model for a representative rabb-almāl. denote the probability of an accident as p that an accident could precipitate a loss. a murārabah contract is assumed to exist primarily due to the initiative of the rabb-al-māl. parties will buy into a murārabah contract if the expected utility of such a contract exceeds the utility of keeping the financier’s capital under his sole control: (2) a rabb-al-māl will take on new murārabah proposals for implementation, subject to the availability of his capital, as long as the ratio of the probability of profitable and loss-making events is compatible with the actuarially fair odds.2 what level of deployment of capital would maximize the expected utility of the rabb-al-māl? in the model itself, both the odds of a profit or loss and the amount of the profit/loss could be admitted to be variable. in the context of answering this question, we fix the odds, and we ought to be able to do so without any loss of generality. it would be easy to treat the quantity of capital that any given rabb-al-māl would deploy on the basis of mudārabah as given. before proceeding any further, we reformulate the foregoing equation by replacing both p and l as more appropriate profit (or loss) functions, as follows: (3) where f(k,l) merely represent the underlying production function, embodying the arguments of capital and labor-services. on purpose, we omit from the profit function the other variables, such as the price of the product in question, etc., are uncertain, in order to reduce notational clutter. we intend to do this without any loss of generality. for the time being, we are using the role, incentives and effects on the rabb-al-māl as defining the focal point of this analysis. the expected utility would be maximized when the first-order condition with respect to the foregoing equation is equal to zero: (4) 2 in terms of the terminology of the insurance-market literature, actuarially fair odds correspond to a situation in which the insurer ends up with zero profits, as the premia paid by the insured exactly equals the indemnity paid against the policy. pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 35 that implies the following: that further implies this finally implies that the expected utility from forming mudārabah would be maximized when the partnerships are being formed along the actuarially fairodds line. the easiest situation to set up is a 2-state set up with p 1 = p and p 2 = 1− p. an interesting way to represent preferences in this case is with a standard consumer model over two goods, consumption in state 1 and consumption in state 2. as usual, an indifference curve is given implicitly by setting utility to a fixed value and treating one variable (say x2) as a function of the other (x1). individuals maximize e [u (x)] = pu (x1) + (1 − p) u (x2) (5) where, the superscript of ‘ic’ is meant to convey the fact that (w+p) is being assumed to be a function of (w-l).we formally substitute x1 for (w-l) and x2 for (w+p). differentiating this condition implicitly once gives the condition: (6) along the 45-degree line of a graph putting x1 against x2, where x1 = x2, then = . along such a line, marginal rate of substitution between (w-l) and (w+p) will be equal to the odds-ratio, no matter the curvature of the utility function. this is one of the implications of expected utility theory, as applied to the two-state diagram. the characteristic of the risk-pool is important to the outcome in terms of the inter-personal distribution of the payoff and any possibility of externality. the first thing that needs to be examined is about the homogeneous risk pool: this is the base case. the base case also provides an opportunity to look at the individual elements of the state diagram. it is to the examination of such a base case that we now turn our attention. 3.1 integrity of using state-diagram from mainstream economics before proceeding any further, addressing the validity of implanting analytical construct(s) from microeconomics is helpful. one could argue that, whereas certain similarities and analogues may exist between the competitive equilibria in insurance markets in the presence of heterogeneity and incomplete information, the symmetry in the islamic murārabah markets however breaks down at certain crucial junctures. for example, in the insurance markets, incentive-compatibility arises from the insured paying premiums into a pool, in anticipation of a quid pro quo in the form of the payment of the indemnity in the event of an accident. pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ pr(1-p) = w + p wt = ------------ pr(p) = w – l (1–p). u(w + p) + p. u(w–l) > u(w) e (u) = ((1– p). u(w + f (k.l) + p. u (w – f (k, l)) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ( )] ( ) ( ) ( ) [ ( )] ( ) ( ) (( ) ( )) ̅ ⁄ ( )⁄ chowdhury: mush?rakah and mud?rabah instuments 36 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 whenever the premium is set at a higher level than the actuarially fair premium, the degree of insurance coverage chosen by individuals is lower. in fact, the algebraic formulation of the ‘fair-odds’ line specifically includes both the (i) the probabilities of the loss and non-loss events; and (ii) the expected utility associated with the payment of the premiums and the indemnity. the institutional set-up of the murārabah match-up does not include such a hardwired dove-tailing of material incentives. arguably, implanting the diagram containing fair-odds line, and indifference curves that intersect only once giving rise to the single-crossing property may be illegitimate. further discussion is needed. first, it is incontrovertible that the rabbal-māl and the muḍārib in islamic economics will be interested to pool their resources only and solely with a desire to bring into better alignment their respective income or consumption or utility levels across two states of nature namely the occurrence of loss and the occurrence of profit. in short, the very proposition of the mudārabah is about trading between two collections of economic prospects with different degrees of potentially measurable uncertainty, which is called risk. the mainstream microeconomics witnessed vigorous development of analytical constructs that are especially relevant for grasping the drivers and policy implications arising from the trade-offs in such risky bundles (cutler & zeckhauser, 1998). this has been especially so since the conventional economics has been prodded by the discovery of the pervasiveness of incomplete information (in the sense of some information being hidden, selectively) and the challenges it entailed for much of the neo-classical economic analysis. in the analytical development, therefore, we have to depend necessarily on the constructs that have been used in the conventional microeconomics texts to probe the achievement of equilibrium under agent heterogeneity and under asymmetric information. crucially, in doing so, one has to diligently bear in mind the role of the institutional differences between conventional risk markets and islamic murārabah markets. not every aspect of the expected-utility theory or the definition of risk-aversion/risk-neutrality would readily fit into a discussion of choices to be made in an islamic economy. this parameter is adhered to very consciously in the writing of this paper. secondly, in seeking to improve the distribution of wealth amid riskiness, both the rabb-al-māl and the mudārib are parting with what is valuable to both; for one it is general-purpose money and for the other his work skills. but even the mudārib agrees to take the lids off from his commercial code of proprietary secrecy in operational matters when he agrees to pool resources with the investor because she has access to capital that would bring the output to reality as well as the profit share. it is the mudārib who is informationally advantaged to some extent. the entrepreneur may also be possessed of specialized or highly specialized technical or process-related skills (windows technology of bill gates). being willing to train employees in the murārabah will be tantamount to his being willing to share from his cache of trade secrets. that said, that particular effect may also mean partial, even substantial, neutralization of the international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 37 mudārib’s own competitive differentiation in future as the number of potential competitors swells through copycat replications (china syndrome). erosion of one’s own essential competitiveness in informal markets, with loose intellectual property rights protection, for technical or even business skills, will often be channeled through informal training programs. there is a lot of evidence for that. for instance, nathan rosenberg of stanford university has argued that scientific prowess is partial to publicity and assiduously seeks disclosure, publication, and dissemination. in sharp contrast, technical or process-related, or technological, prowess is publicity-shy, assiduously avoiding disclosure, duplication, free replication, and copying (rosenberg, 1973). scientists who stumble upon ground-breaking discoveries in areas of pure or natural science set great store by publishing quickly hoping that their craft will achieve the status of a classic. in contrast, innovators, who are after big payoff that follows commercialization, set great store by stealth, secrecy and successful rapid patenting of the essence of the technical or process innovation before the competition has even a wind of what is going on. in general, the more proprietary the technical or process related skill that the muḍārib brings to bear during the course of implementation of the business plan of the partnership enterprise, the more valid is the argument made in the previous paragraph. as long as information relating to the loss and profit probabilities, and the distribution of expected profits and losses in the relevant events are available, the constructs of indifference curve, fair-odds line and incentive-compatibility, are all applicable with necessary changes. of course, there will be one major difference between competitive equilibria between conventional insurance markets and the islamic mudārabah markets. in the former, given that individuals are risk averse and insurance companies are risk neutral, the first best is characterized by full insurance. by assumption, insurance companies will enter into the industry as long as positive profits can potentially be made by selling insurance. the impulse towards less-than-full insurance will be undercut by competition until the zero-profit equilibrium is reached. in contrast, under the analysis conceived in this paper, the landmark evolutionary stage in market’s development is about a full subscription of the rabb-al-māl’s capital by the muḍāribs. a sample of mudāribs could either be identical in terms of the individual probability of loss. or the sample could harbor heterogeneous probability of loss. the issue needs to be framed diagrammatically in the interest of greater clarity. the underlying issues run the gamut from actuarially fair odds to incentives compatibility, all of which can be handled diagrammatically. the ratio of the probability of profit occurring compared with a loss is akin to the relative prices of two commodities being compared. odds are actuarially fair when the likely return and risk are being traded at a rate equal to the ratio of the probability of profit versus loss occurring. 3.2 when the risk pool is homogeneous the domain of risk-pool includes all the mudāribs in the economy. the mudāribs could arguably be all identical in terms of their trade-related competencies, that chowdhury: mush?rakah and mud?rabah instuments 38 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 is, equally knowledgeable, shrewd, diligent, customer-friendly, well-tuned to the state of the market, and with identical leadership qualities. this would be a case of the rabb-al-māl facing a group of mudāribs with identical probability of loss. it is already assumed that all losses are equal to l. the loss probability (p) and the amount of the loss (l) are the only two variables in the present analysis. we shall want to keep l fixed in the following analysis. fixing l is without loss of generality. we only need one free parameter here, either p or l: we will be using p below. there will be an initial endowment with coordinates of (w, w l). this initial endowment is represented by the point e in figure 1 below. the fair-odds line extends with slope equal to the odds ratio between the loss and no-loss states. a rabb-al-māl will optimally allow full subscription of his entire capital, as long as profit rates offered by the mudāribs and the probabilities of occurrences of profit or loss do not fall shy of the rate commensurate with actuarially fairodds line. following from the von neumann morgenstern expected utility property, the highest indifference curve tangent to the fair-odds line has slope at its point of tangency with the fair-odds line, which is where it intersects the 45 0 line. at this point, wealth is equalized across states. in figure 1, the loss probability is shown on the vertical axis and the residual probability of profit on the horizontal axis. by initial assumption, both the mudārib and the rabb-al-māl have identical utility functions. each indifference curve is about how either the mudārib or the rabb-al-māl trades between a state of risk-i.e., a pecuniary loss¾and a state of return¾i.e., a profit. the rabb-almāl will presumably face a multiplicity of demand for his capital, because by assumption there is a shortage of capital, the economy suffers from a scarcity of capital, and therefore capital-owners enjoy a sought-after status in the economy.3 the full employment of all available capital in the economy is therefore also an imperative. situations of the full subscription of the rabb-al-māl’s capital (in other words, the full coverage of capital provisioning) corresponds to points on the 45-degree line while situations of incomplete coverage lie to the right of the bisetrix (where w1 >w2). the expected value of the loss in any given year could alternatively be represented by present value of future tax credits expected to materialize due to those losses, given the so-called loss-carry forward allowed universally in most tax codes. it bears repetition that losses racked up in the given year under discussion are by assumption being represented by present value of expected future tax credits receivable against them. this institutional assumption underlies this paper. 3 of course, both the mudārib and the rabb-al-māl will have mouth(s), including his own, to feed. by assumption, the former only has business skills which do not translate readily into command over food: if he does not have savings from past periods, the pressure upon the mudārib to team up with a rabb-al-māl and start up a murābahah will be immense. in contrast, the rabb-al-māl has cash resources on hand. he can just go to the market and buy what he needs, paying down with cash: he is not desperate for a mudārabah deal. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 39 figure 1: a two-state analysis of mushārakah/muḍārabah contract equilibria each investment project will be about trading expected losses in terms of expected profits. the indifference curve for risk is about the marginal rate at which expected loss is traded into units of expected profits¾the marginal rate of substitution between losses and profit. the ‘actuarially fair-odds line’ is the budget line. a budget line identifies the best budget constraint available to the agents (both the rabb-al-māl and muḍārib) for trading risk into returns. put it differently, the budget line is about marginal rate of transformation of risk relative to returns. the slope of an investment project’s budget line would be different from others, of course. a budget line would have a steep slope if either the underlying enterprise or its mudārib is inherently more risky: it takes a larger amount of losses (expected future tax credits) to translate into a given amount of expected profit. a new enterprise in an industry with significantly greater startup difficulties or with high failure rates would exemplify such higher risks. in contrast, certain genetic, personality or background traits arguably predispose certain human beings to taking undue risks and thus more risky behavior. the ratio of the probability of loss to unity minus the probability of profit occurring is equal to the slope of the line representing the actuarially fair odds: this is the ratio . when the slope of the budget line is equal to the slope of the indifference curve, the risks and returns are evenly balanced.4 the capital of the rabb-al-māl would be fully placed into investment: the number of the e c odds for profit f o dd s of l os s ( )⁄ 4 in the insurance literature, such a situation is said to lead to zero profit for the marginal insurer, given full and free entry into the markets. i 2 i3 i 1 chowdhury: mush?rakah and mud?rabah instuments 40 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 mudāribs looking for capital will be equal to the number of murārabah contracts that arbāb-al-māl are looking to enter into. for enterprises in a particular industry, the probability of the occurrence of loss in any given year of operation will be given. given such a probability, the budget line corresponding to actuarially fair odds will represent an equilibrium. this will be a situation of full subscription of the capital that all the arbāb-al-māl put together will wish to bring on the market. point e in figure 1 corresponds to a situation of no deals involving mudārabah. each point to the northwest of e represents a specific murārabah contract uniquely identified by a certain investment by the rabb-al-māl and an uncertain profit or loss (loss when there is a loss). the segment ef represents the actuarially “fair” odds line. the net payout from an uncertain ‘game’ is actuarially fair if the expected monetary gain, whether in cash or in putative (notional) terms, equals zero. corresponding to each outcome of actuarially fairodds there is a budget line. the budget constraint that reflects the opportunities presented by an actuarially fair contract is called by an actuarist a fair-odds line. along such a budget constraint, the expected value of all ‘states’ of income is equal, and no one group of transactors will be subsidized ex ante at the expense of any other group. in the insurance literature, this line is called the actuarially fair-odds line, or the zero-profit (for the insurance company) line. of those two expressions, the first one, the actuarially fair odds line is probably more appropriate. this line identifies a locus of the break-even budget constraints available to the rabb-almāl and the mudārib for trading outcomes in the two (risky and good) states of affairs. the break-even is in a sense that it just brings the material interests of the two kinds of partners into the kind of alignment that makes it just worthwhile to team up.5 trading income in the two states at a rate equal to the slope of ef leaves the rabb-al-māl and the mudārib with a harmonious balance: no new mudārib has any incentives for disturbing the status quo by offering the rabb-al-māl a more attractive profit rate. starting from point e, any point to the south-west of ef entails combinations in which risk trades into return at a rate more preferential to risk-taking by the rabb-al-māl. there is always the supply of new candidates among mudāribs and are not spurned by the arbāb-al-māl. the underlying risk-return nexus is propitious for risk-taking. that said, such cannot be an environment conducive to the maintenance of equilibrium contracts (as any given profit ratio can always be undercut by a new aspiring murārabah contractee that attracts other arbāb-al-māl and still earns positive profits). in 5 the one major difference between the terminology of the insurance markets and the sharī’ahcompliant financial products markets is here. in the insurance markets, the fair-odds line always corresponds to situations of zero profit for the insurer, which is evocative of an unambiguous aura of ‘fairness’ in the match-up between premia paid and the indemnity in prospect. the corresponding analogue in sharī’ah-compliant markets can only be envisioned more in the abstract, as the concept of zero profit is no longer relevant. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 41 contrast, any point to the north-east of ef entails risk-return combinations that are skewed against risk-taking and it is therefore not feasible. so, given free entry and perfect competition in long-term equilibrium, the arbāb-al-māl and mudāribs find their equilibrium contract along the set of contracts on the ef line. this notion of equilibrium contract connotes only a preference among all the arbāb-al-māl and muraribs not to upend the status quo. we can represent the preferences facing a proposed partnership enterprise with a map of indifference curves. given risk-aversion by both prospective partners, the indifference curves are convex. given any indifference curve, all the points to the northeast entail higher utility and all the points to the southwest entail lower utility. equilibrium is on the highest indifference curve compatible with the expected ‘budget constraint’ provided this point also lies on the bisetrix, ie. point c. in c, the slope of the indifference curve is equal to the slope of ef. thus, when both the arbāb-al-māl and the mudāribs are risk averse, and there is complete information,3 the first best is characterized by full utilization of capital open to mudārabah.4 whenever the marginal rate of transformation of risk into return is lower than the marginal rate of substitution of risk for returns, the appetite for risk remains relatively brisk, and the arbāb-al-māls’ capital that is set aside for the murārabah format will begin to move toward fuller deployment in risk-sharing. gradually, the supply of such a quantity of mudārabah funds will also be equal to the amount of the funds for which the mudāribs have effective demand. the base case is unlikely to be an accurate description except in a minority of the cases. more realistically, the mudāribs are likely to be heterogeneous. a situation of incomplete information is typically in for the ride when the riskpool is heterogeneous. the presence of agent heterogeneity typically triggers the advent of private information. we now turn to this aspect of the problem. 3.3 heterogeneous risk and adding private information the mudāribs could conceivably be divided into two groups. one of these groups comprises mudāribs who are more risk-prone and impulsive, impetuous and single-track in their information-gathering, insufficiently attuned to both the shortand the long-term factors concerning competitive obsolescence of the equipment and technologies used in the project(s) under study. when they buy supplies, they fail to spot opportunities in the market for bargains, and to buy economically, running unduly bloated purchase costs. they are naïve and gullible, credulous ‘dummies’ who would readily put their trust even in strangers. they may not be the best judge of readily concluding that ‘an offer 3 complete information is said to exist when, in particular, principal-agent relationships do not labor under the weight of information frictions. 4 note that only the murārabah-specific portion of the capital of the rabb-al-māl seeking deployment is relevant to this paper. mushārakah -specific capital is not of relevance here. as well, given complete information, it is the points on the fair-probabilities line that the rabb-al-māl will become agreeable to deploying the whole of the murārabah-only portion of his capital. this is intuitively acceptable. chowdhury: mush?rakah and mud?rabah instuments 42 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 may be too good to be true’, and may fall unsuspectingly into a trap that may be deliberately sprang upon them. in short, some muḍāribs may have a mental make-up that makes them, even when they are entirely earnest, a loss-leader. let us assume that the other group is made up of mudāribs who are even-headed and thoughtful, comprehensively networked and thus have reliable access to a relevant and wide array of information, are prudent and methodical, take a decision only after due processes and much careful reflection. the two classes of mudāribs are set apart from each other. there is a category of mudāribs who are prospective loss leaders. there is a second category of mudāribs who are prospective profit leaders. under the watch of the first group of mudāribs, the murārabah is more likely to stumble into operating losses in the ventures they take up. the probability that the murāribāh in question turns up a loss for any given unit of time is higher than elsewhere in the economy, allowing for the specific characteristics of the sector of interest. whereas under the watch of the second group of mudāribs, the probability that the mudārib in question will turn up a profit for any given unit of time is higher than elsewhere in the economy. full deployment of capital of the rab-al-māl will necessarily involve trading profit income between occurrence of profits to be shared, and losses to be split. equilibrium will be impracticable without the rab-al-māl being able to sort out the loss-leaders from the profit-leaders. the former need to be crowded out compared with the profit-leaders. within each industry, the rab-al-māl ought to offer capital to the profit-leaders at a relative profit discount compared with the loss-leaders: profit ratios offered need to allow for an ‘industry-effect’. it is necessary to introduce the industry of origin as a qualifier because industries fundamentally differ in terms of the risk-return nexus: an explicit industry effect is an allowance that needs to be made when it comes to profit differentials among economic agents with different characteristics, such as susceptibility to losses. in addition, we admit private information into the discussion. this means that we assume that individual mudāribs know their risk type but this information is not known to the rab-al-māl. this assumption makes intuitive sense, too. private information without heterogeneity is not meaningful; if everyone is identical, there is no private information. given the two risk types, ‘h’ and ‘l’, and the asymmetric information about these types, there are two possible classes of equilibria in the model. the first of these two is about pooling equilibrium. that is about a situation where the rabb-al-māl does not distinguish between risk-types, which are however for real. it is to this aspect that we turn first now. 3.4 pooling equilibrium figure 2 is drawn with the indifference curve corresponding to the high-risk (‘h’-type) mudāribs being flatter compared with low-risk (‘l’-type) mudāribs. the marginal rate of substitution (mrs) of the high-risk mudāribs is hereby being assumed to be lower compared with the mrs of the low-risk muḍāribs. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 43 the certainty equivalent of the initial without-any-mudārabah-contract state is higher for ‘l’ than for ‘h’ types since the type l have lower odds of a loss. because the probability of loss is lower for type l, type l would want to raise their reservation profit ratio in their prospective contracts signed compared with high-risk mudāribs. likewise, the low-risk types will feel entitled to lower the maximum loss ratios in their prospective contracts compared with high-risk types. the incentive-compatibility constraint of the low-risk mudāribs will entail that they would likely demand higher profit ratio in the event of a profit, and a lower ratio of loss in the event of a loss, both compared with the high-risk mudāribs. figure 2 the two risk types result in two different fair-odds lines for the two classes of mudāribs. that said, there will also be an aggregate fair-odds line, which will be some kind of an average of the two type-specific fair-odds lines. in a pooling equilibrium, both types of mudāribs get away with the same type of treatment by the rabb-al-māl. the achievement of equilibrium requires a construct in which such a treatment lies on the aggregate fair odds line. the outcome of a pooling equilibrium would be represented by the point ‘a’, and that point must lie on the aggregate fair odds line. if the point ‘a’ lies above that line, it would be unattractive to the rabb-al-māl and so would not exist in equilibrium. if it lies below such a fair odds line, the underlying risk proposition would be so attractive to the high-risk segment of the muḍārib population as to swamp the volume of capital to be forthcoming from the rabb-al-māl. fair-odds line: highrisk fair-odds line: low-risk fair-odds line: highrisk fair-odds line: low-risk fair-odds line: highrisk fair-odds line: low-risk fair-odds line: highrisk fair-odds line: low-risk fair-odds line: highrisk fair-odds line: low-risk fair-odds line: highrisk fair-odds line: low-risk fair-odds line: highrisk fair-odds line: low-risk fair-odds line: highrisk fair-odds line: low-risk fair-odds line: highrisk fair-odds line: low-risk chowdhury: mush?rakah and mud?rabah instuments 44 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 so, at least for a while, there will be a tendency for some sort of pooled equilibrium, centered on the equivalent, in a sharī’ah-compliant financing market, of the point a. but this pooled equilibrium will not be sustainable, and so a position like a would not ultimately exist in equilibrium. the pooled equilibrium will be sustained fundamentally by the perception that the mudāribs and rabb-al-māl have identical probabilities for causing an outcome of loss, whereas this assumption will in reality be untrue. the reality shall often be that some mudāribs have safer hands for managing the money than others: the latter will be riskier partners to manage money with. if this latter perception is the more valid one, then this pooled ‘equilibrium’ will also lead to an implicit cross-subsidization of the riskier sub-sample of the population, at the expense of the safer sub-sample. even if the rabb-al-māl were initially to be unaware of the differences between the two risk-types, thus leading, pro tem, pooling equilibrium to prevail in the markets, there will come a time eventually when the low-risk mudāribs will be able to disabuse arbāb-al-māl of the fallacious symmetry. if all mudāribs have equal access (in both quantity and quality) to private information concerning their own risk-types, and if they reflect their private information fully in their contracting with the rabb-al-māl, then pooled equilibria are unlikely to exist in the murārabah market in a sustainable manner. this is because the low-risk type will have material incentives to stake claims for high reservation profit ratios, and also lower loss ratios versus ‘h’-type mudāribs. could they predictably carry the power of persuasion with the rabb-al-māl? we have already said that the latter is equally credulous lot when it comes to private risk information. but even so, that would not prevent the low-risk types from offering convincing arguments for why they deserve higher profit ratios compared with their higher-risk competitors. the main take-away from this discussion is that the issue of material incentives, unleashed by differences in incentive-compatibility between the high and low risk mudārib types, would likely create larger frequency of contracting frictions when it comes to pooling equilibria. on that basis, therefore, such pooling equilibria are more likely to be of interest as episodic, transitional, significance in murārabah markets. the combination of heterogeneity and equal access to private risk information among the mudāribs would likely spawn beneficial externalities of competitive disclosures by the low-risk mudāribs preparatory to ‘educating’ the arbāb-al-māl. cross-subsidizing of the high-risk mudāribs at the expense of the low-risk ones is likely to be the defining pitch of such competitive disclosures by the latter. the systematic differences between the lowand high-risk mudāribs in terms of the underlying material incentives will render the incentivecompatibility constraint of the high-risk types binding. if separating equilibria are not enforced, whatever contracts will be acceptable to low-risk types will also be acceptable to the high-risk types. the challenge facing the policy-makers in islamic economy is to devise and market murārabah contracts that could be offered to the low-risk murārabah types that would also not attract high-risk types. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 45 3.5 separating equilibrium in figure 3, points ‘al’ and ‘ah’ are the points that correspond to the fullsubscription of the arbāb-al-māl’ capital for the two risk groups. the low-risk mudāribs ought to have higher wealth because their probability of experiencing a loss is lower. figure 3: separating equilibrium the point labeled ‘c’ on the fair odds line for the ‘l’ group is where the indifference curve from the full-subscription point for the ‘h’ group crosses the fair odds line for the ‘l’ group. the indifference curve ‘u l ’ that intersects point ‘c’ is steeper than the corresponding curve for the ‘h’ group. the marginal rate of substitution corresponding to the ‘h’ group is smaller compared with the ‘l’ group. at ‘c’, one is at the best outcome one could offer to the ‘l’ types that would not also attract ‘h’ types: this is because ‘c’ does not admit any tangency between the fair-odds line for the high-risk types of mudāribs. the outcome c+ on the figure would be strictly preferred by the ‘l’ types, because the point c+ is closer to the full-subscription point than c. odds for profit o dd s fo r lo ss c chowdhury: mush?rakah and mud?rabah instuments 46 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 the issue is that the outcome c+ would also be preferred by the ‘h’ types, presumably for the same reason, which would create pressures for bringing the pooling equilibrium back in. the outcome of cis one which would be rejected by ‘h’ types, but ‘l’ types would strictly prefer outcome c, the original outcome. so, any outcome like cis dominated by c. it would be neat if outcomes ‘a h ’ and ‘c’ could somehow be offered in the muḍārabah markets, and, in addition, type ‘h’ would always choose ‘a h ’ and type ‘l’ would always choose outcome ‘c’. both outcomes have the desirable property of lying on the fair odds line for the group whose capital requirements are being met. but the complicating issue is that because the percentage split is the same for both profits and also losses, and also because that split does not differentiate among entrepreneurs based on whether they are loss-leaders or profit-leaders, low-risk mudāribs are likely to be only partly covered, relative to high-risk ones. compared with low-risk mudāribs, high-risk users of capital receive a certain amount of preferential treatment. but if it were practicable to fully cover the capital needs of the lowrisk mudāribs, it was not going to be practicable to inoculate them from the high-risk ones. the main problem is that preferences of high-risk mudāribs are binding and shape up as the constraint on the murārabah market. the challenge before the policy-makers is that the decision rules of capital allocation (that is, the sharing ratios for profits as well as loss) must maximize the well-being of low-risk mudāribs subject to the constraint that the high-risk ones can be weaned from the low-risk ones. this is the essence of separating equilibrium.5 following this chain of argument, separating murārabah contracts are designed for two separate kinds of contracts tailor-made to reflect the different risk profiles of the two groups of mudāribs. as well, there is a need to ensure that the two groups find it in their interest to self-select into buying the contract tailor-made for it. the crucial issue is how to do both of those two things. the short answer is sharī’ah-compliant variable-return schemes direly depend upon certain institutional, informational and behavioral structures, allowances and sanctions that make it imperative for both highand low-risk agents to separate themselves in the way in which they sort out the murārabah contracts they sign up. due to space limitations for this paper, just what those structures, allowances and sanctions can be is left as the subject-matter for a follow-up research. we have invoked heterogeneity and information frictions among prospective participants, and established self-selection of mutually-exclusive types of murārabah and mushārakah contracts that would endure as separating equilibria a ‘second-best’ solution. separating ‘equilibria’ will require the establishment of both voluntary signaling and enforced screening to be wellarticulated institutional features of the market. this will be so that the problems posed by hidden information (adverse selection) and hidden actions (moral hazard) are substantively neutralized, if not altogether eliminated. however, 5 high-risk muḍāribs are no better off for the harm they do to the low-risk group. one group loses but no group gains. this is the opposite of pareto improvement—and potentially a large social cost. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 mushārakah and mudārabah instruments: theory development to assess their characteristics: 25-48 47 the foregoing points at quite a formidable exercise. this paper is already long enough. it will be more appropriate to attempt the policy/interventions that are preparatory to any solution of the problem posed in this paper in the form of a research note to be written in the near future. 4. conclusions the main reason why mushārakah and mudārabah are the smallest portion of the sharī’ah-compliant mix of financing modes currently used is because the incentives of the users of capital are not adequately aligned with those of the providers of the capital. equity always represents hard-earned savings of some people. in a murārabah, agents who have typically been starved in their occupational careers of the access to capital are handed close access to it. this can be a compelling experience, triggering ambition and unseemly dispatch in getting to where ambition leads. in addition, it is these agents that enjoy a predictable cache of capability not possessed by other co-transactors. many of the mudāribs have the virtue of informational advantage compared with the rabb-al-māl. the two factors, ambition and advantage, combine to result in a rather potent influence. the situation is made even more volatile in that some muḍāribs represent steadier and safer hands in which to put management of valuable businesses than others. the apparent irreconcilability of incentives that is very much intrinsic to mushārakah/mudārabah can be moderated if there were some built-in features that could induce low-risk and high-risk mudāribs into channels of self-selection representing different terms and conditions of the murārabah that could be custom-made. in the language of microeconomics, what is needed is a separating equilibrium of mudāribs representing differences in risk types or in the proclivity to indulge in his informational advantage to advance his own financial interest to the exclusion of his partner’s. however, such volitional selfselection does not come in a costless or easy manner. an institutional framework is needed in the islamic finance industry to stay the hands of entrepreneur to engage the financiers to reduce the risk of asymmetric information and incentivecompatibility. it is possible to devise such an institutional framework for the advancement of these core financing modes to become more acceptable as a means to spur financing to entrepreneurs. the root causes of the high private costs of securing separating equilibrium have to do with (i) the barriers that wall-off islands of private information from the public square, (ii) the hierarchical nature of socio-economic institutions which impedes the equality of both the elites and the ordinary before the law and (iii) the preponderance of archaic, manual, methods for capturing economically relevant disclosures. all these difficulties create a kind of obstacle course in the economic setting. that in turn makes it especially difficult to intervene with some variant of signaling/screening combinations, which, as in other risky economic settings, have been deployed with good results. a detailed chowdhury: mush?rakah and mud?rabah instuments 48 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 account of the policy interventions must, however, be the subject of a future research. author information: dr. naeem chowdury worked as a senior research fellow at the bangladesh institute of economic development, dhaka, bangladesh. these days he is a freelance writer. his next book project is about exploring the ‘scientific’ prescience in verses in the learned al-quran that especially are aimed at the skeptics and atheists. chowdhury may be contacted via e-mail: naeemchy@gmail.com. references belal, a. r. abdelsalam, o., & nizamee, s. s. (2010). a longitudinal case study of social reporting in islami bank bangladesh limited (ibbl). birmingham: aston business school. dar, h. a. (2007). incentive compatibility of islamic financing. in m. kabir hassan and mervin k. lewis (eds). handbook of islamic banking. cheltenham, uk and northampton, ma: edward elgar. cutler, d. m., & zeckhauser, r. j. (1998). adverse selection in health insurance. in alan a. garber (ed.), frontiers in health policy research vol. 1. cambridge: mit press. goud, b. (2012). should islamic banks use muḍārabah and mushārakah? retrieved from http://investhalal.blogspot.ca hassan, m. k., & lewis, m. k. (eds). (2007). handbook of islamic banking. cheltenham, uk and northampton, ma: edward elgar. iqbal, m., ahmed, a., & khan, t. (1998). challenges facing islamic banking. jeddah: islamic research and training institute. iqbal, m., & molyneux, p. (2005). thirty years of islamic banking: history, performance and prospects. houndmills, uk and new york: palgrave mcmillan. ray, d. (1998). development economics. princeton: princeton university press. rosenberg, n. (1973). perspectives on technology. new york: cambridge university press. rothschild, m., & stiglitz, j. e. (1976). equilibrium in competitive insurance markets: an essay on the economics of imperfect information. quarterly journal of economics, 90(2): 629-649. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 5 bank owned life insurance: a critical examination of banking strategy 15 the international journal of banking and finance, volume 9 (number 3) 2012: pages 15-26 bank owned life insurance: a critical examination of banking strategy leyuan you and krishnan dandapani texas state university and florida international university, united states of america ____________________________________________________________ abstract this paper is an investigation of a rampant insurance practice in the us banking sector, namely the permitted practice of employer-paid insurance policy. under this policy, employee’s life policy paid for by the employer, pays large sums as policy benefits to the employer not to the employee’s family. employers suggest that taking insurance covers the possible monetary loss value of an employee departing the firm, and hence the benefit is quite permissible, kosher. our findings show its widespread occurrence in beefing up the earnings and even the capital base of the us banks. it calls into question if this practice, though legal, is socially responsible since an employee with such employer-paid policy would be deprived of tax deduction for his own selfpaid policy cover, in most instances. banking sector’s prevalent use of this practice to cover even low-paid workers such as janitors brings into focus the ethics of this banking practice. key words: bank owned life insurance, legal practice, ethics in finance jel classification: g21 _____________________________________________ 1. introduction to boli great attention has been given to the bank owned life insurance (boli) recently in the united states (us). boli is the life insurance that a bank purchases on the life of its officers (or sometimes senior employees) whose lives banks has a beneficial actuarial interest. over the last decade, many banks have purchased boli to offset a variety of employee benefit obligations such as financing supplemental executive retirement, deferred compensation programs, health insurance, tuition reimbursement, and insuring against the risk of losing an important officer, all which constitute a major expense for most financial institutions. 16 in a typical case, the bank evaluates each employee by census information: age, sex, compensation, etc. to calculate the present value of benefit expenses. then the bank buys an insurance policy on the life of a group or groups of key managers or employees and pays the premium.1 generally, the bank is also the sole or joint beneficiary of the insurance. thus, when an adverse event happens, the bank makes earnings minus the insurance charges that are paid till that date; most of the traditional bank investments generate taxable interest income. in contrast, due to favorable tax laws in force for insurance settlements, boli earnings produce no current income tax liability for banks. two sources of boli’s income are tax free: growth of the cash value in the policy and the death benefit received upon the death of the insured. by structuring the deal in a strategic fashion, the tax-deferred earnings can be used to offset employee benefit. the value of boli holdings is reported as “other asset”, the earnings on the holdings are reported as “other noninterest income”, and the death benefit is recorded as an extraordinary income by the bank. broadly, the boli can be categorized into two classes: the more typical case, permanent life insurance, provides life insurance for the entire life of the insured employee, and the temporary or term insurance, provides life insurance for a specified time period. since the term insurance does not usually generate cash surrender, it is a less popular tool to offset costs. 1.1 what is boli and why boli? in the 1960s, u.s. companies have used corporate owned life insurance (coli) to finance corporate liabilities. coli have been so successful that some corporations have purchased them more for its investment value than for employee benefit. in the late 1980s and early 1990s, large banks started to follow suits by creating banked owned life insurance. during the past several years, many banks net interest margins and deposit growth have been under constant pressure due to competition and margins have been eroding. banks have responded to this challenge by diversifying their source of income through non-traditional earning assets that can increase their non-interest income. boli has been the most popular strategy and it is proven to generate more returns and stability than even coli. 1 the bank has to get consent from the insured and in several states the insured employee has the right to cancel the insurance if he/she is leaving the position. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 2 17 boli has proven to be a cost-efficient and effective medium for banks to offset rising employee benefit costs. boli is an attractive earning asset for community banks, yielding after tax returns of 200-300 basis points in excess of traditional bank investments. according to meyer-chatfield, a boli consultant company, boli provides historical roas of nearly 300 basis points and historical roes of 30% for banks using this strategy. boli is classified as an “other asset” in financial statements and provides other noninterest income, which is considered important to most investors. the cash surrender value of life insurance (csvli) has become a significant asset for several institutions, and in some cases approaching 20 per cent of tier 1 capital. it can generate large returns because the cash value grows tax deferred and death benefits are tax-free insurance policies that are held to maturity according to the insurance banks. the cost the earnings, the insurance company’s charge on the policy, is the least expensive term insurance rate. when properly implemented, there is no loss of deductible funding cost as with municipal bonds. besides, the credit risk for boli is very low as the crediting rate is generally better than rates of other investments like tbills. according to the general accounting office, in a 2004 report, by the end of 2002 at least one third of banks and thrifts in the u.s. held boli. this number increased to 41% by the end of 2004 according to a leading boli provider. in a recent survey, boli assets hit $126.1 billion in 2008, this increased by 5 per cent from 2007 for 880 top tier bank holding companies and 7,495 commercial and savings banks. when boli is designed properly, it is more than insurance for the bank; it has become a sound investment strategy. 1.2 regulatory history there are four regulators for financial institutions in the u.s.a. the federal reserve board (frb), the federal deposit insurance corporation (fdic), the office of thrift supervision (ots) and the office of the comptroller of the currency (occ). while the occ pioneered the supervisory standards to boli, the fdic and ots have implemented new regulations regarding boli purchases and standards in the last several years. prior to 1991, during the evolutionary days of boli there were common rules with regard to boli. regulators evaluated the boli plan submitted by banks and an after study and appraisal informed the banks if it was approved. then, in june 1991, office of comptroller of 18 currency (occ) released banking circular 249, the first several formal regulations which basically limit the life insurance for non-investment purposes. it initiated a set of standards of the boli purchasing: insure against loss of key employees, insure the life of a borrower, as a security for a loan, and finance employee benefits. later, in august 1993, the fdic issued a set of its own guidelines which adopted and expanded on the previous circular 249. it required the board of directors of the banks to ensure the soundness of boli in several areas such as reasonableness of benefits, concentration and other risk considerations, and accounting issues. in september 1996, the occ issued another set of rules, bulletin 96-51, which provided general guidelines to ensure that the purchase of boli is consistent with safe and sound banking practice. these rules details include a pre-purchase analysis of life insurance in very specific issues. bulletin 96-51 provides more flexibility for banks to structure boli, allowing the bank to have more applications of life insurance as long as they have valid business reasons. in july 2000, occ replaced the 96-51 with a revised circular 2000-23. this set of rules remains largely the same but expand to other areas such as price risk. additionally, in may 2002, occ released 2002-19 to provide supplemental guidance and reinforce the legitimateness of life insurance. in august 2002, ots issued regulatory bulletin 32-26 applicable to ots regulated federal and state thrifts. in december 2004, bank regulators2 issued the interagency statement that sets forth supervisory expectations for senior management and boards of directors to understand and mitigate the risks associated with boli. basically, it outlined elements of “prudent risk management process” which include the following four major areas: 1. effective senior management and board oversight, making sure those directors understand the risks which include: • liquidity risk –boli is not a highly liquid investment. the longer the time elapse between the purchase of the insurance and the death of the insured, the greater the tax free income to the bank. 2 occ bulletin 2004-56 was issued by the office of the comptroller of the currency (occ), federal deposit insurance corporation (fdic), the board of governors of the federal reserve system (frb), and the office of thrift supervision (ots). international journal of banking and finance, vol. 9, iss. 3 [2012], art. 2 19 • transaction/operational risk –the risk to earnings or capital arising from problems with service or product delivery. understanding and appropriate monitoring is needed to make sure it is properly implemented. • tax and insurable interest implications risk consultation with professionals is essential to make sure it is done right or the consequences can be dire. • reputation risk – the newly defined risk in 2004-56, failure of the insurance company providing boli could affect the ability of institution to meet its obligations. also, failure to establish boli in a proper manner (employees insured without their consent) can reflect poorly on the institution. • credit risk – the risk arising from an obligor’s failure to meet the terms of a contract. generally, the carriers have high credit ratings while boli has low credit risk. • interest rate risk the risk arising from movement in interest rates. banks do not have much control over interest rate risk-boli interest rates fluctuate after the first year and these annual "resets" reduce interest rate risks, but boards should review the past performance over business cycles. • compliance/legal risk – the risk arising from violation and non-conformance with laws rules and regulations. failure may involve regulatory and liability consequences. 2. comprehensive internal policies and procedures including cash surrender value (csv) should not exceed its legal lending limit nor in the aggregate exceed 25% of its tier 1 capital. 3. a thorough pre-purchase analysis to identify the need of boli and its benefit, to assess vendor qualification, and to analyze risk and alternatives. 4. an effective, ongoing system of risk assessment, monitoring and internal control including appropriate frame works for internal audit and compliance. this new economic phenomenon also results in social problems, such as moral and tax issues. in the next section, we take a look at the economic effect of boli. specifically, how and how much does the boli benefit the banks and what are the issues of boli. 2. empirical analysis: data and analysis we chose the top 10 largest banks in the u.s. to analyze the impact of boli. we obtained the data of these banks from the fdic official website. this website also provides balance sheets 20 and other financial reports of its member banks. since life insurance is reported under “other assets”, and the earnings from it is reported under the “other noninterest income”, this is not required to be reported elsewhere, therefore, the information on life insurance is not reported in the regular annual report given to investors. banks not insured by the fdic do not have this information on fdic’s website. as of september 10, 2008, the link to obtain the information on life insurance can be found at the following link: https://cdr.ffiec.gov/public/managefacsimiles.aspx. typically, information for boli is obtained from “call reports and thrift financial reports”. the fdic updates its reports every three months. table 1: boli and its earnings $ 0000 for bank of american, december 31, 2002 from the balance sheet total assets (sum of items 1 through 11 in table 2) rcfd 2170 565,382,000 from the explanation for the “other asset” item in the balance sheet cash surrender value of life insurance rcfd c009 9,650,000 from the income statement a. net income (loss) (sum of items 10 and 11 in table 2) iriad 4340 8,216,000 from the explanation for the other non interest income in the income statement b. earnings on/increase in value of cash surrender value of life insurance iriad c014 510,000 an illustration is given in table 1. it provides the value of boli asset, the total asset, the earnings from boli and the total net earnings for bank of america, charlotte, on december 31, 2002. the value of the life insurance for bank of america in 2002 is $9.65 billion, 1.7 per cent (consider banking tier 1 capital is merely 4%) of the total assets, which is $565 billion. the earnings of the life insurance given the income statement are $510 million, 6.21 per cent of the $8.216 billion total net income. with 1.7 per cent of the assets making 6.21 per cent of the net income, this shows that earnings on life insurance can be very profitable. but the earnings cannot be realized every year; most of the time the cash value of the life insurance does not change. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 2 https://cdr.ffiec.gov/public/managefacsimiles.aspx https://cdr.ffiec.gov/public/searchfacsimiles.aspx https://cdr.ffiec.gov/public/searchfacsimiles.aspx 21 we obtain the most recent quarterly report in 2008 for the seven banks, found on the reports of june 30, 2008. we then compared it to the third quarter report in 2007, 2006, 2005 and 2004. table 1: top ten largest banks in u.s. in 2008 rank name (city, state) information consolidated assets provided (in millions of u.s. dollars) 1 citigroup (new york, n.y.) no $2,199,848 2 bank of america corp. (charlotte, n.c.) 3510 1,743,478 3 j. p. morgan chase & company (columbus, ohio) 628 1,642,862 4 wachovia corp. (charlotte, n.c.) 33869 808,575 5 taunus corp. (new york, n.y.) no 750,323 6 wells fargo & company (san francisco, calif.) 34075 595,221 7 hsbc north america inc. (prospect heights, ill.)* 57890 493,010 8 u.s. bancorp (minneapolis, minn.) no 241,781 9 bank of the new york mellon corp. (new york, n.y.) 639 205,151 10 suntrust, inc. (atlanta, ga.) 867 178,986 11 citizens financial group, inc. (providence, rhode island) ** no 161,759 note: as of may 30, 2008. source: federal reserve system, national information center. *hsba usa is used instead of hsbc north america since the latter is not found in fdic website. ** the eleventh bank is shown for comparison purposes. panel a in table 3 gives the value and the earnings of life insurance for the seven banks for the third quarter in every year from 2004 to 2008. total assets and net income are provided; the life insurance value and earnings on life insurance are also expressed in percentages of total assets and net income respectively. panel b of table 3, shows the average values over the five quarter periods of the value and earnings on boli in percentages of total assets and net income. 22 table 3 shows that of the seven banks, all except wells fargo have boli. of the of six banks using boli, only bank of america and wachovia have positive earnings on boli in our sample period, the former in 2008 and the latter from 2004 to 2007 . table 3: summary of boli for the large seven banks from 2004-2008 panel a historical pattern of boli for the seven banks for the past five years bank of america jp morgan chase wachovia wells fargo hsbc bank of new york suntrust average across banks 6/30/2008 boli assets 16,501 7,561 14,575 na 213 1,820 611 5,897 total assets 1,350,269 1,378,468 660,496 176,606 130,062 168,683 552,083 boli/total assets 1.22% 0.55% 2.21% na 0.12% 1.40% 0.36% 0.84% earnings on boli 328,146 0 0 0 0 0 0 net income (loss) 3,391,859 earnings/net income 9.67% 0 0 0 0 0 1.61% 6/30/2007 boli assets 14,131 7,327 13,430 na 206 1,755 584 5,348 total assets 1,264,470 1,252,369 525,811 164,881 95,641 175,010 496,883 boli/total assets 1.12% 0.59% 2.55% na 0.12% 1.83% 0.33% 0.94% earnings on boli 0 0 279,000 0 0 0 0 net income (loss) 3,193,000 earnings/net income 0 0 8.74% 0 0 0 1.46% 6/30/2006 boli assets 13,614 6,993 10,728 na 196 1,690 553 4,825 total assets 1,145,895 1,150,905 480,134 165,593 93,466 179,265 459,323 boli/total assets 1.19% 0.61% 2.23% na 0.12% 1.81% 0.31% 0.89% earnings on boli 0 0 219,000 0 0 0 net income (loss) 3,065,000 earnings/net income 0 0 7.15% 0 0 0 1.19% 6/30/2005 boli assets 13,160 0 9,848 na 0 1,690 0 3,528 total assets 1,067,486 0 452,786 141,452 93,466 0 250,741 boli /total 1.23% 0.00% 2.17% na 0.00% 1.81% 0.00% 0.75% international journal of banking and finance, vol. 9, iss. 3 [2012], art. 2 23 assets earnings on boli 0 0 231,000 0 0 0 net income (loss) 3,055,000 earnings/net income 0 0 7.56% 0 0 0 1.26% 6/30/2004 boli assets 10,125 0 7,349 na na 1,690 0 3,194 total assets 706,888 654,641 368,871 93,466 0 303,978 boli/total assets 1.43% 0.00% 1.99% 1.81% 0.00% 0.87% earnings on boli 0 0 199,000 na na 0 0 net income (loss) 2,229,000 earnings/net income 0 0 8.93% na 0 0 0 1.49% panel b boli statistics averaged across last five years values averaged over 5 years bank of america jp morgan chase wachovia wells fargo hsbc bank of new york suntrust average across banks boli 13,506 4,376 11,186 0 154 1,729 350 4,472 life insurance/total assets 1.24% 0.35% 2.23% 0 0.09% 1.73% 0.20% 0.83% earning 65,629 0 185,600 0 0 0 0 35,890 earnings /net income 1.93% 0.00% 6.47% 0 0 0 0 1.20% note: the total assets are in unit of million and the earnings in unit of thousand. when the qualified event occurs, the earnings on boli make up a considerable percentage of net income. for example, with 1.22 per cent of total assets invested in boli, the earnings on boli account for 9.67 per cent of net income in the third quarter of 2008for bank of america. the boli assets increased from an average of $3,194 to $3,528 in 2005, $4,825 in 2006, $5,348 in 2007, and $5,897 million in 2008. the average quarterly earnings grew from $39,800 thousand to $54,690 thousand in 2008, however, it does not grow at a steady rate. panel b shows that the average value of the boli is $4,472 million for the six banks over the last five years, which accounts for 0.86 per cent of the banks total assets, and the average 3rd quarter earnings on boli over the last five years is $35,890,000, accounting for 1.4 per cent of the net income. in general, table 3 shows that boli can be profitable to banks. 24 4. regulatory implications: how boli affect camel rating of banks the camel methodology was originally adopted by north american bank regulators to evaluate the financial and managerial soundness of u.s. commercial lending institutions. the camel reviews and rates five areas of financial and managerial performance: capital adequacy, asset quality, management, earnings and liquidity. take bank of america, national association at charlotte for example, boli accounts for 1.7 per cent of the total assets on dec 2002. this number increased to almost 16.67 per cent in march 2004. it greatly improves the capital adequacy for the bank. the earnings of boli accounts for 5.29 per cent of the total net income on march 2002, it increases to 6.02 per cent in 9 months. 4.1 other issues capital adequacy, basle and regulation issue: before 1991, there were no regulations addressing the issues around boli. the first several regulations came out around may 1991. they generally require that there are only 4 purposes of boli: insure loss of key employee, insure the life of a borrower in an account equal to the bank’s exposure to loss, as a security for a loan, and finance employee benefits. later on the occ loosened the rules. still possible regulatory review of boli products can occur if not implemented properly. federal tax issue: the increasing audits by irs show that boli has also become a hot issue with respect to tax. under applicable law, the bank must have an “insurable interest” in order to qualify for tax free insurance proceeds. even though the insurance meets the occ requirements, it may not qualify for insurable interest requirement for tax purposes. the irs can challenge whether the bank is entitled to the favorable tax treatment based on the information it collected. it requires exhaustive details on boli and inquiries into all aspects of boli. if the bank doesn’t provide adequate documents to show its legitimate purpose of the insurance, the irs can invalidate the contract as insurance. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 2 25 the irs has expressed its concerns about banks that purchase policy coverage that far exceeds the risk of loss may use it for business purpose. recent cases on coli, though different from boli, rendered some lessons for banks. in order to enjoy tax deferred build up of the boli cash value, the insurance policies must be under the ownership of the insurer, not the bank, meaning the cash values are not controlled by the bank. the irs has established some criteria for separating accounts for boli. however, it is highly dependent on specific circumstances. social, moral and ethical issue making profit out of insurance on people’s lives is a controversial issue. social commentators have argued that it is the banks not the employees who are being insured and their family who get the benefits. while banks do have an actuarial interest in the lives of employees, many controversial practices necessitate a clear and pre-defined path for banks to follow. although these issues may be beyond the scope of this paper here, it needs to be investigated. 5. conclusion boli can benefit banks with present regulations and tax rules. it increases bank earnings, and helps to improve the ratings of banks within the camel context. it also produces risks and social issues. but because boli is a relatively new product, it needs to be further studied and possibly regulated to avoid implicit conflict of interest with the personal welfare of the insured employee. author information: leyuan you is a faculty member in the department of finance and economics, mccoy college of business administration, texas state university san marcos, texas, 78666, united states of america. her contact detail: email ly17@txstate.edu. krishnan dandapani is a professor at the department of finance and real estate, florida international university, miami, florida, 33199, united states of america. he may be contacted at e-mail: dandapan@fiu.edu. mailto:ly17@txstate.edu 26 references brew, jay, (2007). taxes: the forgotten piece of bank profits, american bankers association. aba banking journal. 99: 22-24. collins, patrick j., dieter, jurkat, (2005). the decision to replace trust owned life insurance policies, the banking law journal, 122: 975 mcdonald, randall a., (1998) consolidated supervision of banks, bankofengland.co.uk nurnberg, hugo, douglas p. lackey, (2010). the ethics of life insurance settlements: investing in the lives of unrelated individuals, journal of business ethics 96:513–534. smith, brian w., kimberly m. boylan, gregory s. feder, and joseph e yesutis, (2001). bankowned life insurance: pre-purchase analysis and post-purchase compliance, the banking law journal, 118: 208. smith, brian w, wendy l. morris, (1997). take another look at bank-owned life insurance, american bankers association. aba banking journal 89: 22-24. wamberg, warren t., (1995). the theory & practice of bank-owned life insurance, t. w. o. pubication, nj, united states. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 2 http://proquest.umi.com/pqdweb?rqt=572&vtype=pqd&vname=pqd&vinst=prod&pmid=22987&pcid=35962661&srchmode=3 http://proquest.umi.com/pqdweb?rqt=318&pmid=22987&ts=1292629175&clientid=11421&vinst=prod&vname=pqd&vtype=pqd http://proquest.umi.com/pqdweb?rqt=318&pmid=22987&ts=1292629175&clientid=11421&vinst=prod&vname=pqd&vtype=pqd javascript:void(0); javascript:void(0); http://proquest.umi.com/pqdweb?rqt=318&pmid=27123&ts=1292628292&clientid=11421&vinst=prod&vname=pqd&vtype=pqd http://proquest.umi.com/pqdweb?index=7&did=71782232&csp=581685%2c569734%2c562370%2c562180%2c443153%2c15553%2c9777&srchmode=2&sid=2&fmt=2&vinst=prod&vtype=pqd&rqt=590&vname=pqd&ts=1292627539&clientid=11421 javascript:void(0); javascript:void(0); javascript:void(0); http://proquest.umi.com/pqdweb?rqt=318&pmid=27123&ts=1292627539&clientid=11421&vinst=prod&vname=pqd&vtype=pqd http://proquest.umi.com/pqdweb?rqt=318&pmid=27123&ts=1292627539&clientid=11421&vinst=prod&vname=pqd&vtype=pqd http://proquest.umi.com/pqdweb?index=12&did=11170647&srchmode=1&sid=2&fmt=4&vinst=prod&vtype=pqd&rqt=309&vname=pqd&ts=1292627545&clientid=11421 mutual fund performance and the impact of superannuation and financial regulation: 1-20 1 mutual fund performance and the impact of superannuation and financial regulation james m. cooper diligentia venture partners ltd russell gregory-allen massey university *corresponding author: r. gregory allen @ massey.ac.nz ________________________________________________________________ a r t i c l e i n f o _____________________________ article history: received 22 february 2018 revised 8 july 2018 accepted 8 august 2018 ____________________________ keywords: mutual fund performance, mutual fund regulation, regulation changes. jel code: g11, g15, g23, g28 a b s t r a c t _________________________________ financial innovation such as a new superannuation scheme can allow for broader participation in retirement savings by individuals, but might also impact existing investments. on the other hand, mutual fund regulation involves a balancing act between protecting investors, and allowing fund managers to exercise their skills. some recent changes in the fund environment of new zealand allows an examination of the impact on performance from those changes in a small, open economy. using a sample of new zealand mutual funds, we compared performance before and after the introduction of two significant changes in the financial environment of new zealand. in 2007, a state-sponsored investment scheme called kiwisaver was introduced, providing significant incentives for more and more new zealanders to save. participation was substantial, and by 2015 kiwisaver funds under management had exceeded traditional open-end funds. at the time of kiwisaver’s introduction, mutual fund regulation was quite lax, particularly in the area of financial disclosure. however, in 2013 a new law was introduced, substantially increasing the disclosure requirements for those funds the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 1 2 the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 participating in the kiwisaver scheme. first we examined, the impact on the new zealand mutual fund industry upon the introduction of kiwisaver, and then on the introduction of the increased kiwisaver regulation, in order to determine if these harmed the overall new zealand mutual fund industry. we found that the new zealand mutual funds which focused on new zealand or australian equities experienced some negative performance after the introduction of kiwisaver, but the impact on the overall industry was not significant. we also found that the increased regulation had some positive impact on performance, particularly for those funds emphasising global equities. 1. introduction understanding the impact on mutual funds as a result of the introduction of new financial innovation and regulation is a matter of vital importance because all parties – investors, fund managers and regulators – want to know if performance has been positively or negatively impacted and whether outcomes will likely have implications for future policies. this study investigates the impact of regulatory changes in new zealand, comparing results before and after the introduction of two important events in the financial environment. new zealand’s mutual fund industry has undergone significant changes in the last decade with the introduction of new financial regulation and initiatives. on july 2, 2007 kiwisaver, new zealand’s voluntary work-based retirement (pension) savings scheme was introduced, focused mainly on employee and employer periodic contributions1. subsequently, on july 1, 2013 the new kiwisaver periodic disclosure regulation came into effect. the new zealand mutual fund industry is very small compared to the us, however since the introduction of the kiwisaver scheme, total funds under management have grown substantially and investment participation has taken off. kiwisaver was introduced in july 2007, and by june 2015 kiwisaver had 2.53 million investors2, with nzd $32.8 billion under management (see table 1). as of june 2015 the total population of new zealand was 4.59 million3 with 3.16 million in the workforce4, implying an 80% participation rate in kiwisaver. prior to kiwisaver’s introduction less than 30% of the active mutual fund performance and the impact of superannuation and financial regulation: 1-20 3 labour force was covered by a retirement plan (kritzer, 2007). furthermore, as noted in table 1, as at the end of kiwisaver’s first full year (june 2008) there were $710 million under management compared to $12.8 billion in typical mutual funds. however, by june 2015, kiwisaver had surpassed mutual funds, $32.8 billion to $26.4 billion. table 1. relative growth in new zealand’s open-end funds vs. kiwisaver and other from morningstar’s “fund size-comprehensive” in millions $ nzd, as of 30 june each year open-end kiwisaver other super-annuation 2008 12,809 710 2,905 2009 13,769 2,547 3,745 2010 15,169 5,046 4,079 2011 16,283 7,885 5,300 2012 19,115 11,082 4,989 2013 19,452 17,181 5,249 2014 23,779 23,719 5,471 2015 26,429 32,832 6,326 since the introduction of kiwisaver, both the global and local financial regulatory environments have changed significantly. after the global financial crisis in 2008, financial regulation came under considerable political review. the outcome in most developed economies was increased oversight and changes to regulatory structures and powers. new zealand had the financial markets authority act 2011 (fma act) which saw the establishment of the financial markets authority (fma) in 20115, and in 2013 the financial markets conduct act 2013 (fmc act) was introduced. while all changes are likely to have impacted the new zealand mutual fund industry, this study’s second focus is on kiwisaver (periodic disclosure) regulation 2013 (later revamped to the financial market repeals and amendments act 2013)6 which came into force on july 1, 2013. as the name suggests, the kiwisaver periodic disclosure regulation require fund managers to produce quarterly disclosure statements and annual disclosure statements, also known as qds and ads, respectively, for each kiwisaver fund. qds must be made publicly available by fund managers within 15 working days the end of each quarter and ads within 60 working days. disclosure statements help investors to evaluate fund performance and fees, to know what the major investments of the fund are, and to make comparisons with other funds. this is easier under the new regulation, due to an increased consistency of information. there is an extensive financial literature on mutual fund performance dating back many decades. however, a dearth of academic studies exists on new zealand mutual fund performance. with the relatively recent introduction 4 the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 and growing size and importance of kiwisaver, a few empirical studies have appeared in recent years focussed on kiwisaver performance. our focus, however, is on the wider new zealand mutual funds industry, and how the introduction of kiwisaver has affected the rest of the new zealand mutual fund landscape. our main research questions are: how did risk-adjusted fund performance change as a result of 1) the introduction of kiwisaver and 2) the introduction of the kiwisaver periodic disclosure regulation? we pose these questions to help build our understanding of the impact of financial market innovation and regulation on mutual fund performance. ultimately, further studies could shed more light on many other important detailed questions7, but in this study our interest is to better understand how markets perform with respect to financial regulation. to examine the effect of the introduction of kiwisaver (ks) on new zealand mutual fund performance, we examined the eight year period before and after the september 1, 2007 introduction of ks. the second event is the introduction of the kiwisaver periodic disclosure regulation (kspdr), evaluated with two year periods preand post-kspdr introduction on july 1, 2013. to measure risk-adjusted performance, we use the familiar capm, fama and french (1993) three-factor model, and carhart (1997) four-factor model. following frijns and tourani-rad (2015) this study evaluates fund performance by measuring exposure to new zealand and australasian equity markets. we extend this method by grouping equity funds by regional investment focus to select an appropriate benchmark on four groups: new zealand, australia, australasia (funds investing in new zealand and australia), and global. we find that the overall impact of introducing the kiwisaver funds has been positive for the industry, with more investment participation by new zealanders, and little performance impact. we also find that the impact of the increased regulation has had no negative impact, and may have helped in investment participation. it is our view that this implies that extending similar regulation to the rest of the industry is probably good for the industry as well. 2. literature review 2.1 mutual fund excess performance evaluation a substantial collection of academic studies have examined mutual fund performance. they provide a well-considered pathway of literature, methodology and results to review. at the heart of mutual fund performance evaluation, with respect to methodology, are the capital asset pricing model (capm) and jensen’s alpha. the development of the capm is largely attributed to sharpe (1964), lintner (1965) , mossin (1966), treynor (1961a) and treynor (1961b) all mutual fund performance and the impact of superannuation and financial regulation: 1-20 5 independently developing similar theories. capm derives the expected or required rates of return on risky assets based on the assets’ systematic risk relative to a market portfolio. jensen (1968) created a portfolio performance measure we now call jensen’s alpha which measures a fund manager’s investment ability to underperform or outperform the market. this is done by rearranging the capm and utilising an intercept that we call alpha (α). a statistically significant and positive alpha indicates superior performance or outperformance, while a negative alpha highlights inferior performance or underperformance. multifactor models have since extended the capm and jensen’s alpha model. the two most influential are the three-factor model of fama and french (1993) and the four-factor model of carhart (1997). the fama and french (1993) three-factor model seeks to specify microeconomic risk with the addition of two factors, smb and hml. smb (small minus big) is a proxy designed to capture firm size risk elements by building and measuring the return on a portfolio of small-cap stocks minus the return on a portfolio of large-cap stocks. hml (high minus low) is a proxy based on risk differentials regarding growth versus value stocks. the return on a portfolio of high book-to-market (value) stocks minus the return on a portfolio of low bookto-market (growth) stocks yields the proxy hml. carhart (1997) took the fama and french (1993) three-factor model and combined it with the work of jegadeesh and titman (1993) by adding a fourth factor, momentum (wml). wml (winners minus losers) is a price momentum factor which takes the past year price performance of the highest performing equal-weighted group minus the lowest performing equal-weighted group. 2.2 united states research results prior academic studies have uncovered a variety of results. in the under performance camp jensen (1968), carhart (1997), fama and french (2010), otten and thevissen (2011), ferreira, keswani, miguel, and ramos (2012) , plus many others – all found that mutual funds underperform the market on a risk-adjusted basis. however, in the outperformance camp there were also notable studies such as by kosowski, timmermann, wermers, and white (2006) and petajisto (2013). performance persistence is one particular subset of mutual fund performance that is extensively researched, also with mixed results. grinblatt and titman (1992), elton, gruber, and blake (1996), goetzmann and ibbotson (1994), s. brown and goetzmann (1995), elton, gruber, das, and hlavka (1993) and wermers (1997) all found some evidence of positive performance persistence. conversely, jensen (1969) and carhart (1997) found no evidence of performance persistence. 6 the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 in recent years specific issues have been examined that look to explain performance by other factors such as risk and style choices, fund size, fund family size, fund flows. chen, hong, huang, and kubik (2004), pollet and wilson (2008) gil–bazo and ruiz–verdú (2009) found that mutual fund performance is positively related to the size of the family to which the fund belongs. 2.3 new zealand research results research on new zealand mutual fund performance is relatively scarce. vos, brown, and christie (1995) conducted a test of performance persistence on 14 new zealand and 12 australian equity mutual funds available to new zealand investors during the period, 1988 to 1994. in this small sample they found that past fund performance ranking for both raw returns and risk-adjusted returns had no predictive value for the next period ranking. bauer, otten, and rad (2006) conducted a performance test of 143 funds from 1990 to 2003 controlling for survivorship-bias and found no evidence of outperformance by new zealand mutual funds. however, they did find evidence of significant underperformance for balanced funds. in further tests they discovered a positive relationship between fund size and risk-adjusted performance, as well as for fund size and expense ratio. the recent research trend has been to focus specifically on kiwisaver funds. fund and member flows were examined by thomas and matthews (2014), who found a positive association between performance and both fund flow and member flow. they also found some rather inconclusive results relating to fund size with respect to fund flow. zhang (2014) examined the impact that financial advice had on asset allocation decisions of investors and determined that there was marginal difference between fund performances of those that were advised versus those that were not advised. the individual kiwisaver growth fund performance for 19 funds was evaluated by frijns and tourani-rad (2015). they found no evidence of underperformance when regressing excess fund returns on the excess returns of the nzx50 index and the asx200 index. in fact they found outperformance by six of the 19 funds. they also compared the performance of each find using the capm, the fama-french three-factor model and the carhart four-factor model. in all but one case, the alphas were negative, suggesting underperformance once we control for the specific factors in each model. the fama-french three-factor model is a better fit than the capm with improved 2r results. in many cases, the three-factor model better explains returns than the regression based on the two local market indices. however, the addition of the momentum factor in the carhart four-factor model does not increase the goodness of fit in most cases. trainor (2014) evaluated the performance of the top 10 kiwisaver providers, making up 97.8% of assets. a total of 68 funds were assessed and categorised as either equity, balanced or conservative based on percentage mutual fund performance and the impact of superannuation and financial regulation: 1-20 7 exposure to equity. a value-weighted category average return was calculated using total assets, then compared against an engineered value-weighted asset allocation benchmark. as a whole, he found underperformance, with equity funds performing less than desirable. trainor (2014) also performed regression analysis controlling for a number of other variables, again, he found no evidence of outperformance. however, interestingly he did find a significant positive correlation with the asset allocation benchmark, a lagged benchmark factor and the smb factor. 2.4 regulation and mutual funds there is a substantial amount of research on regulatory changes concerning disclosures. bhojraj, cho, and yehuda (2012) studied the effect of significant regulatory changes on the previously noted positive relationship between mutual fund family size and fund performance. specifically, they considered regulatory changes as being designed to reduce selective access to firm information, late trading and market timing opportunities. they found a significant weakening of that positive relationship after the regulatory changes. tang (2013) assessed the impact of business connections from when a fund manager had previously worked as a sell-side analyst, and found that outperformance on those selected stocks decreased after regulation fair disclosure (aka reg fd) in 2000. ge and zheng (2006) and parida and teo (2016) evaluated the impact that more frequent portfolio disclosures had on mutual fund performance, after the may 2004 securities and exchange commision (sec) regulation increased portfolio holdings disclosures frequency from semi-annual to quarterly. they found that mutual fund performance was adversely affected by the increase in disclosure frequency. also examining that sec change was a comprehensive and detailed analysis conducted by agarwal, mullally, tang, and yang (2015) on stock liquidity and performance impact. they found that, in general, the new disclosure requirements deteriorated the performance of funds, but liquidity improved. others, such as verbeek and wang (2013), frank, poterba, shackelford, and shoven (2004), chen, gallagher, and lee (2017), brown and schwarz (2011), schwarz and potter (2016), shi (2017), and aragon, hertzel, and shi (2013) examined potential detrimental impact on funds’ return from mimicking or front-running. the evidence was mixed, but there seemed to be some negative return impact from more frequent disclosure. there are many us-based studies relating to disclosure, but few specifically on australia or new zealand. using privately collected data, (chen et al., 2017) examined copycat (mimicking) behaviour in australia, finding potential for differential performance, depending on whether a top or a bottom performing fund was being considered. 8 the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 new zealand did not have disclosure regulation for mutual funds until 2013, and australia still does not. brown and gregory-allen (2012), later updated as gregory-allen, ozer balli, and thompson (2018), evaluated the potential effects of mandated holdings disclosure for australian and new zealand mutual funds. this was done by utilising data from funds that were voluntarily disclosed and comparing them to those that did not. they found that voluntary disclosures did not harm fund returns and concluded that mandatory disclosures benefited investors on the whole. the current study differed from others in that we examine the impact of a new pension fund scheme, and then the new regulation, of the overall mutual fund industry in new zealand. 3. data we drew monthly returns (net of fees) data from the morningstar database for all8 new zealand open ended, active, equity mutual funds during the period of september 1999 to august 2015, including 30 funds which terminated during that period. using morningstar’s “global category”, funds were classified according to each fund’s investment focus: a) new zealand – primarily new zealand stocks b) australian – primarily australian stocks c) australasian9 – both new zealand and australian stocks d) global – multinational in each case, this was the region that the fund reported as their focus of investment. we examined all equity funds (including “dead” funds), only excluding those with fewer than 12 monthly returns. in our final sample there were 109 funds, of which 21 were new zealand, eight were australian, 32 were australasian (funds investing in both australia and new zealand), and 48 global. we did not include kiwisaver funds, as we wanted to examine the impact of kiwisaver scheme’s introduction on the rest of the industry. tables 1 – 3 show how the fund industry has grown over this period, both in dollar terms (table 1) and in the number of funds. for example, for funds investing in new zealand only, there were as few as four prior to kiwisaver, but as many as 16 in the period after the scheme’s introduction. regarding total investment, in kiwisaver’s first full year there was a total of $16.4 billion nzd of invested funds (including funds invested in kiwisaver), and this quadrupled to $65.6 billion nzd in 2015. mutual fund performance and the impact of superannuation and financial regulation: 1-20 9 table 2. descriptive statistics before and after introduction of kiwisaver (ks) total num of funds number in any given month monthly returns min max avg obs mean sd panel a: 16 year period (8 years pre & 8 years post-ks) new zealand equity 21 4 16 11.4 192 0.36 3.44 australian equity 8 2 8 5 192 0.26 4.55 australasian equity 32 5 26 14.7 192 0.54 3.30 global equity 48 13 37 24.8 192 -0.22 3.96 panel b: 8 year period (pre-ks) new zealand equity 4 12 8.5 96 0.59 3.46 australian equity 2 6 3.4 96 0.49 4.07 australasian equity 5 13 7.5 96 0.69 3.44 global equity 13 21 17.6 96 -0.42 4.55 panel c: 8 year period (post-ks) new zealand equity 12 16 14.3 96 0.14 3.42 australian equity 6 8 6.6 96 0.03 5.00 australasian equity 13 26 21.8 96 0.39 3.16 global equity 21 37 31.9 96 -0.03 3.29 table 3. descriptive statistics before and after introduction of kiwisaver periodic disclosure regulation (kspdr) total num of funds number in any given month monthly returns min max avg obs mean sd panel a: 4 year period (2 years pre & 2 years post-kspdr) new zealand equity 21 13 16 14.6 48 0.92 2.01 australian equity 8 6 8 7.0 48 0.38 2.94 australasian equity 32 22 26 24.0 48 0.90 1.97 global equity 48 30 37 32.9 48 0.69 2.25 panel b: 2 year period (pre-kspdr) new zealand equity 13 15 14 24 0.88 2.29 australian equity 7 7 7 24 0.30 3.35 australasian equity 23 25 24.1 24 0.80 2.28 global equity 31 34 32.1 24 0.38 2.79 panel c: 2 year period (post-kspdr) new zealand equity 14 16 15.1 24 0.95 1.73 australian equity 6 8 7.1 24 0.47 0.47 australasian equity 22 26 23.8 24 1.01 1.65 global equity 30 37 33.6 24 1.00 1.55 10 the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 4. method to examine the impact of the introduction of kiwisaver (ks), and the introduction of the kiwisaver periodic disclosure regulation (kspdr), we estimated regressions for the overall new zealand fund industry before and after each of these events. as we had sufficient data available, for ks, we estimated regressions eight years before and after 1 sep 2007 (hence sep 1, 1999 to aug 31, 2007, and sep 1, 2007 to aug 31, 2015). with less available data, for kspdr we used two years before and after 1 july 2013 (hence july 1, 2011 to june 30, 2013 and july 1, 2013 to june 30, 2015). for both the eight and two year periods, there were an equal number of observations before and after our two “events” of 1 sep 2007 and 1 july 2013. in each case, for each category described above a) new zealand equity, b) australian equity, c) australasian equity, d) global equity – we formed an equally weighted portfolio of all funds in that category. we then estimated regressions based on models from capm, famafrench (1993) and carhart (1997): where ri,t is the excess return (of the nz 90-day bank bill rate) on the relevant portfolio of funds, rmrf is the excess return on a global market portfolio, smb, hml and wml are factor proxies for size, book-to-market and momentum (winners-losers) from fama-french (1993) and carhart (1997), respectively. then, are the capm (aka jensen) alpha, fama-french alpha and carhart alpha, respectively. we used these alphas to measure risk-adjusted outperformance. in order to gauge the impact of ks and the kspdr, we also examined the change in this outperformance from before and after our two “events”. specifically, we used the difference-in-means test, which takes into account possible differences in the distributions, before and after. so, for example, the test for difference in the capm (jensen) alpha across the introduction of kiwisaver is: 5 in each case, for each category described above a) new zealand equity, b) australian equity, c) australasian equity, d) global equity – we formed an equally weighted portfolio of all funds in that category. we then estimated regressions based on models from capm, famafrench (1993) and carhart (1997): , , j i t i i t i tr rmrf! " #= + + , , ff i t i i t i t i t i tr brmrf ssmb hhml! "= + + + + , , c i t i i t i t i t i t i tr brmrf ssmb hhml wwml! "= + + + + + where ,i tr is the excess return (of the nz 90-day bank bill rate) on the relevant portfolio of funds, rmrf is the excess return on a global market portfolio, smb, hml and wml are factor proxies for size, book-to-market and momentum (winners-losers) from fama-french (1993) and carhart (1997), respectively10. then, ji! , ff i! , and c i! are the capm (aka jensen) alpha, fama-french alpha and carhart alpha, respectively. we used these alphas to measure risk-adjusted outperformance. in order to gauge the impact of ks and the kspdr, we also examined the change in this outperformance from before and after our two “events”. specifically, we used the difference-in-means test, which takes into account possible differences in the distributions, before and after. so, for example, the test for difference in the capm (jensen) alpha across the introduction of kiwisaver is: 1 2 2 1 2 1 2 1 2 1 2 2 1 2 j j j j j j j j j j se n n t se ! ! ! ! ! ! ! ! " " ! ! # # # # $ % $ % & ' & '= + & ' & ' ( ) ( ) = 5. empirical results in table 1 we show how total dollars invested has increased from the time the kiwisaver was introduced until 2015. in 2008, the first full year of the kiwisaver availability, $nzd 16.4 billion was invested in the combination of open-end funds, kiwisaver funds, and other superannuation funds, of which only $710 million was in the kiwisaver. the kiwisaver investment grew rapidly nearly matching open-end funds in 2014 at just under $24 billion, and then in 2015 surpassing open-end with $32.8 billion to open-end’s $26.4 billion. we also noted that with the introduction of the kiwisaver regulations, the amount invested in the kiwisaver almost doubled from 2013 to 2015. tables 2 and 3 provide descriptive statistics about the fund portfolios, ks, and kspdr. in general, raw returns were lower in the post-ks period, which was not surprising given that this period comprised almost entirely of the global financial crisis. in the post-kspdr period, raw returns were higher, which may be mostly due to increased participation. however, it is important to note that in some periods some portfolios had a very small number of funds. this is especially true for the australia portfolio where, in the early years of the pre-ks period (table 2), there were as few as two funds. this suggests that these results should be taken with a substantial grain of salt. as noted above, raw returns are shown in both tables 2 and 3, and risk adjusted returns are reported in tables 4 and 5. in the period during the introduction of the kiwisaver (table 4), there were not many significant alphas, especially when we consider factors 3 and 4 of the carhart alphas, only australasian funds over the entire period were barely significant (at the 10% level). for kspdr, many more of the alphas were significant, and most improved in the post-kspdr period. for example, the entire group (“all equity”) went from an insignificant 0.27 to a significant 0.62. however, our real interest was in the significance of the changes in these alphas. for that we used the difference-in-means test, the results of which are reported in tables 6 and 7. 10 data obtained from the fama/french forum website http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html. the global market is a value-weighted aggregate global market proxy including 23 developed countries: australia, austria, belgium, canada, denmark, finland, france, germany, greece, hong kong, ireland, italy, japan, netherlands, new zealand, norway, portugal, singapore, spain, switzerland, sweden, united kingdom, united states. 5 in each case, for each category described above a) new zealand equity, b) australian equity, c) australasian equity, d) global equity – we formed an equally weighted portfolio of all funds in that category. we then estimated regressions based on models from capm, famafrench (1993) and carhart (1997): , , j i t i i t i tr rmrf! " #= + + , , ff i t i i t i t i t i tr brmrf ssmb hhml! "= + + + + , , c i t i i t i t i t i t i tr brmrf ssmb hhml wwml! "= + + + + + where ,i tr is the excess return (of the nz 90-day bank bill rate) on the relevant portfolio of funds, rmrf is the excess return on a global market portfolio, smb, hml and wml are factor proxies for size, book-to-market and momentum (winners-losers) from fama-french (1993) and carhart (1997), respectively10. then, ji! , ff i! , and c i! are the capm (aka jensen) alpha, fama-french alpha and carhart alpha, respectively. we used these alphas to measure risk-adjusted outperformance. in order to gauge the impact of ks and the kspdr, we also examined the change in this outperformance from before and after our two “events”. specifically, we used the difference-in-means test, which takes into account possible differences in the distributions, before and after. so, for example, the test for difference in the capm (jensen) alpha across the introduction of kiwisaver is: 1 2 2 1 2 1 2 1 2 1 2 2 1 2 j j j j j j j j j j se n n t se ! ! ! ! ! ! ! ! " " ! ! # # # # $ % $ % & ' & '= + & ' & ' ( ) ( ) = 5. empirical results in table 1 we show how total dollars invested has increased from the time the kiwisaver was introduced until 2015. in 2008, the first full year of the kiwisaver availability, $nzd 16.4 billion was invested in the combination of open-end funds, kiwisaver funds, and other superannuation funds, of which only $710 million was in the kiwisaver. the kiwisaver investment grew rapidly nearly matching open-end funds in 2014 at just under $24 billion, and then in 2015 surpassing open-end with $32.8 billion to open-end’s $26.4 billion. we also noted that with the introduction of the kiwisaver regulations, the amount invested in the kiwisaver almost doubled from 2013 to 2015. tables 2 and 3 provide descriptive statistics about the fund portfolios, ks, and kspdr. in general, raw returns were lower in the post-ks period, which was not surprising given that this period comprised almost entirely of the global financial crisis. in the post-kspdr period, raw returns were higher, which may be mostly due to increased participation. however, it is important to note that in some periods some portfolios had a very small number of funds. this is especially true for the australia portfolio where, in the early years of the pre-ks period (table 2), there were as few as two funds. this suggests that these results should be taken with a substantial grain of salt. as noted above, raw returns are shown in both tables 2 and 3, and risk adjusted returns are reported in tables 4 and 5. in the period during the introduction of the kiwisaver (table 4), there were not many significant alphas, especially when we consider factors 3 and 4 of the carhart alphas, only australasian funds over the entire period were barely significant (at the 10% level). for kspdr, many more of the alphas were significant, and most improved in the post-kspdr period. for example, the entire group (“all equity”) went from an insignificant 0.27 to a significant 0.62. however, our real interest was in the significance of the changes in these alphas. for that we used the difference-in-means test, the results of which are reported in tables 6 and 7. 10 data obtained from the fama/french forum website http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html. the global market is a value-weighted aggregate global market proxy including 23 developed countries: australia, austria, belgium, canada, denmark, finland, france, germany, greece, hong kong, ireland, italy, japan, netherlands, new zealand, norway, portugal, singapore, spain, switzerland, sweden, united kingdom, united states. then, ji , ff i , and c i mutual fund performance and the impact of superannuation and financial regulation: 1-20 11 5. empirical results in table 1 we show how total dollars invested has increased from the time kiwisaver was introduced until 2015. in 2008, the first full year of kiwisaver availability, $nzd 16.4 billion was invested in the combination of open-end funds, kiwisaver funds, and other superannuation funds, of which only $710 million was in kiwisaver. kiwisaver investment grew rapidly nearly matching open-end funds in 2014 at just under $24 billion, and then in 2015 surpassing open-end with $32.8 billion to open-end’s $26.4 billion. we also note that with the introduction of kiwisaver regulation, the amount invested in kiwisaver almost doubled from 2013 to 2015. tables 2 and 3 provide descriptive statistics about the fund portfolios, ks, and kspdr. in general, raw returns were lower in the post-ks period, which is not surprising given that this period was almost entirely during the global financial crisis. in the post-kspdr period, raw returns were higher, which may be mostly due to increased participation. however, it is important to note that in some periods some portfolios had a very small number of funds. this is especially true for the australia portfolio where, in the early years of the pre-ks period (table 2), there were as few as two funds. this suggests that these results should be taken with a substantial grain of salt. as noted above, raw returns are shown in both tables 2 and 3, and risk adjusted returns are reported in tables 4 and 5. in the period around the introduction of kiwisaver (table 4), there are not many significant alphas, especially once we consider 3 and 4 factors. of the carhart alphas, only australasian funds over the entire period are barely significant (at the 10% level). for kspdr, many more of the alphas are significant, and most improved in the post-kspdr period. for example, the entire group (“all equity”) went from an insignificant 0.27 to a significant 0.62. however, our real interest is in the significance of the changes in these alphas. for that we used the difference-in-means test, the results of which are reported in tables 6 and 7. from the eight year period, prior to ks to the eight year period after (see table 6) most of the alphas decreased substantially. however, a note of caution is warranted; as with most analysis of new zealand stocks, our sample size is quite small. for example, as noted earlier there were on average only five funds primarily covering australia, with as few as two in the early years of our sample. there were more funds primarily covering new zealand – on average about 11 – but as few as four in the early years. the global funds do show improvement over these periods, but only for the capm and fama-french risk models. considering all the funds, there are about 56 funds on average and none of the changes in alpha are significant at the 1% level. this suggests that although kiwisaver has had a substantial impact on the number of new zealanders investing and there may have been isolated instances of some traditional funds experiencing detrimental performance. for the overall industry what impact there was can probably be attributed to the gfc. 12 the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 ta bl e 4. p er fo rm an ce a na ly si s im pa ct o f i nt ro du ct io n of k iw is av er (k s) ; f or e ac h pe ri od pa ne l a : 1 6 ye ar p er io d (8 y ea rs p re & 8 y ea rs p os tk s) c a pm f am afr en ch 3 -f ac to r m od el c ar ha rt 4 -f ac to r m od el α r m r f r α r m r f sm b h m l r α r m r f sm b h m l w m l r 2 n ew z ea la nd e qu ity 0. 19 0. 47 5* ** 0. 40 4 0. 15 1 0. 48 1* ** 0. 01 8 0. 08 2 0. 40 8 0. 16 4 0. 47 5* ** 0. 02 7 0. 07 5 -0 .0 17 0. 40 8 (0 .9 84 ) (1 1. 35 1) (0 .7 65 ) (1 1. 36 5) (0 .1 9) (1 .0 84 ) (0 .8 16 ) (1 0. 57 ) (0 .2 75 ) (0 .9 52 ) (0. 35 9) a us tr al ia n e qu ity 0 0. 71 7* ** 0. 52 6 -0 .0 61 0. 71 4* ** 0. 29 3* ** 0. 03 8 0. 54 4 -0 .0 3 0. 70 2* ** 0. 31 4* ** 0. 02 1 -0 .0 39 0. 54 5 (0 .0 02 ) (1 4. 51 4) (0. 26 7) (1 4. 52 9) (2 .6 96 ) (0 .4 28 ) (0. 12 9) (1 3. 44 3) (2 .7 84 ) (0 .2 3) (0. 70 9) a us tr al as ia n e qu ity 0. 35 3* * 0. 50 4* ** 0. 49 6 0. 32 3* 0. 50 4* ** 0. 12 8 0. 02 5 0. 50 3 0. 29 0* 0. 51 7* ** 0. 10 6 0. 04 3 0. 04 2 0. 50 5 (2 .0 78 ) (1 3. 68 5) (1 .8 64 ) (1 3. 56 3) (1 .5 55 ) (0 .3 77 ) (1 .6 43 ) (1 3. 12 2) (1 .2 45 ) (0 .6 24 ) (1 .0 06 ) g lo ba l eq ui ty -0 .4 44 ** 0. 60 6* ** 0. 49 7 -0 .2 93 0. 58 9* ** -0 .1 92 ** -0 .2 75 ** * 0. 53 2 -0 .2 21 0. 56 0* ** -0 .1 44 -0 .3 14 ** * -0 .0 92 * 0. 54 1 (2. 17 8) (1 3. 70 2) (1. 45 4) (1 3. 6) (2. 00 3) (3. 55 5) (1. 08 4) (1 2. 27 9) (1. 46 5) (3. 94 6) (1. 89 2) a ll eq ui ty -0 .0 96 0. 56 9* ** 0. 60 9 -0 .0 39 0. 56 1* ** -0 .0 37 -0 .1 15 ** 0. 61 6 -0 .0 04 0. 54 7* * -0 .0 14 -0 .1 34 ** -0 .0 44 0. 61 9 (0. 62 7) (1 7. 18 8) (0. 25 2) (1 6. 86 6) (0. 49 8) (1. 93 7) (0. 02 7) (1 5. 52 5) (0. 18 0) (2. 17 8) (1. 17 6) pa ne l b : 8 y ea r p er io d (p re -k s) c a pm f am afr en ch 3 -f ac to r m od el c ar ha rt 4 -f ac to r m od el α r m r f r α r m r f sm b h m l r 2 α r m r f sm b h m l w m l r 2 n ew z ea la nd e qu ity 0. 41 6 0. 49 4* ** 0. 31 8 0. 26 3 0. 54 2* ** 0. 04 7 0. 12 7 0. 32 8 0. 27 8 0. 53 6* ** 0. 05 7 0. 12 2 -0 .0 14 0. 32 8 (1 .4 11 ) (6 .6 23 ) (0 .8 07 ) (6 .3 12 ) (0 .3 86 ) (1 .1 45 ) (0 .8 27 ) (5 .8 82 ) (0 .4 31 ) (1 .0 54 ) (0. 19 7) a us tr al ia n e qu ity 0. 27 5 0. 62 9* ** 0. 37 4 0. 08 2 0. 65 7* ** 0. 30 3* * 0. 09 4 0. 40 7 0. 18 4 0. 61 8* ** 0. 37 2* ** 0. 05 5 -0 .0 92 0. 41 7 (0 .8 28 ) (7 .4 96 ) (0 .2 28 ) (6 .9 33 ) (2 .2 63 ) (0 .7 65 ) (0 .5 01 ) (6 .1 9) (2 .5 66 ) (0 .4 38 ) (1. 22 3) a us tr al as ia n e qu ity 0. 50 4* 0. 52 8* ** 0. 36 8 0. 39 5 0. 54 4* ** 0. 17 1 0. 05 3 0. 38 3 0. 33 9 0. 56 5* ** 0. 13 4 0. 07 4 0. 05 0. 38 7 (1 .7 88 ) (7 .3 98 ) (1 .2 71 ) (6 .6 49 ) (1 .4 85 ) (0 .5 05 ) (1 .0 61 ) (6 .5 28 ) (1 .0 64 ) (0 .6 79 ) (0 .7 66 ) (c on tin ue d) mutual fund performance and the impact of superannuation and financial regulation: 1-20 13 c a pm f am afr en ch 3 -f ac to r m od el c ar ha rt 4 -f ac to r m od el α r m r f r 2 α r m r f sm b h m l r 2 α r m r f sm b h m l w m l r 2 g lo ba l eq ui ty -0 .7 02 ** 0. 83 3* ** 0. 52 5 -0 .3 83 0. 74 7* ** -0 .2 05 -0 .2 36 * 0. 54 9 -0 .2 52 0. 69 7* ** -0 .1 17 -0 .2 85 ** -0 .1 18 0. 56 2 (2. 17 5) (1 0. 19 6) (1. 09 4) (8 .1 01 ) (1. 57 7) (1. 97 7) (0. 70 8) (7 .2 2) (0. 83 4) (2. 33 2) (1. 61 0) a ll eq ui ty -0 .1 39 0. 68 4* ** 0. 58 1 -0 .0 51 0. 65 6* ** -0 .0 27 -0 .0 73 0. 58 4 0. 01 7 0. 63 1* ** 0. 01 8 -0 .0 98 -0 .0 61 0. 59 (0. 58 6) (1 1. 42 5) (0. 19 4) (9 .4 94 ) (0. 28 0) (0. 81 4) (0 .0 61 ) (8 .6 47 ) (0 .1 73 ) (1. 06 4) (1. 10 1) pa ne l c : 8 y ea r p er io d (p os tk s) c a pm f am afr en ch 3 -f ac to r m od el c ar ha rt 4 -f ac to r m od el α r m r f r 2 α r m r f sm b h m l r 2 α r m r f sm b h m l w m l r 2 n ew z ea la nd e qu ity -0 .0 39 0. 46 4* ** 0. 49 7 -0 .0 35 0. 46 3* ** -0 .0 58 0. 02 2 0. 49 8 -0 .0 2 0. 45 4* ** -0 .0 68 -0 .0 11 -0 .0 44 0. 5 (0. 15 5) (9 .6 4) (0. 13 8) (9 .0 45 ) (0. 34 8) (0 .1 46 ) (0. 07 8) (8 .4 75 ) (0. 40 2) (0. 07 1) (0. 59 4) a us tr al ia n e qu ity -0 .2 65 0. 76 8* ** 0. 63 7 -0 .3 3 0. 80 5* ** 0. 25 3 -0 .3 86 ** 0. 66 2 -0 .3 21 0. 80 0* ** 0. 24 8 -0 .4 04 ** -0 .0 24 0. 66 3 (0. 85 6) (1 2. 83 6) (1. 08 5) (1 3. 12 4) (1 .2 6) (2. 16 4) (1. 04 8) (1 2. 44 5) (1 .2 22 ) (2. 11 2) (0. 27 2) a us tr al as ia n e qu ity 0. 19 9 0. 49 1* ** 0. 65 4 0. 19 9 0. 49 1* ** 0. 00 8 0. 00 2 0. 65 4 0. 19 4 0. 49 4* ** 0. 01 1 0. 01 3 0. 01 5 0. 65 4 (1 .0 41 ) (1 3. 31 9) (1 .0 27 ) (1 2. 51 9) (0 .0 59 ) (0 .0 18 ) (0 .9 91 ) (1 2. 01 8) (0 .0 84 ) (0 .1 08 ) (0 .2 61 ) g lo ba l eq ui ty -0 .2 13 0. 47 5* ** 0. 56 3 -0 .2 14 0. 47 6* ** -0 .0 94 -0 .0 04 0. 56 5 -0 .2 12 0. 47 5* ** -0 .0 95 -0 .0 06 -0 .0 04 0. 56 5 (0. 95 4) (1 1. 01 4) (0. 94 4) (1 0. 41 6) (0. 63 1) (0. 02 8) (0. 92 9) (9 .9 19 ) (0. 63 0) (0. 04 5) (0. 05 6) a ll eq ui ty -0 .0 66 0. 50 3* ** 0. 67 4 -0 .0 73 0. 50 7* ** -0 .0 28 -0 .0 38 0. 67 5 -0 .0 7 0. 50 6* ** -0 .0 3 -0 .0 43 -0 .0 07 0. 67 5 (0. 35 4) (1 3. 95 2) (0. 38 2) (1 3. 24 5) (0. 22 7) (0. 34 0) (0. 36 7) (1 2. 6) (0. 23 6) (0. 35 9) (0. 12 0) tst at is tic s a re r ep or te d be lo w th e co ef fic ie nt s in p ar en th es es . c oe ffi ci en ts m ar ke d w ith * ** , * *, a nd * a re s ig ni fic an t at th e 1% , 5 % , a nd 1 0% le ve l, re sp ec tiv el y 14 the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 ta bl e 5. p er fo rm an ce a na ly si s im pa ct o f i nt ro du ct io n of k iw is av er p er io di c d is cl os ur e r eg ul at io n (k sp d r ); fo r e ac h pe ri od pa ne l a : 4 y ea r p er io d (2 y ea rs p re & 2 y ea rs p os tk sp d r ) c a pm f am afr en ch 3 -f ac to r m od el c ar ha rt 4 -f ac to r m od el α r m r f r 2 α r m r f sm b h m l r 2 α r m r f sm b h m l w m l r 2 n ew z ea la nd e qu ity 0. 76 2* ** 0. 19 4* * 0. 13 1 0. 77 6* ** 0. 18 5* * 0. 01 8 0. 1 0. 13 6 0. 71 4* * 0. 19 6* ** 0. 02 0. 13 9 0. 07 8 0. 14 3 (2 .7 29 ) (2 .6 31 ) (2 .7 05 ) (2 .4 ) (0 .0 83 ) (0 .5 01 ) (2 .3 32 ) (2 .4 58 ) (0 .0 91 ) (0 .6 59 ) (0 .6 1) a us tr al ia n e qu ity 0. 02 2 0. 45 0* ** 0. 32 8 -0 .0 18 0. 46 0* ** -0 .1 97 -0 .1 44 0. 34 0. 06 5 0. 44 6* ** -0 .2 -0 .1 97 -0 .1 05 0. 34 6 (0 .0 6) (4 .7 43 ) (0. 04 9) (4 .6 54 ) (0. 72 5) (0. 56 6) (0 .1 67 ) (4 .3 79 ) (0. 72 9) (0. 73 1) (0. 64 3) a us tr al as ia n e qu ity 0. 65 4* ** 0. 31 0* ** 0. 34 7 0. 66 1* ** 0. 30 2* ** -0 .0 2 0. 08 5 0. 35 1 0. 61 4* * 0. 31 0* ** -0 .0 19 0. 11 4 0. 05 9 0. 35 5 (2 .7 53 ) (4 .9 41 ) (2 .7 12 ) (4 .5 98 ) (0. 11 3) (0 .5 ) (2 .3 59 ) (4 .5 73 ) (0. 10 4) (0 .6 38 ) (0 .5 45 ) g lo ba l eq ui ty 0. 38 0* 0. 38 5* ** 0. 41 0. 35 5* 0. 37 1* ** -0 .3 29 ** 0. 10 3 0. 45 4 0. 28 3 0. 38 3* ** -0 .3 27 ** 0. 14 9 0. 09 1 0. 46 2 (1 .4 72 ) (5 .6 54 ) (1 .3 87 ) (5 .3 89 ) (1. 73 4) (0 .5 82 ) (1 .0 38 ) (5 .4 14 ) (1. 71 5) (0 .7 96 ) (0 .8 01 ) a lle qu ity 0. 50 1* ** 0. 33 3* ** 0. 46 0. 49 1* * 0. 32 4* ** -0 .1 65 0. 07 2 0. 47 8 0. 44 2* * 0. 33 2* ** -0 .1 64 0. 10 3 0. 06 1 0. 48 4 (2 .4 86 ) (6 .2 59 ) (2 .4 05 ) (5 .9 ) (1. 09 0) (0 .5 09 ) (2 .0 32 ) (5 .8 72 ) (1. 07 3) (0 .6 86 ) (0 .6 73 ) pa ne l b : 2 y ea r p er io d (p re -k sp d r ) c a pm f am afr en ch 3 -f ac to r m od el c ar ha rt 4 -f ac to r m od el α r m r f r 2 α r m r f sm b h m l r 2 α r m r f sm b h m l w m l r 2 n ew z ea la nd e qu ity 0. 75 4* * 0. 21 9* ** 0. 19 0. 78 1* 0. 20 3* * 0. 16 1 0. 20 5 0. 21 3 0. 69 6* 0. 22 6* * 0. 24 6 0. 24 5 0. 11 9 0. 22 9 (1 .7 4) (2 .2 7) (1 .6 61 ) (1 .9 92 ) (0 .3 77 ) (0 .6 85 ) (1 .4 04 ) (2 .0 61 ) (0 .5 42 ) (0 .7 9) (0 .6 34 ) a us tr al ia n e qu ity -0 .0 24 0. 55 2* ** 0. 56 4 -0 .1 42 0. 55 3* ** -0 .3 56 0. 01 8 0. 57 7 -0 .0 68 0. 53 3* ** -0 .4 31 -0 .0 17 -0 .1 05 0. 58 3 (0. 05 2) (5 .3 35 ) (0. 28 3) (5 .0 47 ) (0. 77 5) (0 .0 56 ) (0. 12 7) (4 .5 11 ) (0. 88 1) (0. 05 2) (0. 51 9) a us tr al as ia n e qu ity 0. 59 5* * 0. 34 5* ** 0. 47 6 0. 58 9* 0. 33 6* ** 0. 03 0. 12 8 0. 48 4 0. 52 1 0. 35 4* ** 0. 09 8 0. 16 0. 09 5 0. 49 4 (1 .7 16 ) (4 .4 74 ) (1 .5 54 ) (4 .0 81 ) (0 .0 86 ) (0 .5 3) (1 .3 04 ) (4 .0 02 ) (0 .2 66 ) (0 .6 38 ) (0 .6 25 ) (c on tin ue d) mutual fund performance and the impact of superannuation and financial regulation: 1-20 15 c a pm f am afr en ch 3 -f ac to r m od el c ar ha rt 4 -f ac to r m od el α r m r f r 2 α r m r f sm b h m l r 2 α r m r f sm b h m l w m l r 2 g lo ba l eq ui ty 0. 12 9 0. 42 5* ** 0. 48 1 -0 .0 37 0. 38 4* ** -0 .2 92 0. 57 3* * 0. 59 4 -0 .0 23 0. 38 1* ** -0 .3 07 0. 56 6* * -0 .0 2 0. 59 4 (0 .3 06 ) (4 .5 15 ) (0. 09 0) (4 .3 04 ) (0. 78 0) (2 .1 84 ) (0. 05 2) (3 .9 25 ) (0. 76 3) (2 .0 59 ) (0. 12 3) a ll eq ui ty 0. 37 8 0. 37 4* ** 0. 56 2 0. 3 0. 35 2* ** -0 .1 17 0. 31 9* 0. 61 1 0. 27 7 0. 35 8* ** -0 .0 94 0. 33 0* 0. 03 2 0. 61 3 (1 .1 93 ) (5 .3 15 ) (0 .9 14 ) (4 .9 38 ) (0. 39 3) (1 .5 23 ) (0 .7 93 ) (4 .6 35 ) (0. 29 4) (1 .5 06 ) (0 .2 44 ) pa ne l c : 2 y ea r p er io d (p os tk sp d r ) c a pm f am afr en ch 3 -f ac to r m od el c ar ha rt 4 -f ac to r m od el α r m r f r 2 α r m r f sm b h m l r 2 α r m r f sm b h m l w m l r 2 n ew z ea la nd eq ui ty 0. 82 3* * 0. 12 8 0. 04 3 0. 82 8* * 0. 12 4 -0 .0 75 -0 .0 17 0. 04 7 0. 79 6* * 0. 11 6 -0 .1 1 0. 06 2 0. 09 7 0. 05 6 (2 .1 84 ) (0 .9 89 ) (2 .0 01 ) (0 .8 7) (0. 30 5) (0. 05 8) (1 .8 53 ) (0 .7 91 ) (0. 41 4) (0 .1 76 ) (0 .4 14 ) a us tr al ia n e qu ity 0. 28 6 0. 17 8 0. 03 9 0. 20 1 0. 20 4 -0 .1 97 -0 .2 82 0. 07 2 0. 23 8 0. 21 3 -0 .1 57 -0 .3 74 -0 .1 12 0. 07 7 (0 .5 17 ) (0 .9 42 ) (0 .3 36 ) (0 .9 89 ) (0. 55 3) (0. 66 5) (0 .3 83 ) (1 .0 01 ) (0. 40 8) (0. 72 7) (0. 33 3) a us tr al as ia n e qu ity 0. 79 1* * 0. 21 3* * 0. 13 1 0. 83 5* * 0. 19 5* -0 .0 83 0. 07 7 0. 14 1 0. 80 3* * 0. 18 7* -0 .1 18 0. 15 8 0. 09 8 0. 15 1 (2 .3 12 ) (1 .8 19 ) (2 .2 34 ) (1 .5 12 ) (0. 37 2) (0 .2 9) (2 .0 7) (1 .4 11 ) (0. 49 3) (0 .4 91 ) (0 .4 65 ) g lo ba l eq ui ty 0. 72 9* * 0. 26 7* ** 0. 23 3 0. 63 0* * 0. 29 2* ** -0 .4 51 ** * -0 .4 08 ** 0. 53 2 0. 51 2* * 0. 26 3* ** -0 .5 79 ** * -0 .1 14 0. 35 7* ** 0. 67 8 (2 .4 16 ) (2 .5 87 ) (2 .4 31 ) (3 .2 58 ) (2. 91 2) (2. 22 4) (2 .2 82 ) (3 .4 2) (4. 17 1) (0. 61 4) (2 .9 29 ) a ll eq ui ty 0. 71 9* ** 0. 21 8* * 0. 22 0 0. 68 3* ** 0. 22 4* ** -0 .2 54 * -0 .1 83 0. 33 8 0. 62 0* * 0. 20 9* * -0 .3 22 ** -0 .0 28 0. 18 9* 0. 39 6 (2 .8 22 ) (2 .4 94 ) (2 .6 44 ) (2 .5 17 ) (1. 64 8) (1. 00 3) (2 .4 11 ) (2 .3 75 ) (2. 02 2) (0. 13 1) (1 .3 51 ) tst at is tic s a re r ep or te d be lo w th e co ef fic ie nt s in p ar en th es es . c oe ffi ci en ts m ar ke d w ith * ** , * *, a nd * a re s ig ni fic an t at th e 1% , 5 % , a nd 1 0% le ve l, re sp ec tiv el y 16 the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 for the two years before and after the kspdr, there is again the problem of small samples for some of the categories. however, most of the categories, notably the global funds, do have an improvement in performance. table 6. impact of introduction of kiwisaver (ks) 8 years before and after 1 sept, 2007 avg num of funds capm alpha famafrench alpha carhart alpha new zealand equity 11.4 -0.45 (-.29) -0.3*** (-3.23) -0.3*** (-3.16) australian equity 5.1 -0.54*** (-5.32) -0.41*** (-3.91) -0.51*** (-4.72) australasian equity 14.7 -0.31*** (-4.01) -0.2** (-2.39) -0.14* (-1.73) global equity 24.8 0.49*** 0.17* 0.04 (5.57) (1.82) (.42) all 55.9 0.07 -0.02 -0.09 (1.07) (-.3) (-1.17) table 7. impact of introduction of kiwisaver periodic disclosure regulation (kspdr) 2 years before and after 1 july, 2013 avg num of funds capm alpha famafrench alpha carhart alpha new zealand equity 14.5 0.07 0.05 0.1 (.54) (.34) (.68) australian equity 7.1 0.31* (1.92) 0.34** (1.96) 0.31* (1.67) australasian equity 23.9 0.2* (1.8) 0.25** (2.07) 0.28** (2.26) global equity 32.9 0.6*** (5.17) 0.67*** (6.14) 0.53*** (4.86) all 78.3 0.34 0.38 0.34 (.55) (.67) (.66) t-statistics reported below coefficients, in parentheses. coefficients marked with ***, **, and * are significant at the 1%, 5%, and 10% level, respectively mutual fund performance and the impact of superannuation and financial regulation: 1-20 17 6. conclusion this study examines the impact on the mutual fund industry of introducing a major superannuation scheme into a small economy, and the impact of increased regulation on the scheme. kiwisaver, new zealand’s new superannuation scheme, has dramatically increased the investment participation rate of the labour force, resulting in funds invested in the scheme overtaking standard mutual funds by 2015, with a four-fold increase in total overall fund investment. with regard to mutual fund performance, a comparison of the risk-adjusted returns in the eight year period after kiwisaver to the prior eight years shows a significant deterioration in some categories of the funds. however, especially due to small sample sizes, our overall evidence for a significant impact is weak. given the dramatic improvement in investment participation by new zealanders, we can readily conclude that overall, the introduction of kiwisaver funds has been positive. we draw a similar conclusion for the new law, the kiwisaver periodic disclosure regulation. in comparing mutual fund performance in the two year period after the regulation change to the prior two years, we found some significant improvement and no evidence for a negative impact. this shows that the increased regulation have not harmed fund performance and may well have contributed to the rapid inflow of invested dollars into kiwisaver funds. obviously, these results for new zealand are not conclusive for other small economies, but they are suggestive. while it is clear that kiwisaver has substantially improved participation in retirement savings throughout the workforce, and that kiwisaver periodic disclosure regulation improved the kiwisaver’s overall safety, this study has shown that both have not had a broad, deleterious effect on the existing mutual fund environment. end notes 1 there is an initial $1000 contribution from the government, then a (currently) minimum combined contribution rate of 6%. 2 kiwisaver statistics available from http://www. kiwisaver. govt. nz/statistics/ 3 population statistics as of june 30, 2015 available from statistics new zealand http://www.stats.govt.nz/browse_for_stats/population/estimates_and_ projections/nationalpopulationestimates_hotpat30jun15.aspx 4 data from bloomberg 5 more information about the fma and its role https://fma.govt.nz/fmas-role/ what-we-do/our-role/ 6 for legislative information http://www.legislation.govt.nz/regulation/public/ 2013/0047/latest/whole.html 18 the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 7 for example, what is impact of regulatory change on fund managers performance?, does increased regulation lead to elimination of small funds or industry consolidation, or increased fund administrative costs?, what is effect of fund size, flows, or characteristics? 8 to avoid survivor bias we include all funds. 9 although we use the term “australasian” it includes only australia and new zealand; we use this term (rather than, say ‘combined’) to avoid confusion with a combination of the first two sets. references agarwal, v., mullally, k. a., tang, y., & yang, b. (2015). mandatory portfolio disclosure, stock liquidity, and mutual fund performance. the journal of finance, 70(6), 2733-2776. aragon, g. o., hertzel, m., & shi, z. (2013). why do hedge funds avoid disclosure? evidence from confidential 13f filings. journal of financial and quantitative analysis, 48(5), 1499-1518. bauer, r., otten, r., & rad, a. t. (2006). new zealand mutual funds: measuring performance and persistence in performance. accounting & finance, 46(3), 347-363. bhojraj, s., cho, j. y., & yehuda, n. (2012). mutual fund family size and mutual fund performance: the role of regulatory changes. journal of accounting research, 50(3), 647-684. brown, k., & gregory-allen, r. (2012). the potential effects of mandatory portfolio holdings disclosure in australia and new zealand. available at ssrn 2131992. doi:https://papers.ssrn.com/sol3/papers.cfm?abstract_ id=2131992 brown, s., & goetzmann, w. n. (1995). performance persistence. the journal of finance, 50(2), 679-698. brown, s., & schwarz, c. (2011). the impact of mandatory hedge fund portfolio disclosure. available at ssrn 1683628. carhart, m. m. (1997). on persistence in mutual fund performance. the journal of finance, 52(1), 57-82. chen, j., hong, h., huang, m., & kubik, j. d. (2004). does fund size erode mutual fund performance? the role of liquidity and organization. american economic review, 94(5), 1276-1302. chen, z., gallagher, d. r., & lee, a. d. (2017). testing the effect of portfolio holdings disclosure in an environment absent of mandatory disclosure. accounting & finance, 57(s1), 101-116. elton, e. j., gruber, m. j., & blake, c. r. (1996). the persistence of risk-adjusted mutual fund performance. journal of business, 133-157. mutual fund performance and the impact of superannuation and financial regulation: 1-20 19 elton, e. j., gruber, m. j., das, s., & hlavka, m. (1993). efficiency with costly information: a reinterpretation of evidence from managed portfolios. the review of financial studies, 6(1), 1-22. fama, e. f., & french, k. r. (1993). common risk factors in the returns on stocks and bonds. journal of financial economics, 33(1), 3-56. fama, e. f., & french, k. r. (2010). luck versus skill in the cross section of mutual fund returns. the journal of finance, 65(5), 1915-1947. ferreira, m. a., keswani, a., miguel, a. f., & ramos, s. b. (2012). the flowperformance relationship around the world. journal of banking & finance, 36(6), 1759-1780. frank, m. m., poterba, j. m., shackelford, d. a., & shoven, j. b. (2004). copycat funds: information disclosure regulation and the returns to active management in the mutual fund industry. the journal of law and economics, 47(2), 515-541. frijns, b., & tourani-rad, a. (2015). on the performance of kiwisaver funds. pacific accounting review, 27(3), 266-281. ge, w., & zheng, l. (2006). the frequency of mutual fund portfolio disclosure. gil–bazo, j., & ruiz–verdú, p. (2009). the relation between price and performance in the mutual fund industry. the journal of finance, 64(5), 2153-2183. goetzmann, w. n., & ibbotson, r. g. (1994). do winners repeat? journal of portfolio management, 20(2), 9-18. gregory-allen, r., ozer balli, h., & thompson, k. (2018). the impact of portfolio holdings disclosure on fund returns working paper. grinblatt, m., & titman, s. (1992). the persistence of mutual fund performance. the journal of finance, 47(5), 1977-1984. jegadeesh, n., & titman, s. (1993). returns to buying winners and selling losers: implications for stock market efficiency. the journal of finance, 48(1), 65-91. jensen, m. c. (1968). the performance of mutual funds in the period 1945–1964. the journal of finance, 23(2), 389-416. jensen, m. c. (1969). risk, the pricing of capital assets, and the evaluation of investment portfolios. the journal of business, 42(2), 167-247. kosowski, r., timmermann, a., wermers, r., & white, h. (2006). can mutual fund “stars” really pick stocks? new evidence from a bootstrap analysis. the journal of finance, 61(6), 2551-2595. kritzer, b. e. (2007). kiwisaver: new zealand's new subsidized retirement savings plans. soc. sec. bull., 67, 113. lintner, j. (1965). the valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets. review of economics and statistics, 47, 13-37. mossin, j. (1966). equilibrium in a capital asset market. econometrica: journal of the econometric society, 768-783. 20 the international journal of banking and finance, vol. 13. number 2, 2017: 1-20 otten, r., & thevissen, k. (2011). does industry size matter? revisiting european mutual fund performance. available at ssrn 1741197. parida, s., & teo, t. (2016). the impact of more frequent portfolio disclosure on mutual fund performance. journal of banking & finance. petajisto, a. (2013). active share and mutual fund performance. financial analysts journal, 69(4), 73-93. pollet, j. m., & wilson, m. (2008). how does size affect mutual fund behavior? the journal of finance, 63(6), 2941-2969. schwarz, c. g., & potter, m. e. (2016). revisiting mutual fund portfolio disclosure. the review of financial studies, 29(12), 3519-3544. sharpe, w. f. (1964). capital asset prices: a theory of market equilibrium under conditions of risk. the journal of finance, 19(3), 425-442. shi, z. (2017). the impact of portfolio disclosure on hedge fund performance. journal of financial economics, 126(1), 36-53. doi:10.1016/j. jfineco.2017.06.001 tang, y. (2013). business connections and informed trading of mutual fund managers. available at ssrn 1741197. thomas, c., & matthews, c. (2014). kiwisaver member behaviour: a quantitative analysis. pacific accounting review, 26(3), 248-273. trainor, w. j. (2014). assessing kiwisaver fund providers. new zealand journal of applied business research, 12(2), 1. treynor, j. l. (1961a). market value, time and risk. unpublished manuscript. treynor, j. l. (1961b). toward a theory of market value of risky assets. unpublished manuscript, 6, 831-868. verbeek, m., & wang, y. (2013). better than the original? the relative success of copycat funds. journal of banking & finance, 37(9), 3454-3471. vos, e., brown, p., & christie, s. (1995). a test of persistence in the performance of new zealand and australian equity mutual funds. accounting research journal, 8(2), 19-35. wermers, r. (1997). momentum investment strategies of mutual funds, performance persistence, and survivorship bias. documento de trabalho, university of colorado. zhang, a. c. (2014). financial advice and asset allocation of individual investors. pacific accounting review, 26(3), 226-247. efficiency of domestic and foreign banks in thailand since the asian financial crisis the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 101 efficiency of domestic and foreign banks in thailand since the asian financial crisis tosporn chotigeat nicholls state university, united states abstract during the 1990’s, two simultaneous phenomena had converged, greatly changing the financial services industry in thailand. at the national level, thai banks had to restructure in response to the financial reforms implemented as a result of the financial crisis of 1997. at the global level, large multinational banks were taking advantage of worldwide, financial deregulation and rapid technological advances by offering a full range of financial products and services in order to fiercely compete, both domestically and globally. using quarterly financial time-series data of domestic and foreign banks in thailand from 1997 to 2003, this paper seeks to analyze the cause of their efficiency. the findings indicate that both the efficiency ratio and loan loss provisions influenced the negative performance of domestic banks, while only loan loss provisions had negatively influenced the performance of foreign banks. keywords: efficiency, domestic/foreign banks, loan loss provision, roe, financial crisis jel classification: c32, e44, f23, g15 1. introduction since the asian financial crisis ten years ago, two simultaneous phenomena had converged to greatly change the financial services industry landscape, both at the regional and global level. in the wake of the crisis, those afflicted countries had adopted (whether voluntarily and/or under pressure from the imf and other international agencies) new policies emphasizing high levels of transparency and the liberalization of capital markets. at the same time, major players in the global financial services sector have continued leveraging advances in telecommunications and information technology, while increasing the amount of cross-border mergers and acquisitions. this has led to an intensified increase of competition, both in the crisis countries and in the rest of the world. at this point, the question then arises as to whether or not these domestic banks and foreign banks are equally competitive; and if not, what are the salient factors that have contributed to the differentiation in their respective levels of performance. ijbf 102 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 since 2000, with restructuring near completion of the thai financial sector, the economy had experienced a successful recovery with favorable key economic indicators, both internally and externally, displaying a comfortable level of stability. (appendix a). thus, domestic and foreign banking operations have been selected for this study, for they offer the best scope regarding the health and stability of the financial sector in a post-crisis thailand. this paper attempts to empirically investigate and compare each group’s performance by using quarterly data from 1997 to 2003. section 2 will provide a background description of thai commercial banks and their evolution since the financial crisis, as well as an analysis of the global integration of the thai markets with an increased international competition in the financial services industry. section 3 will review the literature related to banks’ performance in various countries, while the model and time-series data being used to test the efficiency of banks will be provided in section 4 with great detail. lastly, the empirical results and conclusions will be provided in the remaining two sections. 2. background 2.1 reorganization of the thai financial sector after the crisis immediately after the financial crisis of 1997, the thai government had closed down 64 of its 91 domestic financial companies, leaving only 27 open with 13 local banks in operation. those remaining banks and financial companies were allowed to create 300 billion baht for new capitalization by issuing preferred shares and new debt debentures in exchange for 10-year government bonds. financial institutions were also required to increase their reserve funds for the existing large non-performing loans (npls) (see bangkok post, 1999]. although the npls were reduced to 12.65 percent by the end of 2003, they still continue to burden the financial system in 2007. in addition, the financial sector restructuring authority (fra) and asset management corporation (amc) were created to handle the assets of now defunct financial companies. the fra seized the assets of defunct financial companies and banks, then selling them at an average bid of 37 percent of book value. the process was very slow with many assets lacking the full, proper, and official documentation. by the end of 2003, there were only 13 domestic commercial banks and 18 foreign banks operating in thailand. 2.2 global synergy & international competition of the financial services industry since the inception of a global integration, international competition in the financial services industry around the world has greatly intensified. during the last 20 years, only a few firms have swiftly advanced their cause like that of the banking industry. the competitive sector in which they had competed took advantage of the financial deregulations and rapid technological advances worldwide in the 1990s. with deregulation removing the regulatory barriers to entry and technological advances reducing the cost of managing affiliates in distant markets, banks had undertaken a the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 103 strategy of growth by merger & acquisition to accomplish two objectives: defend a position within their domestic stronghold by merging with other domestic competitors, thereby fending off potential foreign competitors establish presence in foreign markets by acquiring interest in foreign institutions1 while the extent of the industry’s outreach into the global arena through international acquisitions is debatable, 2what is clear is the full effect of deregulation and technological advances (i.e., cross-border, mega-size mergers & acquisitions recently undertaken by multinational banks). the large multinational banks now offer a full range of financial products and services, fiercely competing in both the domestic and global markets. 3. bank performance literature on other countries many specialists have a mixed point of view on the effects of deregulation and technological advances. for us banks, berger & mester (2003) found that cost productivity worsened even though profit margins substantially improved for banks engaging in mergers. focarrelli and pozzolo (2001) and williams (2003) had evaluated the profitability of international banks in the oecd, as well as that of domestic and foreign banks in australia, respectively. both studies had argued that efficient banks expand abroad seeking higher profits, but that profits are negatively correlated to their market share. using italian data, focarelli, panetta, and salleo (2002) had shown that mergers increase income from expanded services and product lines, but the increase is offset by higher costs. however, the return on equity improves because of a decrease in the capital base of the acquiring bank. similar studies of commercial banks in other countries have been undertaken, i.e., studies of japanese banks by drake & hall (2003); indian banks by kumbhakar and sarkar (2003); portuguese banks by mendes and reblo (1999); turkish banks by isik and hassan (2003); and spanish banks by grifell-tatje and lovell (1996). while the analytical designs and econometric variables used in these studies had considerably varied, one common theme had emerged: following the deregulation or technological change, the measures of profitability and/or productivity had generally declined for banks in these diverse countries. 4. measure of global bank performance and determinants properly gauging all aspects of profitability is a generally accepted way to measure the performance of businesses, specifically commercial banks. however, banks vary considerably in terms of size, types of products offered, and types of assets 1 see larosiere (2000) for an excellent discussion on banking consolidation in europe and why most european banks favor merger and acquisition for their growth. 2 examining how more than 2,000 affiliates of large mnc’s patronize banks in 20 european nations, berger, dai, ongena and smith (2002) argue that the banking industry may never become fully globalized. 104 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 owned, among many other factors. therefore, some standard form of measuring profitability is needed. two potential measures of bank performance are return on assets (roa) and return on equity (roe). due to the nature of bank assets (largely loans), roe seems to be the more appropriate measure, at least from a purely logical perspective. for that very reason, roe is used as the dependent variable for this analysis. finally, regression analysis is the generally accepted tool for measuring the relationship among variables. similar studies of international banks have used five variables (total assets, efficiency ratio, tier1 capital ratio, loan loss provisions, and non-interest income) to explain bank performance, as noted by the likes of allen and raj (1996), kim and singer (1997), and brimmer (1998). for purposes of comparing thai domestic banks and foreign banks in thailand, this analysis follows similar lines. efficiency can be defined as the ratio of a given bank’s non-interest expense to total revenue. it is hypothesized that the coefficient of this variable will be negative because higher ratios imply the reduced efficiency of bank operations. second, a measure of risk was ascertained by calculating the ratio of tier1 capital to risky assets. the tier1 capital ratio is defined as core capital (essentially the book value of a bank’s equity) divided by risk-adjusted assets (basically total assets, weighted more heavily for more risky assets). the coefficient of the tier1 capital ratio is hypothesized to be negative. thirdly, loan loss provisions are used to represent bad debts. here a negative coefficient for the variable is anticipated since increases in the loan loss provisions suggest that the quality of the bank’s loan portfolio is reduced. the non-interest income variable is expected to have a positive impact on a bank’s performance. finally, the total asset value is considered as the other potential explanatory variable, but the coefficient is not expected to produce statistically significant results. an empirical study done by grosse and gart (2001) reports that there are considerable variations in the key factors that influence the performance of multinational banks from different countries. for american banks, efficiency is the only significant variable, whereas, for uk banks, total assets, efficiency, and loan loss provisions are significant. for german banks, the total assets and noninterest income are significant, whereas the efficiency and tier1 capital ratio are significant for japanese banks. surprisingly, none of these variables were found to be a significant determinant for the performance of swiss banks. from 1993 to 1999, chotigeat et al. (2004) had empirically studied the performance of the following three major french banks: societe general, bnp paribus, and credit lyonnais. it was found that the factors determining the banks’ profit are total assets, efficiency, tier1 capital ratio, and loan loss provisions. 5. data and model for this study, all commercial banks that operate in thailand were selected (i.e., 13 thai domestic banks and 18 foreign banks). the data used originates from the quarterly reports of the bank of thailand from 1997 to 2003, a period in which: thailand reconstructed its financial system after the financial crisis of july 1997, the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 105 which included a policy of transparency and capital market liberalization there were significant cross-border investments taking place in the world’s banking industry. the six, key variables calculated from the raw time-series data to be used in this study are: return on assets (roa), return on equity (roe), total assets, efficiency ratio (non-interest expense to total revenue), the tier1 capital (equity capital divided by adjusted-risk assets), loan loss provisions, and non-interest income. 5.1 the model the least-square model used to measure a bank’s performance is expressed as: roe it = α + β 1 asst it + β 2 eff it + β 3 tier1 it + β 4 nii it + β 5 llp it + ε it (1) where, roe = the return on equity, llp = loan loss provisions. asst = total assets, tier1 = tier1 capital ratio, eff = efficiency ratio, nii = non-interest income, and ε it (error term) ~ n (0, σ2). 6. empirical results table 1 provides the performance of domestic and foreign banks in thailand. foreign banks perform better than domestic banks in terms of total income (both interest income & non-interest income) and pre-tax profit, even though foreign banks had higher operating costs and loan loss provisions. in fact, domestic banks earned a negative value in pre-tax profit. the domestic banks’ poor performance reflects the impact felt by the thai financial crisis of july 1997 and its aftermath of adopting new, liberal financial policies. table 1: performance of thai banks vs. foreign banks in thailand, major u.s. banks, and european banks (in percent) thailand (domestic) thailand (foreign) u.s.* euro zone* france* spain* u.k.* 1. net interest margin 1.60 4.88 3.22 1.68 0.93 2.66 2.19 2. non-interest income 1.04 3.08 2.65 1.19 0.89 1.36 1.75 3. total income: (1) + (2) 2.64 7.96 5.87 2.87 1.82 4.02 3.98 4. operating cost 2.24 3.96 3.80 1.98 1.26 2.67 2.59 5. loan loss provision 2.48 2.56 0.39 0.31 0.24 0.39 0.22 6. pre-tax profit: (3)-(4)-(5) -2.08 1.44 1.67 0.57 0.32 0.97 1.13 *data covered 1996-98 from larosiere (2000). 106 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 table 2 compares and presents the mean values of the key factors (i.e., roe, roa, efficiency ratio, loan loss provisions/total assets, and tier1 capital ratio) that affect bank performance. clearly, thai banks show negative values for both roe and roa, whereas they have a higher ratio of loan loss provision to total assets. on the contrary, foreign banks in thailand perform better. their key factors are all in small positive values, yet, they still did not meet the banking industry’s standard measure of good performance:i.e., roa > 1 (ritter et al. (2000, p. 225). in fact, only a handful of banks worldwide have their roa > 1, such as, lloyds and chase manhattan (chotigeat et al. 2004). table 2: performance data for thai banks and foreign banks in thailand roe roa efficiency ratio loan loss provisions/ total assets tier 1 capital thai banks (1997-2003) -.24 -0.02 -0.01 0.03 0.28 foreign-branch banks in thailand (1997-2003) 0.12 0.31 0.12 0.01 0.38 table 3 provides a matrix of correlation coefficients among the variables being used in equation (1). table 3: correlation coefficient matrix panel a. thai commercial banks eff asst tier1 roe nii llp eff 1.00 -0.16 -0.10 -0.40 0.09 0.05 asst -0.16 1.00 -0.56 0.36 0.44 -0.40 tier1 -0.10 -0.56 1.00 -0.29 -0.17 0.35 roe -0.40 0.36 -0.29 1.00 0.33 -0.66 nii 0.09 0.44 -0.17 0.33 1.00 -0.35 ll 0.05 -0.40 0.35 -0.66 -0.35 1.00 panel b. foreign-branch banks in thailand eff asst tier1 roe nii llp eff 1.00 -0.06 -0.12 0.27 0.16 0.01 asst -0.06 1.00 -0.67 -0.11 0.05 0.45 tier1 -0.12 -0.67 1.00 0.05 0.23 -0.47 roe 0.27 -0.11 0.05 1.00 0.17 -0.55 nii 0.16 0.05 0.23 0.17 1.000 0.01 llp 0.011 0.45 -0.47 -0.55 0.01 1.00 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 107 in each panel, only one of the coefficients (in absolute value) is greater than 0.50. the correlation coefficient between total assets and tier1 capital ratio for thai commercial banks is -0.56, respectively, while for foreign banks in thailand, the correlation coefficient for the same relationship is –0.67, respectively. to avoid multi-collinearity issues, we chose not to use both the total assets and tier1 capital ratio at the same time in an equation. thus, we formulated two models: model 1 has total assets, efficiency, tier 1 capital, loan loss provision, and non-interest income as the independent variables, while model 2 is the same as model 1 without the variable of total assets. table 4: results from regression model 1 thai banks foreign banks (in thailand) dependent variable: roe dependent variable: roe variable coefficient variable coefficient c 3.81 (0.82) c -0.21 (-0.13) asst -0.25 (-0.71) asst 0.02 (0.16) eff -9.18* (-3.14) eff 0.03 (0.26) tier1 -2.81 (-1.05) tier1 0.23 (0.25) nii 0.01 (1.30) nii 6.01e-05 (0.80) llp -0.08* (-3.78) llp -0.01* (-2.07) ar(1) 0.496* (2.37) r-squared 0.62 r-squared 0.55 adj. r-squared 0.54 adj. r-squared 0.41 s.e. of reg. 0.10 s.e. of regression 0.05 ssr 0.25 ssr 0.06 log likelihood 26.24* (f-stat=7.42) log likelihood 43.54* (f-stat=4.12) d-w stat 2.24 d-w stat 2.10 t-stat is in parenthesis; * 1% significance level; ** 5% significance level the ls regression model of equation (1) was estimated for model 1 with the estimated parameters provided in table 4. in the left two columns of table 4, the coefficients show that thai domestic bank performance is statistically influenced negatively by only two variables: the efficiency ratio and loan loss provision. the 108 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 two coefficients have the appropriate signs in relation to the hypotheses postulated. for coefficients of the other three variables (total assets, tier1 capital ratio, and non-interest income), each has an appropriate sign, yet they are not statistically significant. overall, all the estimated parameters conform to the hypothesis in terms of sign, but only two were found to be statistically significant. the right two columns of table 4 provide the estimated coefficients of foreign banks in thailand. the model was modified to correct the autocorrelation of the error-term in the model; thus, the specification of equation (1) that includes ar (1) is appropriate. only the coefficient of loan loss provision shows a negative sign (as was expected) whose significance is statistically viable. the coefficient of non-interest income has an appropriate positive sign, but it is not statistically significant. table 5 provides the estimated parameters of model 2 (again, model 1 without the total assets variable). the pattern of coefficient estimates has not changed for thai domestic banks, nor has it for foreign banks operating in thailand. table 5: regression model 2 thai bank foreign bank (in thailand) variable coefficient variable coefficient c 0.52* (2.26) c 0.04 (0.48) eff2 -8.48* (-3.11) eff2 0.02 (0.24) tier1m -1.76 (-0.80) tier1m 0.13 (0.17) niin 0.01 (1.12) niinm 6.32e-05 (0.90) llp -0.07* (-3.76) llpm -0.01* (-2.18) ar(1) 0.49* (2.79) r-squared 0.61 r-squared 0.55 adj. r-squared 0.55 adj. r-squared 0.44 s.e. of regression 0.10 s.e. of regression 0.05 ssr 0.25 ssr 0.06 log likelihood 25.9* (f-stat=9.34) log likelihood 43.51* (f-stat=5.18) d-w stat 2.11 d-w stat 2.11 t-stat is in parentheses; * 1% significance level; ** 5% significance level the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 109 our findings are consistent with those found in other countries that were reported in other studies.3 the determinants of the thai domestic banks and foreign banks in thailand are compared to those of other top banks around the world (see table 6). for thai domestic banks, the significance of the efficiency ratio is similar to that of japanese, british, american, and french banks; loan loss provision is similar to that of british and french banks. for foreign-branch banks in thailand, loan loss provision is similar to that of british, french, and thai domestic banks. table 6: determinants of efficiency: individual country banks comparison thai (domestic) foreign (in thailand) france usa uk germany japan total assets x x x efficiency x x x x x tier1 capital ratio x loan loss provisions x x x x non-interest income x (x indicates a statistical significance of the coefficient.) source: chotigeat et al. (2004, p. 14). overall, our findings confirm the previous studies that efficiency and loan loss provision are the dominant factors that determine the profit performance of banks worldwide. for foreign banks, despite the fact that they… are from different countries operate under different banking systems under different organizational structures have different corporate cultures operate under varying national economies, they still share many common variables with global competitors from around the world. 7. conclusion with new, financial systems that emphasize & enforce a policy of liberalization & transparency, many foreign banks are now seeking to operate in thailand. what 3 focarelli, panetta and salleo (2002); grifell-tatje and lovell; and kumbhakar andh sarkar 110 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 is of great interest is how thai banks had performed after the crisis and how their performance compared to that of foreign banks that operated in the same environment in thailand. their performance (specifically, the value of the roe capital ratio) was negatively influenced by the efficiency ratio and loan loss provisions, where each negative sign confirms our hypothesis. for the foreign banks in thailand, only the loan loss provision factor confirms our hypothesis (negative sign) as a determinant of performance. when we compared the determinants of performance for both groups in this study with those of other top banks around the world, we found that the significance of efficiency for the thai banks’ performance is similar to that of french, us, british, and japanese banks. the loan loss provision is similar to that of french, british, and other foreign banks in thailand. lastly, foreign banks in thailand only have loan loss provisions as their determinant, which is similar to that of french and british banks. on balance, loan loss provision is the only common factor that is applicable to the performance evaluations of the thai domestic banks and foreign banks operating in thailand. globally, efficiency, total assets, and loan loss provisions are the three determinants of banks’ performance. we think foreign banks may have an edge over domestic thai banks because foreign banks have recognized the competitive intensity of the financial services business. simply put, multinational banks worldwide have had to react to the changing business environments brought forth by global deregulation by increasing their m&a activity. therefore, thai domestic banks could possibly benefit by learning form their foreign counterparts. author statement: tosporn chotigeat, department of economics and finance, nicholls state university. e-mail: ecfi-tc@nicholls.edu. he is the editor of the global business and finance review. references allen, linda, and annop rai. 1996. operational efficiency in banking: an international comparison. journal of banking & finance, 20: 655-672. (bangkok post. 1999. “the 1999 mid-year economic review.” july. berger, allen n., qinglei dai, steven ongena, and david c. smith. 2002. to what extent will the banking industry be globalized? a study of bank nationality and reach in 20 european nations. journal of banking & finance, 27, 383415. berger, allen n., and loretta j mester. 2003. explaining the dramatic changes in the performance of u. s. banks: technological change, deregulation, and dynamic changes in competition. journal of financial intermediation, 12, 57-95 brimmer, andrew. 1998. competition and the integration of the markets for banking and financial services. north american journal of economics and finance, 9: 187-202. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 111 cabral, ines, frank dierick, and iukka vesala. 2002. banking integration in the euro area. occasional paper no. 6, european central bank chotigeat, t., sebastien kramer, and c.s. pyun. 2004. efficiency and resilience of french multinational banks: evidence from the pre-euro era, multinational business review, 12-1, 3-17. drake, leigh, and maxmilan j. b. hall. 2003. efficiency in japanese banking: an empirical analysis, journal of banking and finance, 27, 891-917. the economist, feb 2, 2002, “deutsche’s american dream ‘, pp. 69-70. focarelli, dario, fabio panetta and carmelo salleo. 2002. why do banks merge? journal of money, credit, and banking, 34-4, 1047-1066. focarelli, d., and a. pozzolo. 2001. the pattern of cross-border bank margin and shareholdings in oecd. journal of banking and finance, 25, 2305-2337. goldberg, lawrence, and gerald hanweck. 1991. the growth of the world’s 300 largest banking organizations by country. journal of banking & finance, 15: 207-223. grifell-tatje, e., and c. a. k. lovell. 1996. deregulation and productivity decline: the case of spanish savings banks. european economic review, 40, 12811303. grosse, robert, and alan gart. 2001. international banks’ performance: a fivecountry comparison. global business and finance review, 6(2): 1-15. isik, ihsan, and m. kabir hassan. 2003. financial deregulation and total factor productivity change: an empirical study of turkish commercial banks. journal of banking and finance, 27, 1445-1485. kim, seung, and robert singer. 1997. us and japanese banks: a comparative and evaluative analysis. the bankers magazine, march-april, 56-61. kumbhakar, subal c., and subrata sarkar. 2003. deregulation, ownership, and productivity growth in the banking industry: evidence from india. journal of money credit and banking, 35, 403-424. larosiere, jacqueo de. 2000. banking consolidation in europe, the per jacobson foundation lecture series. bank for international settlement, lucerne, switzerland, 18-44. mendes, v., and j. reblo. 1999. productive efficiency, technological change and productivity in portuguese banking, applied financial economics, 9, 513521. pindyck, robert s., and daniel l. rubinfeld, 1991. econometrics models & economic forecasts, 3rd edition, new york: mcgraw-hill inc. ritter, l.s., w.l. silber, and g.f. udell. 2000. money, banking, and financial markets. 10th edition. new york: addison-wesley. sanford, charles. first quarter 1994. financial markets in 2020. economic review. kansas city: federal reserve bank of kansas city, 19-28. spiegel, john, alan gart, and steven gart. 1996. banking redefined. chicago: irwin. williams, barry. 2003. domestic and international determinants of bank profits: foreign banks in australia. journal of banking and finance, 27, 11851210. 112 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 101-112 appendix a key economic indicators of thailand (1997-2003) internal balance country year gdp government cpi stock interest % change budget % change index rate (%) (bn baht) short-term 1997 -0.4 -15.16 8.0 376.2 14.59 1998 -8.0 -128.95 4.3 355.8 13.02 1999 0.2 -154.19 7.2 391.5 3.70 2000 6.0 -108.06 1.8 327.5 3.25 2001 4.2 -122.99 1.3 303.5 2.75 2002 6.0 -102.07 0.4 353.48 2.00 2003 5.8 -105.01 1.1 567.1 2.00 thailand external balance trade financial foreign currency balance account reserves per us$ us$bn us$bn us$bn 1997 1.60 -5.4 26.9 47.00 1998 13.10 -9.8 29.5 41.37 1999 10.50 -7.9 34.8 37.84 2000 5.60 -10.3 32.7 40.15 2001 2.50 -5.1 33.0 44.48 2002 3.40 -4.0 38.9 43.00 2003 4.50 -7.0 39.2 40.01 international journal of banking and finance 3-1-2008 efficiency of domestic and foreign banks in thailand since the asian financial crisis tosporn chotigeat recommended citation banking competition and efficiency in jordan: a note 77 the international journal of banking and finance, volume 9 (number 2) 2012: pages 77-88 banking competition and efficiency in jordan: a note khaldoun al-qaisi amman arab university, jordan ____________________________________________________________ abstract the financial economics literature contains numerous research papers which examine issues that concern the banking industry. one of these issues is banking competition. indeed, this issue is important because of its implications to financial stability and the growth of the borrowing firms. the purpose of this paper is to assess the competitive behavior of the jordanian banking sector during the period ranging from 1999 to 2008 using the nonstructural test developed by panzar and rosse. in more specific terms, this paper examines the overall competitive condition during the period 1999 – 2008 and how it has evolved over time. based on the empirical findings, it is expected that a number of policy recommendations may be provided. the objective of these recommendations is to enhance the regulation of the banking sector in jordan and improve their performance. key words: financial economics, banking competition, jordanian banking sector, financial stability, economic growth jel classification: g21, g32, e44, e51 _____________________________________________ 1. introduction it is common knowledge that the theoretical predictions of the financial development and economic growth nexus are well supported by many cross-country, industry-level, and firmlevel research papers. based on this vast literature, one can state that the role of finance in economic growth has become a stylized fact. indeed the literature has shifted its attention to other and related issues such as the determinants of financial development and the development of new measures of financial development and economic growth. the issue of financial development is relatively important for study in the jordanian case. this is based on a number of observations. first, all jordanian banks are private and listed on the local stock exchange (amman securities exchange). second, the total number of bank branches has increased from 25 in 1970 to 438 by 2008. third, the ratio of total bank 78 assets to gdp is equal to 33.4 per cent in 1970. this ratio has improved to 209 per cent by the end of 2008. finally, the ratio of bank credit to the private sector as per cent of gdp has increased from 19.9 per cent to 93 per cent over the same period. the fact is that financial intermediaries are economically important and finance literature contains a large number of papers examining an equally large number of banking issues. these include, for example, the determinants of net interest margin, determinants of bank capital, impact of foreign bank entry on the performance of local banks, and bank competition. against this background, the primary objective of this research is to provide answers to the following questions: (i) what is the degree of competition within the jordanian banking industry during the period 2000-2008?; and (ii) has there been a shift in the intensity of competition within the jordanian banking industry? the rest of the paper is organized as follows. after a brief description in the next section about the growth in size and intensity of banking sector over a 50-year history, we provide in section 3 the data description and the methodology. the results are also discussed in the same section. conclusions are drawn in section 4. 2. the jordanian banking sector the banking system in jordan consists of the central bank of jordan (cbj), licensed banks, and specialized credit institutions. the establishment of licensed banks in jordan dates back to 1925 when the ottoman bank (later to become british bank) started its operations. following this, and due to the 1948 arab-israeli war, the arab bank, which was established in palestine (jerusalem) in 1930, moved its headquarters to amman (jordan). in other words, since the late 1950s and early 1960s, the jordanian economy has been witnessing the growth of privately licensed banks. indeed, since that time period, the total number of banks has grown to reach thirteen national banks, two islamic banks, and eight foreign banks. in table 1, we report data on some basic information to reflect the development of the jordanian banking sector. based on the reported figures, we can make the following observations. first, the size of jordanian banks is extremely large. for example, the ratio of the total assets of all jordanian banks has increased from 33.4 per cent (1970) and 91.9 per international journal of banking and finance, vol. 9, iss. 2 [2012], art. 5 79 cent (1980) to more than 200 per cent by the end of 2002. since then, this ratio has been consistently above the 200 per cent mark. second, it is interesting to note that jordanian banks hold high proportions of their assets in the form of foreign assets and these are largely deposits in foreign banks operating outside jordan. indeed, by the end of 2008, this proportion has reached 19.5 per cent. finally, it is useful to note that the available figures indicate that during the period jordanian banks accounted for about 90 per cent of the assets, deposits, and credit of all licensed commercial banks in the country. in other words, non-jordanian banks are relatively small and, as mentioned above, geographically concentrated in a limited number of cities (mostly in the capital). table 1: assets of jordanian banks, 1964 – 2008 year total assets (million jd) total assets / gdp foreign assets (million jd) foreign assets / total assets 1970 76.4 0.334 6.1 0.080 1980 1070.5 0.919 205.2 0.192 1990 4090.0 1.481 1045.7 0.256 2000 12913.5 2.153 3711.7 0.287 2002 15119.3 2.225 4492.7 0.297 2004 17821.1 2.203 5002.1 0.281 2006 24237.6 2.304 6203.4 0.256 2007 26815.6 2.288 6516.5 0.243 2008 29796.6 2.100 5810.3 0.195 source: various central banks of jordan annual reports. table 2: deposits and credit year deposits / gdp private sector deposits / gdp credit to private sector / gdp 1970 0.253 0.201 0.199 1980 0.694 0.613 0.484 1990 0.957 0.849 0.675 2000 1.373 1.135 0.759 2002 1.379 1.192 0.755 2004 1.429 1.268 0.765 2006 1.387 1.266 0.928 2007 1.364 1.245 0.964 2008 1.276 1.156 0.930 source: various central bank of jordan annual reports. al-qaisi: banking competition and efficiency in jordan: a note 80 on the liabilities side, it is noticeable that all licensed banks have growth in importance relative to the size of the national economy. for example, we can see in table 2 that the ratio of total deposits to gdp (total private sector deposits to gdp) has increased from about 96 per cent in1990 to reach 127.6 per cent by the end of 2008. similarly, the ratio of total credit provided by all licensed banks to the private sector has also increased significantly to reach 93 per cent of gdp as at 2008. as expected, a large proportion of the credit which is provided to the private sector is made in the local currency. as reported in table 3, less than 5 per cent of the credit provided to the private sector is denominated in foreign currencies. indeed, this is expected due to the higher risk involved in lending in foreign currencies. table 3: proportion of credit to private sector in foreign currency to total credit year proportion 2000 0.048 2002 0.051 2003 0.055 2004 0.041 2005 0.029 2006 0.032 2007 0.027 2008 0.041 source: various central banks of jordan annual reports. table 4: the distribution of total credit sector 2000 2002 2004 2005 2006 2007 2008 agriculture 0.028 0.020 0.018 0.014 0.014 0.014 0.016 mining 0.022 0.018 0.013 0.007 0.004 0.006 0.004 industry 0.150 0.154 0.145 0.127 0.112 0.119 0.121 commerce 0.244 0.244 0.238 0.205 0.196 0.216 0.221 construction 0.164 0.149 0.154 0.150 0.160 0.172 0.172 transport 0.029 0.032 0.028 0.028 0.030 0.031 0.027 tourism 0.034 0.034 0.025 0.023 0.020 0.023 0.028 services 0.053 0.068 0.080 0.072 0.065 0.065 0.064 financial 0.034 0.027 0.016 0.023 0.025 0.034 0.032 individuals 0.241 0.253 0.284 0.351 0.373 0.320 0.316 total 1.00 1.00 1.00 1.00 1.00 1.00 1.00 source: various central bank of jordan annual reports. in addition to the above, statistics in table 4 provide information on the distribution of the credit provided by licensed banks to the various sectors of the economy. again, based on these figures, one can observe a number of interesting points. for example, the agricultural international journal of banking and finance, vol. 9, iss. 2 [2012], art. 5 81 sector accounts for a very small proportion of the total credit (1.4 per cent). this observation is not surprising if one realizes that the agriculture sector accounts for a small proportion of the national economy (about 2 per cent of gdp). in contrast to the agricultural sector, it is interesting to note that the largest proportion of the credit goes to private individuals (31.6 per cent). these are mostly short term bank credit and overdraft facilities. 3. literature review on bank competition in the literature, there are three approaches to empirically evaluate the degree of competition in, thus the efficiency of the banking industry. these are: panzar and rosse (1982 and 1987), lau (1982, and hall (1988). however, the h-statistic approach of panzar and rosse is the most popular non-structural model which has been used by many researchers to examine the competitive structure of the banking industry in various countries. the panzar-rosse methodology develops a test statistic (named h-statistic) to distinguish between oligopolistic, competitive, and monopolistic markets. the h-statistic is computed from the reduced-form revenue equation at the bank level and this can serve as a measure of the competitive behaviour of banks under certain assumptions. these assumptions include (i) profit maximization at bank and industry levels, (ii) equilibrium in the industry, and (iii) conventional cost and demand structures. market power is evaluated by the extent to which changes in factor prices are reflected in the equilibrium revenues. the sum of the elasticities of the reduced-form revenues with respect to input prices defines the competitive indicator (the h-statistic). the magnitudes of the estimated values of the h-statistic indicate the market structure of the industry and a summary. if the market structure is characterized by monopolistic competition, the h-statistic lies between zero and one, as revenues increase less than proportionately to changes in input prices. if the market is characterized by perfect competition, the h-statistic is equal to one. indeed, any increase in input prices will increase both marginal and average costs without changing the equilibrium output of any individual bank. this is the case because any bank which cannot cover the increase in its input prices will be forced out of the market. an important feature of the h-statistic is that it must be performed on banks which are in long-run equilibrium. this can be tested by regressing the return on asset (roa) instead of bank revenue on input prices. if the sum of the elasticities is equal to zero, this indicates longal-qaisi: banking competition and efficiency in jordan: a note 82 run equilibrium (e-statistic). if the sum of the elasticities is negative, this indicates disequilibrium (e-statistic). the panzar-rosse approach has been widely used in the literature, so a study of jordanian banks using this method should provide reliable comparisons. indeed, the number of papers applying this method is very large and we review only a few. however, some of the main papers, which examined competition in the emerging country banks are examined by yildrim and philippatos (2007), prasad and ghosh (2005). t is reported (turk-ariss, 2009) that banks in the middle east and north africa (mena) region operate under conditions of monopolistic competition. this result is also supported by a much earlier paper by murjan and ruza (2002). 4. data, methodology and results to assess the competitive behaviour of the jordanian banking sector during the period 2000 to 2008, a total of 12 listed jordanian banks are analysed. based on the international evidence, the following two models are estimated to arrive at the measures of competitiveness in the jordanian banking industry: lnrevit = α0+α1lnpeit+α2lnieit+α3lnooeit+α4lncapit+α5lnsizeit +α6lnriskit+α7lngrot+ εit (1) lnintrevit=α0+α1lnpeit+α2lnieit+α3lnooeit+α4lncapit+α5lnsizeit +α6lnriskit+α7lngrot + εi (2) where ln stands for the natural logarithm, rev is the ratio of bank revenue (interest and noninterest) to total assets, intrev is the ratio of interest revenue to total assets, pe is the ratio of personnel expenses to total assets, ie is the ratio of interest expenses to total deposits, ooe is the ratio of other operating expenses to total assets, cap is the ratio of equity capital to total assets, size is total assets, risk is net loans to total assets, and gro is the real gdp growth rate. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 5 83 the subscripts i and t denote banks (i = 1, …, n) and time (t = 1, …, t) respectively. in expression (1), the dependent variable (rev) is used to measure market competitiveness with respect to the overall banking activities. in expression (2), the dependent variable (intrev) is used to focus on the core banking activity (lending). based on the above two expressions, the h-statistic is given by h = α1 + α2 + α3. in other words, the sum of these three coefficients indicates how bank total revenues (rev) or interest revenues (intrev) react to changes in the three input prices (pe, ie, and ooe). as far as the other independent variables are concerned, they account for bank-specific (cap and risk) and market-specific (growth) variables. in addition, and as mentioned above, the equilibrium condition is modelled as follows: lnroait = α0 + α1lnpeit + α2lnieit + α3lnooeit + α4lncapit + α5lnsizeit + α6lnriskit + α7lngrot + εit (3) where roa is the pre-tax return on assets and the independent variables are as defined above. the market is in long-run equilibrium if e = α1 + α2 + α3 = 0. in other words, bank’s rates of return should not be significantly correlated with input prices. a constant is added to roa to avoid the natural logarithm of a negative number. table 5: descriptive statistics rev is the ratio of bank revenue (interest and non-interest) to total assets, intrev is ratio of interest revenue to total assets, roa is return on assets, pe is ratio of personnel expenses to total assets, ie is ratio of interest expenses to total deposits, ooe is the ratio of other operating expenses to total assets. measure mean median minimum maximum rev 0.071 0.071 0.030 0.103 intrev 0.054 0.055 0.029 0.083 roa 0.018 0.018 -0.041 0.061 pe 0.011 0.010 0.003 0.029 ie 0.015 0.014 0.003 0.046 ooe 0.026 0.025 0.006 0.075 to assess the competitive behaviour of the sector, the estimations are carried out for a balanced panel of banks over the period 2000-08. in addition, the sample period is divided 84 into two sub-periods: these are 2000-04 and 2004-08. this analysis should enable us to conclude whether there has been any change in the competitive conditions in the jordanian banking sector. finally, to estimate the regressions 1-3, the method that we use is period seemingly unrelated regression (sur) – pooled estimated generalized least squares (egls). in table 5, we report the summary data for the dependent and cost variables which are used in the empirical analysis. as one might expect, the reported statistics reflect the presence of some variations between the 12 banks and this can be seen from the reported minimum and maximum values. note that the roa is pretty much comparable to the statistics across international studies. in tables 6, 7, and 8, we present the estimation from models (1), (2), and (3). in addition, we report in table 8 the sum of the three coefficients (personnel expenses, interest expenses, and other operating expenses) to judge the efficiency levels. table 6: model 1 estimation results lnrevit = α0 + α1lnpeit + α2lnieit + α3lnooeit + α4lncapit +α5lnsizeit + α6lnriskit + α7lngrot + εit. ln is natural logarithm, rev is the ratio of bank revenue (interest and noninterest) to total assets, pe is ratio of personnel expenses to total assets, ie is ratio of interest expenses to total deposits, ooe is ratio of other operating expenses to total assets, cap is ratio of equity capital to total assets, size is total assets, risk is net loans to total assets, and growth is real gdp growth rate. variable coefficient t-statistic lnpe -0.057 -0.637 lnie 0.220 2.499* lnooe 0.001 0.043 lncap -0.025 -1.227 lnsize -0.086 -10.404* lnrisk 0.039 3.516* lngro -0.131 -2.078** adjusted r2 0.897 1.870 744* durbin-watson sts. f-statistic as is evident from the statistics in model (1) results in table 6, the model fit is well established with high adjusted r-square values and f-ratio. the same could be said of the results in table 7 for model (2) and in table 8 for model (3). the estimation of the three critical parameters in the three sets of results suggest that the jordanian banking sector is not competitive since the summed coefficients fall short for international journal of banking and finance, vol. 9, iss. 2 [2012], art. 5 85 that conclusion to be made. the statistics suggest that the banking sector is monopolistic competition, which is indicative of its relative inefficiency. table 7: model 2 estimation results lnintrevit = α0 + α1lnpeit + α2lnieit + α3lnooeit + α4lncapit +α5lnsizeit + α6lnriskit + α7lngrot + εit. ln is natural logarithm,, intrev is ratio of interest revenue to total assets, pe is ratio of personnel expenses to total assets, ie is ratio of interest expenses to total deposits, ooe is ratio of other operating expenses to total assets, cap is ratio of equity capital to total assets, size is total assets, risk is net loans to total assets, and growth is real gdp growth rate. variable coefficient t-statistic lnpe -0.192 -2.627* lnie 0.371 4.799* lnooe -0.042 -2.566* lncap -0.026 -1.946*** lnsize -0.075 -18.040* lnrisk 0.079 9.733* lngro -0.467 -21.769* adjusted r2 0.898 1.907 555* durbin-watson sts. f-statistic table 8: model 3 estimation results lnroait = α0 + α1lnpeit + α2lnieit + α3lnooeit + α4lncapit +α5lnsizeit + α6lnriskit + α7lngrot + εit. ln is natural logarithm, rev is the ratio of bank revenue (interest and noninterest) to total assets, intrev is ratio of interest revenue to total assets, pe is ratio of personnel expenses to total assets, ie is ratio of interest expenses to total deposits, ooe is ratio of other operating expenses to total assets, cap is ratio of equity capital to total assets, size is total assets, risk is net loans to total assets, and growth is real gdp growth rate. variable coefficient t-statistic lnpe -0.011 -1.464 lnie 0.023 2.984* lnooe -0.022 -9.544* lncap 0.001 0.499 lnsize -0.001 -2.689* lnrisk 0.001 1.332 lngro 0.009 3.010* adjusted r2 0.863 1.927 113* durbin-watson st f-statistic 86 the panzar-rosse statistics are obtained from the three regressions for the whole period and for two sub-periods of the study. these are summarized in table 9. it is evident from these statistics that the desired competition level is not yet achieved despite the banks being operated by private sector and fast phased development over the test period, in fact impressive gains in size variables. table 9: panzar-rosse statistics rev is the ratio of bank revenue (interest and non-interest) to total assets, intrev is ratio of interest revenue to total assets, roa is return on assets 2000-2008 2000-2004 2004-2008 rev 0.164 0.277 0.082 intrev 0.137 0.009 0.070 roa -0.010 -0.016 -0.006 in summary then an international comparison would place jordanian banks as being monopolistic rather than fully competitive as are such developing economies as malaysia and korea. the banking sector has to develop to be at least oligopolistic before becoming competitive. for that pointed reforms are needed in the management of the banks in jordan. 5. summary and conclusions the issue of financial development cannot be overstressed in jordan. indeed, this is based on many factors including the fact that while economic growth averaged 7.4 per cent a year in real terms during the 2004-08 test period, the national economy could not reduce its unemployment rate, which is around 12 to 14 per cent. it is higher for the young people (20 to 24 years): it is equal to 27 per cent. this is why successive governments have persevered with some consistent policies including the commitment to the promotion of the role of the private sector. this policy (private sector promotion) makes the private sector, at once a major player, an instrumental actor, and a direct beneficiary of any expected economic gains. indeed, we can state that the banking system in jordan constitutes a major part of this private sector. this is based on a number of observations. first, all jordanian banks are private and listed on the local stock exchange. second, the total number of banks has increased and third, international journal of banking and finance, vol. 9, iss. 2 [2012], art. 5 87 the ratio of total bank assets to gdp has also increased. finally, the ratio of bank credit to the private sector has increased as well. the fact that the economy faces a number of economic challenges, the jordanian banking system is relatively expanded, financial intermediaries in general are economically important. the costs of any bank failure are much greater than that of other businesses. has the banks, while growing larger, how do they fare in terms of competition to provide the welfare maximization to the economy and the people? the empirical results show that the jordanian banking system operates under monopolistic competitive conditions, and has to move towards more efficient status to provide welfare maximization. in addition, it is concluded that the time period 2000-2008 has not witnessed any change in this condition. the implications of this finding are straight forward. jordanian policy-makers must realize the status of competitiveness in the banking sector so that its competitive efficiency has to be increased. that should be the major focus for the period ahead. indeed, this is probably why during the past few years, a total of six arab and foreign banks have been allowed to operate, to increase competition. in other words, it would be interesting to examine the impact of foreign bank entry on the competitive nature of all banks operating at a future date. author information: khaldoun m. al-qaisi is an assistant professor of finance in amman arab university in jordan. his address is: p.o. box 1825, amman (11821) jordan. he may be contacted via e-mail: khaldoun_21@yahoo.com. mailto:khaldoun_21@yahoo.com 88 references hall, r., (1988). the relation between price and marginal cost in us industry, journal of political economy 96: 921-947. lau, l., (1982). on identifying the degree of competitiveness from industry price and output data, economic letters 10: 93-99. murjan, w., and c. ruza (2002). the competitive nature of the arab middle eastern banking markets, international advances in economic research 4: 267-274. panzar, j., and j. rosse (1982). structure, conduct and comparative statistics, bell laboratories economics discussion paper. panzar, j., and j. rosse (1987). testing for ‘monopoly’ equilibrium, journal of industrial economics 35: 443-456. prasad, a., and s. ghosh (2005). competition in indian banking, imf working paper no. 141. turk-ariss, r., (2009). competitive behaviour in middle east and north africa banking sustems, the quarterly review of economics and finance 49: 693-710. yildirim, h., and g. philippatos (2007). competition and contestability in central and eastern european banking markets, managerial finance 33: 195-209. yuan, y., (2006). the state of competition of the chinese banking industry, journal of asian economics 17: 519-534. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 5 the international journal of banking and finance, vol. 7. number 2: 2010: 1-18 1 influence of transaction costs on foreign exchange option contracts: intra-daily tests ariful hoque, meher manzur, and geoffrey poitras university of southern queensland, curtin business school, and simon fraser university ________________________________________________________________ abstract this paper tests the impact of transaction cost speci� cation on deviations from lower boundary and put-call parity properties. using phlx traded foreign exchange options, prices for puts and calls are matched to the nearest � ve minutes. the results indicate how boundaries on the arbitrage pro� t function determined by alternative measures of transactions costs can impact the interpretation of deviations from distribution free properties of options such as put-call parity. keywords: put-call parity; market ef� ciency; arbitrage; european options jel classi� cation: g13, g15, f31 ________________________________________________________________ 1. introduction this paper examines the impact of transaction costs on the distribution-free properties of european foreign exchange options traded on the philadelphia exchange (phlx) from aug. 2005-july 2006. empirical tests were conducted using intra-daily option quotes observed at � ve minutes intervals. deviations from put-call parity (pcp) were examined to determine the role of alternative transaction cost measures. a more formal statistical procedure was also employed to test the factors determining deviations from pcp. the results extended previous studies that examined the impact of transaction costs on pcp deviations for stock index options, e.g., vipul (2008), garray, ordonez and gonzalez (2003) and wagner, ellis and dubofsky (1996), and hoque, chan and manzur (2008) for foreign currency options. the paper is organized as follows. in section 2, transaction cost measures are speci� ed and discussed. section 3 gives an overview of the distribution free properties of option prices being considered: a lower boundary condition; and, put-call parity. section 4 describes the data used in this study. section 5 provides empirical ijbf 1 hoque et al.: influence of transaction costs on foreign exchange option contrac produced by the berkeley electronic press, 2010 2 the international journal of banking and finance, vol. 7. number 2: 2010: 1-18 results for deviations from distribution free conditions. section 6 contains a review of a more formal econometric analysis. the � nal section presents conclusions of the paper. 2. transaction cost measures at least since gould and galai (1974), the role of transaction costs in determining deviations from the pcp relationship have been of interest. in foreign exchange options markets, these costs are of various types: bid-ask spreads commissions; and, other types of costs such as clearing fees, exchange fees, and exercise fees. the actual impact of different types of transaction costs are market and trader speci� c and costs associated with the marginal trader are dif� cult to determine due to the lack of precise data about speci� c trades. studies going back to demsetz (1968), phillips and smith (1980) and stoll (1989), identi� ed the bidask spread as the most important implicit transaction cost. however, in assessing deviations from put-call parity, it is not apparent when a trade is initiated to exploit an observed deviation whether the trade will be closed out prior to the expiration date. holding the trade to expiration and delivering on the underlying positions involve different transaction costs than a trade closed out prior to that date which requires additional transactions costs, especially bid-ask spreads, to be assessed. to study the impact of different transaction cost measures associated with put-call parity deviations, the following different categories of transaction costs are considered: (a) minimum transactions costs, where only the bid-ask spreads on the initial trade are considered; (b) costs in (a) plus additional estimated bid-ask spreads associated with closing out the call, put and exchange rate positions prior to expiration; and, (c) maximum transactions costs, including bid-ask spreads from (a) and (b) plus ‘other transaction costs’ that includes commissions and other � xed costs. recognizing that ‘other transaction costs’ are more dif� cult to estimate, the identi� able costs for a one-lot, phlx options contract are taken as: options commission $1.75 per contract (minimum per order) phlx forex options exchange fees $0.44 per contract option clearing corporation (occ) fees: clearing fees $0.05 per contract exercise fees $1.00 per contract (per line item) clearing member authorization stamp $23.00 per stamp (minimum 1 contract) total $26.24 per contract in turn, because ‘maximum transaction costs’ re� ect costs associated onelot trading, this measure can be considered as the upper bound for transaction costs associated with put-call parity deviations. 2 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 1 http://epublications.bond.edu.au/ijbf/vol7/iss2/1 in� uence of transaction costs on foreign exchange option contracts: intra-daily tests: 1-18 3 3. two distribution free properties in this section, two distribution-free properties of currency option prices are examined: a lower boundary condition (lbc); and, put-call parity (pcp). assessing the role of transactions costs requires the trading strategies associated with these relationships to be speci� ed using appropriate bid and ask rates. active execution requires that purchases (borrowing) be made at the ask and sales (lending) be made at the bid. lower boundary condition the lower boundary condition used in this study is obtained from the standard lower boundary condition for a european currency option. in the case of a call, this condition can be stated, e.g., poitras (2002, p.378): � � � � � � � �tttt rbxrbstxsc �� *,0max,, (1) where: � � �� tt ctxsc ,, call option price in domestic (us) currency at time t; t = time to expiration time of the option measured in fractions of a year; tr and *tr are the time t domestic and foreign currency interest rates with maturity date t; � � � �tt rrb �� exp and � � � �** exp tt rrb �� , are the continuously compounded present value of us$1 or fx$1 to be paid at t; s t = foreign exchange rate ($us/fx$) at time t; and, x = option exercise price in domestic currency. violation of this condition allows for: � �*tt rbs units of fx to be borrowed at *tr ; the funds converted to domestic currency at st to purchase the call at c t , and invest � �trbx . by assumption that the lbc is violated, there is an arbitrage pro� t remaining because the investment will mature to x which will be used to exercise the call for s t which is used to settle the loan. the lbc is a relatively weak arbitrage condition compared to pcp. if actual time dated prices used by traders could be continuously observed, it is almost certain that few if any lbc deviations would be found. as such, tests of the lbc condition serve to benchmark sample noise that can arise from a variety of factors such as data error, lack of synchronous observations, variations in accuracy across data providers and so on. the european lower boundary restriction on call and put option prices can empirically be re-expressed as inequalities: � � � � � �xsxscrbtxsc ttttt ������ ,0max,, 1 (2) � � � � � �ttttt sxsxprbtxsp ������ ,0max,, 1 (3) where: put option price in domestic currency; and, is the future value of $1 borrowed at t. to achieve equivalence with the distribution free lower boundary condition for a european currency option, the � � �� tt ptxsp ,, � � � 1� trb 3 hoque et al.: influence of transaction costs on foreign exchange option contrac produced by the berkeley electronic press, 2010 4 the international journal of banking and finance, vol. 7. number 2: 2010: 1-18 empirical observation that � � � � 1*1 �� tt rbrb is used (see table 3) to get the lbc used in this study: � � � � xsrbtxsc ttt ���1,, � � � � ttt sxrbtxsp ���1,, (4) due to � � 1�trb these conditions are slightly weaker upper bound on option prices than the standard european lbc. put-call parity the pcp arbitrage condition is a distribution free property for put and call options with the same exercise price and expiration date, e.g., poitras, veld and zabolotnyuk (2009). more precisely: � � � � � � � �txsprbsrbxtxsc ttttt ,,,, * � (5) the two arbitrage trading strategies associated with pcp depend on whether the call or put is overpriced. when c > sb* + p xb then a conversion strategy is executed where the call is sold and xb is borrowed, with the proceeds used to buy the put and the balance converted at s t and used to buy the foreign b*. this generates an arbitrage pro� t by assumption at t. this trade is sometimes referred to as the long pcp strategy. similarly, the arbitrage when the put option is overpriced, p > xb + c sb* is the reversal strategy also known as the short pcp strategy. deviation test framework bid and ask prices play a critical role in determining the impact of transactions costs on arbitrage trading. in particular, a conversion strategy would involve: borrowing the present value of the exercise price at the domestic offer rate with maturity date t; writing a call at the bid price; buying a put with the same x and t at the ask price; exchanging the remaining balance into foreign currency at ask rate (us$/c$); and, investing the proceeds into an appropriate � xed income security with maturity date t at the foreign bid rate.1 given this, using minimum transactions costs (a) absence of arbitrage requires: 1. bid and ask rates for foreign exchange depend on the fx quoting method. when the us$ is the domestic currency, then the exchange rate into fx$ is quoted as $us/$fx. for option prices and prices in general, the bid must always be not greater than the ask. however, for foreign exchange transactions, whether the bid rate for selling us and buying fx is greater than the ask rate for buying us and selling fx depends on the quotation method. for example, when $us/c$ = .9575 is the bid, then $us/c$ = .9600 could be the ask; in this case bid > ask. however, selling $us and buying canadian takes place at the ask rate because the bid will receive more c$ per us$. in effect, the bid (ask) quote based on c$/us$ is translated into ask (bid) quotes based on us$/c$. 4 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 1 http://epublications.bond.edu.au/ijbf/vol7/iss2/1 in� uence of transaction costs on foreign exchange option contracts: intra-daily tests: 1-18 5 � � � � � 0* ��� � ltatatstbtconv rbsprbxct (6) conversely, a reversal strategy involves: borrowing the present value of one unit foreign currency at ask rate; exchanging these funds from foreign into domestic currency at the bid rate (us$/c$); writing a put at the bid price; buying a call with the same x and t at ask price; and, investing the balance into an appropriate � xed income security with maturity date t. the absence of arbitrage condition for the reversal strategy is: (7) in both cases, the assumption is that the arbitrage will be held until the expiration and maturity date (t). in calculating the empirical deviations when other transactions costs are also included, the following arbitrage pro� t function is used for the conversion strategy: (8) similarly, for the reversal strategy: (9) where tc is � xed transactions costs per trade and i refers to one of two cases (b and c) for calculating such costs. in contrast to minimum transaction costs (a), case b includes bid-ask costs for trades closed out prior to maturity. this is a more complicated task because transactions costs for actual close-out trades cannot be accurately determined, e.g., bodurtha and courtadon (1986, 1995). in practice, there is a timing option that permits traders to sell at the ask and buy at the bid for some, but not necessarily all, of the positions involved in the trade. with this in mind, additional bid-ask transaction costs (b) for the conversion and reversal trades are determined as: (10) for the maximum transaction costs case (c), $26.24 per contract is added to tc b , to obtain tc c where the per contract cost is appropriately adjusted to re� ect the notional principal of the contract. tests of arbitrage relationships such as pcp depend fundamentally on the quality of the data employed. in particular, any study involving bid and ask option prices has to confront the dif� culty of determining, say, the ‘ask’ price from the available spectrum of market and limit quotes available at a given point in time. being a less restrictive condition than pcp, tests of the lbc serve to benchmark the data quality used in pcp tests. with this in mind, the following absence of arbitrage lbc conditions can be developed for both puts and calls: � � � � � 0* ��� � ltatstbtbtrev rbxcrbspt � � � � � � 0,*, ���� � tiltatatstbticonv tcrbsprbxct � � � � � � 0,*, ���� � tiltatstbtbtirev tcrbxcrbspt � � � � � � �btatbtatbtattb ssppcctc � � ��, � 5 hoque et al.: influence of transaction costs on foreign exchange option contrac produced by the berkeley electronic press, 2010 6 the international journal of banking and finance, vol. 7. number 2: 2010: 1-18 � � � 0, ���� xsrbc btltatcalllbc � � � 0, ���� atltatputlbc sxrbp (11) for deviations associated with transaction cost measures b and c, these lbc conditions are augmented to include tcl b and tcl c (e.g., tcbcalllbcbcalllbc tcl ,,,,, � ) where: � � � �btatbtattcb sscctcl � ��,, � � � � b t a t b t a ttpb sspptcl � ��,, (12) in other words, a distinction is made for transactions costs associated with the lbc condition for calls, which uses the bid-ask spread for calls, and conversely for puts. in both cases, the two arbitrage pro� t functions using tcl c involve addition of $26.25 per contract to tcb tcl ,, and tpbtcl ,, . 4. data this study uses intra-day price quotes for the british pound, swiss franc and euro options traded on the philadelphia stock exchange (phlx). these data are obtained from reuter’s database through sirca for the trading period from 1 august 2005 to 31 july 2006. each intra-day observation contains the options price (bid and ask), strike price, maturity date and trading time indicated with a time-stamp. put and call data are matched at � ve minute intervals. for example, put-call pairs traded at 09:05, 09:10, 09:15 and so on are included in the sample. the structured query language (sql) was used to obtain the synchronised putcall pairs as given in tables 1 and 2. due to signi� cant differences in trading volume, the number of pair-wise put-call observations varies for each currency. for the british pound, a ‘small sample’ of 618 synchronized put-call pairs were obtained from 13,414 and 285,211 observations for calls and puts, respectively. for the swiss franc, there are 41,212 put-call pairs and a ‘large sample’ of 197,815 put-call pairs for the euro. available option maturities are for 3, 6, 9, and 12 months. the expiration months are march, june, september and december. if the expiration month has 5 fridays, the options expire on the third friday, otherwise it is the second friday of the expiration month. the option contract size was £10000, sf10000 and £10000 for british pound, swiss franc and euro, respectively. signi� cantly, the daily closing spot fx rate (bid and ask) and daily closing eurodollar and eurocurrency interest rates (bid and ask), obtained from datastream, were employed instead of the corresponding time dated spot fx rate and interest rates. while the prices for put and call pairs were time synchronized, because the fx and interest rates used to determine the pcp 6 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 1 http://epublications.bond.edu.au/ijbf/vol7/iss2/1 in� uence of transaction costs on foreign exchange option contracts: intra-daily tests: 1-18 7 deviations were not synchronized, the deviations did not necessarily represent actual arbitrage opportunities. however, due to phlx market organization and trading conventions used to price and trade phlx foreign exchange options,2 the use of synchronized data for rates would produce deviations that are primarily due to noise arising from, say, dif� culties in determining actual bid and ask prices at which trades could actually be executed. descriptive statistics for spot fx rates and interest rates based on daily data are presented in table 3. for the british pound, the mid-spot fx rate mean, i.e., [(spot bid mean + spot ask mean)/2] was $1.7857, slightly higher than the strike price mean reported in table 2 by $0.0857 (= $1.7857 – $1.7000). in contrast, the difference between the mid-spot price mean and strike price mean is $0.0069 and $0.0183 for swiss franc and euro, respectively. this indicates that, on average, these put-call option pairs are typically trading at or near the money. table 1: synchronized put-call data* options sample period option type nob synchronized put-call pairs british pound 01/08/2005 31/07/2006 call 13,414 618 put 285,211 swiss franc call 583,230 41,212 put 897,258 euro call 3,397,196 197,815 put 3,547,101 * the synchronized put-call pairs are obtained as the matching puts and calls data for same trading date and time, maturity date and strike price at an interval of 5 minutes during trading hours. 2. the phlx demutualized in 2003 and, circa sept. 2005, was owned by 505 shareholders. investment banks and brokerage � rms collectively owned 45 percent of the phlx, with an option to purchase additional ownership. merrill lynch, citadel derivatives group and morgan stanley, each paid $7.5 million for 10 percent of the phlx. citigroup, credit suisse first boston and ubs each paid s3.75 million for 5% (scotti, 2005). each partner had the option to double its ownership if it delivered a certain level of order � ow to the exchange. the amount of order � ow sent to the phlx determined the cost to exercise its option to double ownership. during the sample period examined in this study, high foreign currency option volume was generated by the trade through protection provided by reg nms that dictated when the best quoted price in the national market system was from phlx, that quote would have priority over all others in the nms. the phlx was purchased by the nasdaq group in 2008. 7 hoque et al.: influence of transaction costs on foreign exchange option contrac produced by the berkeley electronic press, 2010 8 the international journal of banking and finance, vol. 7. number 2: 2010: 1-18 table 2: descriptive statistics: call, put and strike prices, intra-daily data* currency statistical measures call put strike pricebid ask bid ask british pound mean 5.67 5.93 4.17 4.42 170 median 5.71 5.96 4.15 4.40 170 skewness 0.12 0.12 -0.02 -0.02 na kurtosis 3.27 3.27 2.48 2.48 na jb 3.36 3.36 6.94* 6.94* na swiss franc mean 1.58 1.62 1.44 1.51 77.83 median 1.38 1.44 1.27 1.3 78.00 skewness 0.40 0.40 0.40 0.39 -0.41 kurtosis 2.06 2.05 2.07 2.06 2.12 jb 2.64e3* 2.63e3* 2.57e3* 2.56e3* 2.43e3* euro mean 2.58 2.71 2.57 2.70 124.06 median 2.51 2.66 2.47 2.62 123 skewness 0.42 0.38 0.40 0.37 0.34 kurtosis 3.22 3.14 2.76 2.72 2.35 jb 6.23e3* 5.03e3* 5.82e3* 5.16e3* 7.23e3* * the jarque-bera (jb) statistic follows a chi-square distribution with 2 degree of freedom. the critical value of the chi-square distribution is 5.99 at the 5% level of signi� cance. the statistical signi� cance level at 5% is denoted by *. na denotes not applicable. table 3: descriptive statistics: spot exchange and interest rates, daily data* currency statistical measures spot foreign exchange rate interest rate bid ask bid ask british pound mean 1.7857 1.7861 4.6426 4.6915 median 1.7729 1.7733 4.6000 4.6500 skewness 0.5269 0.5268 0.5307 0.5315 kurtosis 2.0847 2.0842 1.9986 2.0115 jb 21.1857* 21.1930* 23.1567* 22.9141* swiss franc mean 0.7849 0.7854 1.4276 1.4751 median 0.7812 0.7816 1.3900 1.4400 skewness 0.4695 0.4689 -0.2977 -0.3025 kurtosis 2.0359 2.0357 1.9851 1.9835 jb 19.6959* 19.6775* 15.0576* 15.2158* (continued) 8 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 1 http://epublications.bond.edu.au/ijbf/vol7/iss2/1 in� uence of transaction costs on foreign exchange option contracts: intra-daily tests: 1-18 9 currency statistical measures spot foreign exchange rate interest rate bid ask bid ask euro mean 1.2222 1.2225 2.8462 2.8763 median 1.2132 1.2135 2.8600 2.8900 skewness 0.4119 0.4118 -0.1912 -0.1916 kurtosis 2.0136 2.0136 1.8516 1.8519 jb 17.9602* 17.9580* 15.9329* 15.9331* u.s. dollar mean 4.9169 4.9469 median 4.8800 4.9100 skewness -0.1792 -0.1792 kurtosis 2.0718 2.0718 jb 10.7666* 10.7666* * see notes to table 2 5. empirical results for deviations this section presents empirical results for lbc and pcp deviations for different categories of transaction costs. the number of lbc deviations is reported in table 4 and summarized in table 5. the mean pro� t ($) per contract for individual currencies is calculated as the average of the arbitrage pro� t ($) for each lbc deviation, multiplied by the contract size. similarly, mean pro� t ($) for all currencies is calculated as summation of total arbitrage pro� t ($) for each currency (british pound, swiss franc and euro) divided by total number of pcp deviations for all currencies. as can be seen in the minimum transactions cost case (panel a), the lbc deviations for british pound, swiss franc and euro, respectively, are: for call options are 0, 0.45 percent, and 0.53 percent and, for all currencies jointly, 0.52%; for put options, lbc deviations represent 0, 2.56, and 2.69 per cent and, for all currencies taken together, 2.64%. consistent with shastri and tandon (1985), the results for all currencies jointly indicate that lbc deviations in put prices are higher than those for call prices. for all currencies taken together, the average pro� t for both call and put options is about $us25 (or about 1.5 percent) per contract due to the deviation from the lbc for calls and puts. comparing these results with round-trip bid-ask spreads (b) and total transactions costs (c), the number and percentage of lbc deviations is progressively reduced with the ratio of % put deviations to % call deviations staying relative constant at approximately � ve times more put deviations than for calls. table 3: descriptive statistics: spot exchange and interest rates, daily data* 9 hoque et al.: influence of transaction costs on foreign exchange option contrac produced by the berkeley electronic press, 2010 10 the international journal of banking and finance, vol. 7. number 2: 2010: 1-18 table 4: lbc tests for call and put option price* options on lbc violation and arbitrage pro� t ($) per contract call option price put option price no of violation violation % mean pro� t ($) no of violation violation % mean pro� t ($) a: with one-way bid-ask spreads british pound 0 0 0 0 0 0 swiss franc 187 0.45 17.70 1053 2.56 13.46 euro 1055 0.53 25.24 5278 2.69 29.25 all currencies 1242 0.52 23.10 6331 2.64 26.62 b: with round trip bid-ask spreads british pound 0 0 0 0 0 0 swiss franc 119 0.29 14.75 487 1.18 13.20 euro 633 0.32 24.42 3269 1.65 26.25 all currencies 752 0.31 22.89 3756 1.57 24.56 c: with total transaction costs british pound 0 0 0 0 0 0 swiss franc 19 0.05 10.43 73 0.18 14.65 euro 248 0.13 20.50 1356 0.69 21.88 all currencies 267 0.11 19.78 1429 0.60 21.51 * the sample size across currencies is 618, 41212, 197815 and 239645 for british pound, swiss franc, euro and all currencies, respectively. table 5: comparison of lbc tests results* options on sample size percentage of lbc violation call option price put option price panel a panel b panel c panel a panel b panel c british pound 618 0 0 0 0 0 0 swiss franc 41212 0.45 0.29 0.05 2.56 1.18 0.18 euro 197815 0.53 0.32 0.13 2.69 1.65 0.69 all currencies 239645 0.52 0.31 0.11 2.64 1.57 0.60 * panel a, panel b and panel c represent the situations of lbc tests in absence of transaction costs, in presence of bid-ask spreads and in presence of total transaction costs. the sample size across currencies is 618, 41212, 197815 and 239645 for british pound, swiss franc, euro and all currencies, respectively. 10 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 1 http://epublications.bond.edu.au/ijbf/vol7/iss2/1 in� uence of transaction costs on foreign exchange option contracts: intra-daily tests: 1-18 11 examination of deviations from the lbc condition provides an initial check on the quality of the pcp deviations being examined. by introducing moneyness and the relationship between the prices of put call pairs, pcp provides a more demanding arbitrage condition than lbc. given this, the impact of transactions costs on the number of pcp deviations is reported in table 6 and summarized in table 7. using minimum transaction costs, the percentage of pcp deviations due to overpricing of call is 0.32, 39.61, and 33.97 per cent for british pound, swiss franc and euro, respectively. for all currencies jointly, the corresponding number is 34.86%. due to overpricing of puts, the pcp deviations represent 53.69, 21.40, and 23.75% for british pound, swiss franc and euro, respectively. for all currencies, the corresponding number is 23.43%. overall, calls tend to be overpriced more frequently than the puts. these results are in agreement with el-mekkaoui and flood (1998), wagner et al. (1996), and berg et al. (1996). when round-trip bid-ask spreads are used to measure transaction costs (b), the percentage of pcp deviations drops dramatically for all currencies. using total transaction costs (c), pcp deviations fall almost to the level of lbc violations. table 6: pcp tests for conversion and reversal strategy* options on pcp violations and arbitrage pro� t conversion strategy reversal strategy no of violation violation % mean pro� t ($) no of violation violation % mean pro� t ($) a: minimum transaction costs british pound 2 0.32 9.96 338 53.69 41.74 swiss franc 16323 39.61 17.13 8820 21.40 18.78 euro 67205 33.97 28.18 46987 23.75 30.21 all currencies 83530 34.86 18.42 56145 23.43 30.24 b: with round trip bid-ask spreads british pound 0 0 0 117 18.93 36.57 swiss franc 6291 14.26 16.15 3587 8.70 17.53 euro 25,632 12.96 28.24 20122 10.17 24.89 all currencies 31,923 13.32 25.85 23826 12.60 23.84 c: with total transaction costs british pound 0 0 0 75 12.14 22.51 swiss franc 1,183 2.87 12.11 871 2.11 13.76 euro 10,785 4.45 23.95 8332 4.21 20.19 all currencies 11,968 4.99 22.78 7411 3.09 24.54 * the sample size is 618, 41212, 197815 and 239645 for british pound, swiss franc, euro and all currencies, respectively. the mean pro� t ($) is per options contract. 11 hoque et al.: influence of transaction costs on foreign exchange option contrac produced by the berkeley electronic press, 2010 12 the international journal of banking and finance, vol. 7. number 2: 2010: 1-18 table 7: comparison of pcp tests results* options on sample size percentage of pcp violation conversion strategy reversal strategy panel a panel b panel c panel a panel b panel c british pound 618 0.32 0 0 54.69 18.93 12.14 swiss franc 41,212 39.61 14.26 2.87 21.40 8.70 2.11 euro 197,815 33.97 12.96 4.45 23.75 10.17 4.21 all currency 239,645 34.86 13.32 4.99 23.43 12.60 3.09 * panel a, panel b and panel c represent pcp tests situations in absence of transaction costs (a), in presence of bid-ask spreads (b) and in presence of total transaction costs (c), respectively. the sample sizes are 618, 41212, 197815 and 239645 for british pound, swiss franc, euro and a currencies, respectively. decomposition of deviations to determine whether pcp deviations are systematically related to the moneyness of options, the pcp deviations for minimum transaction costs are decomposed into where the calls and puts are in-the money (itm), out-of-the-money (otm) and at-the-money (atm). the results under conversion strategy are summarized in table 8 and the reversal strategy in table 9. in table 8, the pcp deviations are evaluated in three states. in state 1, calls and puts are itm and otm, respectively. in this situation, pcp deviations are created by the difference between call bid price and put ask price being larger than the difference of present value of spot fx ask and strike price. in state 2, calls and puts are atm at same strike price; the call bid price should be equal to the put ask price. in this case, pcp is violated if call bid price is observed to be larger than put ask price. in state 3, calls and puts are written otm and itm, respectively and pcp deviations arise in three possible circumstances: case (i), where call bid price is observed to be higher than put ask; case (ii) where the call bid price is equal to put ask price; and, in case (iii) where the put ask price and call bid price difference is smaller than the difference of present value of strike and spot price. table 8: decomposition of pcp violations for conversion strategy* deviation states pcp violation british pound swiss franc euro state 1: when calls and puts are itm and otm, respectively at the same strike price, pcp is violated if call bid and put ask price difference is larger than the difference of present value of spot ask and strike price. 2 (0.32%) 2532 (6.15%) [¢0.12] 14893 (7.53%) [¢0.24] (continued) 12 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 1 http://epublications.bond.edu.au/ijbf/vol7/iss2/1 in� uence of transaction costs on foreign exchange option contracts: intra-daily tests: 1-18 13 deviation states pcp violation british pound swiss franc euro state 2: when calls and puts are atm at same strike price, pcp is violated if call bid price is larger than put ask price. 0 142 (0.34%) [¢0.20] 116 (0.06%) [¢0.13] state 3: when calls and puts are otm and itm, respectively at same strike price with, pcp is violated in the following circumstances: (i) if call bid price is higher than put ask price. (ii) if call bid price is equal to the put ask price. (iii) if put ask and call bid price difference is smaller than the difference of present value of strike and spot price. total pcp violations under situation c total pcp violations for all situations average arbitrage pro� t 0 5975 (13.50%) [¢0.19] 16468 (8.32%) [¢0.32] 0 72 (0.17%) [¢0.17] 199 (0.10%) [¢0.25] 0 0 2 (0.32) 0 7602 (18.45%) [¢0.17] 13649 (33.12%) 16323 (39.61%) $17.00 35529 (17.96%) [¢0.28] 52196 (26.38%) 67205 (33.97%) $24.40 * itm, atm, and otm represent in-the-money, at-the-money, and out-of-the-money, respectively. see notes to table 7. the percentage of pcp deviations and amounts of option overpricing (in us cents) are given in the parentheses, and brackets respectively. the results in table 8 indicate that pcp deviations in state 1 are 0.32, 6.15, and 7.53 percent for the british pound, swiss franc and euro, respectively. figures in squared parentheses indicate the call is overpriced on average by 0.12 and 0.24 cents for swiss franc and euro. in state 2, pcp deviations for swiss franc (0.34%) and euro (0.06%) tend to be small and no deviation is observed for british pound, not surprising given the small number of observations for this currency. in state 3, deviations are considerably larger for swiss franc (a total of 33.12%) and euro (a total of 26.38%). in the last row, the average arbitrage pro� t amount ($) is calculated as the average of the call overpriced amount ($) for each state multiplied by the contract size (reported in data section). for example, the average arbitrage pro� t of us$17 for swiss options is calculated as: (¢0.12+¢0.20+¢0.19+¢0.17+¢0.17)/100 = $0.0085/5 = $0.0017 (average) x table 8: decomposition of pcp violations for conversion strategy* 13 hoque et al.: influence of transaction costs on foreign exchange option contrac produced by the berkeley electronic press, 2010 14 the international journal of banking and finance, vol. 7. number 2: 2010: 1-18 10,000 (contract size) = $17.00. average pro� ts for swiss franc and euro are close to the mean pro� t ($) per contract under the conversion strategy reported in panel a of table 6. in state 3, otm calls have a higher overpricing tendency than the itm counterparts in state 1. for the swiss franc, the magnitude of overpricing for otm calls is in the order of more than � ve times that for itm calls. the results for the euro were similar. in market equilibrium, the overpricing tendency of otm calls is a re� ection of option market speculators wanting to make a leveraged speculation by buying foreign currency options at lower premium cost which creates trading opportunities for option writers to offer higher otm call ask prices than indicated by pcp. results for put prices, given in table 9, provided a different picture: it is itm puts that have a higher tendency to be overpriced than otm puts. for swiss franc options, the percentage of observations with overpricing, i.e. pcp deviations, for itm puts is almost twice that of otm puts. the results for euro are similar. in market equilibrium, the itm put option overpricing tendency re� ects the option purchaser’s preference for trading itm put option to obtain effective downside currency protection. this is consistent with the use of put options for insurance in currency risk management, e. g. poitras (2002, p. 523). option writers provide the needed liquidity by quoting higher put ask prices than indicated by pcp. this result is consistent with the well-known volatility smile arising from the impact of moneyness on the variation of implied volatilities. table 9: decomposition of pcp violations for reversal strategy* deviation states pcp violations british pound swiss franc euro sate 1: when puts and calls are itm and otm, respectively at the same strike price, pcp is violated if put bid price and call ask price difference is larger than the difference of present value of strike and spot exchange rate. 0 5701 (13.83%) [¢0.16] 29884 (15.11%) [¢0.28] state 2: when puts and calls are atm at same strike price, pcp is violated if put bid price is larger than call ask price. 0 0 20 (0.10%) [¢0.12] state 3: when puts and calls are otm and itm, respectively at same strike price, pcp is violated in the following circumstances: (i) if put bid price is higher than call ask price. 0 206 (0.50%) [¢0.44] 622 (0.31%) [¢0.69] (continued) 14 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 1 http://epublications.bond.edu.au/ijbf/vol7/iss2/1 in� uence of transaction costs on foreign exchange option contracts: intra-daily tests: 1-18 15 deviation states pcp violations british pound swiss franc euro (ii) if put bid price is equal to the call ask price. 0 13 (0.03%) [¢0.26] 32 (0.01%) [¢0.54] (iii) if call ask and put bid price difference is smaller than the difference of present value of spot and strike price . 338 (53.69%) [¢0.42] 2900 (7.04%) [¢0.22] 16429 (8.31%) [¢0.33] total pcp violations under situation c total pcp violations for all situations average arbitrage pro� t 338 (53.69%) 3119 (7.57%) 17083 (8.64%) 338 (53.69%) 8820 (21.40%) 46987 (23.75%) $42.00 $21.60 $39.20 * itm, atm, and otm represent in-the-money, at-the-money, and out-of-the-money, respectively. the sample size for british pound, swiss franc and euro is 618, 41212 and 197815, respectively. the pcp deviations and amounts of put overpricing (in us cents) are given in the parentheses, and brackets respectively. 6. econometric analysis following hoque et al. (2008), a more formal statistical analysis was employed to examine the impact of transaction costs on the validity of pcp. under the null hypothesis that pcp is valid, regression equations were estimated for the conversion and reversal strategy, respectively. each equation is estimated using all the sample data, e. g. 197,815 put-call pairs are used to estimate each of two euro regressions. after conducting stationarity tests using the standard augmented dickey-fuller (adf) and phillips-perron (pp) tests to check whether a unit root is present in the data series involved, regressions for the conversion strategy were speci� ed as: � � � � tstltatatbt rbxrbspc ��� �� �� � � � �� *10 (13) regressions for the reversal strategy were also estimated using: � � � � tstbtltatbt rbsrbxcp ��� �� �� � � � �� *10 (14) where the equation errors are speci� ed to accommodate serial correlation and garch effects. a table 9: decomposition of pcp violations for reversal strategy* 15 hoque et al.: influence of transaction costs on foreign exchange option contrac produced by the berkeley electronic press, 2010 16 the international journal of banking and finance, vol. 7. number 2: 2010: 1-18 lagrange multiplier (lm) test was conducted for the presence of garch(r,s) errors in _t . once the presence of garch errors was con� rmed by the lm test, an appropriate lag order was determined by further diagnostic tests. the regressions for the conversion and reversal strategies were initially estimated using ordinary least squares regression. the regression equations were then re-estimated, incorporating adjustments for serial correlation and garch effects. with appropriate error term adjustments, the estimate of � 0 is not statistically different than 0 for any currency, while�� 1 is statistically less than 1 being: .980 [.00051] for the swiss franc; and, .974 [.00029] for the euro with [coef� cient standard errors] adjusted for serial correlation and garch effects. as with the conversion strategy results,�� 0 = 0 cannot be rejected at any reasonable signi� cance level. the estimates of � 1 are again statistically less than 1 being: .991 [.00041] for the swiss franc; and,. 967 [.00034] for the euro. to interpret these results observe that the independent variable is, more-orless, the moneyness of the options. the zero intercepts indicate that bid prices are systematically higher than warranted by pcp, likely to account for market marker pro� ts. to see this, consider atm options in the conversion trade. because the bid price is below the ask for a given option, the use of ab pc � as the dependent variable subtracts the higher ask price from the lower bid price. because atm means the independent variable will be more-or-less zero, it is expected that the intercept would be negative in this case; just as a positive intercept would be expected if ba pc � were used as the independent variable. interpretation of the regression slopes being slightly less than 1 is more complicated. given that most options are short dated, and foreign and domestic interest rate levels are not substantially different, symmetric pricing with respect to moneyness for puts and calls implies a coef� cient equal to 1 when the intercept is zero. because all the sample data were used in each regression, the estimated value cannot be attributed to the convex impact of moneyness on option pricing arising from a bias in the selection of speci� c put-call pairs for the sample. all the sample data is used for each regression. recalling that the results in tables 8 and 9 indicated overpricing of otm calls and itm puts, the regression results extend these observations to the full sample where the number of deviations for the conversion strategy outnumber the reversal deviations (e.g. 67,205 to 46, 987 for the euro). this provides a larger (smaller) number of observations where it is most likely that otm calls (itm puts) will be over-priced. 7. conclusions this paper empirically examined two distribution free properties for currency options traded on philadelphia stock exchange (phlx) from 1 august 2005 to 31 july 2006. the two fundamental no-arbitrage conditions are a lower boundary condition (lbc) and put-call parity (pcp). deviations from arbitrage conditions were investigated using alternative measures of transaction costs. 16 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 1 http://epublications.bond.edu.au/ijbf/vol7/iss2/1 in� uence of transaction costs on foreign exchange option contracts: intra-daily tests: 1-18 17 three measures of transaction costs were considered: (i) minimum transaction costs that involve only initial bid-ask spreads, (ii) transaction costs associated with trades closed out prior to expiration; and (iii) a total transaction costs measure. even using minimum transaction costs calculated with closing spot fx and interest rate quotes, less than 1% (3%) of call (put) option prices involve deviations from the lbc. the lbc deviations are measured in fractions of a percent as alternative, more expensive, measures of transaction costs are employed. in contrast, pcp deviations for the conversion and reversal trades are 33.86% and 23.43%, respectively, for all currencies jointly. using round trip transactions costs, the pcp deviations are reduced on average, to 13.32% and 12.60% percent, respectively; reducing to 4.99 and 3.09 percent of all put-call pairs when total transaction costs are taken into account. author information: the submitting author is professor geoffrey poitras of simon fraser university, canada. he may be contacted at poitras@sfu.ca. the co-authors of the paper are respectively teaching staff ariful haque of university of southern queensland and assoc professor meher manzur of curtin university, meher.manzur@cbs.curtin.edu.au. references berg, e., brevik, t., and saettem, f. (1996). an examination of the oslo stock exchange options market. applied financial economics, 6, 103-113. bodurtha, j., and courtadon, g. (1986). ef� ciency tests of the foreign currency options market. journal of finance, 41, 151-162. bodurtha, j., and courtadon, g. (1995). probabilities and values of early exercise: spot and futures foreign currency options. journal of derivatives, 3, 57-75. demsetz, h. (1968). the cost of transacting. quarterly journal of economics, 82, 33-53. el-mekkaoui, m., and flood, m. (1998). put-call parity revisited: intra daily tests in the foreign currency options market. journal of international financial markets, institutions and money, 8, 357-376. garay, u., m. ordonez and gonzalez, m. (2003). tests of the put-call parity relation using options on futures on the s&p 500 index. derivatives use, trading & regulation, 9, 259-280. gould, j., and galai, d. (1974). transactions costs and the relationship between put and call prices. journal of financial economics, 1, 105-129. hoque, a., chan, f. and manzur, m. (2008). ef� ciency of the foreign currency options markets. global finance journal, 19, 157-170. phillips, s., and smith, c. (1980). trading costs for listed options. journal of financial economics, 8, 179-201. poitras, g. (2002). risk management, speculation and derivative securities. new york: academic press. 17 hoque et al.: influence of transaction costs on foreign exchange option contrac produced by the berkeley electronic press, 2010 18 the international journal of banking and finance, vol. 7. number 2: 2010: 1-18 poitras, g., veld, c. and zabolotnyuk, y. (2009). european put-call parity and the early exercise premium for american currency options. multinational finance journal, 13, 39-54. scotti, m., (2005 september 20). “phlx gains new life”. traders. shastri, k., and tandon, k. (1985). arbitrage tests of the ef� ciency of the foreign currency options market. journal of international money and finance, 4, 455-468. stoll, h. (1989). inferring the components of the bid-ask spread: theory and empirical tests. journal of finance, 44, 115-134. vipul, (2008). cross-market ef� ciency in the indian derivatives market: a test of put–call parity. journal of futures markets, 28, 889-910. wagner, d., ellis, d., and dubofsky, d. (1996). the factors behind put-call parity violations of s&p 100 index options. financial review, 31 535552. 18 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 1 http://epublications.bond.edu.au/ijbf/vol7/iss2/1 recommended citation international journal of banking and finance 8-1-2010 influence of transaction costs on foreign exchange option contracts: intra-daily tests ariful hoque meher manzur geoffrey poitras multi-country study of bank credit risk determinants 135the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152 ijbf multi-country study of bank credit risk determinants nor hayati ahmad and mohamed ariff university utara malaysia and bond university, australia abstract this paper presents fresh findings about key determinants of credit risk of commercial banks in emerging economy banking systems compared with developed economies. australia, france, japan and the us represent developed economies; emerging economies are india, korea, malaysia, mexico and thailand. credit risk theories and empirical literature suggest eight credit risk determinants. we find anywhere from two to four factors are alone significantly correlated with credit risk of any one banking system. regulatory capital is significant for banking systems that offer multi products; management quality is critical in the cases of loan-dominant banks in emerging economies. contrary to theory or studies, we find leverage is not correlated with credit risk in our test period. data transformations and statistical corrections ensured these results are reliable: model robustness was tested using aic. the model developed here could be applied to test more emerging economy banking systems to generalize our findings to other economies. jel classification: f28 key words: bank risk management; credit risk; regulatory capital; emerging economies. 1. introduction a survey of the literature shows that, among the studies of bank credit risk determinants, there are currently few publications on (a) multi-country comparisons that (b) incorporate a multi-factor modeling to study credit risk determinants. studies on bank credit risk are mostly in a single country setting; secondly, much attention is directed to developed country banks. this paper presents a comparative study of all factors contributing to credit risk of commercial * editorial note: this paper was one of three best papers selected by the previuos editor of ijbf at the fma-asian finance association’s 2005 17th conference. the co-author of this paper commenced duties as co-editor of ijbf in 2007, and did not have any part in the selection/review of this paper. the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152136 banks in a multi-country setting: australia, france, japan and the u.s. represent developed economies while emerging economy banking systems are represented by india, korea, malaysia, mexico and thailand. the findings in several studies lend support to the theory that changes in international and national macroeconomic variables, which are systematic in nature, and a set of unsystematic bank-specific factors influence the formation of bank credit risk (see demirguckunt, 1989; corsetti, pesenti and roubini, 1998; neal, 1996; ariff and marisetty, 2001, bikker and metzemakers, 2004; cebenoyan and strahan, 2004, and kraft and jankov, 2005). we take these findings as our starting point for our research process, and report significant new findings in this paper. this study is limited to the bank-specific variables except for few market-wide factors. among the factors used to identify risk, non-performing loan (npl) or impaired loan is a factor that has received central focus in the analysis of how credit risk built-up after the 1997 asian financial crisis (takayasu et al., 2000) in the affected banking systems. this variable is widely used as credit risk. table table 1: non-performing loans of financial systems . non-performing loan is defined as the percentage of loan values that are not serviced for three or more months. banking crises were identified in each country. 137the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152 1 illustrates this: npl in developing economies is almost 10 times higher compared to that in the u.s. banking system or 5 times higher than that of japanese and korean banks. among external factors, economic downturn is a significant determinant of credit risk (fisher, gueyie and ortiz, 2000; ahmad, 2003; bikker and metzemakers, 2004 and kraft and jankov, 2005). during such downturns, the quality of bank assets is likely to deteriorate, which would increase the risk of the bank, and it could require an increase in capital requirements, which if the banks are weaker, are very costly or may simply be unobtainable (bikker and meltzemakers, 2004) during a period of crisis. most banks in economies such as thailand, indonesia, malaysia, japan and mexico experienced high nonperforming loans and significant credit risk rises during financial and banking crises, which resulted in the closing down of several banks in indonesia and thailand. in the case of malaysia, no bank was closed down but the weaker banks were required to merge with others to strengthen their capital base, which was eroded by huge accumulated losses due to impairment of loans. subsequently, stronger commercial banks (which were later known as the ten anchor banks) were required by the central bank to acquire these banks. the exercise involving injection of new experienced top management resulted in a massive restructuring of the entire banking system in 1999-2000 (ahmad, 2003). although a larger number of banks in developing (in particular, asean) economies are worst hit by credit risk in the test period compared to their developed counterparts, not much research attention has been given to the factors responsible for high credit risk in these countries in that period. most existing studies of credit risk determinants were those relating to developed markets (see hassan et al., 1994; gallo et al., 1996; galloway et al. 1997; fischer et al., 2000; bikker and meltzemakers, 2004, jimenez and saurina, 2004; kraft and jangkov, 2005). other studies on changes in risk premiums were conducted in the u.s. (see elyasiani and mansur, 1998) and ariff and marisetty, 2001) using a multi-country risk premium modeling of firms in 22 oecd countries. this study has opted for a wider approach to this research issue by investigating all factors contributing to credit risk across two sets of countries, developed and developing. we also develop a more comprehensive credit risk model by incorporating a different and comprehensive set of variables. the data set contains 23,499 bank years of observations covering nine countries over a crisis-prone period, 1996 to 2002. this multi-factor and multi-country approach allows us to compare and then identify whether there exists major differences in credit risk determinants across countries in different developmental stage. the rest of the paper is organised as follows. section 2 contains a review of theory and related literature. section 3 describes the methodology and that is followed by an analysis of results in section 4. section 5 concludes the paper. the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152138 2. literature review on credit risk and determinants 2.1 theory this study is conducted against the backdrop of three main theories on risk: sharpe’s (1963) capital asset pricing theory (capm), markowitz’s (1959) portfolio theory and hamada’s (1972) theory on risk and leverage. the capm suggests that price or expected return of an asset is related to its risk-free rate, the systematic risk and the expected risky market’s risk-premium. hence, e(r j ) = r f +β j [e (r m -r f )] (1) where, e(r j ) : the expected return on the asset j = 1, …., n, r f : the risk-free rate of return measured as treasury bill/bond yield, r m : the expected return for the risky market portfolio, βj : the individual asset’s systematic risk relative to the risky mar ket’s portfolio, and e (r m -r f ) : the expected risk-premium of risky market portfolio. applied to a portfolio of bank loans, bank managers may be said to maintain a combination of loans with varying risk levels. the portfolio of bank loans would have to be such that the overall risk of the loans is diversified given the covariance of returns from each pair of loans is likely to be such that the correlation coefficient is closer to 0 rather than exactly equal to +1.00 as suggested by the markowitz for diversifying individual asset’s (loan) risk. equation 1 denotes that, for risky loans, bank managers would charge a premium equal to the difference between the overall risk-premium applicable in the ‘market’ for all the loans in the market (r t -r mt ) in order for the banks to compensate themselves for the additional risk of a particular loan. in essence, the capm infers that the required rate of return demanded by banks is equal to a risk-free rate (equivalent to the yield of a treasury instrument with the same term as that of the loan) plus a premium as determined in the market for the total loan holding. accordingly, a bank will price its loans according to the level of risk perceived whereby a perceived higher (lower) risk loan will attract a higher (lower) price for lending. since individual loan risk can be eliminated through loan diversification as stated by markowitz, the risk that a bank is concerned with in its loan-pricing decision is the market risk and the price of risk is: price of risk = unit of risk x risk-premium = β j [e (r m )-r f ] (2) as a highly-levered firm, a bank has to incorporate in its loan pricing, other riskrelated costs for example, tax and bankruptcy cost. bankruptcy costs will arise if 139the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152 a bank indulges in excessive risk-taking lending or trading. thus, risk increases if debt to equity ratio increases for a bank. incorporating the leverage effect (hamada op cit.) into equation (2), beta of an individual bank, j, with loan capital can be expressed as: ß jl = ß ju [1 + d/e j (1-tc) ] (3) where, ß jl = beta of levered firm, ß ju = beta of unlevered firm, d/e = debt-equity ratio, tc = tax rate applicable to income streams of a bank. a banking firm’s risk increases as leverage increases. thus, capital structure is likely to affect credit risk. 2.2 review of empirical literature two strands in the literature on bank credit risk continue to receive increasing attention. one strand appears to suggest internal variables as potential determinants of risk measured as the unsystematic risk (see hassan, 1993; brewer et al., 1996; gallo et al., 1996; berger and deyoung, 1997; angbazo, 1997). the other strand highlights changes in external variables in the financial markets, regulations and economic conditions as affecting the systematic bank risk (hassan et al., 1994; corsetti et al., 1998). both streams provide evidence of significant relationships among the internal variables, external factors and bank risk. chang (2001) claims that there exists a confusion as to what causes crises in emerging markets and states that the knowledge about crises in emerging markets is inadequate for analyzing risk in recent banking crises. he cites that one of the causes was the maturity mismatch: short-term international liabilities as shown by short-term foreign debts exceeding short-term international reserves. table 2 provides statistics relating to the maturity position of five asian economies. the statistics suggest that indonesia and south korea had this mismatch providing a huge build-up of risk as far back as 1990. june 1997 saw thailand facing a very high risk in this regard, and the market participants’ trading action, given this very dangerous development, led to a fall in currency first in thailand, then korea and elsewhere as contagion. thus was born the 1997-8 asian crisis. kraft and jankov (2005) find that in croatia, rapid loan growth increased the probability of credit quality deterioration and perpetrated the current account table 2: asean 5: short-term foreign debt/international reserves the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152140 and foreign debt problems. rapid loan growth takes place during economic boom times and such lending booms have been identified as a significant factor that increases the risk of a crisis (caprio and klingebiel, 1996). ariff and marisetty (2001) indicate that gdp is negatively related to bank risk in a study covering oecd and asian countries. similar negative relationship between gdp and credit risk of commercial banks was also obtained by fischer, gueyie and ortiz (2002) in the u.s. and canada. gdp is also significantly negatively related to credit risk of banks in malaysia (ahmad, 2003). as to bank-specific credit risk determinants, a recent study by jimenez and saurina (2004) in spain show that collateralised loans have a higher probability of default .they find that banks, whose probability of loan defaults is protected by collateral, have less incentives to undertake adequate screening and credit assessment at the time of loan approval. type of loan is also a significant determinant of credit risk where loans given by saving banks are riskier and are prone to default. a close bank-to-borrower relationship is also found to be significantly positively related to credit risk where this relationship increases the banks’ willingness to take more credit risk by lending to riskier firms. galloway et al. (1997) used standard deviation of weekly share returns as a proxy for bank risk while market to book value, operating leverage, capital, real size and highrisk, a dummy variable, were five independent criterion variables used. the study shows that capital is significantly negatively related to risk. size (measured by market value of equity) is positive and is significantly related to risk during the regulatory period but is negatively significant during a deregulatory regime when restrictions were imposed on banks following the failure of too-bigto fail policy. in contrast, gallo, apilado and kolari (1996) find no significant relation between leverage and risk. berger and deyoung (1997) find lagged capital (measured by equity capital to total assets) shows mixed results for different types of banks. in the case of thinly capitalized banks, the coefficient of lagged capital is significantly negatively related to credit risk. this finding supports the moral hazard hypothesis, and suggests that, on an average, thinly capitalized banks take more risky loans, which potentially could lead to higher problem loans. however, for all-banks sample, lagged capital coefficient is positive and significant. the researchers suggest that a possible reason could be that banks may raise the capital in advance to provide a cushion against possible loan loss arising from npl increases. shrieves and dahl (1992) also find similar result but cummins and sommer (1995) discover an inverse relationship between capital asset ratio and credit risk. this finding suggests that better capitalized banks take more risk. further evidence of the impact of credit risk on bank performance is shown by angbazo (1997). in this study, banks with a larger risky loan portfolio appear to require higher net interest margin to compensate the higher risk of default. this result suggests the significant relationship between net interest margin and 141the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152 credit risk. the findings of hassan, et al. (1994) support the earlier findings that size and diversification are negatively related to risk. brewer et al. (1996) find that loan sectors have a significant association with risk. fixed-rate mortgage loans, investment in service corporations and real estate loans are found to be significantly negatively related to credit risk. however, non-fixed rate mortgage loan is significant and positively related to risk. bikker and metzemakers (2004) investigated how bank provisioning behavior is related to business cycle. they find that banks make substantially higher provisions against potential loan loss or higher credit risk when gdp growth is low. thus, higher provision indicates higher credit risk potentials. all macroeconomic variables tested (gdp growth, loan growth, and bank specific factors such as loan loss provision to total assets, earnings before tax and provision, loan to total assets and capital to total assets) are significant determinants of loan loss provisions. ahmed et al. (1998) find loan loss provision (llp to total asset) to be positively significantly associated with npl. hence, a higher llp indicates an increase in credit risk and deterioration in loan quality. fisher, gueyie and ortiz’s (2002) study of banks in nafta countries finds similar results with llp positively related to risk despite being in a different economic setting or stage of development or banking system, the malaysian banks exhibit similar results where their llp is positively and significantly related to credit risk (ahmad, 2003). fisher, gueyie and ortiz (2002) find that, in canada and mexico, leverage is significantly positively related to bank risk . in contrast, galloway, lee and roden, 1997 find mixed results for the u.s. operating leverage is positively related to risk in pre-deregulatory period; negatively in de-regulatory periods. however, it is significantly positively related to risk in post-deregulatory or re-regulation periods. fisher, gueyie and ortiz (2002) and hassan et al. (1994) find that bank size (measured by natural logarithm of total assets) is significantly negatively related to risk of u.s. banks. 3. data, hypotheses and methodology 3.1 data we use cross-sectional data of individual bank balance sheet and income statement items of commercial banks of selected countries. besides being in different economic settings, these countries are chosen because they operate under different banking systems and market structures. the data for the analysis is obtained from the bankscope data stream at the monash university. we employ data of commercial banks only to obtain a more homogeneous group of financial institutions. the annual date pertain to the years in the study period stated previously. the major hypothesis is that the literature-based eight factors will be highly correlated with credit risk of banks in each of the nine economies, more so in the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152142 emerging economies. next, we predict (since ours is large comparative study with all factors) that only a sub-set of the eight factors will be significant in a given economy. third, we expect more factors in emerging economies and fewer factors in developed economies will be relevant to risk. 3.2 variables the dependent variable is changes in impaired loan (or non-performing loans) to total gross loans as a measure for credit risk. the independent variables are bank-specific factors are: management efficiency, loan-loss provision, loanto-deposit ratio, leverage, regulatory capital, funding costs, liquidity, spread and total assets. the definitions of each variable for hypotheses testing are shown in table 3. 3.3 the model the equation for the full model used in this study is: cr it = λ 0 + λ 1 mgt i,t +λ 2 llp i,t +λ 3 ld i,t + λ 4 lev i,t + λ 5 regcap i,t + λ 6 fcost i,t +λ 7 liq i,t + λ 8 spread i,t + λ 9 lnta i,t + _ i,t (4) the dependent variable, cr it , is theimpaired loans or the ratio of non-performing loans to total loans as at the end of financial year of bank j in time t. the definitions of the eight independent variables are as stated in table 3. multi-country data items are standardized by the developers of the bankscope database and, therefore, the variables are similarly defined across all countries. table 3: definitions of independent variables 143the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152 3.4 treatment on econometric issues in order to ensure that all regression assumptions are met, each variable is tested for linearity and normality: see tables 4 and 5. initial tests indicate that the variables are heteroscedastic and some variables are non-linear. following hair et al. (1998: 75), heteroscedaticity and non-linearity are corrected through data transformation. thus, all variables were transformed into first level differences to ensure that the variables are homoscedastic. correlation analysis and collinearity diagnostic were then carried out to assess the extent of multicollinearity among independent variables. this problem is detected using variance inflation factor (vif) with a cut-off threshold of 10 and tolerance value of 0.1: (see hair et al. 1998: 193). variables with large vif values are excluded. the results of collinearity diagnostics and normality tests for each country’s test model are in tables 4(a) to (f). table 5 shows no multi-collinearity problem between the risk predictors and no violation of normality assumption. this enables us to proceed with identifying the key credit risk and total risk determinants of deposit-taking institutions in the selected countries. all independent variables are regressed against credit risk proxy for the whole model. subsequently, using aic criterion within the stepwise regression procedure, a parsimonious model was identified and thus we determine key credit risk predictors for each country. all estimates are based on generalized least square regressions after corrections for heterokedasticity using white heteroskedastiity consistent standard errors and covariance technique in eviews programs. 4. analysis of results 4.1 descriptive statistics table 6 reports a summary of descriptive statistics of credit risk determinants. as for credit risk level, thailand recorded the highest credit risk of 18.66 per cent over the study period. the next highest credit risk is for banks in malaysia (10.03%). this finding appears to show that thailand and malaysia incurred credit risk level about 9 times and 5 times respectively higher than the 2 per cent, which is the international standard. the acceleration in credit risk in these two countries is also attributable to the 1997 asian financial crisis. the other countries in our sample such as korea experienced a lower level of mean credit risk of 3.5 percent and 4.5 percent respectively. mexico and france recorded higher mean credit risk level of 5.1 percent and 4.8 percent respectively. table 6 and figure 1 highlights the significant differences in the mean credit risk between the two economic regions, where lower credit risk is shown for developed economies, notably the u.s. statistics in table 6 shows that most banks both in emerging and developed countries maintain a mean proportion of earning assets to total assets (mgt) of 65 percent. this suggests that the banks had two-thirds of their asset portfolio in interest-bearing, interest rate sensitive assets. the earning assets to total assets the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152144 ratio reflects a bank’s management efficiency in managing its assets to earn interest income (angbazo, 1997). it is postulated that the higher the ratio the higher would be the management efficiency in generating interest income, and the lower is the credit risk. malaysia’s ratio of 65 per cent was within the international norm. table 4: multicollinearity diagnostic statistics 145the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152 figure 1:credit risk: multi-country comparison (1996-2002) banks in korea and thailand maintained a mean loan loss provision (llp) of about 2 percent per annum. this shows as a common practice among most emerging country banks. however, banks in france and mexico show a higher mean llp ratio of 3 percent. judging from the lower credit risk experienced by banks in these two countries, the higher llp ratio suggests a more prudent loan loss provisioning policy adopted by the french and mexican banks over the study period. the u.s. banks maintained a much better mean llp of 1.13 percent, the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152146 table 5: diagnostic statistics on normality normality test results shapiro-wilk test figure 2: bank spread: multi-country comparison 147the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152 which appears to be consistent with its low credit risk level (mean = 0.0096). malaysia and japan are noted to have a 1 percent mean llp although the mean credit risk is higher (malaysia = 10%; japan = 4%). the analysis further highlights that mexican banks earned a spread of an average of 8.96 per cent, which is the largest spread in our sample. in contrast, thailand banks recorded the smallest spread of an average 1.6 per cent during the study period. figure 2 illustrates the mean spreads of banks in some of the countries in the sample: malaysia (2.8%), japan (1.9%), korea (4.1%), thailand (1.6%), mexico (8.9%) and france (3.5%). table 6 highlights that the variable with the largest standard deviation among banks is the leverage. this is noted in malaysia, mexico, japan and france. the standard deviation in funding cost (defined as total interest expenses plus noninterest expenses to total assets) is noted also as highest among korean banks although the mean funding cost is highest for banks in mexico (20.2%). this suggests marked differences in operating costs levels among korean banks. 4.2 regression results on credit risk determinants the statistics in table 7 suggest a set of important credit risk determinants across the 9 countries based on the results prior to a search for a parsimonious model: table 8 is from the parsimonious model identification using aic. based on the r-squared statistics, the best accountability for the variations in credit risk is found for mexico, malaysia and australia with adjusted r-squared values 0.5190, 0.3427 and 0.3632 respectively. the model fit is good as indicated by the f-ratio obtained (not shown). several variables emerge as significant determitable 6: descriptive statistics of the nine variables entering the regression tests the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152148 the test model is: cr it = λ 0 + λ 1 mgt i,t +λ 2 llp i,t +λ 3 ld i,t + λ 4 lev i,t +λ 5 regcap i,t + λ 6 fcost i,t +λ 7 liq i,t + λ 8 spread i,t + λ 9 lnta i,t + ε i,t . the variables are defined in table 3. of the eight independent variables, the dependent variable, credit risk of commercial banks in the named economies, is significantly correlated with a number of risk factors. table 7: credit risk determinants, whole model, for developed and developing economies nants of credit risk of commercial banks, despite some differences in credit risk determinants across countries. mgt (denoted by earning assets to total assets ratio) is significantly related to credit risk of banks in three countries, malaysia, india and france. the positive signs of this coefficient for malaysia (1.3454 with t= 2.6485) and france (0.0627 with t = 8.2478) suggest that the higher the proportion of earning assets (which largely consist of loans) the greater is the tendency for banks in these countries to incur potentially high credit risk. conversely, mgt is significantly negatively related to credit risk of banks in india. the possible explanation for this obverse finding could be that the earning assets of the banks during the test period comprised more interest-earning assets other than loans such as investment securities and incomes from dealing in securities, which did not incur any non-performing incidence or high default rates. that is plausible given the more oligopolistic banking structure in that country with very large banks acting as allpurpose banks, although these banks are commercial banks. hence, the higher the bank’s earning assets the lower would be their credit risk. the negative relationship between mgt and credit risk supports the findings of angbazo (1997). loan loss provisions to total loans (llp) is significantly positively related to credit risk of banks in australia, japan, mexico and thailand,. the results indicate that higher llp ratio signals potentially higher credit risk as banks need to 149the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152 make greater provisions against potentially greater non-performing loans. this finding is consistent with past findings (ahmed et al. 1998; fisher, gueyie and ortiz, 2002; ahmad, 2003 and bikker and metzemakers, 2004). the loans to deposit ratios (ld) predicts that the bigger the loan portfolio relative to deposit size the higher should be the probability of loan default. ld ratio is a significant positive determinant of credit risk in malaysia, in the u.s., and france. however, contrary to theory and past evidence for different test periods, leverage (measured by total liabilities to total assets as used in many studies) is not a significant factor for credit risk formation in all countries except for mexico in this study. another contrasting finding is that we find the coefficient estimate for liquidity ratio (liq) is significantly positively related to credit risk in several countries: australia, india, korea and the us. this finding suggests that the banks with more liquid assets tend to have higher credit risk. a possible explanation for this contrary-to-theory empirics is that banks are motivated to short term assets or lend by means of short-term interest rates to hold more liquid asset (especially in periods of crises) to ward off rapidly changing interest rate markets in crisis periods while the banks may earn higher income from converting these liquid assets into cash at higher rates, the high interest rates might cause many bank borrowers to experience cash flow problems, forcing them to default their loan repayments, thus increasing credit risk for mexico, liq is significant but the coefficient is negative. the result in this situation suggests that the banks might have maintained a higher proportion of their funds in liquid assets, less fund to provide for loans. thus, the banks might have potentially lower credit risk based on a smaller table 8: credit risk determinants in developing and developed economies – parsimonious model the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152150 loan portfolio. this could also be the banks’ strategy to counter high problem loans incurred from the preceding years. spread emerged as a significant credit risk determinant. however, the evidence for the relationship between spread and credit risk in this study is mixed: negative for banks in india and thailand but positive for banks in france. this inverse relationship implies that a bank that charges higher interest to its borrower to earn higher spread might deter less qualified borrowers (who are potential loan defaulters) to borrow, thus curtailing the banks’ credit risk exposure. another key credit risk determinant common to many countries in the study is the regulatory capital. (the evidence on the impact of regulatory capital on credit risk is mixed as discovered by berger and deyoung, 1997.) for japan, malaysia, and mexico, capital is significantly positively related to credit risk. this result would have us believe that banks would be required to increase their capital as a cushion to absorb potential losses that might arise from an increase in credit risk. for our test period, this is intuitively the response expected. berger and deyoung (1997) and cummins and sommer (1995) find the same result further, and rationalized that large banks are willing to take higher risk or lend to risky borrowers for lucrative returns because they have the capability to absorb the losses as they enlarge their capital. in contrast, the results for australia and india suggest that under-capitalized banks take more risks, as the relationship between credit risk and regulatory capital is significant but negative. this finding is consistent with that provided in shrieves and dahl (1992), park (1997) and berger and deyoung (1997). 6. conclusions this study investigated credit risk determinants across two types of banking systems. although several studies on credit risk determinants have been conducted in the past, they are mostly confined to single country. eight factors as potential risk determinants of credit risk of banks were incorporated in two test models to reveal which of the eight are major contributors of credit risk in emerging economy banking systems compared to those in developed economies. our major findings can be summarized: anywhere from two to four factors are significant determinants of credit risk of any one banking system. we find regulatory capital is important for banking systems that offer multi products; management quality is critical in the cases of loan-dominant banks, example in emerging economies. contrary to theory and some earlier studies, we find leverage is irrelevant to credit risk of banks in several economies during our test period. an increase in loan loss provision consistently emerges, as in prior studies, to be a significant determinant of potential credit risk. this study also highlights that credit risk in emerging economy banks is higher than that in developed economies and that risk is formed by a larger number of bank-specific factors in emerging economies compared to their counterparts. acknowledgement: this article is an edited version of one of three best papers 151the international journal of banking and finance, vol. 5.(number 1): 2007: 135-152 selected after blind review process at the 17th fma-asian finance association conference, malaysia, 2005. we acknowledge the helpful comments of the editor in revising this paper to this published version. author statement: nor hayati ahmad is an associate professor at the university utara malaysia, sintok, kedah, malaysia. her teaching and research interests are in banking. ariff holds a chair in finance, bond university, australia, and his research interest is on asian capital markets behaviour. references ahmad, n.h., (2003). “credit risk determinants: by institutional type”, proceedings of malaysian finance association conference, 2003. ahmed, a.s., c. takeda and t. shawn 1998. bank loan loss provision: a reexamination of capital management, earnings management and signaling effects. working paper, department of accounting, syracuse university, 1-37. angbazo, lazarus, (1997). “commercial bank net interest margins, default risks, interest rate risks and off-balance sheet banking”, journal of banking and finance 2: 55-87. ariff, m. and marisetty, v.b., (2001). “a new approach to modelling multicountry risk premium using panel data test method”, proceedings of mfs conference in cyprus, 2003. berger, allen n and deyoung, robert, (1997). “problem loans and cost efficiency in commercial banks”, journal of banking and finance 21: 849-870. brewer, e.iii, jackson, w. e.iii and mondschean, t.s., (1996). “risk, regulation, and s & l diversification into nontraditional assets”, journal of banking and finance 20: 723-744. caprio, g. jr. and d. klingebiel, (1997). “bank insolvency: bad luck, bad policy and bad banking”, michael bruno and boris pleskonic, eds. annual bank conference on development economic 1996, the world bank. chang, roberto, (2001). “understanding recent crises in emerging markets”, economic review, federal reserve bank of atlanta, 2nd quarter ,georgia, u.s.a.: 6-16. cebenoyan a. sinan and philip e. strahan, (2004). “risk management, capital structure and lending at banks”, journal of banking and finance 28: 19-43. corsetti, g, p. pesenti, and n. roubini, (1998). “fundamental determinants of the asian crisis: a preliminary empirical assessment”, paper prepared for the jimf-fordham university conference on “perspectives on the financial crisis in asia”. cummins, j.d. and sommer, d.w., (1996). “capital and risk in property-liability the international journal of banking and finance 3-4 (special issue): 2005-2006: 135-152152 insurance markets”, journal of banking and finance 20: 1069-1092. demirgunc-kunt, a., (1989). “deposit institution failures: a review of the empirical literature”, federal reserve bank of cleveland, economic review, (qtr. 4). elyasiani, e. and mansur, i., (1998). “sensitivity of the bank stock returns distribution to changes in the level and the volatility of interest rate: a garch-m model,” journal of banking and finance 22:535-563. fischer, k.p., gueyie, j.p. and ortiz, e., (2000). “risk-taking and charter value of commercial banks’ from the nafta countries”, paper presented at the 1st international banking and finance conference, nikko hotel, kuala lumpur, malaysia. gallo, j.g, apilado, v.p. and kolari, j.w., (1996). “commercial bank mutual fund activities: implications for bank risk and profitability”, journal of banking and finance 22: 1775-1791. galloway, t.m., lee, w.b. and roden, d.m., (1997). “banks’ changing incentives and opportunities for risk taking”, journal of banking and finance 21: 509527. hamada, robert s., (1972). “portfolio analysis, market equilibrium and corporate finance”, the journal of finance xxiv (1): 13-31 hassan, k.m., (1993). “capital market tests of risk exposure of loan sales”, quarterly journal of business & economics 32 (1): 27. hassan, k.m, karels, g.v. and peterson, m.o., (1994). “deposit insurance, market discipline and off-balance sheet banking risk of large u.s. commercial banks”, journal of banking and finance 18: 575-593. kraft, evan and ljubinko, jankov (2005). does speed kill? lending booms and their consequences in croatia, journal of banking and finance 29 : 105121. neal, r.s., (1996). “credit derivatives: new financial instruments for controlling credit risk”. frb of kansas city”, economic review 81 (2): 15-57. park,s., (1997). “risk-taking behaviour of banks under regulation”, journal of banking and finance 21: 491-507. sharpe, w.f., (1963). “a simplified model for portfolio analysis”, management science 9: 277-93. shrieves, ronald, e and dahl, d., (1992). “the relationship between risk and capital in commercial banks”, journal of banking and finance 16: 439457. takayasu, k., and yosie, y., (2000). non-performing loan issue crucial to asia’s economic resurgence, sakura investment research: 1-6 . international journal of banking and finance 1-1-2007 multi-country study of bank credit risk determinants nor hayati ahmad mohamed ariff recommended citation cetak a4 ijbf6(2)09.indd the international journal of banking and finance, 2008/09 vol. 6. number 2: 2009: 155-165 155 market risk var historical simulation model with autocorrelation effect: a note wantanee surapaitoolkorn sasin chulalunkorn university, thailand abstract the modern market risk model using value at risk (var) method in the banking area under the basel ii accord can take different forms of simulation. in this paper, historical simulation will be applied to the var model comparing the two different approaches of geometric brownian motion (gbm) process and bootstrapping methods. the analysis will use correlation plots and examine the effects of the autocorrelation function for stock returns. keywords: basel ii accord, market risk model, var model, stochastic process, historical simulation, bootstrapping jel classifi cation: c15, c51, c88 1. introduction in all aspects of life, we often want to have the ability to predict the future with some certainty. the most reliable techniques used for forecasting the future are probability theory and statistical applications. in the banking and fi nance world, we often combine quantitative engineering techniques such as quantum physics, classical econometrics theory and operational research techniques to help with predictions of the movement of liquidity price in fi nancial markets. in the area of fi nancial market risk, the main technique used to predict the movement of prices is based on the so-called value at risk (var) model. this model has been used widely in the banking industry after the completion of the basel i accord in 1996 and ever since it was publicly introduced by j. p. morgan in 1997 (see pearson (2002) for more details). the mathematical methods behind this model are based on the stochastic process using geometric brownian motions which will be introduced in the next section. more detailed explanations of the basel accord and the market risk model can be found in surapaitoolkorn (2007). ijbf ht tp :// ijb f.u um .e du .m y 156 market risk var historial simulation model with autocorrelation effect: a note: 155-165 the main feature of the var model is the performance by calculating risk statistics using a bank’s portfolio value with a correlation method like the parametric var or simulation methods using the historical or the monte carlo (mc) techniques. two essential criteria for choosing the simulation method are the accuracy of the estimated output and the timing required to create such output (see glasserman (2004) for more details). simulation method requires fast computer machine and software technology as well as the best risk model specialist (often referred to as “fi nancial engineer”) who can handle the data applications, the stochastic process techniques, the simulation methods and who can understand the market risk drivers well. in this analysis, the historical simulation method will be used. it will be explained in section 3. one of the most important concepts in modelling market risk is the use of adequate future fi nancial market price returns. financial market data required for the estimation of var often come in terms of market prices and spread rates such as interest rates, foreign exchange (fx) rates, equity index rates, and implied volatilities. these prices and rates shall be referred to as ‘market return rates’. market price returns provided by the bank often come in daily high frequency date rates. in this quantitative analysis, the equity index rates will be used. they are introduced in section 4. in this paper, the estimation of future expected value of fi nancial assets using fi nancial market data will be analysed using the autocorrelation (acf) produced from the historical simulation of the var model. the stochastic processes involved in the var model will be introduced in the next section. the aim is to see the effects of the simulation path generated from fi nancial market data. 2. stochastic process for market risk var model there are a lot of var applications used for the calculation of risk factors in measuring fi nancial market risk. among them is mcneil et al. (2005). the purpose of using the var model is to fi nd the expected loss of a portfolio over some time horizon with a given level of probability at the maximum level using stochastic processes. in this section, we begin with the key theory behind the var model known as the stochastic process. in the world of fi nance, the uncertainty of the future can be quantifi ed using the so-called stochastic process. many genius mathematicians and scientist like gauss, wiener, levy, and ito have contributed the most extraordinary mathematical techniques such as gaussian distribution, wiener process, brownian motions, levy processes, and ito’s lemma respectively to the modern world of fi nancial banking. see mikosch (2006) for further details. all of these techniques are based on the method of stochastic process which is proven to be the most powerful tool for forecasting the future of important fi nancial applications in the banking sector. stochastic process is used in simulation methods to obtain the maximum values of the required market risk statistics. the simplest stochastic process can be defi ned using mathematical notations as ht tp :// ijb f.u um .e du .m y market risk var historial simulation model with autocorrelation effect: a note: 155-165 157 (1) where y t is the random variable which can be any market return rate; t refers to time interval frequency type of data like monthly, daily, hourly, second-bysecond (or tick-by-tick); and a is defi ned as some space in the process. a. geometric motion processes one of the most useful stochastic processes that are well used in fi nancial engineering is the geometric motion. the word geometric often refers to exponential form of mathematics. there are three useful geometric motions used in the market risk banking system. we will defi ne them in this section. in the market risk system, y t can take different forms of stochastic process depending on the type of market data returns. suppose that y t represents the stock price process at time t. in fi nancial mathematics, the stock price process y t is often of the geometric exponential form given by (2) or (3) where, x t is defi ned as the rate of log-return of the stock price. the log-return process x t can be modelled by a stochastic diffusion equation (sde), for example an ornstein-uhlenbeck (ou) process, or a levy process, for example variance gamma process. the three geometric motions are defi ned as followed: (i) geometric brownian motion (gbm) process geometric brownian motion (gbm) process is named after robert brown, the biologist whose research dates to the 1820s. it was norbert wiener (1923) who introduced brownian motion to the mathematical world. if x t is a brownian with or without drift, then equation (3) is called geometric b r o w n i a n motion (gbm). the sde for y t can be written as (4) where, z t is a standard brownian motion. in this case, y t has a log-normal distribution, since x t has a normal distribution. therefore, equation (4) is sometimes called log-normal model. in practice, we often use this geometric process for equity index rates or stock price returns. ,tttt dzydtydy attyy tt ,),( ,0 t x t eyy ),log( tt yx ht tp :// ijb f.u um .e du .m y 158 market risk var historial simulation model with autocorrelation effect: a note: 155-165 (ii) geometric ornstein-uhlenbeck (ou) process the ornstein-uhlenbeck (ou) process is named after leonard salomon ornstein and george eugene uhlenbeck, the dutch physicist whose research dates to the 1920s. it is also known as the mean reverting stochastic process. in any market risk model, the log-return x t is an ou process with dynamics sde as follows (5) or (6) where, z(t) is a standard brownian motion, α is the speed of reversion, µ is the (log) mean reversion level, and σ is the volatility. then equation (3) may be called geometric ou process. in practice we often use this geometric process for interest rates. (iii) geometric levy process the levy process is named after the french mathematician paul levy. this is the stochastic continuous time process based in probability theory. wiener process is one of the most well-known examples of the levy process which has to have stationary independent increments as well as starting the value at zero. another good example of the levy process is the poisson process. if x t is a levy process, for example x t is a variance gamma process, then a brownian motion is a special levy process, therefore (4) is also a geometric levy process. 3. historical simulation and bootstrapping method historical simulation is one of the most well-known simulation methods used for market risk modelling. in historical simulation, rather than generating random numbers from a machine, the actual past history of market data are used so as to refl ect realities like fat tails, persistence, and other common stylized facts of volatility. in historical simulation, we assume that past history is repeated and predict the future using the past data. the data used can be referred to as “nonoverlapping” or “overlapping” data. in this section, we fi rstly analyse these two types of data using the historical simulation. bootstrapping method used to help with the correlation between paths as well as the limitation of data series will be introduced at the end of this section. ,)log(()log( ttt dzdtyyd ,)( ttt dzdtxdx ht tp :// ijb f.u um .e du .m y market risk var historial simulation model with autocorrelation effect: a note: 155-165 159 (i) non-overlapping d ata for non-overlapping data, consider the following: • the 1st simulation run is to be constructed from [1,…,n] • the 2nd simulation run is to be constructed from [n+1,…, 2n] • and so on until….. • the nth / n (if n is a multiple of n) simulation run is to be constructed from [nn+1 ,…, n] in practice, more paths are generated in order to increase accuracy of the var. this is where the overlapping historical periods can be used. (ii) overlapping data for overlapping data, consider a n day var calculation (e.g. 10 days). obviously, the simulation must construct many paths composed of n observations. assume that for each market risk factor there exist n historical observations. currently the paths will be sampled as follows. • observations [1,…,n] can be used to construct the fi rst path. • observations [2,…, (n+1)] can be used to construct the second path, and so on until • the n-n+1 path is constructed from observations [(n-n+1), … , n]; so that n-n+1 (> n/n) simulation paths are generated from this overlapped data. let take a discrete example. suppose that the stock prices for the last 3 days were 100, 105.13, 111.63 and today’s stock price is 119.72, so that the daily stock returns for the past 3 days were 5%, 6% and 7%. if this data is used to run the historical simulation and predict the stock price for tomorrow and 3 days from today, we have: • at time node 1 for day 1 : 119.72 * exp(0.05) = 125.86, and • at time node 2 for day 3 : 125.86 * exp(0.06) * exp(0.07) = 125.86 * exp(0.13) = 143.33 so the rate change data used for time node 1 is 5% and that for time node 2 is 13%. for example, assume that there are 9 days of past data and we want to estimate the values for the next 3 days. from the data of days 1, 2 and 3, we can generate one path for future 3 days, from days 4, 5, 6 for the 2nd path, and from days 7, 8, 9 for the 3rd path. now, we have 3 historical simulation paths, and at each future time node of day 1, 2 and 3, we have 3 samples each so that we can compute the statistics for the values. in historical simulation, the process often allows for overlapping. this means that it is possible to use [days 1, 2, 3] for the 1st path, [days 2, 3, 4] for the 2nd path, [days 3, 4, 5] for the 3rd path, etc. this way, more simulation paths are ht tp :// ijb f.u um .e du .m y 160 market risk var historial simulation model with autocorrelation effect: a note: 155-165 generated. the only disadvantage is that the paths are correlated. the purpose of this analysis is to see if any distortion results due to this correlation. to avoid the correlation problem, (for example, 3-day data can be randomly picked from 9 days of data) the bootstrapping methods may be introduced. bootstrapping method in historical simulation method, the main problem often found is the lack of historical data. to avoid this problem, we introduce a statistical technique known as bootstrapping. bootstrapping method assumes that daily returns are independent and can be used in our historical simulation. for example, we can randomly sample the rate of returns from the historical data and repeat many times to extract more information from the given data. the basic idea of the bootstrapping method is to take some real data and then create a large number of replicate data sets by sampling with replacement (with equal probability) from the original sample. thus each new, replicate data set is a slightly perturbed version of the original one. sampling with replacement means that every sample is returned to the data set after sampling. therefore a particular data point from the original data set could appear multiple times in a given bootstrap sample. 4. quantitative analysis for market risk model in this section, we will outline the correlation and autocorrelation functions, proposed analysis and the analysis carried out for this paper. the us equity returns using the 10 assets of 1000 observations is used as fi nancial market data. matlab software is chosen to help with constructing the market risk var model in order to generate the simulated paths using the three different methods of sampling from the historical data. (i) correlation and autocorrelation functions (acf) the linear correlation of two variables x, and y can be defi ned as , where (7) the correlation value lies between –1 to +1, where high positive or negative correlation is possible depending on the market movements. let us consider the following three examples. • in the overlapping data, if the market is keep moving up, so that the rate of return for a week is 0.1, 0.2, 0.4, 0.5, 0.7, then the one-day lagged data could be 0.2, 0.4, 0.5, 0.7, 0.8. obviously they are highly positively correlated. • if the market moves up and down, the correlation between the initial data and the lagged data could be negatively correlated. for example, 2 1 varvar ),( yx yxcov xy yexeyxeyxcov ..),( ht tp :// ijb f.u um .e du .m y market risk var historial simulation model with autocorrelation effect: a note: 155-165 161 if the daily returns are 0.1, 0.0, 0.2, 0.1, 0.2, and the one-day lagged data are 0.0, 0.2, 0.1, 0.2, 0.0, then the correlation is highly negative. • for non-overlapping real market data, it is possible to obtain either positive or negative correlation values. if the rates were moving up last week and moving up again this week, then there is a positive correlation between last week and this week. if the rates were moving up last week and moving down this week, then there is a negative correlation between last week and this week. the auto-correlation function (acf) or serial correlation occurs when we know that observations that are recorded sequentially over time are not independent of one another. for example, if a series x = (x t , x t-1 ), then x t , x t-1 share information. this happened in overlapped historical periods. the acf can be defi ned as (8) (ii) proposed analysis it is proposed that the autocorrelation effect can be quantifi ed by constructing a simple stand-alone simulation of this process in matlab. in particular, historical market equity data will be used to construct the sample paths (in the manner outlined above). the terminal value of each sample path will be used to value a simple contract (for example a vanilla option) and the var of this contract calculated. the simulation will run for calculating the var for a 10-day horizon. we will then analyse the resulting sample paths generated for acf or serial autocorrelation. in addition to this, each of the simulated paths will be analysed for evidence of systematic bias in their correlations. for example, the correlations between paths and their autocorrelation functions can be determined. this method will then be contrasted with the three further procedures for sampling from the historical data provided. • (i) selecting non-overlapping periods. that is, observations (1...n) will be used to construct the fi rst path, observations (n+1)...,2n will be used to construct the second path, and so on until the n / n th path is constructed from observations. n-n+1, … , n. • (ii) a sampling procedure based on the bootstrap method will be utilised (where samples of n observations will be chosen repeatedly at random from the set of n observations). • (iii) each of these procedures can also be undertaken using simulated equity data generated from a gbm process as detailed in subsection 2.1. , )var()var( ),( 2 1 1 1 tt tt t xx xxcov ht tp :// ijb f.u um .e du .m y 162 market risk var historial simulation model with autocorrelation effect: a note: 155-165 (iii) analysis for procedures (i) and (ii) using the non-overlapping and the overlapping periods via the bootstrapping method, there exists a new window n w for each generated new path. the number of new windows can be considered as (9) where tob represents the number of total observations, horizon represents the sample size within each window, and tau represents number of increments for the next simulation path (i.e. 1 day, 2 days,…etc.). note that when horizon equals tau, we have a non-overlapping path. for procedure (iii), by generating simulated data using a gbm process, we will perform the same analysis on this simulated data (which has the same moments as the actual market data) as we do for the historical data for comparison (so that the length of the simulated time series is identical to that of the historical data). the gbm process can be considered as follows: (10) where s t – s t-1 = ∆s is the change in stock price, s is the small interval of time, ∆t , and ε is a random drawing from a standardised normal distribution. a parameter µ represents the expected rate of return per unit time from the stock and a parameter σ represents the volatility of the stock price. 5. empirical results for bootstrapping methods, simulated values of variables mentioned in equation (9) for the historical paths give a mixture of negative and positive values. using the two following examples, the results are shown below. (i) taking the ‘tau’, ‘horizon’ to be 500, and the ‘tob’ to be 2500, there are the non-overlapping historical periods. the number of windows is 5. the correlation values is displayed below: 1. using the gbm process, 1 0.071894 0.090361 -0.020968 -0.0013005 0.071894 1 0.025145 0.026059 -0.056777 0.090361 0.025145 1 -0.011327 -0.02862 -0.020968 0.026059 -0.011327 1 -0.015625 -0.0013005 -0.056777 -0.02862 -0.015625 1 1 tau horizontob nw , ttss tt 2 exp 2 1 ht tp :// ijb f.u um .e du .m y market risk var historial simulation model with autocorrelation effect: a note: 155-165 163 2. using the bootstrapping method, 1 0.041655 0.098772 0.033606 0.08557 0.041655 1 -0.015867 -0.016676 0.016893 0.098772 -0.015867 1 -0.026674 -0.008783 0.033606 -0.016676 -0.026674 1 0.010165 0.08557 0.016893 -0.008783 0.010165 1 by examining the correlation plots (not shown) for these outputs it was observed that the plots are not adequate. this shows that we should increase the number for variables tau and horizon. the second analysis is detailed below. (ii) taking the tau = 10, horizon = 1500, and tob = 2500 there are overlapping historical periods. the number of window is 101. the correlation values for the fi rst 5 columns (i.e 5 windows) and the fi rst 5 paths are displayed below. 1. using gbm process, 1 0.015563 0.039555 0.038064 0.0037525 0.015563 1 0.015907 0.038379 0.034823 0.039555 0.015907 1 0.012851 0.038529 0.038064 0.038379 0.012851 1 0.0078944 0.0037525 0.034823 0.038529 0.0078944 1 2. using the bootstrapping method, 1 0.0081298 -0.013902 -0.032916 0.017007 0.0081298 1 0.013858 -0.011403 -0.034304 -0.013902 0.013858 1 0.021605 -0.010686 -0.032916 -0.011403 0.021605 1 0.022989 0.017007 -0.034304 -0.010686 0.022989 1 the correlation plots displayed in figure 1 represent the simulation of path 1. these plots suggest that there exists a serial correlation between paths around zero. the extreme point of value 1 comes from the diagonal matrix of 1 where each path is correlated to itself. this is just one example of one particular path; however, it is true for all historical simulation paths. at this stage, let us examine the autocorrelation (acf) plots below to see whether our model is a suitable model for our data set. by taking the lag at 50, the acf plots for path 1 show little correlation for the paths at all lag. slightly more auto-correlation at lag 14 occurred in the second plot below. again, the acf plots for all paths are similar to the path 1. this means that the historical simulation model may be of value in fi tting with our fi nancial stock data. ht tp :// ijb f.u um .e du .m y 164 market risk var historial simulation model with autocorrelation effect: a note: 155-165 figure 1: correlation plots using the gbm process and the bootstrapping method figure 2: acf plots using the gbm process and the bootstrapping method 0 50 100 150 -0.5 0 0.5 1 tob gbm: 1st window 0 50 100 150 -0.5 0 0.5 1 tob gbm: 2nd window 0 50 100 150 -0.5 0 0.5 1 tob bootstrapping: 1st window 0 50 100 150 -0.5 0 0.5 1 tob bootstrapping: 2nd window 0 5 10 15 20 25 30 35 40 45 50 -0.5 0 0.5 1 lag sa m pl e au to co rre la tio n sample autocorrelation function (acf) for gbm 0 5 10 15 20 25 30 35 40 45 50 -0.5 0 0.5 1 lag sa m pl e au to co rre la tio n sample autocorrelation function (acf) for bootstrapp ht tp :// ijb f.u um .e du .m y market risk var historial simulation model with autocorrelation effect: a note: 155-165 165 6. conclusion these empirical results suggest that the gbm process and bootstrapping methods provide similar output in terms of correlation values and the acf plots. that is, the correlation values are not consistent. the non-overlapping paths should provide correlation values to be relatively larger than the overlap paths. although this is not always the case, it seems that the values of correlation are very similar throughout (i.e. around zero) for all methods. author statement: submitting author is wantanee surapaitoolkorn, sasin graduate institute of business administration, chulalongkorn university, bangkok, thailand, 2007. tel: +66-2-2183893; e-mail: wantanee. surapaitoolkorn@sasin.edu. references glasserman, p. (2004). monte carlo methods in financial engineering, application of mathematics: stochastic modelling and applied probability, springer. mcneil, a. j., frey, r., and embrechts p. (2005). quantitative risk management: concepts, techniques, tools. princeton series in finance, princeton university press, princeton and oxford. mikosch, t. (2006). elementary stochastic calculus with fi nance in view. advanced series on statistical science & applied probability vol. 6. pearson, n. d. (2002). risk budgeting: portfolio problem solving with value-atrisk. john wiley & sons. saunders, a., boudoukh, j., and allen, l. (2003). understanding market, credit and operational risk: the value at risk approach. blackwell publishing. surapaitoolkorn, w. (2007, july-september). quantitative review on the present and future of fi nancial risk modelling and the role of basel ii accord. scms journal of indian management, 4 (3). ht tp :// ijb f.u um .e du .m y thai banking: a note on technological change and technological capabilities 165the international journal of banking and finance, vol. 5.(number 1): 2007: 165-184 ijbf thai banking: a note ontechnological change and technological capabilities jarunee wonglimpiyarat the national science and technology development agency (mtec), thailand abstract this paper reports on the technological capabilities and learning of thai banking system. it identifies innovation development of the system as it evolved and how the learning process took place. this study is based on a leading technological regime change literature and the results are based on the study of five commercial banks: bangkok bank, siam commercial bank, thai farmers bank (kasikorn bank), krung thai bank, and bank of ayudhya. mass automation of work procedures occurred during the 1960s and 1970s while the smart automation regime began in the early 1970s. the ways in which the banks improved their technological capabilities via electronic banking services is explored. the results also show that the use of technology in the mass automation regime is carried through to the smart automation regime, showing that the technological change in the banking sector is not revolutionary, but ocurred slowly, i.e. evoled via slow learing process. jel classification: g21 key words: technological change, technological capabilities, revolutionary, evolutionary, mass automation, smart automation. 1. introduction there is a wealth of literature on how organisations learn and how technological changes lead to learning but little attention is given to the issues of whether change is perpetrated as evolutionary or revolutionary technological regime changes in banking industry: for a leading article see buzzacchi et al. (1993), the paper which this study closely follows. most studies have been concerned with manufacturing sector leading to just few reports on service organisations. it is, therefore, useful to investigate this neglected area relating to banking, which is recognised as an important sector in the use of information the international journal of banking and finance, vol. 5.(number 1): 2007: 167-186166 technology (it) (barras, 1990) in that the use of it plays an important role in accumulating technological capabilities within individual banks. for very good practical reasons, thai banking has been increasingly challenged by a high degree of it adoption since it plays a similar important role, as in other countries as well, in improving the performance of the banking business. the objectives of this paper in studying the technological capabilities of thai banking system are: • to examine the learning concept theories as ways to accumulate technological capabilities by an organisation. • to test the concepts of technological learning and the technological regime as in buzzacchi applied to the case of thai banking. the research uses a case study methodology to examine the technological capabilities. the case studies (banking in a developing country) are used to test the model, which states that there is a weak evolutionary link between the mass and smart automation regimes, causing the stock of knowledge and skills used in the first to be unusable in the subsequent regime. the following research questions are the subject of this inquiry: q1. is the technological change in thai banking evolutionary or revolutionary? q2. can cumulative learning in the mass automation regime be used in the smart automation regime? the rest of the paper is organized as follows. section 2 reviews the concepts of learning and technological change – that includes a critical examination of the adopted analytical approach. the theoretical framework on technological accumulation leading to learned capabilities provides a basis for the analysis. section 3 provides an overview of thai banking to provide a background understanding of the process of creating technological capabilities. section 4 presents the analysis of findings using five major commercial banks; the bangkok bank (bbl), the krung thai bank (ktb), the thai farmers bank (kasikorn bank tfb), the siam commercial bank (scb), and the bank of ayudhya (bay). section 5 ends this paper with suggestion for future research. 2. theoretical framework 2.1 review of technological learning concepts to understand the learning process, to accumulate technological capabilities of banking, it is necessary first to consider the various concepts of learning that precedes technological change. the learning may be defined in a number of different ways. according to bell and scott-kemmis (1985), learning concerns the mechanisms and processes which bring about technological progress and innovation. they emphasise the acquisition of additional technical skills and knowledge by individuals and by organisations. other theorists, such as senge of mit and argyris of harvard see learning as a collective, diagnostic process. 167the international journal of banking and finance, vol. 5.(number 1): 2007: 165-186 table 1: alternative concepts of learning in the literature table 1 shows the different views of authors on what is meant by learning. cohen and levinthal (1990) along with amsden and hikino (1992) view learning as the acquisition of knowledge from outside sources while adler and clark (1991), including bessant and buckingham (1993), view learning as any activity leading to incremental improvement (learning curve improvement). according to rosenberg (1993), malerba (1982), and von hippel and tyre (1995), the concept of learning is defined as an integrated process of knowledge acquisition which ultimately enhances the stock of knowledge. however, the last concept of learning according to dosi and freeman (1994) is broad since it covers any type of technical change plus the acquisition of any type of knowledge. in this study, the third concept is used in order to suggest that learning is a process of enhancing the stock of knowledge. lall (1982) divided the different concepts of learning into elementary, intermediate, and advanced forms. elementary learning includes basic learning like learning by doing what is stated as in von hippel and tyre (1995) and arrow (1962). intermediate learning includes processes undertaken to achieve incremental adaptations and improvements, such as learning by training. lall refers to advanced learning as the skills and know-how required for carrying out major technological adaptations in both product and process technology, like learning by searching. yet, there are problems in classifying the degree of increased improvement which some scholars like freeman and perez (1988) have tried to do a taxonomy of innovations. the adoption of different learning mechanisms, to some extent, determines the pace of progress or improvement along the technological trajectory of individual banks. improvement is not constant over time but depends on a firm’s strategy for the allocation of resources. bell 1984), malerba 1993), and hobday (1995) suggest that learning requires resources and is thus a rather costly process. the international journal of banking and finance, vol. 5.(number 1): 2007: 167-186168 at the level of low resource allocation, the learning mechanism includes the neoclassical notion of learning by doing (1962) a problem-solving process usually plugged into a production function. lall (1994), von hippel and tyre (1993), and von tunzelmann (1995) also view this type of learning as a process of trial and error and problem-solving. this is in accordance with rosenberg’s view of learning by using, which seems to be a series of minor improvements in sharp contrast to major innovations of schumpeterian magnitude. bell argues that improvement in productivity as represented by learning curve improvement may have nothing to do with an increase in skill and/or knowledge. however, it is important that this type of learning has involved some element of tacit knowledge since it is tied to the usage or activation of codified knowledge. tacit knowledge is also a key feature of technological accumulation because a large part of technology is embodied in people and institutions while such knowledge is not easily transferred (see mansel and when (1998)). other types of learning which seem to be costless, by-products of doing, include learning by operating and learning by changing. at the level of greater resource allocation, learning mechanisms include system learning, learning by training, learning by searching and learning by hiring. they are channels calling for intellectual engagement, rather than repetition like doing-based learning. such progress does not occur passively or constantly through time but is the result of deliberate resource allocation, quite often the result of firms’ strategies. however, the capacity to absorb knowledge together with the ability to recognise, assimilate, and use external competencies with a view of incorporating them into internal development seems to be more difficult, particularly in the sense of how to internalise knowledge (absorptive capacity) under certain market conditions. even so, the development of new competencies cannot be assured through learning mechanisms because learning channels are still difficult to incorporate and ‘technologies’ are so product-specific (as in tunzelman (1995). 2.2 learning approach accumulation of technological capabilities the term capabilities emphasises the key role of strategic management in the process of adapting, integrating, and re-configurating internal and external organisational skills, resources, and functional competence in a changing environment as in teese and pisano (1994). the term technological capability is more technically focused, and has different definitions. for example, dahlman and westphal (1982) define it as having the technical knowledge necessary to carry out production engineering projects and create new technology. lall (1982) defines it as the general ability to undertake a broad range of tasks in the setting up, operating, improving and expanding of production facilities. it seems that these scholars view technological capability as a production function. a wider notion is held by bell and pavitt (191993) that the resources needed to generate 169the international journal of banking and finance, vol. 5.(number 1): 2007: 165-186 and manage technical change, including the skills, knowledge, and experience are the institutional structures and linkages. i will assume that the resources needed are the ways of learning and assimilation which are far from automatic, requiring extensive commitment to developing technological capabilities. lall categorises firm-level technological capabilities into 3 parts: investment capabilities, production capabilities, and linkages capabilities with a focus on manufacturing services. he emphasises that the basic core functions in each major category has to be internalised by the firm to ensure successful commercial operation. this is in accordance with bell and scott-kemmis’ view that certain core technological capabilities should be built within the firm and by firm. that is, they should have the ability to make the necessary adaptations to induce a new generation of technology. this means that technological capabilities have their roots in various processes set up by the firm’s own efforts and are conditioned by the firm’s learning approach. such accumulation can be undertaken actively or passively. active accumulation is an intentional effort to make modifications, adaptations, and improvements to technological activities, as in an empirical case study of learning by training in the korean petrochemical industry (see dahlman and fonseca, 1987) and by searching and training as in the case of the usiminas steel plant in brazil. in contrast, passive accumulation refers to low effort intensity, like passive learning by doing. it has a limited value as a mechanism for technological accumulation and may result in a shallow learning curve. however, the complementarities from these learning processes will yield enhancements to the stock of knowledge according to bell’s learning curve theory. teece and pisano (1994) emphasise path dependencies as factors that can influence the way ahead. in other words, what a firm can do in the future is constrained by what it has been able to do in the past. it appears that the development of technological capabilities is cumulative and dependent on firms’ histories. this is in accordance with nelson and winter’s view that the firm’s competence building is path-dependent because it is shaped by its routines. barras, bell, bell and pavitt (1993) imply that learning is cumulative in character the learning in one stage is dependent on the learning path in the past. past learning is not wiped out but provides the foundation for further knowledge to be added to it or to reinforce it in subsequent periods. such a view of enhancing knowledge seems to put aside the technologically deterministic view of buzzacchi et al. who rejects the cumulative nature of learning by arguing that learning in the smart automation regime has nothing to do with learning in the mass automation regime. this nature of things also effaces the demarcation of the separate sets of knowledge identified by them (see further discussed in section 2.3.2) that the shift in technological regime causes a vast chasm of two sets of technological knowledge learnt in the mass and smart automation regime. in this sense, the writer is led to argue that learning is complex the international journal of banking and finance, vol. 5.(number 1): 2007: 167-186170 and multidimensional. yet in models of economic growth whether neoclassical or evolutionary learning is usually reduced to a single feature of a learning curve. learning is an input which combines with other inputs to set in motion the process of continuing technological capabilities. however, learning is also an output of existing knowledge as well as of activities. new capabilities based on past learning propel more developed capabilities in product and process innovation. it is clear that a firm’s technological capabilities are not only conditioned by its learning approach at the level of the firm. organisational and institutional levels also pose challenges and opportunities for the sustainability of firms’ activities in the continuous production of new developmental knowledge of capabilities. external factors such as infrastructure, government policy and market mechanisms also play a significant role in firms’ behaviour. this suggests that the national system of innovation is important for corporate competitiveness in the sense that it affects the progress of firms, which ultimately results in firmspecific technological capabilities. 2.3 technological change and learning implications evolutionary or revolutionary echnological change? the conceptual model of innovation in the banking industry developed by buzzacchi et al. (1993) is derived from an econometric analysis of italian commercial banks. they argue that the innovation process in the ‘mass automation’ regime, coinciding with the automation of back-office procedures, is quite distinct from the ‘smart automation’ regime that followed, which focused mainly on the provision of electronic banking (eb) services; cash dispenser (cd), automatic teller machine (atm), point of sale (pos), remote banking (rb), home banking (hb), and corporate banking (cb). accordingly, the technical change in banking exhibits revolutionary characteristics triggered by the availability of distributed data processing and data transmission networks. in the discussion of moving through the regimes, buzzacchi et al. (1993) and tushman and anderson (1986) have in common the view that technical change represents discontinuities between old and new technologies. the schumpeterian view also emphasises discontinuity. however, thomas hughes (a famous us historian of large technical systems) argues that there is no such thing as radical innovation by looking at the evolution of large systems (e.g. electricity networks, teleco networks, and banking networks could similarly be included). in contrast, freeman and perez (1988) argue that there are radical technological changes or discontinuities, probably because in their view, technology (e.g. it) affects the suppliers of components and changes their industrial boundaries. von hippel ((1988) argues that innovating is not always radical from the perspective of innovative organisation, that there are incremental improvements through all organisational activities. 171the international journal of banking and finance, vol. 5.(number 1): 2007: 165-186 the different views of the conceptual implications arise from consideration from different perspectives. on the one hand, buzzacchi et al. consider the product level as the unit of analysis (eb consisting atm, pos, rb, hb) within the product life cycle pattern of product substitution. they consider the development of the microprocessor in the early 1970s as a trigger which progressively enhanced product capacity enabling product innovations and substitutions for prior product/process technologies leading to revolutionary technological change. buzzacchi et al. on the other hand, interpreted barras’s rpc model as having evolutionary characteristics when applied to the banking sector. they may regard barras’s unit of analysis as the firm level within the context of the innovation life cycle (a progression from process to product innovation) because the rpc model is basically generated from case studies of it-based service organisations like retail banking, insurance, accountancy, and local government. so, they see the adoption of it in these organisations as a continuous progression through three stages: improved efficiency, improved quality, and new products, all of which has an evolutionary character. although buzzacchi et al. tried to make a conceptual distinction, the two concepts are not mutually exclusive. barras’s concept of the rpc model as progress from stages of improved efficiency and quality to new products can explain buzzacchi et al.’s mass automation regime in the sense that the last stage of new products can be seen as a link to buzzacchi et al.’s smart automation regime, being defined as a radical innovation of electronic banking (eb). this suggests that there is no clear-cut distinction between the two regimes, but there is an interrelated period with a smooth progression from one to the other. learning accumulation along with the technological change? according to dosi (1982), the technological trajectory is defined as a pattern of normal problem-solving activity (or progress) along a technological paradigm. nelson and winter (1977) define it as the direction towards which the problem-solving activity moves, or as the ‘natural trajectories’ the of technical progress. tushman and anderson describe patterns of technological change as a cumulative process until punctuated by discontinuous innovations. this causeing technological shifts, either competence-enhancing or competence-destroying. competence-enhancing discontinuities are order-of-magnitude improvements based on cumulative experience in the use of earlier vintages of technology in contrast to competencedestroying discontinuities which require the mastery of new technology, skills, abilities, and knowledge in both the development and the production of the product. buzzacchi et al. argues that a shift in the technological regime of the mass automation and smart automation regimes lead to a substantial devaluation of the stock of knowledge and skills accumulated by firms in the previous regime. “one important implication of the change in technological regime is that the the international journal of banking and finance, vol. 5.(number 1): 2007: 167-186172 discontinuity largely canceled out the cumulative effects associated with the path previously followed by technology.” in terms of moving through the regime, buzzacchi et al. and tushman and anderson agree that there is a discontinuity in the technological regime and a shift towards revolutionary change results in competence-destroying. this suggests that they understate the importance of cumulative learning. however, pavitt (1986) argues that even when there are radical innovations, they rarely displace all the skills and knowledge related to the earlier technological regime. buzzacchi et al. considers the development of it knowledge through the progression from the mass automation regime to the smart automation regime as a weak evolutionary link. however, the investigation of the model shows that this does not represent a weak link, but rather, a strong evolutionary link for the following reasons. first, their conceptual model of innovation can be viewed as technologically deterministic in the sense that they tie the learning process of a firm to the use of new hardware technology. they determine the learning process as dependent on new hardware technology in the sense that the different hardware structure of eb services requires new knowledge to be mastered. in other words, the learning process is dependent on the new hardware technology. it may be true that new technology demands new skills/knowledge but the former knowledge is not totally destroyed. it is, rather, an interactive process by which the former knowledge is replaced by the new knowledge (the process of continuous learning and enhancement). this suggests that technology is systemic and should not be thought of in isolation. at the same time, the firm has technological choice in selecting new technology compatible with the old technology currently in place in the organisation. second, their conceptual model of innovation presents a fragmented learning process in the sense that there are two independent regimes demanding different knowledge. when learning in the mass automation regime ends, it recommences in the smart automation regime, causing a technological discontinuity. even if it is appropriate for buzzacchi et al. to state their case from the perspective of the product level and then classify the regimes, it seems that there is still a continuous learning process across the regimes. it appears likely that they ignore the importance of the learning that is tacit in nature, embodied in people and institutions in a cumulative way (path dependent). people may not radically change the way they think and interact with others (which is a way of fostering individual learning processes) because such a way involves tapping into people’s perceptions based on their own experiences. insofar as the concept of different technological regimes is concerned, the characteristic of progression seems to be a continuous learning process. 173the international journal of banking and finance, vol. 5.(number 1): 2007: 165-186 3. overview of thai banking a background of the thai banking industry at the macro level is shown in table 2. the major banks in thailand are the bangkok bank (bbl), the krung thai bank (ktb), the thai farmers bank (kasikorn bank tfb), the siam commercial bank (scb), and the bank of ayudhya (bay). the banking innovations developed gradually through the use of technology. awareness that technology is vital to competition in banking began when bangkok bank introduced interbranch online services in 1970. however, pro-active competition through the use of it to improve services across the industry did not occur until table 2:the highlights of the banking industry in thailand the emergence of atm service in the smart automation regime. the introduction of atm services by siam commercial bank in 1983 triggered the banks to invest in technology to build up their technological capabilities. in other words, it was a technology push that pressurised all the banks to enter into a high degree of competition to survive in the banking industry. table 3: major banking services in thailand the international journal of banking and finance, vol. 5.(number 1): 2007: 167-186174 the widespread investment in technology to offer an array of eb services in thailand began in the smart automation regime. in the late 1990s, electronicbased banking began to transform into internet-based banking. the major banking services which are interesting and widely used in the thai banking are shown in table 3. this could be summarised as done in the table under five topics. online services were introduced to facilitate customer contact at the lowest costs. automation via atm services played critical role in speedier response to this table 4:the major adoption of technology to improve technological capabilities 175the international journal of banking and finance, vol. 5.(number 1): 2007: 165-186 critical cash and other service provisions without human contact. the other more sophisticated technology suc as internet banking evolved later. the adoption of technology to improve the technological capabilities in the retail banking business is shown in table 4. our analysis of these banks adoption of it shows clearly that the technology change was widespread, and very early as adoption of newer ways of doing things. the progression from online interface to internet and high-speed and from encrypting internet information shows that the major banks were leading in the adoption of technological change as pioneers. 4. case study analysis of thai banks is technological change evolutionary or revolutionary? this section distinguishes between the two technological regimes to see whether the change from the mass to the smart automation regime is evolutionary or revolutionary. the variables for measuring the change are similar to those employed by buzzacchi et al. in order to do this, seven variables related to the change of technological regime are used as indicators. among these, two variables (the adoption of mainframe computers and off-line batch processing) are concerned with the mass automation regime. the other five (the adoption of online real-time processing, automated teller machine (atm), point of sale (pos), phone banking, and personal computers (pc)) are concerned with the smart automation regime. see table 5. following their attempts to show that there is a shift in technological regime, the analysis in the following table should be interpreted as follows. if the variables table 5: analysis of technological change (evolutionary/revolutionary) note :• : the use of the variable in the stated technological regime n/a : not applicable since the technology has not yet been adopted within the stated regime. mass : mass automation regime. smart : smart automation regime. the international journal of banking and finance, vol. 5.(number 1): 2007: 167-186176 which represent the use of technology in the individual regime are solely employed in the mass or smart automation regime according to the aforementioned indicators, then this represents a shift in technological regime or exhibits a revolutionary characteristic. if not, they seem to be evolutionary, and thus question their argument. the analysis shows that technological change in thai banking is evolutionary for a number of reasons. first, regarding the mass automation regime, all the banks used mainframe computers and off-line batch processing in accordance with buzzacchi et al.’s definition. however, three out of the five major banks bangkok bank, siam commercial bank, and thai farmers bank (kasikorn bank) used online, realtime transaction processing within this regime, in contrast to what they suggested, i.e., the exclusion of online from this regime. in the industry, bangkok bank was the first bank to use online in 1970 for use in credit processing and accounting entries. with bangkok bank’s leading role in online technology, siam commercial bank caught up bangkok bank in the same year by offering an online service to enable customers to make deposits or withdrawals at any branch. this was mainly aimed at catching up with bangkok bank rather than facilitating accounting functions within the bank. in the same year, thai farmers bank (kasikorn bank) followed the leading banks by gradually developing a front-office, online service. the other two banks namely the krung thai bank and bank of ayudhya also tried to catch up, but rather, they lagged behind. it can be seen that the online service was carried out in parallel with off-line batch processing for deposits and withdrawal transactions, showing that online, real-time processing and off-line, batch processing can co-exist in the same technological regime, in sharp contrast to buzzacchi et al.’s characterisation. second, regarding the smart automation regime, the variables representing the adoption of technology – online, real-time transaction processing, atm, point of sale (pos), phone banking, and personal computers seem to be in line with buzzacchi et al.’s definition. however, there are exceptions in that the the mass automation regime did not result in the withdrawal of mainframe computers and off-line, batch processing. the mainframe computers were still in use by all major banks. for example, in the early 1970s, bangkok bank installed an ibm 360/30 mainframe computer at its headquarter to handle the automation of cash deposits and withdrawal functions of savings accounts. siam commercial bank also used a mainframe computer system 4331 for branch development tasks in 1971. in 1979, siam commercial bank introduced the singer mainframe computer system 10 to operate online systems for customer services for deposits and withdrawals at its chidlom branch. thai farmers bank (kasikorn bank) also continued to use the burroughs and ibm 308x mainframe with the operating system of mvs/xa (multivirtual system/extended architecture) for online data processing. its business development information system for its reengineering program also ran on os/2 177the international journal of banking and finance, vol. 5.(number 1): 2007: 165-186 based pentium pcs linked to the mainframe computer system. krung thai bank used tandem, ibm, and philips mainframe computers during this regime and bank of ayudhya also used the mainframe computer system together with nec local area network (lan) 4700 as an enterprise server for branch operations. for the off-line, batch processing used by all 5 major banks during this regime, bangkok bank developed the system 434 batch processing for deposits, credit, accounting, and back-office functions in parallel with the online system. siam commercial bank also carried out batch processing for reports in parallel with the operation of the online system at its chidlom branch. thai farmers bank ran batch processing for back-office banking processes to calculate interest and for accounting transaction entries, as did bank of ayudhya. in addition, since bank of ayudhya was constrained by limited funds for investment when purchasing new equipment, it did not throw away the existing system and turned to the new and open system of unix. the existing system was still capable of handling transactions, representing the technological choice in selecting new technology to be based on an existing technology for reasons of compatibility. the analysis shows that there is no shift in technological regime. to put it another way, the technological change in the thai banking industry is not revolutionary because if it were, there should be no further use of mainframe computers and off-line batch processing within the smart automation regime (the position of buzzachi et al.). as far as the analysis was concerned, the technological trajectory of thai banking did not take a different path from the past because the mainframe computers and off-line batch processing were still used during the smart automation regime. third, the technology used in the mass and smart automation regime was not consistent with that of buzzacchi et al.’s definition. in the case of thai banking, the online, real-time transaction processing which they defined as the technology used in the smart automation regime was already used in the mass automation regime. even within the smart automation regime itself, there were still the continuous use of the mainframe computers and off-line batch processing, as well as the use of the technologies – electronic banking (eb) services defined as appropriate in this regime. can cumulative learning be used in smart automation regime? this sub-section explores whether there is any disruption of cumulative learning with a change in regime. as stated in section 2, buzzacchi et al. argues that the shift from the mass automation regime to the smart automation regime leads to a substantial devaluation of the stock of knowledge and skills accumulated by firms in the previous regime. two variables the continuity of employment of the existing personnel and new employees related to the concept of cumulative learning are used as indicators in order to signify whether their argument is valid in the case of thai banking. the international journal of banking and finance, vol. 5.(number 1): 2007: 167-186178 • the continuity of employment of the existing personnel means the continued employment of personnel from the mass automation regime in the smart automation regime. if the existing personnel can undertake the eb services in the subsequent regime through reinforcing their skills with new training programs or new knowledge, then this represents competence-enhancing. • new employees means the hiring of new personnel to fit the smart automation regime. if the banks have to hire a new set of employees to undertake eb services in the smart automation regime with no help from existing personnel, then this represents competence-destroying. it should be noted that tushman and anderson have not developed any variable to determine competence-enhancing or competence-destroying capabilities. neither have buzzacchi et al. developed any variables. these cannot be determined solely by quantitative methodologies because the number of increasing or decreasing personnel does not directly represent competenceenhancing or competence-destroying capabilities since personnel may leave the table 6: analysis of cumulative learning through the technological regime note : • : the use of the variable in the stated technological regime. bbl : the bangkok bank; scb : the siam commercial bank; tfb : the thai farmers bank (kasikorn bank); ktb: the krung thai bank; bay : the bank of ayudhya. bank to move to other banks, retire, etc. detailed staff histories would be necessary to determine whether the reason for leaving related only to the change of regime. however, the variables used here are not a solid measure but are suggested in order to perform a qualitative analysis: see table 6. the analysis shows that cumulative learning in the mass automation regime can be used in the smart automation regime for the following reasons. first, regarding the mass automation regime, the existing personnel were able to perform system operation and maintenance tasks. in the initial phase of it adoption, working through vendor relationships (learning by interacting with suppliers) was important because the banks lacked a strong it team. so it was necessary to have technical links to support the efficient use of new technologies. 179the international journal of banking and finance, vol. 5.(number 1): 2007: 165-186 in other words, the learning in this regime can be regarded as a process of continuous problem-solving and developing a system to operate properly. for example, bangkok bank (bbl) received training support from the supplier ibm during its first use of computer technology. the accounting personnel who already had background knowledge in accounting and banking systems were trained to be programmers or systems analysts to operate and maintain the new computer system installed by the vendor. siam commercial bank (scb) also operated with technical advice from vendors like ibm and hitachi. the development of the system gradually evolved through the close cooperation with ibm systems representatives to help the bank personnel resolve any technical problems. the same was true of thai farmers bank (kasikorn bank) as ibm not only provided training for bank personnel when selling the software, enabling the system to be maintained through the skills transfer. however, the capabilities of extending, developing, or adding features to obtain optimum use are characterised as learning by doing. although there was a great deal of dependency on the vendors (i.e. krung thai bank on philips and digital technology; bank of ayudhya on ibm and tandem technology), bank personnel still had to study from system manual themselves in order to learn gradually how to improve it because all the software purchased from abroad had to be modified to suit the environment and regulations of the thai banking system. new employment in the mass automation regime was mainly to replace existing staff as well as for newly created posts. since the adoption of it, the 5 major banks had no lay off policy for fear of anti-computer technology protests from the existing personnel. however, the it did reduce manual work, so the bank transferred any excess personnel to undertake the other banking functions or to fill posts in new branch operation. second, in the smart automation regime, the continued employment of existing personnel was to build up skills and knowledge necessary to carry out the eb services. they were aided by the training program to enhance competencies in the use of new it. this is evidenced by the continued employment of the existing personnel in the smart automation regime who had joined the bank in the days of the mass automation regime. this shows competence-enhancing along with technological change1 . if the smart automation regime really resulted in competence-destroying, the existing personnel should not be able to continue the eb services since buzzacchi et al. claimed that there seemed to be no positive relationship between the eb services and the knowledge in the previous regime. although most of the variables for technology use (atm, pos, phone banking, and pc as shown in table 5) began this regime, they were not a totally new technological concept per se, but a functional improvement of the existing technology, since there is an interrelationship between the two technologies. 1 these are the examples in the technological aspect but there may be some other reasons for the continued employment like patronage, personal reasons for the mass of workers. the international journal of banking and finance, vol. 5.(number 1): 2007: 167-186180 indeed, there seems to have been a carryover of learning from one system to another. it is argued that the basic knowledge of off-line, batch processing could be applied to the online real-time processing, but there was not much advance in terms of application theory. the online, real-time processing or atm was an improved technology service from the basic knowledge of deposits and withdrawals, but its novelty was only peripheral. without the online, real-time processing operated before in the mass automation regime (as seen in table 5 for bangkok bank, siam commercial bank and thai farmers (kasikorn bank), the atm in the smart automation was not operable since it was based on the concept of online networking. the mainframe computers were still used to operate online systems using message switching, allowing for the interbranch transactions of deposits and withdrawals. in addition, the use of pcs for office automation in this regime still had to be based on the mainframe, centralised, data processing system by using network switching between the terminals and the host computers, facilitating deposits, withdrawals transactions, and balance enquiries. bangkok bank built up its strong it team by training the existing personnel in knowledge of both the system and the banking products so that they could specialise in handling both the system and service of the business effectively. for krung thai bank, during the transition from off-line, batch processing to online, real-time processing, the existing personnel who had written the programs for batch processing (with the aid of some training) also wrote the programs for the online processing. siam commercial bank underlined the importance of the ‘learning organisation’ by preparing not only the technology, but also the personnel to be ready for competition with foreign banks under the period of financial liberalisation (1990s, onwards). the banks in this regime still needed cooperation from the suppliers to support the skills of their personnel (i.e. thai farmers bank with the support of burroughs and ibm for online programming). however, the effort to build up its own it workforce (i.e. thai farmers bank’s system engineering department, krung thai bank’s kcs, bank of ayudhya’s in-house development) was greatly emphasised. the bank’s in-house technology helped lower costs and increased efficiency since the banking technicians could receive information from, and coordinate with, bank personnel. the banks also sent their personnel to learn more about banking technologies in the advanced countries and to bring the knowledge back to foster their own system development department (i.e. to hong kong for the system of systematic and hogan; u.s.a. in the area of application development). in the smart automation regime, there was a need for new employment patterns with new skills (learning by searching) for specialised and sophisticated banking business. the recruitment of new personnel was likely in areas where the banks lacked technological capabilities. it is argued that the knowledge of system development is partly tacit and embodied in personnel. this tacit knowledge 181the international journal of banking and finance, vol. 5.(number 1): 2007: 165-186 was essential for further building-up of technological capabilities, so the bank had to solve the problem by purchasing it experts (learning by searching) and software (i.e. credit processing function), instead of developing these itself. this type of learning by krung thai bank included the search for foreign experts to speed up software development to catch up with other banks. krung thai bank’s learning, thus, seemed to be pushed by internal problems and was rather different from the other banks’ learning, which put more emphasis and effort intensity into developing the system by self-learning. third, it should be noted that the continuous employment of existing personnel possessing some level of skill and knowledge of the current technology seemed necessary so that they could help new personnel to absorb the new knowledge embodied in new technology. in other words, path dependency still matters because the knowledge is partly tacit for internal learning to happen. the importance of path dependency does not only concerns technological capabilities, but also the managerial capabilities to deal with the socio-economic and market conditions to enable competencies to be enhanced. these are capabilities that intertwine and cannot be built up in isolation by new personnel since these capabilities are the complementary assets necessary for building up technological capabilities under conditions of dynamic competition. 5. conclusions the study assumes the essence of the learning process is cumulative development, which is the central concept of accumulating technological capabilities. regarding the learning concepts in the context of the thai banking industry, the learning mechanisms to strengthen the bank’s technological capabilities are mainly ‘learning by doing’ (e.g. the case of bangkok bank) and ‘learning by interacting with suppliers’ (e.g. the case of siam commercial bank, thai farmers bank (kasikorn bank), krung thai bank and bank of ayudhya). in the era of banking modernisation (late, smart automation regime), the banks realise the necessity to adopt other learning mechanisms like learning by training and learning by searching to build up human resources, as these are concerned with capability accumulation. interestingly enough, this conclusion approaches the thinking of the learning process. the study suggests that continuous learning is essential but the banks should not place too much dependence on suppliers which may limit its own capacity to develop capabilities. today, competitive contention in the banking industry challenges the banks to leverage their it strategy to offer banking products to meet their customers’ needs. banks compete by investing in technology to speed up the banking services and improving operational efficiency of banking operations. the offer of banking services like home banking services, vdo banking, internet banking, and mobile banking (mobile commerce) shows that the international journal of banking and finance, vol. 5.(number 1): 2007: 167-186182 the banks had adopted technology to compete in service performance, which enabled them to build up their technological capabilities. the empirical analysis shows that the technical change in thai banking is evolutionary and the cumulative learning from the mass automation regime can be used in the smart automation regime. this has shown that the conceptual model of buzzacchi et al. arguing that there is a discontinuity between the mass and smart automation regime, and thus the cumulative learning in the mass automation regime cannot be used in the smart automation regime, is not valid in the case of thailand. the argument that there is a substantial devaluation of the stock of knowledge and skills accumulated by firms in the mass automation regime is not applicable to the case of thai banking. the empirical analysis shows that there is a cumulative development of learning processes by the personnel who cross from one regime to the other which shows that cumulative technological knowledge is not cancelled out, but present. therefore, it should not be considered as a weak link, but rather, a strong evolutionary link from the mass automation regime towards the smart automation regime. the continuity of cumulative learning shows that it is not competencedestroying but competence-enhancing between the two regimes. hence, the learning of new technological capabilities in thai banking seems to be in accordance with the third concept of learning, that is, learning is a process of enhancing the stock of knowledge in an evolutionary manner. the study has made a small contribution to technological change literature linked to evolutionary and revolutionary issues, which is a missing element in the literature on learning concepts. technological learning seems to have a close relationship with organisational learning. however, the latter is outside the scope of the present study. it would, therefore, be interesting if future research could explore the technological change and its relationships within organisational structures in the banking industry to understand if there is an interdependent relationship between technological learning and organisational learning. acknowledgment: the author acknwledges with thanks the revisions to the paper following the blind review of the paper. the remaining errors are the responsibility of the author. author statement. jarunee wonglimpiyarat is certified public accountant (cpa) and science policy researcher at the ministry of science and technology thailand. her main research area is financial innovations. her publications include the barclaycard case study in ‘exploring corporate strategy’ (europe’s best-selling strategy textbook) by gerry johnson and kevan scholes strategies of competition 183the international journal of banking and finance, vol. 5.(number 1): 2007: 165-186 in the bank card business. the author wishes to thank the anonymous reviewer as well as the editors of the journal for extensive revisions of the original version fo this paper. references adler, p., and clark, k., (1991). ‘behind the learning curve: a sketch of the learning process’, management science, vol. 37(issue no. 3): 267-281. amsden a. and hikino, t., (1993). ‘borrowing technology or innovating: an exploration of two paths to industrial development’, in fransman, m. and king, k. (editors), technological capability in the third world, macmillan press, london. arrow, k., (1962). ‘the economic implications of learning by doing’, review of economic studies, vol. 29: 155-73. barras, r., (1990). ‘interactive innovation in financial and business services: the vanguard of the service revolution’, research policy, vol. 19: 215-237. bell, m. and scott-kemmis, d., (1985), who learned what by doing what:: a reexamination of ‘learning by doing’ in the air flame and shipbuilding industries, mimeo, science policy research unit, university of sussex. bell, m., (1984). ‘learning and the accumulation of industrial technological capacity in developing countries’, in fransman, m. and king, k. (editors), technological capability in the third world, macmillan press, london. buzzacchi, l., and colombo, m., (1993). ‘technological regimes and innovation in services: the case of the italian banking industry’, research policy, vol. 24: 151-168. bessant, j., and buckingham, j., (1993). ‘innovation and organizational learning : the case of computer-aided production management’, british journal of management, vol. 4: 219-234. bell, m., and pavitt, k., (1993). ‘technological accumulation and industrial growth: contrasts between developed and developing countries’, industrial and corporate change, vol. 2(issue no. 2): 185-203. cohen, w. and levinthal, d., (1990). ‘absorptive capacity: a new perspective on learning and innovation’, administrative science quarterly, vol. 35: 128-152. dosi, g., freeman, c., (1994). ‘the process of economic development: introducing some stylized facts and theories on technologies, firms and institutions’, industrial and corporate change, vol. 3(1): 1-45. dahlman, c., and fonseca, f., (1987). ‘from technological dependence to technological development: the cases of the usiminas steel plant in the international journal of banking and finance, vol. 5.(number 1): 2007: 1-25184 brazil’, in katz, j., (editor), technological generation in latin american manufacturing industries, macmillan press, london. dahlman, c., and westphal, l., (1982). ‘technological effort in industrial development: a survey’, in f. stewart and j. james (editors), the economics of new technology in developing countries, frances pinter, london. freeman, c., and perez, c., (1988). ‘structural crises of adjustment, business cycles and investment behaviour, in dosi g., and c. freeman (editors), technical change and economic theory, pinter publishers, london. hobday, m., (1995). innovation in east asia: the challenge to japan, edward elgar, cheltenham. lall, s., (1994). developing countries as exporters of technology: a first look at the indian experience, macmillan, london. lall, s., (1994). ‘technological capabilities’, in j. salomon et al. (editors), the uncertain quest: science technology and development, united nations university press, tokyo, japan. malerba, f., (1992). ‘learning by firms and incremental technical change’, the economic journal, vol. 102: 845-859. mansel, r., and wehn, u., (1998). knowledge societies : information technology for sustainable development, oxford university press, oxford. nelson, r., and winter, s., (1982). an evolutionary theory of economic change, harvard university press, cambridge. nelson, r., and winter, s., (1977), ‘in search of a useful theory of innovation’, research policy, vol. 6. pavitt, k., (1986). ‘technology, innovation, and strategic management’, in j. mcgee and h. thomas (editors), strategic management research, a european perspective, john wiley & sons, chichester. rosenberg, n., (1982). ‘learning by using’, in rosenberg, n. (editor), inside the black box: technology and economics, cambridge university press, cambridge. teece d., and pisano, g., (1994). ‘the dynamic capabilities of firms: an introduction’, industrial and corporate change, vol. 3(issue no. 3): 537556. tushman, m., and anderson, p., (1986). ‘technological discontinuities and organization environments’, in a. pettigrew (editor), the management of strategic change, blackwell, oxford. von hippel, e., (1988). the sources of innovation, oxford university press, new york, new york. von hippel, e., and tyre, m., (1995). ‘how learning by doing is done: problem identification in novel process equipment’, research policy, vol. 24: 1-12. von tunzelmann, g.n., (1995). technology and industrial progress, edward elgar, aldershot. international journal of banking and finance 1-1-2007 thai banking: a note on technological change and technological capabilities jarunee wonglimpiyarat recommended citation economic freedom, macroeconomic fundamentals and foreign direct investment in fast emerging brics and malaysia 57 the international journal of banking and finance, volume 10 (number 1), 2013: pages 57-73 economic freedom, macroeconomic fundamentals and foreign direct investment in fast emerging brics and malaysia catherine s. f. ho, noryati ahmad and hayati mohd dahan univeristi teknologi mara, malaysia ___________________________________________________________ abstract this study investigates the major factors that determine the inflow of foreign direct investment (fdi) into fast emerging countries: brazil, china, india, russia, south africa (brics) and malaysia. two sets of factors are identified: macroeconomic and country specific fundamentals. the period of analysis is 1977-2010. the study provides empirical evidence that economic growth, government consumption and trade openness are vital for fdi. in addition, country specific infrastructure quality and economic freedom are also critical factors in determining fdi for this group of countries. our findings have significant policy implications for the growth and development of these countries, particularly through foreign direct investments. keywords: fdi, trade openness, economic freedom, macroeconomic fundamentals jel classification: f21, f31, f41, f43. _________________________________________________________________________ 1. introduction the global economy has experienced unprecedented paradigm shifts in the last decade. coupled with the recent crises in developed regions, it may be timely for emerging economies to fill the vacuum of growth left vacant by developed countries in order to drive world economic development. despite decades of civil unrest, political and economic turmoil overshadowing some of these countries, there are strong indications that economic development is fast growing in these regions. although still unable to free themselves from the shackles of corruption and poverty, the success of these economies currently rests on their ability to attract foreign direct investments from developed and developing countries. trade between brics countries and the rest of the world has grown significantly with china and brazil being the world’s fastest growing economies. these countries are set to improve their economies by developing infrastructure, agriculture, manufacturing and the development of small businesses. ranked as the poorest continents in the world, latin america and africa have posted 58 strong growth rates in recent years, drawing increasing inward investment. despite a projected much faster gross domestic product (gdp) for these countries, a more optimistic prospect compared to the debt-ridden economies of the west, brics countries should not take a backseat in terms of creating opportunities for trade and investment. foreign direct investment (fdi) has come to play a major role in the internationalization of business in the past decades. lokesha and leelavathy (2012) defined fdi as the process whereby domestic investors of home country acquire assets for the intention of controlling all the activities of the firm which is located outside the country. changes in technology coupled with growing liberalization of the national regulatory framework governing investment in enterprises and changes in capital markets have led to profound changes in the size, scope and method of fdi. foreign investment usually brings along capital inflows as investments in long-term assets contributing to higher aggregate demand of the economy and therefore eventual growth. corporations within the country tend to be more productive to effectively counter the threat of the competition from abroad brought about by fdi. this contributes to the growth of a country’s income and employment, another positive effect of fdi when a country becomes more productive. employment is also created with increased productivity and competitiveness. new informational technological systems and the decline in global communication costs have also made management of foreign investments far more effective than in the past. changes in trade and investment policies as well as the regulatory environment such as trade and tariff liberalization, easing of restrictions on foreign investment and acquisition, plus the deregulation and privatization of many industries have been significant vehicles for expansion of fdi across the globe. the recent global financial crisis and the previous financial crises in asia and latin america resulted in relying on fdi in order to supplement national savings and promote economic development. emerging economies derive benefits from foreign investments to improve their balance of payments, increase exports to earn more hard currencies, reduce imports to save more of their hard currencies, increase employment to improve their fiscal position, and enhance their access to newer technologies. this study investigates the relations between macroeconomic fundamentals, country specific factors and fdi in fast emerging brics countries. understanding the drivers of foreign investments would provide authorities in respective fast emerging countries with vital information on policy decisions that would enable them to accelerate growth and eliminate poverty. the set of fast emerging countries included in the study are brazil, russia, india, china, south africa and malaysia. the major factors that are considered are divided into two sets: macroeconomic fundamentals consisting of market size, trade openness, financial depth, exchange rate, government consumption, inflation rate, international journal of banking and finance, vol. 10, iss. 1 [2013], art. 4 59 interest rate; and country specific factors of economic freedom, employment, literacy rate and infrastructure quality. 2. prior empirical evidence this section provides some historical analyses of fdi for the countries considered. it also details prior empirical evidence on variables that are relevant to emerging countries but may not yet been commonly researched, for instance, economic freedom, financial development, and trade openness. a. fdi in brics and malaysia vijayakumar and sridharan (2010) conclude that emerging brazil, russia, india, china and south africa (brics) economies exhibit economic strength in the face of the u.s. credit turmoil and growth slowdown by their strong domestic demand growth. three countries -brazil, india and south africa -attracted fdi due to their demographic trends, labor supply dynamics and low urbanization ratios. there is a huge potential for economic growth in these countries and it would exert competition and challenges towards developed countries. the current flow of fdi into brics is extremely intricate and competitive, both in the home and host countries. this study therefore intends to examine if macroeconomic fundamentals, economic freedom and other major determinants are significant in affecting fdi flows into malaysia and brics countries. foreign investors are currently looking into new markets with vast changes in trade and investment preferences. countries including brazil, russia, china, india and south africa are currently the new fdi destinations. these countries offer not only an enormous untapped market but also much lower production costs, especially labor. the availability of human capital, infrastructure as well as skills and education are factors that attract investors to these countries. recent developments in malaysia show a decline in fdi inflows from 2006 to 2009 (figure 1). the country has lost some of its attractiveness from foreign investors to neighboring countries like china and india due to the shortage of human capital and pressure on wages. it is believed that malaysia still has the ability to improve economically by enhancing and stimulating its domestic economy. other parts of the world including china, russia and india are attracting vast amount of fdi (figure 2). many factors may have affected fdi to these countries and their identification and analysis are the aims of this paper. 60 figure 1: fdi inflows for south africa and malaysia source: unctad figure 2: fdi inflows for brazil, russia, india and china source: unctad. b. macroeconomic fundamentals this section, as well as the next one, provides historical support for major variables that are considered important in determining fdi. we rely mainly on the findings in prior contributions. investment return and profits in developing countries are directly affected by market size and its growth. higher rate of market growth indicates potential for a larger market and more promising prospects for products produced. market size is thereby considered as one of the most important determinants for horizontal market-seeking fdi. zhang (2001) shows that market size and international journal of banking and finance, vol. 10, iss. 1 [2013], art. 4 61 infrastructure are key factors in attracting foreign investments. duanmu and guney (2009) find similar results for china and india where fdi into both countries are positively related to host country’s market size. they conclude that both countries’ fdi are attracted to locations with large market size. dynamic economy and appropriate economic growth policies are also crucial in order to attract fdi. torrisi and delaunay (2008) apply generalized least squares regression to identify the determinants of fdi in central europe and find that market size was a critical factor for fdi inflows during the economic transition period from 1989 to 2006. according to quazi (2007), greater market size, measured by per capita real gdp, attracts more fdi in east asia. hailu’s (2010) empirical analysis of the demand side determinants of fdi into african nations finds that the host country’s market has statistically significant positive effect on fdi inflow. kimino et al. (2007) on the other hand, find contradicting results in their research where there exists negative relationship between market size of source countries and fdi flows into japan. their result confirms that market size does not exert statistically significant influence on fdi into japan, a developed country. this clearly shows that investors target local market besides the export market when undertaking fdi decisions. janicki and wunnava (2004) find that the size of the market accurately reflects their theoretical expectations and confirm that market size is a significant determinant of fdi between members of the european union. based on their result, fdi flows are greater in larger economies with well-built markets. in addition, bakir and alfawwaz (2009) identify the determinants of foreign direct investments (fdi) in jordan for the period 1996 to 2007 and confirm that gdp is positively related to fdi. the important role of market size in terms of gdp is found to significantly affect the volume of fdi flow to jordan. kim and rhe (2009) also explore the trends as well as determinants of south korean fdi and find that south korean investors prefer larger markets and strategic assets. a study by russ (2007) on the endogeneity of the exchange rate as a determinant of fdi indicates that the variance of the exchange rate presents an impact towards the decision by multinational companies (mnc) to enter a particular market. artige and nicolini (2006) study the euro-effect on fdi for three european countries and conclude that exchange rate (euro currency) is an important criterion to attract foreign investors to these regions. erdal and tatoglu (2002) find that the instability of exchange rates has a negative relationship with fdi inflows. their study investigates the case of turkey whereby a highly volatile currency would discourage foreign investments. vogiatzoglou (2007) investigates the location determinants of inward fdi in south and east asia find that if the degree of integration of the host country to the international economy rises, inward fdi also increases in the long run. trade openness therefore has significantly positive effect on fdi. in addition, agosin and machado (2007) develop an ordinal index in order to measure the openness of fdi policy regimes in developing countries and find that openness is a factor that permits fdi. aw 62 and tang (2010) as well as oladipo (2010) find that trade openness has significant positive effect on trade and investments. their results suggest that liberalization in nigeria has significant impact on the economy with higher levels of exports due to a more flexible trade policy. similarly, ho and rashid (2011) on asean also find that trade openness is significant in attracting fdi. in a recent study, rehman and raza (2011) apply export of goods and services as a percentage of gdp as a proxy to measure trade openness in pakistan and find that trade openness has negative significant impact on fdi flow in pakistan. another study by sekkat and veganzones-varoudakis (2007) assess the importance of openness in developing countries and find that increased openness as well as improvement in the other aspects of investment have contributed greatly in attracting fdi. their result indicates that efforts towards openness should be initiated or further strengthened in order to make economies attractive to foreign investors. in contrast, vijayakumar and sridharan (2010) find no significant effect of trade openness on fdi inflows in brics countries. they suggest that trade openness may not be a crucial factor in attracting fdi inflows to these fast emerging countries. similarly, mateev (2009) identifies several key determinants of fdi flows into transition economies of central and south-eastern europe and indicates that trade openness is statistically insignificant. borensztein, et al. (1998) test the relations of foreign direct investment (fdi) and economic growth in sixty-nine developing countries and show that financial development has insignificant effect on fdi. woo (2009) investigates the effect of fdi on total factor productivity (tfp) growth in a large sample of countries from 1970 to 2000 and confirms that financial development is positively associated with fdi. oladipo (2010) identifies the determinants of foreign direct investment flow in a small open economy and finds that the level of financial development had negative implication on the growth fdi. it is surprising to note that financial development may not necessary induce fdi in smaller economies. another study by wang and wong (2009) find complementary relation between financial depth and fdi and when financial depth increases, fdi growth increases but at a decreasing rate. a country’s exchange rate indirectly determines the competitiveness of its production. some studies show that the stability of a host country’s currency value attracts more foreign investment into host country (rogoff and reinhart, 2003; brzozowski, 2006; ho and rashid, 2011). alba, et al. (2009) also examines the impact of exchange rates in the united states (u.s.) and finds mixed results. macdermott (2008) determines the relationship between the volatility in exchange rate and fdi and indicates that devaluation of the home currency would lead to a fall in fdi. in addition, volatility in exchange rate discourages foreign investor to invest in a host country and a favorable fdi environment is therefore vital for fdi inflows. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 4 63 inflation affects prices of a country’s produce and therefore directly affects trade and investment. hailu (2010) finds that inflation has negative relation with fdi. when inflation rate increases, fdi decreases. al-nasser (2007) also finds that inflation is an important determinant of fdi with statistically significant negative relationship. a panel data analysis on location determinants of chinese and indian foreign directs investment by duanmu and guney (2009) find that inflation has insignificant relationship towards fdi. in addition, gast and herrmann (2008) identify the determinants of foreign direct investment for organization of economic cooperation and development (oecd) countries from 1991 to 2001 and find that inflation is negatively related to fdi. wijeweera, et al. (2010) suggest that when government increases its spending towards education, quality of human capital improves and results in increasing fdi inflows. kyrkilis and pantelidis (2003) conclude that interest rate is statistically significant in affecting fdi in italy, the netherlands and korea. c. country specific factors some researchers have applied economic freedom index as a determinant for fdi inflows and empirical evidence indicates that there is a positive significant relationship between this index and economic growth (ayal and karras, 1998; heckelman, 2000). according to the heritage foundation,1 economic freedom is defined as the fundamental rights of every human to control his or her own labor and property. in addition to researching the relation between economic growth and economic freedom, researches have also employed economic freedom as a variable in fdi. quazi (2007) indicates that a host country should formulate its fdi strategies by focusing on economic freedom components in order to attract more fdi inflow into the country. according to caetano and caleiro (2009) the risk perceived by investors is crucial in their decision to invest especially when it concerns a particular foreign country. prior results indicate that economic freedom and fdi are positively related. the literature that examines the relation between fdi, wages and productivity of a host country has found empirical evidence that host country with more conducive environment for fdi in terms of fixed entry cost and lower wages tend to attract more foreign investor (lipsey and sjoholm, 2010; hayakawa, et al. 2010). jensen and rosas (2007) find that fdi and wages in mexico are positively significant. mexico is among the fast emerging countries where labour rate is still very low relative to other emerging countries and therefore it could still attract foreign investments into the country with increasing wage rate. moreover, blanton and blanton (2007) find that foreign investors prefer countries with not only well trained and skilled labor but also low cost of wages. 1 the heritage foundation is the official website that provides information on the world economic freedom. the organization works with wall street journal to monitor the movements of economic freedom around the world. their result is released via the economic freedom index. 64 according to suh and boggs (2011), ict infrastructure has positive significant relation with investments in developed market but the results are insignificant in recent years. hailu (2010) also shows that infrastructure is positively related to fdi flows in african countries. al-nasser (2007) confirms that infrastructure is one of the important factors in attracting foreign investor in latin america and asia. in contrast, quazi (2007) finds insignificant relation between infrastructure and fdi in east asia. similar to vijayakumar, et al. (2010), rehman and raza (2011) also find positively significant relation between infrastructure and fdi for pakistan, and zhang (2001) for china. again, similar to gast and herrman (2008) on oecd countries, hailu (2010) finds that human capital has insignificant effect on fdi in africa. further, kyrkilis and pantelidis (2003) and blanton and blanton (2007) also indicate that human capital has positive effect on fdi flows into non-oecd countries. 3. data and methodology annual data from 1977 to 2010 for brazil, russia, china, india, south africa and malaysia are from imf international financial statistics (ifs), and department of statistics and central bank of each country. economic freedom index is from the heritage foundation. eleven variables that consist of macroeconomic fundamentals and country specific variables are designated as independent variables. the changes in the variables are computed and are used in the analysis in order to avoid spurious correlations. four techniques including ols, seemingly unrelated regression (sur), fixed country effect panel, and granger-causality tests are applied in order to determine significant results. panel data analysis uncovers dynamic relationships and blends intercountry differences and intracountry dynamics with both cross-sectional and time-series data. it provides more accurate inferences of model parameters wih less multicollinearity problems and hence improve the efficiency of econometric estimates while taking into account individual country effects. fixed effect panel model using generalized least square (gls) produces robust estimates. macroeconomic fundamentals include market size, trade openness, financial development, exchange rate, interest rate, government consumption and inflation rate. these are used under the label of model 1. the country specific variables are economic freedom, wages, human capital, and infrastructure quality. these enter into the analysis labeled as model 2. the list of variables and their proxies are included in table 1. macroeconomic fundamentals (model 1) 0 1 2 3 4 5 6 7it it it it it it it itfdi gdp open findepth er ir ifr govα β β β β β β β ε= + + + + + + + + (1) international journal of banking and finance, vol. 10, iss. 1 [2013], art. 4 65 country specific factors (model 2) 0 1 2 3 4it it it it itfdi a w free lr inqλ λ λ λ δ= + + + + + (2) table 1: proxy of each variable and expected relationship with fdi inflow both augmented dickey-fuller (adf) and kwiatkowski-phillips-schmidt-shin (kpss) unit root tests for individual countries as well as panel unit root tests are applied in order to check for robustness of the time series. panel unit root test results are shown in table 2.2 all the macroeconomic fundamental and country specific time series are transformed to ensure there is no unit-root problem and that all the time series used in the tests are stationary. table 2: panel unit root test results levin, lin & chu t* im, pesaran and shin w-stat adf fisher chisquare pp fisher chisquare fdi 589.196 -7.04907* 70.2228* 101.423* gdp -5.79735* -4.92477* 47.2639* 63.5560* open -5.63255* -5.38188* 51.3382* 86.8010* fin depth -5.84644* -4.98788* 51.2975* 85.6724* er -0.44364 -4.53172* 44.2583* 281.100* gov -4.82138* -4.23113* 40.0638* 60.6726* ifr -1.04584 -3.94876* 38.2805* 162.164* ir -3.87420* -7.18459* 72.2365* 119.459* w -1.600145*** -3.71962* 37.1395* 67.6616* lr 56.1898 -2.14977** 30.6987* 70.6587* free -4.36504* -3.57767* 35.2047* 76.8976* inq -30.5461* -15.5459* 48.7461* 96.9415* note: llc, ips and adf and pp-fisher tests have null hypothesis of the existence of a unit root in any of the series in the panel. probabilities for fisher tests are computed using an asympotic chi-square distribution. all other tests assume asymptotic normality. for a group or pool unit root test, the eviews default is to use automatic selection methods: information matrix criterion based for the number of lag difference terms (with automatic selection of the maximum lag to evaluate), and the newey-west method for bandwidth selection. *, ** and *** denote statistical significance at 1, 5 and 10 %. 4. findings and interpretation individual results for models 1 and 2 on fundamental and country specific factors are included in tables 3 and 4. 2 detailed individual country unit root test results are available from the authors. variables proxy expected relationship total foreign direct investment (fdi) annual fdi market size (gdp) annual gdp positive trade openness (open) ( x + m) / gdp positive financial development (fin depth) m2/gdp positive exchange rate (er) exchange rate/us dollar negative interest rate (ir) annual interest rate negative government consumption (gov) government consumption negative inflation rate (ifr) consumer price index positive wages (w) % of total employed negative human capital (lr) literacy rate positive economic freedom (free) index economic freedom positive infrastructure quality (inq) log per capita electricity positive 66 the results for brazil show that there is marginally significant positive relation between interest rate and fdi. an increase in the country’s interest rates would result in an increase in the country’s fdi. besides that, positive relation is also found between fdi and financial depth, trade openness, economic growth, government consumption and exchange rate. the overall result for country specific variables in model 2 is statistically significant (probability of f-significance of 0.074). the results for brazil show that economic freedom and infrastructure quality (inq) have statistically significant positive relation with fdi. improvement in domestic infrastructure and policies to allow more freedom to make decisions would improve the potential for attracting foreign investments into host countries. table 3: macroeconomic factors and fdi inflow for brics variables brazil russia india china s africa malaysia c 0.088 (0.738) 0.328 (0.169) -0.126 (0.678) --0.296 (0.417) 4.058 (0.675) -1.438 (0.034)** gdp 0.057 (0.981) 4.156 (0.016)** 3.524 (0.409) 5.217 (0.095)* 27.211 (0.667) 15.424 (0.0215)** open 2.246 (0.287) 0.961 (0.595) 2.235 (0.118) 0.0735 (0.931) -9.718 (0.788) 6.8678 (0.087)* fin depth 0.205 (0.774) -0.582 (0.366) 0.313 (0.315) 0.090 (0.997) 3.641 (0.432) ifr -0.020 (0.939) -0.425 (0.162) 0.203 (0.964) -1.771 (0.595) -90.133 (0.367) 7.252 (0.712) gov 1.431 (0.505) -4.521 (0.021)** -0.545 (0.873) -2.438 (0.109) -34.447 (0.578) -6.756 (0.163) ir 0.984 (0.085)* 0.259 (0.568) 1.158 (0.376) -0.477 (0.060)* -8.862 (0.659)** -0.837 (0.784) er -0.011 (0.692) -0.239 (0.616) 4.493 (0.004)*** 4.387 (0.163) adjusted r2 0.035 0.435 0.001 0.527 0.001 0.460 f-significance 0.383 0.089* 0.500 0.041** 0.950 0.008*** note: * indicates 10% significant level, **5% significant level, and ***1% significant levels. market size (gdp) and government expenditure are significant in affecting fdi flows for russia where the potential of a large market for goods and services is a vital factor in attracting investments. the overall model for russia is also statistically significant in explaining fdi into russia. government consumption has negative effect on fdi where contractionary fiscal policy is inversely related to foreign investments and foreigners perceive higher government consumption as detrimental to investments. financial depth is excluded in model 1 for russia due to multicollinearity effects with gdp, and robustness test performed confirm that it is not significant in affecting fdi. in addition, results for model 2 verify that literacy rate and infrastructure quality are important considerations for foreign investment. improvement in accessibility to markets and export destinations would encourage investors locating themselves in certain geographically challenged areas. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 4 67 based on theoretical expectation, exchange rate is significant in affecting fdi decisions in india. the results however suggest that a fall in the currency value would encourage foreign investors into india. the results for country specific factors are not significant, though there exists positive relation between fdi, wages and infrastructure, and negative relation between fdi, literacy rate and economic freedom. the model for china is statistically significant and affirms that market size is significant in affecting fdi flows. foreign investors seeking new markets in china are attracted by the potential of a large market size. in recent years, china has attracted a substantial amount of fdi which enable the economy to grow at tremendous rate relative to the rate of world growth. in addition, interest rate is another macroeconomic fundamental that possess significant relation with fdi in this country where increase in interest rates which increase production costs discourages foreign investments. significant relations are also found for employment, infrastructure and economic freedom for country specific model for china in model 2. higher employment and therefore indirectly higher cost of labor is found to discourage foreign investments in china. the economic freedom index is found to be negatively related to fdi where freedom in decisions would probably reduce the power of foreign firms in dictating the market. this scenario is typically found in newly open economies but not the others. similar to other fast emerging countries, infrastructure improvements would enable foreign organizations to move their goods to market efficiently and this would encourage fdi across all countries dictating positive relationship. the f-statistics for the china model is also statistically significant. table 4: country specific factors and fdi inflow for brics variables brazil russia india china s africa malaysia constant -0.209 (0.383) 0.364 (0.046)** 0.157 (0.725) -0.086 (0.477) -6.955 (0.136) 1.811 (0.162) w 5.957 (0.113) 0.401 (0.409) 0.433 (0.166) -0.796 (0.015)** 31.373 (0.500) -14.956 (0.140) lr -10.128 (0.771) -652.1 (0.028)** -1.233 (0.942) -1.4137 (0.760) 474.640 (0.005)*** 347.944 (0.301) inq 7.083 (0.015)** 2.335 (0.007)*** 1.342 (0.879) 2.384 (0.023)** 12.024 (0.072)** 19.923 (0.028)** free 2.673 (0.066)* 3.470 (0.111) -4.194 (0.316) -1.520 (0.012)** -72.735 (0.144) 34.763 (0.065)* adjusted r2 0.359 0.214 0.001 0.405 0.001 0.326 f-significance 0.074* 0.218 0.787 0.053* 0.506 0.092* note: *10% significant level, **5% significant level, ***1% significant level interest rate is the only macroeconomic fundamental that significantly affect fdi in south africa. fast emerging countries including china and south africa find that increasing interest rate increases the cost of doing business and discourages not only foreign investments but also domestic investments. education is a significant driver of investments when skilled and knowledgeable 68 workforce improves productivity and reduces costs in south africa. in addition, infrastructure improvements also enable emerging countries to accelerate growth and boost productivity through foreign investments. it is important to note that market size is a significant factor in attracting fdi into emerging countries including russia, china and malaysia where the models tested are statically significant. openness to trade indicated by export and import activities enables foreign investors to distribute their goods not only in the domestic economy but also to neighboring markets which may not be attractive for production facilities. trade openness is found to be significant to malaysia only but not to the other fast emerging countries and the overall model is statistically significant. there is generally positive relation between trade openness and fdi in emerging markets. the set of country specific factors which is positively significant for malaysia include infrastructure and economic freedom and the model is statistically significant. similar to the other fast emerging countries, positive significant effect of improvement in infrastructure in attracting fdi is also found for malaysia. this underscores the findings for fast emerging countries and justifies additional government expenditure on infrastructure to improve not only domestic logistics but also increasing the development of the domestic economy through foreign investments. the results of the seemingly unrelated regression (sur) and fixed effect panel data analyses are shown in table 5. the significance and robustness of both macroeconomic fundamentals and country specific determinants of fdi into brics and malaysia are verified in the reults. there is significant relation between market size, trade openness, government consumption and fdi for this group of emerging economies. the larger the market potential, the more attractive it is for multinational companies’ investment due to larger demand, which in turn enables production to achieve cost advantage through economies of scale. attractive trade policy that enables openness to trade in not only imports of capital and raw materials but also exports of finished products to nearby regions would also result in higher level of fdi. in addition, responsible government management in consumption and taxation are also magnets for foreign investments. it is important to note that all the sur and panel models are statistically significant in driving fdi inflows for this group of country as a whole. the analyses also suggest that infrastructure quality and economic freedom are significant country specific determinants of fdi in model 2. multinational companies aim to invest in countries that would enable them to take advantage of not only the market potential with responsible governments but also countries where they can smoothly operate their business with ease. one major attributes they seek is the ability of the host country to support infrastructure development which would prevent logistics issues. developing countries are therefore allocating a large budget for 12 international journal of banking and finance, vol. 10, iss. 1 [2013], art. 4 69 improving their respective infrastructure not only for domestic economic development but also to attract foreign investments. future studies should also include other variables to investigate fdi in addition to those included in this study. in summary, the results are consistent and robust from both the individual and panel data analyses. table 5: sur and fixed effect analysis of fdi in brics and malaysia sur fixed effect model 1 model 2 model 1 model 2 gdp 4.031 (0.001)*** 3.139 (0.000)*** open 3.376 (0.005)*** 1.195 (0.063)* fin depth 0.641 (0.347) -0.006 (0.985) ifr -0.003 (0.995) -0.036 (0.875) gov -1.402 (0.215) -1.646 (0.024)** ir 0.127 (0.841) 0.438 (0.202) er 0.022 (0.623) 0.001 (0.967) w 0.513 (0.121) -0.093 (0.791) lr -4.053 (0.507) -3.377 (0.495) inq 1.269 (0.081)* 2.466 (0.002)* free 0.422 (0.716) -1.739 (0.058)* constant -0.531 (0.020)** -0.127 (0.296) -0.858 (0.000)*** -0.309 (0.000)*** adjusted r² 0.114 0.073 0.157 0.133 f-prob 0.004*** 0.036** 0.002*** 0.015** note: * is 10% significant level, ** 5% significant level, and ***1% significant levels table 6 provides the results of granger causality tests. there is evidence to support market size granger cause fdi in brazil, russia and china. this confirms that market size plays a vital role in attracting foreign investment into these countries. trade openness is also found to granger cause fdi in brazil where open flow of trade attracts investments into the country. financial depth is found to granger cause fdi in china where fdi is attracted by financial market development in the country where liquidity and financing would not pose any problem. in addition, interest rate is found to granger cause fdi in malaysia where foreign investors are very sensitive to financial costs when making investment decisions. there is also some evidence that fdi granger cause inflation and exchange rate in china. fdi granger cause government consumption in india, and interest rate changes in brazil and russia. for 70 the set of country specific factors, there is evidence to conclude that employment granger cause fdi in russia, and infrastructure quality granger cause fdi in china. in addition, it is also found that fdi granger cause education in russia and labor cost in malaysia. table 6: granger causality results for model 1 and 2 brazil russia india china s africa msia panel gdp does not granger cause fdi 2.652 0.092* 4.107 0.042** 0.783 0.468 3.910 0.035** 0.356 0.704 0.362 0.699 0.089 0.915 open does not granger cause fdi 5.596 0.011** 0.378 0.695 0.786 0.467 1.079 0.358 1.046 0.367 1.559 0.233 1.418 0.246 find does not granger cause fdi 0.739 0.488 0.924 0.519 1.113 0.363 4.366 0.040** 0.066 0.936 1.207 0.322 0.059 0.943 fdi does not granger cause ifr 0.593 0.560 0.189 0.829 1.788 0.189 2.731 0.086* 0.531 0.595 1.864 0.178 0.018 0.982 gov does not granger cause fdi 3.005 0.070* 0.043 0.958 0.893 0.424 0.172 0.843 0.416 0.665 0.784 0.469 0.602 0.549 fdi does not granger cause gov 0.402 0.674 1.375 0.285 3.480 0.049** 0.074 0.929 0.038 0.962 0.144 0.867 0.237 0.789 ir does not granger cause fdi 0.778 0.472 1.515 0.256 0.083 0.921 1.723 0.203 0.138 0.871 3.426 0.050** 0.017 0.983 fdi does not granger cause ir 4.042 0.032** 3.182 0.075* 0.461 0.637 0.139 0.871 1.278 0.298 0.098 0.392 1.026 0.361 fdi does not granger cause er 0.833 0.186 0.689 0.524 0.700 0.506 6.501 0.005*** 0.372 0.693 0.657 0.527 0.006 0.994 w does not granger cause fdi 0.765 0.494 11.787 0.008*** 0.187 0.831 1.504 0.243 2.622 0.111 0.489 0.619 0.350 0.705 fdi does not granger cause w 1.410 0.293 0.789 0.496 0.228 0.797 2.408 0.111 0.456 0.644 2.866 0.078* 0.132 0.876 fdi does not granger cause lr 0.240 0.789 2.549 0.100* 0.613 0.551 0.895 0.422 0.276 0.761 1.116 0.344 0.075 0.928 inq does not granger cause fdi 1.478 0.246 0.687 0.521 0.447 0.644 9.715 0.001*** 0.034 0.966 0.560 0.580 0.509 0.602 note: * indicates 10% significant level, ** 5% significant level, and ***1% significant levels. in summary, granger causality tests confirm that there are significant causality effects from market size, trade openness, financial development, government consumption and interest rate to fdi of emerging countries. in addition, for the country specific model, employment and infrastructure quality are found to granger cause fdi in brics and malaysia. 5. conclusion this study provides empirical evidence for the determinants of fdi in brics and malaysia using individual country analyses, seemingly unrelated and panel regressions, and granger causality tests over the period of 1977 to 2010. most of our findings are theoretically supported and statistically significant. economic freedom is a novel variable that is considered and it is found to be statistically significant. findings from this research add new knowledge to the understanding of fdi flows into international journal of banking and finance, vol. 10, iss. 1 [2013], art. 4 71 fast emerging countries. in addition, the results are expected to assist emerging nations to strategize policies for managing fdi flows. the findings suggest that trade openness, economic growth, government expenditure, infrastructure quality are the most critical factors in determining fdi flows in brics and malaysia. we also note with interest that economic freedom is found to significantly affect fdi in brazil, china and malaysia. prior literature on economic freedom has also found significant relation with fdi inflows. further, and again in line with prior studies, an increase in a country’s trade openness seem to exert positive effect on fdi inflow. different factors might affect the countries differently and some factors may be more significant in one country than others. this study has found significant positive relation between infrastructure quality and fdi flows in brazil, russia, china, south africa and malaysia. this variable is one of the critical country specific determinants of fdi flows. the results of sur and fixed effect panel models indicate that economic growth, openness, infrastructure quality and economic freedom are important for fdi flows. robustness tests confirm the significance of these factors for both models 1 and 2. in addition, granger causality tests also indicate that economic growth, trade openness, government consumption, and interest rate granger caused fdi in some countries. for the country specific factors, wage rate and infrastructure quality granger cause fdi flows in russia and china, respectively. exploring other determinants is the subject for future research. increasing the length of the data series, and improving theoretical modeling in line with the data are recommended. the accessibility and availability of other country specific variables would further stimulate interest in this area. ___________________________________________________________ authors information: catherine s f ho is a professor at the universiti teknologi mara, shah alam, malaysia. email: catherine@salam.uitm.edu.my tel:603-5544-4792. she holds concurrent positions in the institute of business excellence and arshad ayub graduate business school. noryati ahmad and hayati mohd. dahan are also staff members at the universiti teknologi mara. the authors acknowledge the research assistance provided by khairunnisa amir, linda nasaruddin sia and nurain farahana zainal abidin. the authors acknowledge with thanks the editorial review suggestions and comments from the 2012 mfa conference as well as the useful editing of the paper by the journal’s co-editors. the authors alone are responsible for any error. tel:603-5544-4792 72 references agosin, m. and machado, r. (2007). openness and the international allocation of foreign direct investment, journal of development studies, 43(7), 1234-1247. al nasser, o. (2007). the determinats of the u.s foreign direct investment: does the region matter? global economic review, 36(1), 37-51. alba, j. d., park, d., and wang, p. (2009). the impact of exchange rate on fdi and the interdependence of fdi over time, adb economics working paper series (164), 1-25. artige, l. and nicolini, r. (2006). evidence on the determinants of foreign direct investment: the case of three european regions, crepp working paper 2006/07, centre of research in public economics and population economics, hec-management school, university of liège. aw, t. and tang, t. (2010). the determinants of inward foreign direct investment: the case of malaysia, international journal of business and society, 11(1), 59-76. ayal, b. e. & karras, g. (1998). components of economic freedom and growth: an empirical study, journal of developing areas, 32(3), 327-338. bakir, a. and alfawwaz, t. (2009). determinants of foreign direct investment in jordan, international management review, 5(2), 66-73. blanton, s. l. and blanton, r. g. (2007). what attracts foreign investors? an examination of human rights and foreign direct investment, the journal of politics, 69(1), 143-155. borensztein, e., gregorio, j. and lee, j. (1998). how does foreign direct investmenti affect economic growth? journal of international economics, 45(1), 115-135. brzozowski, m. (2006). exchange rate variability and foreign direct investment: consequences of emu enlargement, eastern european economics, 44(1), 5-24. caetano, j. and caleiro, a. (2009). economic freedom and foreign direct investment: how different are the mena countries from eu, ibusiness, 1(2), 65-74. duanmu, j. and guney, y. (2009). a panel data analysis of locational determinants of chinese and indian outward foreign direct investment, journal of asia business studies, 3(2), 1-15. erdal, f. and tatoglu, e. (2002). locational determinants of fdi in an emerging market economy: evidence from turkey, multinational business review, 10(1), 21-28. gast, m. and herrmann, r. (2008). determinants of foreign direct investment of oecd countries 1991-2001, international economic journal, 22(4), 509-524. hailu, z. (2010). demand side factors affecting the inflow of foreign direct investment to african countries: does capital market matter? international journal of business and management, 5(5), 104-116. hayakawa, k., lee, h. h. and park, d. (2010). the role of home and host country characteristics in fdi: firm-level evidence from japan, korea and taiwan, ide discussion paper (267), 1-25. heckelman, j. c. (2000). economic freedom and economic growth: a short-run causal investigation, journal of applied economics, 3(1), 71-91. ho, c. s. f. and rashid, a. h. m. (2011). macroeconomic and country specific determinants of fdi, the business review cambridge, 18(1), 219-234. janicki, h. and wunnava, p. (2004). determinants of foreign direct investment: empirical evidence from eu accession candidates, applied economics, 36(5), 505-509. jensen, n. m. and rosas, g. (2007). foreign direct investment and income inequality in mexico 19902000, international organization, 61(3), 467-487. kim, j. and rhe, d. (2009). trends and determinants of south korean outward foreign direct investment, the copanhagen journal of asian studies, 27(1), 126-154. kimono, s., saas, d. and driffield, n. (2007). macro determinants of fdi inflows to japan: an analysis of sources country characteristics, the world economy, 30(3), 446-469. kyrkilis, d. and pantelidis, p. (2003). macroeconomic determinants of outward foreign direct investment, international journal of social economics, 30(7), 827-836. lipsey, r. e. and sjoholm, f. (2010). fdi and growth in east asia: lessons for indonesia. ifn working paper (852), 1-49. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 4 73 lokesha, b. k. and leelavathy, d. s. (2012). determinants of foreign direct investment: a macro perspective, the indian journal of industrial relations, 47(3), 459-469. macdermott, r. (2008). linking exchange eates to foreign direct investment, the international trade journal, 22(1), 3-16. mateev, m. (2009). determinants of foreign direct investment in central and southeastern europe, oxford journal, 8(1), 133-149. oladipo, o. (2010). foreign direct investment: determinants and growth effects in a small open economy, the international journal of business and finance research, 4(4), 75-88. quazi, r. (2007). economic freedom and foreign direct investment in east asia, journal of the asia pacific economy, 12(3), 329-344. rehman, a. and raza, a. (2011). determinants of foreign direct investment and its impact on gdp growth in pakistan, interdisciplinary journal of contemporary research in business, 2(9), 198205. rogoff, k.and reinhart, c. (2003). fdi to africa: the role of price stability and currency instability, imf working paper 03/10. russ, k. n. (2007). the endogeneity of the exchange rate as a determinant of fdi: a model of entry and multinational firms, journal of international economics, 71(2), 344-372. sekkat, k. and veganzones-varoudakis, m. (2007). openness, investment climate and fdi in developing countries, review of development economics, 11(4), 607-620. suh, t. and boggs, d. (2011). communication infrastructure and traditional determinants of inward foreign investment: a comparison of developed and emerging markets, competitiveness review: an international business journal, 21 (2), 205-223. torrisi, c. and delaunay, c. j. (2008). fdi in central europe: determinants and policy implication, journal of international finance and economics, 8(4), 136-147. vijayakumar, n. and sridharan, p. (2010). determinants of fdi in brics countries: a panel analysis. international journal of business science and applied management, 5(3), 1-13. vogiatzoglou, k. (2007). vertical specialization and new determinants of fdi: evidence from south and east asia, global economic review, 36(3), 245-266. wang, m. and wong, m. (2009). foreign direct investment and economic growth: the growth accounting perspective, economic inquiry, 47(4), 701-710. wijeweera, a., villano, r. and dollery, b. (2010). economic growth and fdi inflows: a stochastic frontier analysis, the journal of developing areas, 43(2), 143-158. woo, j. (2009). productivity growth and technological diffussion through foreign direct investment. economic inquiry, 47(2), 226-248. zhang, k. h. (2001). what attracts foreign multinational corporations to china? contemporary economic policy, 19(3), 336-346. ijbf7-marina.indd the international journal of banking and finance, vol. 7. number 1: 2010: 19-34 19 stock market contagion in the early stages of the global financial crisis: the experience of the gcc countries imad moosa monash university, australia and kuwait university _____________________________________________________ abstract this study examined stock market contagion from the united states to the markets of the gcc countries during the period 2007-08. these countries (bahrain, kuwait, oman, qatar, saudi arabia and the united arab emirates) were also experiencing accelerating debt levels, overheated real estate markets, and drying up of liquidity. the main hypothesis under investigation is that the collapse of the gcc stock markets did not result purely from contagion, in the sense that these markets did not follow closely the us market during that period. it is argued that local factors were more infl uential in triggering the collapse and that those markets would have collapsed with or without the global fi nancial crisis. the empirical results show rather limited evidence for the effect of u.s. stock prices on gcc stock prices and a much more important role for oil prices. however, neither of these variables alone can explain the behaviour of gcc stock prices during the period under investigation because of the role played by the domestic factors that caused bubbles and crashes. keywords: contagion, gcc markets, oil prices and stock prices, market bubble jel classifi cations: g15, g11, f30. _____________________________________________________ 1. introduction the global fi nancial crisis has resulted from the u.s. subprime crisis, which arose from excessive growth of subprime lending as fi nancial institutions offered nonstandard mortgages to individuals with dubious credit profi les. the crisis has been described as a “credit crunch” and a “credit squeeze”, two terms that differ only to the extent of the curtailment in credit supply. whether it is a “crunch” or a “squeeze”, the present crisis is characterised by a decline in the terms and availability of credit for consumers and businesses, which has adversely affected ijbf ht tp :// ijb f.u um .e du .m y 20 the international journal of banking and finance, vol. 7. number 1: 2010: 19-34 consumption and investment spending, thus putting the economies of the u.s. and other countries in recession. another characteristic of the crisis is disorder in fi nancial markets, as fi nancial institutions fi nd it diffi cult to determine the true (or fair) value of assets that are no longer traded in suffi cient volumes. this crisis is far more complex than earlier crises because fi nancial innovation has (through securitisation) allowed the creation of complex assets whose risk profi les are diffi cult to assess and which are not easy to price. another characteristic of the crisis is the speed with which it spread from the us to other countries, through a contagious effect that was transmitted via several channels. the fi nancial markets and economies of emerging countries have been adversely affected by the crisis. the objective of this paper is to examine and evaluate the effects of the crisis on the six oil-exporting countries of the gulf cooperation council (gcc): bahrain, kuwait, oman, qatar, saudi arabia and the united arab emirates. specifi cally, this paper deals with stock market contagion during the early stages of the crisis (2007-2008). for the purpose of empirical analysis, contagion is proxied by the effect of changes in us stock prices on stock prices in these countries. the empirical work was based on a sample of daily data, covering the period january 2007 to december 2008, on stock prices in seven stock markets, including two uae markets: abu dhabi and dubai. the reason for choosing two markets from the same country is that they are located in two different states (emirates) with different oil endowments and different development policies (more explosive in dubai than in abu dhabi). since we are also going to examine the effect of oil prices, a sample of daily oil prices covering the same period has been collected. data on the gcc markets were obtained from the gulf investment corporation, whereas the dow jones data were obtained from the yahoo finance website. data on oil prices were obtained from bloomberg. 2. effect of the crisis on financial institutions and markets in gcc countries before we examine the effect of the crisis on the stock markets of the gcc countries, it may be useful to consider its effect on fi nancial institutions and other fi nancial markets in these countries. as is the case anywhere else in the world, fi nancial institutions in the gcc countries have been affected by the crisis via several channels: (i) direct exposure to the us housing market, real estate products or cdos; (ii) exposure to institutions (both as shareholders and creditors) that have direct exposure as in (i); and (iii) the effect of the crisis on the liquidity needs of the gcc fi nancial institutions. we start with the direct and indirect exposure to subprime losses. by the end of 2008, the gcc countries held foreign assets valued at $1.8 to 2 trillion, about 60% of which were dollar denominated. only a limited number of gcc banks have publicly admitted exposure to subprime losses, with estimated losses amounting to $3 billion, although some estimates put the fi gure at $10 billion ht tp :// ijb f.u um .e du .m y stock market contagion in the early stages of the global fi nancial crisis: 19-34 21 (hussain, 2008). it is, however, believed that most of the losses resulting from direct and indirect exposure to the subprime market have been incurred by sovereign wealth funds such as the abu dhabi investment authority (adia) and the kuwait investment authority (kia). according to woertz (2008), these funds represent sophisticated investors with a considerable equity component, which means that their exposure to the current market turmoil must be considerable. they probably invested in risky assets, such as cdos, more than banks. the kia injected massive amounts into merrill lynch, a us investment bank that has experienced a substantial loss of value as a result of direct exposure to “toxic assets”. the unavailability of exact and detailed fi gures prevents us from saying anything defi nitive about the direct and indirect losses incurred by the gcc fi nancial institutions. as far as liquidity needs are concerned, woertz (2008) argued that “while the direct subprime exposure of gcc banks has been limited thus far, the real problem for the gcc banks lies in the direct exposure to increased costs of funding amidst maturity mismatches and credit exposure to local consumers, project and real estate fi nancing”. what makes things worse is that even before the onset of the crisis, there was already a liquidity squeeze resulting from negative real interest rates. foreign funds used for speculating on what was thought to be an imminent revaluation of local currencies against the us dollar were withdrawn as speculators realised that it would not take place. foreign investors also pulled out because of the need for short-term capital to rebalance their portfolios and cover losses, leading to the removal of liquidity from the banking system, particularly from the subsidiaries of foreign banks. as a result of the crisis, fi nancial institutions in the gcc faced (and still do) constrained fi nancing conditions while enduring losses resulting from exposure to the local stock and real estate markets, both of which had experienced bursting bubbles. it seems, therefore, that the status quo has not resulted entirely from, but it has been exacerbated by, the global fi nancial crisis. according to the business monitor international (2008), the current liquidity crunch in gcc countries is not particularly linked to the global fi nancial crisis, and it is the result of strong demand by a rapidly growing private sector. the crisis has also affected gcc fi nancial markets other than stock markets. the crisis has to a certain extent disrupted the syndicated loan markets. in the second half of 2008, interbank rates across the gcc soared. in the uae, the usually liquid one-year market has almost dried up and the three month rates rose 230 basis points in three months. the rise materialised despite cuts in policy rates and the fall in libor, which libor has historically followed. the other liquid regional interbank markets saw similar trends, with gulf spreads on libor moving far above the medium-term norm. also, spreads (over libor) for corporate bonds in the gcc rose from around 145 at the end of 2007 to 500 basis points at the end of 2008. gcc countries were affected through exposure to foreign fi nancial markets that have brought in large private capital fl ows to the region during the recent boom years. ht tp :// ijb f.u um .e du .m y 22 the international journal of banking and finance, vol. 7. number 1: 2010: 19-34 a. stylised facts on the effect of the crisis on gcc stock markets figure 1 displays daily stock prices over the period 2 january 2007-31 december 2008 (or close to the year end) along with the dow jones (numbers on the horizontal axis refer to daily observations). a careful examination of the data reveals that the gcc stock markets have endured losses not only because of the global fi nancial crisis but also because these markets were in a bubble before the recognition of the crisis, a bubble that was created by purely domestic factors, as the previous discussion indicates. the locations of turning figure 1. stock prices relative to the dow jones (2/01/07=100) continued ht tp :// ijb f.u um .e du .m y stock market contagion in the early stages of the global fi nancial crisis: 19-34 23 points in gcc stock prices relative to the dow jones indicate decoupling of the gcc markets from the us market. consider the fi gures in table 1, which reports changes in stock prices in the seven gcc markets normalised on the value of 100 on 2 january 2007. the column “last observation” refers to the price at the end of the sample period. the peak value refers to the highest value of the price and when that value occurred (month/year). we can see that most markets peaked in june 2008, whereas the dow reached its peak value in october 2007. thus, even as the us stock market was declining because of the crisis, the gcc stock markets kept on rising. table 1 stock price changes (02/01/07-31/12/08) market first observation last observation peak value (mm/yy) first to peak (%) peak to last (%) abu dhabi 100 79.75 171.81 (06/08) 71.81 -53.58 bahrain 100 81.04 130.39 (06/08) 30.39 -37.84 dubai 100 39.66 152.52 (01/08) 52.52 -73.99 kuwait 100 77.84 156.58 (06/08) 56.38 -50.28 oman 100 96.72 215.25 (06/08) 115.52 -55.06 qatar 100 91.23 177.33 (06/08) 77.33 -45.46 saudi arabia 100 60.26 149.26 (01/08) 49.26 -59.62 (u.s.) dow jones 100 70.54 113.54 (10/07) 13.54 -38.04 the column “fi rst to peak” refers to the percentage change in stock prices between the fi rst observation and the peak. stock markets of the gcc countries rose much faster than the u.s. market, which registered a rise of 13.54%, compared with the omani market that rose by 115.52%. the last column “peak to last” shows the percentage change in stock prices between the peak and the figure 1. stock prices relative to the dow jones (2/01/07=100) ht tp :// ijb f.u um .e du .m y 24 the international journal of banking and finance, vol. 7. number 1: 2010: 19-34 last observation. apart from bahrain, gcc markets lost more of their values than the u.s. market as represented by the dow. these fi gures provide prima facie evidence for the proposition that the gcc markets experienced their own bubbles and crashes, which were more vibrant (in the case of bubbles) and more severe (in the case of crashes) than the corresponding bubbles and crashes exhibited by the dow. table 2 reports correlations of the gcc markets with the u.s. market. correlations are not particularly high and they differ signifi cantly from one market to another, ranging between zero in the case of oman and 0.52 for saudi arabia and dubai. volatility, as measured by the standard deviation of the daily percentage changes in stock prices, also shows signifi cant variation, with some markets being almost three times as volatile as the u.s. market. again, these fi gures indicate that domestic factors that had no relation to the global fi nancial crisis propelled the gcc markets, causing the bubble and crash behaviour that we have seen. table 2 stock price correlation and volatility market correlation with u.s. standard deviation of daily percentage changes abu dhabi 0.18 1.52 bahrain 0.34 0.71 dubai 0.55 2.10 kuwait 0.34 0.90 oman 0.00 4.80 qatar 0.09 1.89 saudi arabia 0.52 4.96 (u.s.) dow jones 1.00 1.81 3. the effect of contagion a. methodology and empirical results the methodology used to test the contagion hypothesis in this paper was based on the structural time series model of harvey (1989, 1997), which is used to represent an observed time series in terms of its unobserved components. this model may be written as: (1) where p t is the logarithm of stock prices,  t is the trend component,  t is the cyclical component, and  t is the random component, which is assumed to be white noise. the trend component, which represents the long-term movement ttttp ht tp :// ijb f.u um .e du .m y stock market contagion in the early stages of the global fi nancial crisis: 19-34 25 1 harvey (1989, pp. 510-511) lists all of the possibilities. koopman, harvey, doornik and shephard (1995) identify several models including: a constant term model, local level model, random walk with and without drift, local level with fi xed slope, a smooth trend model, etc. 2 for details of the estimation method, see harvey (1989, chapters 4 and 7) and koopman et al. (1995, chapter 14). for applications of this technique in economics and fi nance, see moosa (2006) and moosa and gazos (2009). in a series, is assumed to be stochastic and linear. this component can be represented by:  (2) (3) where  t is a random walk with a drift factor, t , which follows a fi rst order autoregressive process as represented by equation (3). this is a general representation of the trend, which encompasses all other possibilities. for example, this process collapses to a simple random walk with drift if , and to a deterministic linear trend if as well. if, on the other hand, while then the process will have a trend which changes relatively smoothly.1 the cyclical component, which is assumed to be a stationary linear process, may be represented by: (4) where t is time and the amplitude of the cycle is given by (ab2) . in order to make the cycle stochastic, the parameters a and b are allowed to evolve over time, while preserving continuity is achieved by writing down a recursion for constructing before introducing the stochastic components. if disturbances and a damping factor are also introduced, we obtain: (5) such that and  0 = a and  t are t * uncorrelated white noise disturbances with variances and , respectively. the parameters 0 ≤≤ and 0≤≤ are the frequency of the cycle and the damping factor on the amplitude, respectively. once it has been written in this form, the model can be estimated by maximum likelihood, using the kalman fi lter to update the state vector (whose elements are the time series components), as new observations become available. related smoothing algorithms can be used to obtain the best estimate of the state vector at any point in time within the sample period.2 1 2 2 2 * b*0 . t t t t1 1 t t t1 t nid~ ( , )0 2 , and t nid~ ( , )0 2 . 2 0 , 2 0 2 0 2 0 t a t b tcos sin t t t t t t * * * cos sin sin cos 1 1 ht tp :// ijb f.u um .e du .m y 26 the international journal of banking and finance, vol. 7. number 1: 2010: 19-34 3 no seasonal component is included in equation (6) because daily data are used in this study. 4 the modifi ed coeffi cient of determination, ( 2dr ) is calculated on the basis of the fi rst differences rather than the levels of the variables, and it is more appropriate for trended data. a model with a positive 2dr is better than a random walk with a drift model ( 2 dr can be negative). the q statistic is calculated as ),( knq ) where n is the number of autocorrelation coeffi cients and k is the number of estimated parameters. it has a )(2 k distribution. the test statistic for heteroscedasticity, h(h), is calculated as the ratio of the squares of the last h residuals to the squares of the fi rst h residuals, where h is the closest integer to one third of the sample size. it is distributed as f(h,h). to test the contagion hypothesis, the basic structural time series model represented by equation (1) must be modifi ed by introducing an explicit explanatory variable. thus, the modifi ed model is: (6) where p i , t is the logarithm of the stock price index in market i, such that i = 1,2,...,7 and p us,t is the logarithm of us stock prices measured by the dow jones index.3 the hypothesis of contagion is confi rmed if the coeffi cient on p us,t is positive and statistically signifi cant. the beauty of this model is that it allows us to examine the effect of factors other than us stock prices even without identifying them explicitly. the effect of these factors is refl ected in the behaviour of the components μ and . if these components are statistically signifi cant while  is also signifi cant, then this means that factors other than us stock prices affect stock prices in the gcc countries. if, on the other hand, the components are insignifi cant while  is signifi cantly positive, this means that the only factor affecting stock prices in the gcc countries is us stock prices, thus providing very strong evidence for the contagion hypothesis. if, however,  turns out to be insignifi cant while the components are signifi cant, this means that only other factors affect stock prices in the gcc countries, which provides evidence against the contagion hypothesis. table 3 reports the results of estimating equation (6) over the whole sample period. the results include the fi nal state vector (with the t statistics given in parentheses), the coeffi cient of determination (r2), the modifi ed coeffi cient of determination ( ) and three diagnostic test statistics: the durbin watson (dw) statistic, the ljung-box (1978) test statistic for serial correlation (q), and a test statistic for heteroscedasticity (h).4 we can see that the model is reasonably well-determined in terms of goodness of fi t and the diagnostic tests (all equations have positive ). in all cases the level of the trend (which is equivalent to the constant term in a conventional regression) is highly signifi cant, whereas the slope (which is equivalent to the coeffi cient on a deterministic time trend in a conventional regression) is signifi cantly negative in fi ve cases. titustititi pp ,,,,, 2 dr 2 dr ht tp :// ijb f.u um .e du .m y stock market contagion in the early stages of the global fi nancial crisis: 19-34 27 table 3 estimation results of equation (6): whole period* abu dhabi bahrain dubai kuwait oman qatar saudi arabia μ 4.682 4.362 3.102 4.439 3.778 4.275 3.226 (31.19) (62.71) (14.91) (51.38) (22.61) (22.08) (14.83)  -0.008 -0.004 -0.012 -0.008 -0.006 -0.002 -0.004 -3.31) (-3.83) (-3.86) (-3.84) (-2.53) (-0.69) (-1.64)  0.006 -0.005 0.008 -0.008 0.015 -0.008 0.005 (1.45) (-2.79) (1.74) (-3.78) (3.29) (-1.78) (1.69) * -0.004 -0.005 -0.011 0.003 0.012 -0.003 -0.007 -0.92) (-2.93) (-2.34) (1.49) (2.60) (-0.68) (-2.15)  -0.074 0.009 0.134 -0.018 0.183 0.065 0.206 -1.07) (0.54) (2.74) (-0.89) (4.66) (1.43) (3.98) r2 0.90 0.95 0.92 0.99 0.96 0.86 0.88 0.07 0.10 0.08 0.15 0.11 0.03 0.05 dw 1.83 1.85 1.99 1.87 1.78 1.87 1.94 q 26.03 27.63 20.95 21.85 26.87 23.98 25.73 h 1.19 1.24 1.20 1.23 1.26 1.27 1.19 * q is distributed as 2 (18), which gives a 5 % critical value of 28.86. h is distributed as f(165,165), which gives a 5% critical value of 1.29. table 4 estimation results of equation (6): first sub-sample* abu dhabi bahrain dubai kuwait oman qatar saudi arabia  5.120 4.327 4.432 4.896 5.022 4.572 4.786 (14.73) (20.88) (9.62) (26.25) (20.10) (12.68) (8.15)  0.003 0.0004 0.000 0.001 0.006 0.008 -0.001 (1.33) (0.37) (0.17) (0.87) (3.01) (2.32) (-0.30)  0.002 0.000 -0.0005 0.003 0.005 0.003 -0.002 (1.04) (0.01) (-0.26) (3.66) (1.70) (0.75) (-0.25) * 0.006 0.000 0.005 -0.001 0.004 0.0002 -0.002 (3.45) (0.001) (2.50) (-0.39) (1.39) (0.05) (-0.14)  -0.069 0.088 0.043 -0.007 -0.035 0.039 -0.045 (-0.94) (1.02) (0.44) (-0.14) (-0.65) (0.51) (-0.35) r2 0.94 0.92 0.90 0.95 0.89 0.99 0.82 0.05 0.03 0.01 0.06 0.01 0.08 -0.005 dw 1.78 1.92 2.00 1.97 1.77 1.83 1.97 q 12.70 13.58 10.27 7.21 8.80 12.99 4.59 h 1.02 1.36 1.18 0.71 0.89 0.49 0.28 * q is distributed as 2 (18), which gives a 5% critical value of 16.91. h is distributed as f(63,63), which gives a 5% critical value of 1.52. 2 dr 2 dr ht tp :// ijb f.u um .e du .m y 28 the international journal of banking and finance, vol. 7. number 1: 2010: 19-34 5 the fi rst sub-sample extends between 2 january 2007 and 4 october 2007. the second sub-sample covers the rest of the period until the end of 2008. 6it seems, therefore, that the positive effect of us stock prices on the markets of dubai, oman and saudi arabia over the whole sample period is a refl ection of the effect arising in the second sub-sample, when the dow was going down. these markets were affected by the decline, but not by the rise, of the dow, which means that the effect is asymmetric. abu dhabi bahrain dubai kuwait oman qatar saudi arabia  4.668 4.388 3.029 4.446 3.879 4.266 3.129 (25.23) (53.78) (11.32) (40.84) (18.45) (17.52) (12.02)  -0.008 -0.005 -0.003 -0.007 -0.005 -0.002 -0.004 (-2.82) (-3.46) (-2.13) (-3.12) (-2.33) (-0.72) (-1.28)  0.007 -0.005 0.007 -0.008 0.026 -0.007 0.005 (1.310 (-2.37) (1.27) (-3.01) (-1.32) (-1.35) (1.15) * -0.004 -0.005 -0.001 0.003 0.053 -0.002 -0.007 (-0.69) (-2.36) (-1.95) (1.29) (1.92) (-0.47) (-1.76)  -0.070 0.003 0.152 -0.019 0.170 0.068 .225 (-1.61) (0.14) (2.41) (-0.76) (3.47) (1.18) (3.72) r2 0.93 0.97 0.88 0.99 0.98 0.84 0.90 0.08 0.12 0.05 0.14 0.13 0.03 0.07 dw 1.62 1.80 1.94 1.58 1.91 1.71 1.91 q 21.48 21.03 17.28 20.57 21.83 20.74 20.70 h 1.32 1.21 1.23 1.27 1.19 1.20 1.01 * q is distributed as χ2 (18), which gives a 5% critical value of 22.36. h is distributed as f(99,99), which gives a 5% critical value of 1.39. the cyclical component is also signifi cant in some cases. the coeffi cient on the explanatory variable is signifi cantly positive in three cases only. there is, therefore, more evidence for the role played by other factors than for the role played by us stock prices, providing only weak evidence for contagion. it has been established that stock markets tend to be more strongly correlated when they are falling (particularly when they are falling rapidly). for this reason, equation (6) was estimated by dividing the sample period into two sub-samples, the fi rst extending between the beginning of the sample and observation number 194, which marks the peak of the dow jones, whereas the second sub-sample covers the subsequent period.5 the results are reported in tables 4 and 5, for the two sub-samples. again, the model seems to be well determined in terms of the goodness of fi t and diagnostics, but again the results show only limited evidence for contagion. there is no effect of us stock prices on gcc stock prices in the fi rst sub-sample, but a signifi cant effect on these markets in the second sub-sample, when the markets were going up.6 for nearly eight months stock prices in gcc countries were rising while the dow was falling, providing some indication that other factors must have been infl uencing stock prices in gcc markets. table 5 estimation results of equation (6): second sub-sample* 2 dr ht tp :// ijb f.u um .e du .m y stock market contagion in the early stages of the global fi nancial crisis: 19-34 29 b. evidence on the effect of oil prices the question that arises from the examination of the results presented in the previous section is the following: what are the other factors that infl uence stock prices in gcc countries if us stock prices played a minor role? one variable that is worthy of consideration is bound to be oil prices. figure 2 displays stock prices in gcc markets relative to the price of crude oil measured in dollars per barrel. in all cases we see remarkable similarity of the time paths of stock prices and that of the oil price. figure 2 shows the correlation coeffi cients of stock prices with the price of oil, as compared to correlation with the dow jones. it is obvious that gcc stock prices are much more highly correlated with the price of oil than with the dow and that the dow is very weakly correlated with the oil price. for this reason it seems to be worthwhile to examine the effect of the price of oil formally. this can be done by re-writing equation (6) as: (7) where z t is the logarithm of the price of oil. figure 2. correlations with the dow jones and oil price tittititi zp ,,,, ht tp :// ijb f.u um .e du .m y 30 the international journal of banking and finance, vol. 7. number 1: 2010: 19-34 table 6 reports the results of estimating equation (7) in which the explanatory variable is the price of oil. we can see that the coeffi cient on the explanatory variable is signifi cantly positive in all cases, implying positive effect of the price of oil on stock prices in gcc countries. however, we can still see statistically signifi cant components, particularly the level of the trend. the implication of these results is that while it is evident that the price of oil had a more important effect on stock prices in gcc countries than us stock prices, the effect of other unidentifi ed factors is even more signifi cant. table 6 estimation results of equation (7): whole period* abu dhabi bahrain qatar dubai kuwait oman saudi arabia μ 4.159 4.434 4.557 3.691 4.279 4.512 4.30 (53.10) (111.42) (49.99) (33.93) (68.48) (48.98) (35.00)  -0.007 -0.005 -0.003 -0.013 -0.008 -0.006 -0.007 (-3.26) (-3.87) (-0.93) (-3.82) (-4.01) (-2.53) (-2.29)  0.013 -0.005 0.015 .006 -0.009 0.013 -0.009 (1.08) (-2.73) (1.28) (1.26) (-4.21) (2.82) (-2.91) * 0.032 -0.005 -0.006 -0.011 0.003 0.009 -0.002 (2.47) (-2.95) (-0.54) (-2.22) (1.23) (1.99) (-0.46)  0.059 0.043 0.061 0.048 0.038 0.041 0.036 (2.73) (2.32) (2.65) (2.61) (2.26) (2.38) (2.21) r2 0.95 0.96 0.87 0.84 0.99 0.85 0.80 0.10 0.11 0.06 0.05 0.26 0.05 0.02 dw 1.92 1.87 1.89 2.01 1.88 1.83 1.92 q 23.29 25.64 25.93 17.71 26.97 25.33 15.36 h 1.27 1.26 1.28 1.13 1.26 1.24 1.24 * q is distributed as 2 (18), which gives a 5% critical value of 28.86. h is distributed as f(165,165), which gives a 5% critical value of 1.29. c. non-nested model selection tests to demonstrate further the importance of variables other than us stock prices and the price of oil, we conduct a series of non-nested model selection tests, using conventional regression analysis that allows no role for factors other than these two. non-nested model selection tests are used here to fi nd out if either the dow or the price of oil can on its own explain gcc stock prices. consider the following two models, m1 and m2: 2 dr ht tp :// ijb f.u um .e du .m y stock market contagion in the early stages of the global fi nancial crisis: 19-34 31 m1: 1 10 k t,ikt,us k t,i pp m2: 1 ,10, k tikt k ti zp p t = x     p t = y       where x is an observation matrix on us stock prices, y is an observation matrix on the price of oil,   and   are unknown regression coeffi cient vectors, and   and   are disturbance vectors. m1 and m2 are specifi ed as: (10) (11) in which gcc stock prices are explained by a geometrically declining distributed lag of us stock prices (m1) and the price of oil (m2). the two models are said to be non-nested if the regressors of one model cannot be expressed as an exact linear combination of the regressors of the other. obviously, m1 and m2 as represented by equations (10) and (11) are non-nested because the explanatory variables are different. table 7 non-nested model selection test results* abu dhabi bahrain qatar dubai kuwait oman saudi arabia m1 vs m2 n -117.2 -182.2 -687.2 -65.0 -216.2 -3941.7 -75.2 nt -112.6 -173.8 -509.8 -63.8 -200.8 -703.2 -71.5 en 1054.8 1723.0 3852.7 598.6 1751.2 4000.1 366.6 m2 vs m1 n -19.1 -16.3 4.0 -44.2 -29.2 4.75 -41.9 nt -18.4 -16.1 4.0 -43.6 -28.9 4.78 40.6 en 180.8 126.5 24.5 403.7 259.3 70.95 215.2 *n and nt have t distribution, whereas en has f distribution. three model selection tests were used: n is the cox test derived in pesaran (1974), nt is the adjusted cox test derived in godfrey and pesaran (1983), and en is the encompassing test proposed, inter alia, by mizon and richard (1986). the n and nt test statistics have t distribution, whereas the encompassing test has f distribution. the tests were run both ways for m1 versus m2 and m2 ht tp :// ijb f.u um .e du .m y 32 the international journal of banking and finance, vol. 7. number 1: 2010: 19-34 versus m1. when m1 was tested versus m2, the null hypothesis was that m1 is a better model (in terms of specifi cation) than m2. a signifi cant test statistic indicates that m1 is not a better model than m2. when m2 was tested against m1, the null is that m2 is a better model than m1. a signifi cant test statistic indicates that m2 was not a better model than m1. if we obtain signifi cant test statistics both ways, this means that the two models are misspecifi ed (that is, neither us stock prices nor the price of oil can explain gcc stock prices on their own). the econometrics of non-nested model selection tests can be found in pesaran and pesaran (1997). the results of the non-nested model selection tests are reported in table 7. 7 as we can see, the tests statistics are signifi cant both ways in all seven cases. the implication of these results is that a model in which gcc stock prices are determined by u.s. stock prices alone or the price of oil alone, without allowing a role for other factors, are misspecifi ed. notwithstanding the superiority of the price of oil as an explanatory variable over us stock prices, other factors played perhaps a more important role in shaping gcc stock prices during the period under examination. 4. discussion of the results and concluding remarks the price of oil goes some way in explaining disparities in the behaviour of gcc stock markets, but it cannot explain the full picture. for example, it can be argued that the stock market of abu dhabi rose more than the stock market of dubai when oil prices were rising because abu dhabi is more dependent on oil than dubai. likewise, the stock market of bahrain recorded the lowest rise when oil prices were rising and the lowest decline when oil prices were declining because bahrain is very low on oil. however, the price of oil cannot explain why the stock market of oman rose by over 100% in 2007. other possible factors that may explain the empirical results, particularly the signifi cance of the components, include the behaviour of foreign investors, exposure to the domestic real estate market, leverage and the liquidity position. since no precise fi gures are available on foreign participation in gcc stock markets, it is not possible to say anything defi nitive. however, one proposition that makes some sense is that the gcc stock markets kept on rising after the collapse of the dow because investors who liquidated their positions in the u.s. (and other developed markets) moved to emerging markets, including the gcc markets, thus sustaining the bubbles there. but when foreign investors started to pull out, because they were in need of dollar liquidity, gcc markets collapsed. it is more plausible, however, to explain the behaviour of the gcc stock markets in terms of domestic factors, including exposure to overheated real estate markets, unsustainable levels of leverage and shallow liquidity positions. 7 equations (10) and (11) were estimated on the assumption that 60. . the higher the value of  (ranging between zero and one), the greater the weight given to more recent observations of the underlying explanatory variable. ht tp :// ijb f.u um .e du .m y stock market contagion in the early stages of the global fi nancial crisis: 19-34 33 these factors may explain why the stock market of dubai declined much more than any of the other six markets. dubai has been experiencing unsustainable growth in the real estate sector fi nanced by foreign borrowing. when foreign fi nancing became diffi cult and liquidity started to dry up, the real estate market was badly hit and listed companies started to incur huge losses, dragging the stock market down. the same chain of events took place in other markets, but to a lesser extent than in dubai. but irrespective of what the most valid explanation is, the empirical results of this study clearly show that contagion was not the reason (at least not a major reason) for the bubble and crash behaviour exhibited by the gcc markets. the main objective of this paper was stated at the outset as the desire to fi nd out evidence for or against stock market contagion from the us to the gcc markets. the evidence undermines, if not rejects, the contagion hypothesis. author information: imad moosa is a professor in the department of accounting and finance, faculty of business and economics, monash university, melbourne.phone: +61399031472. e-mail: imad.moosa@buseco. monash.edu.au references business monitor international (2008). credit crunch gets local: key implications, 18, 6-7. godfrey, l.g., and pesaran, m.h., (1983). tests of non-nested regression models: small sample adjustments and monte carlo evidence. journal of econometrics, 21, 133-154. harvey, a.c., (1989). forecasting: structural time series models and the kalman fi lter. cambridge: cambridge university press. harvey, a.c., (1997). trends, cycles and autoregressions. economic journal, 107, 192-201. hussain, s., (2008, november). challenges in gcc banks: impact of the fi nancial ccrisis. paper presented at the smes conference, bahrain. retrieved from http://www.oecd.org/dataoecd/45/27/41728432.pdf. koopman, s.j, harvey, a.c., doornik, j.a. and shephard, n., (1995). stamp 5.0: structural time series analyser, modeller and predictor. london: chapman and hall. ljung, g.m., and box, g.e.p., (1978) on a measure of lack of fi t in time series models. biometrika, 65, 297-303. mizon, g.e., and richard, j.f., (1986). the encompassing principle and its application to non-nested hypotheses. econometrica, 54, 657-678. moosa, i.a. (2006) structural time series modelling: applications in economics and fi nance. hyderabad: icfai university press. ht tp :// ijb f.u um .e du .m y 34 the international journal of banking and finance, vol. 7. number 1: 2010: 19-34 moosa, i.a., and gazos, t., (2009). the monetary view of exchange rate determination under hyperinfl ation. hyderabad: icfai university press. pesaran, m.h. (1974). on the general problem of model selection. review of economic studies, 41, 153-171. pesaran, m.h., and pesaran, b. (1997). working with microfi t 4.0: interactive econometric analysis. oxford: oxford university press. woertz, e., (2008, october) impact of the us fi nancial crisis on gcc countries. grc report. ht tp :// ijb f.u um .e du .m y how defined, benefit pension assets affect the returns and volatility of the sponsor's stock the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 87 how defined, benefit pension assets affect the returns and volatility of the sponsor’s stock * brooks marshall, timothy b. michael, david m. maloney and *faramarz damanpour * james madison university, university of virginia and university of houston abstract in its valuation of firms with defined benefit plans, the stock market combines changes in the valuation of pension assets with changes in the valuation of the net core assets. unfortunately, aggregating the two disparate asset classes in valuation discards information about both classes. this work shows that by extracting the pension component of returns, two types of insights result: first, an enhanced understanding of the underlying risk and return of the firm’s net core assets; and, second, an enhanced perspective of the potential benefit from incorporating pension asset allocation into overall risk management. keywords: pension, risk management, valuation, assets jel classification: j32, m52 1. introduction when funding defined, benefit pension plans, firms are required by federal law to place their sponsored assets into a trust. any analysis of the sponsor’s stock needs to recognize that stock returns reflect the performance operational and pension plan assets. insights from such analysis can often be enhanced if the pension returns are extracted from the overall returns. as a result, such insights can improve both the firm’s external and internal analysis. if, for instance, an investment analyst desires to evaluate the market’s perception of a firm’s competitive nature, , removal of the pension returns could sharpen the analyst’s understanding of the performance visa-vis its competitors. likewise, focusing on the firm’s operations could provide a more rational basis towards proper incentive compensation. the analytical problems stemming from a valuation that combines different asset groups into one stock price are particularly problematic in the case of defined benefit pension plans. in general, pension assets are comprised of a well-diversified ijbf 1 88 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 portfolio of publicly traded securities. a close approximation of the portfolio held by a firm’s pension fund could be readily gathered and held by other pension funds. the operational assets on the other hand, define the firm’s business, competitors, and customers. pertaining to its operational assets, the firm’s management team distinguishes itself from its competitors through sound articulation and implementation of competitive strategies.. in addition, management’s decisions can modify to a certain degree (at a specific level) the firm’s industry affiliation. moreover, in contrast to pension assets, operational assets tend to be rather illiquid and difficult to value since the strongest part of a firm can be comprised of attributes not reflected in the financial statements (i.e.,customer loyalty and innovative capabilities). 2. literature review successfully “removing” the return on pension assets from market returns depends upon whether and how the market impounds pension returns in equity valuation. since the latter part of the 1970s, the economic ownership of pension assets had not been resolved. some had thought that the assets belongingto the employees were completely separate from the firm’s equity valuation, while others considered the association of the firm’s assets to its economic fortunes. a series of studies conducted by the likes of oldfield, (1977), feldstein and seligman(1981), daley (1984), landsman (1986), barth, beaver, and landsman (1992), coronado and sharpe (2003), and franzoni and marin (2006) had reasoned that the value of the pension assets is reflected in the value of the firm’s common stock. the relationship is statistically significant and economically substantive with results suggesting that pension plan assets are valued on a one-to-one basis with the operating assets of the firm. the findings suggest that the returns on pension assets are impounded into equity returns. the empirics of the valuation studies are generally consistent with the idea that us$1 of pension assets corresponds to us$1 of market value. however, the 1:1 relationship overstates theoretical expectations. black (1980) had posited that the returns on pension assets would be impounded into the stock price on an after tax-basis because of the tax deductibility of the pension contribution. because the sponsoring firm can deduct the contribution from taxable income, only the aftertax proportion (1tax rate) of corporate cash is required to create $1 in the pension plan. consequently, us$1 of returns produced by pension assets should impact the market value of the firm’s stock by us$1(1 marginal tax rate). bulow, morck, and summers (1987) had argued that us$1 of pension returns may have us$1 of equity value because the contributions to pension funds are constrained at the liability level. in this paper, we assume that pension returns are reduced by the tax rate when they are impounded into the stock’s return. if the 1:1 relationship holds for a given firm, our analysis of the impact would qualitatively apply, but the proportions would change. both corronado and sharpe (2003), as well as franzoni and marin (2006), have challenged the effectiveness of the market’s impounding. neither study had the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 89 explicitly evaluated the returns of pension assets, but instead, had evaluated the net pension liability. corronado and sharpe had found that the market’s valuation tended to rely more on the accounting aspect of the pension rather than that of the economic view. franzoni and marin’s results did not support the broad claims of corronado and sharpe; they had found that the stock market did not incorporate pension-related, future cash flow information for severely underfunded plans. zion (2002) had found that effective impounding is hindered by cumbersome and often time non-economic accounting representations. the outcry from the financial community regarding defined benefit reporting resulted in the promulgation of financial accounting standard 158 in september 2006 (financial accounting standards board, 2006). being that fas 158 addresses many of the issues that may have impeded the economic interpretation of pension results, the likelihood of the capital markets correctly impounding the pension performance is enhanced, going forward. jin, merton, bodie (2006) (jmb) had provided an additional support for the equity market’s impounding pension valuations, but had differed from the prior studies in two ways. firstly, the authors had focused on the relationship between the systematic risk of the firm’s stock returns (instead of the firm’s market value) and the systematic risk of the pension plan. secondly, jmb characterized their study as “dynamic,” wherein their variables were based on changes instead of levels. jmb had found that the beta of the firm’s stock was influenced by the beta of the pension assets. the impact was theoretically and statistically significant. like jmb, this paper specifically seeks to examine the dynamic returns of pension funds, but differs from jmb and other studies in that the focus strictly lies on the performance of pension assets without netting the impact of pension liabilities. the rationale is that pension liabilities stem from operational costs (i.e., the salaries of employees). funding these liabilities before payments are due through the contribution to the pension fund is distinct from the firm’s operations, albeit, legally required. in addition, the returns on the pension assets are much more subject to the sponsor’s specific investment policy, whereas changes in the economic value of pension liabilities stem more from economy-wide changes in interest rates. in sum, we argue that pension returns are the component of pension plans that most confound the interpretation of how the equity market perceives the firm’s operational performance. moreover, the focus on pension returns allows for greater interpretation of the funding decision. we structured the model to enable a measurement of the firm’s performance as if the pension was funded strictly on a pay-as-you-go basis. 3. relative size of pension assets and market capitalization 3.1 data and methodology characteristics of pension plans are related to several industry-specific factors, including the age of the sponsoring firm and employees, as well as the financial status of the sponsoring firm (ballester, fried, and livnat (1998); friedman (1983); blitzer, silverblatt, and guarino (2005)). by focusing on a single industry, we 90 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 are able to control some of these factors by paralleling the focus with an analyst’s perspective (either internally or externally), in that the particular industry is often the initial point of reference. the impact of pension returns developed below, where us$1 of pension returns increases the stock price by us$.65 (given the marginal corporate tax rate of 35 percent), depends on the assumption that the firms in the industry are financially healthy. for a firm with a questionable financial status, us$1 of pension returns would have some chance of benefiting creditors and pension beneficiaries instead of shareholders. the automotive, steel, and airline industries are prime examples of industries where additional dollars of pension returns would benefit creditors or beneficiaries, and the returns would not accrue to the sole benefit of shareholders. we selected the oil industry because it has sizably defined benefit plans with unquestionable operational health. our sample contains firms with the same fourdigit sic code (2911) from the oil and gas industry. we also limited the group to those firms of the standard & poors 500 index, which had data and published annual repots reported by standard & poors compustat between 1991 and 2004. five firms meet these criteria – amerada hess (hes), chevron-texaco (cvx), exxon mobil (xom), conocophilips (cop), and sunoco (sun). stock returns were obtained from crsp, where as pension returns were obtained from the firms’ 10-k reports. here, beta is used to measure the systematic risk of the stock using a regression between the stock’s annual returns and that of the standard & poors 500 index over the 12-year period from 1992-2004. being that pension returns are only reported annually, this analysis will use annual returns for the beta.. 3.2 sample size of pension assets panel a of table 1 reports the size of the pension assets (pa) for each firm in the sample in absolute terms. cop’s fund experienced the most rapid growth during the study period at a rate of 17 percent, while sun had experienced negative growth at –1 percent. in 2004, the funds had represented a substantial portfolio in absolute terms with an average size of over $6 billion. panel b of table 1 presents the ratio of the size of the pension fund (pa) relative to the total market capitalization (mv). this ratio represents the potential for the pension fund to impact the sponsor’s stock. comparing the data of 1991 and 2004 in panel b indicates that all firms except hes experienced a decline in pension assets relative to market capitalization (pa/ mv). this holds as panel a demonstrates how the absolute level of pension assets has grown between 1991 and 2004. for example, xom’s pa/mv decreased from 10.0 percent in 1991 to 5.5 percent in 2004, even though the us dollar value of pension assets grew at an annual rate of 6.9 percent. simply put, market value grew faster than pension assets. understanding “why” would provide an underpinning for the dynamics of the industry. factors affecting the relative level of pension assets include labor intensity; contributions to the pension fund; returns achieved in the pension fund; and the substitution of defined contribution plans for defined benefit plans. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 91 table 1: pension assets and market capitalization panel a: pension assets (pa) year hes cvx xom cop sun 1991 $276 $4,507 $7,554 $412 $1,313 1992 283 3,899 6,880 377 1,207 1993 309 3,832 7,509 374 1,272 1994 289 3,626 7,278 507 1,183 1995 397 4,033 8,300 680 1,222 1996 429 4,163 8,840 819 1,242 1997 482 4,454 9,383 999 1,277 1998 477 4,741 10,098 1,162 1,350 1999 534 4,673 16,654 1,230 1,439 2000 543 4,225 14,575 1,097 1,287 2001 495 5,947 12,170 1,113 1,110 2002 487 4,835 11,351 2,260 930 2003 626 6,573 16,486 2,763 1,071 2004 750 8,410 17,972 3,328 1,158 annual growth rate 8.0% 4.9% 6.9% 17.4% -1.0% pension assets as a proportion of market capitalization (pa/mv) (%) year hes cvx xom cop average without sun sun average for all 5 firms 1991 7.2 18.8 10.0 6.6 10.7 40.6 16.6 1992 6.6 17.3 9.1 5.8 9.7 40.5 15.9 1993 7.4 13.5 9.6 4.9 8.9 40.6 15.2 1994 6.8 12.5 9.6 5.9 8.7 38.5 14.7 1995 8.1 11.8 8.3 7.6 9.0 60.3 19.2 1996 8.0 9.8 7.3 7.0 8.0 69.8 20.4 1997 9.6 8.8 6.2 7.8 8.1 42.9 15.1 1998 10.6 8.8 5.7 10.8 9.0 41.4 15.5 1999 10.4 8.2 5.9 10.3 8.7 68.1 20.6 2000 8.4 7.8 4.8 7.6 7.2 42.5 14.2 2001 8.9 6.2 4.5 4.8 6.1 39.4 12.8 2002 9.9 6.8 4.8 6.9 7.1 36.5 13.0 2003 13.1 7.1 6.1 6.2 8.1 27.8 12.1 2004 9.9 7.6 5.5 5.5 7.1 20.4 9.8 average over 14 years 9.0 43.5 15.3 92 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 the most striking relation seen in panel b is how sun’s pa/mv dominates the ratio for other firms. in any given year, sun’s pa/mv ratio is from 3 to 8 times that of other firms, with sun’s ratio being between 40 and 70 percent in most years, dropping to the 20 percent range in 2003 and 2004. in contrast, pension assets for the other plans tended to be less than 10 percent of market capitalization. we chose not to include sun in our subsequent analysis of pension impact for two reasons. firstly, sun appears to be a different type of firm when compared with the other firms. it appears that sun’s financial assets play a much larger role in its performance. secondly, sun’s pension fund appears to be a statistical outlier, disproportionately skewing any averages towards sun’s results. as a result tables 2 through 5 use hes, cvx, xom, and cop to demonstrate how the impact of pension returns can be removed from stock market results and how the adjusted returns can be evaluated. 4. the impact of pension assets on the stock’s return and risk panel of table 2 provides the returns on the firm’s common stock, while panel b highlights the returns on the pension assets.. panel c (net core returns) shows the implied returns on the stock with the pension returns removed. stock returns were obtained from crsp, while pension returns were taken from the 10-k reports of the respective firms. however, disaggregating the pension returns from the overall stock returns required assumptions about what is impounded by the stock market. as discussed in the literature review, our model assumes each us$1 of pension returns increases the market value of the firm by us$0.65, given the 35 percent federal income tax rate. table 2: stock returns, pension returns, and net core returns (%) hes cvx xom cop overall 1992 1.48 6.97 5.94 11.09 1993 1.13 28.23 7.94 20.77 1994 4.62 8.02 1.99 18.63 1995 17.72 21.21 34.08 9.26 1996 11.60 25.95 23.42 31.34 1997 -2.01 21.22 26.20 13.78 1998 -5.87 12.56 21.81 -7.83 1999 16.30 10.15 14.21 16.03 2000 33.04 5.36 11.26 30.12 2001 -9.97 10.18 -6.90 10.83 2002 -2.78 -23.85 -7.18 -17.00 2003 2.13 32.41 19.93 35.76 2004 49.49 23.59 25.16 31.98 average 8.99 14.00 13.68 15.75 13.11 vol (sd) 16.78 14.45 12.91 15.46 14.90 beta 0.028 0.162 0.598 0.419 .393 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 93 panel b: pension returns hes cvx xom cop average 1992 7.01 6.86 5.40 1.70 1993 12.41 12.11 17.51 6.63 1994 -3.02 1.62 -0.43 0.00 1995 23.18 20.08 19.52 24.06 1996 10.31 12.47 14.28 10.00 1997 14.82 16.74 16.01 20.63 1998 11.27 15.15 14.21 13.71 1999 13.27 15.19 35.16 12.91 2000 -2.44 2.35 1.18 -0.57 2001 -7.18 -7.36 -7.35 -10.03 2002 -8.48 -7.11 -11.48 -14.29 2003 21.36 18.80 21.50 15.97 2004 11.82 12.44 12.45 11.83 average 8.03 9.18 10.61 7.12 8.74 vol (sd) 10.28 9.21 12.77 11.36 10.90 beta 0.511 0.873 1.244 0.748 .844 panel c: net core returns (%) hes cvx xom cop overall 1992 1.23 6.98 5.97 11.45 1993 0.56 29.78 7.30 21.24 1994 4.97 8.59 2.15 19.37 1995 17.42 21.30 34.91 8.49 1996 11.67 26.87 23.88 32.36 1997 -3.13 21.49 26.63 13.41 1998 -7.14 12.40 22.10 -9.46 1999 16.51 9.87 13.37 16.25 2000 35.08 5.52 11.59 31.71 2001 -10.14 10.92 -6.89 11.50 2002 -2.37 -25.67 -7.03 -17.29 2003 0.29 33.90 19.86 37.57 2004 52.11 24.19 25.65 33.58 average 9.00 14.32 13.81 16.17 13.32 vol (sd) 17.80 15.21 13.11 16.30 15.60 beta -0.005 0.131 0.553 0.395 .36 (continued) 94 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 beta is one of the most widely used measures of risk. here it measures the systematic risk of the stock using a regression between the stock’s annual returns and the annual returns for the standard & poors 500 index return over the 19922004 time period. beta is typically computed using monthly returns; however, this analysis uses annual returns because pension returns are only reported annually. in order to isolate the influence of pension assets in equity returns, our model of the firm is as follows: m v f = m v ca + m v pa where: mv f : the equity value/price per share of the firmmultiplied by the total number of shares outstanding mv ca : the firm’s assets and liabilities (including the pension liability), excluding pension assets mv pa : year-end market value as reported in the annual reports of the sponsoring firmsthe model is analogous to a two-asset portfolio where the overall portfolio value (market value of firm) and one portfolio component (pension assets) are observable, but the other asset (core assets) is not directly observable. returns and rates of return can also be depicted in the context of the two-asset portfolio: e ror = (weight ca * net core ror) + (weight pa * pension asset ror ) where: e ror : annual rate of return realized by shareholders. net core ror : the firm’s assets and liabilities (including the pension liability) minus pension assets. weight ca : proportion of market value in core assets or 1(pa(1-tx) / mveq. weight pa : proportion of market value in pension assets or pa(1-tx) , with tx mveq equal to the maximum marginal corporate tax rate of 35 percent. pension asset r : annual rate of return for pension assets, presented in accordance with fas 158, with reported pension returns divided by beginning of the year pension assets. the net core ror can then be specified in terms of observable asset returns, as in equation 1: net core ror = equity ror – (weight pa * (1-tx) * pension asset ror) (1) weight ca the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 95 for example, the return on cop’s net core assets in 2004 is computed as: net core ror = [31.98% (5.5% *.65 * 11.83%)]/[1-(5.5% * 0.65%)] = 33.58% the first two panels in table 2 allow direct comparison of the stock returns in panel a and pension returns in panel b. for the period 1992-2004, stocks outperformed the pension funds for all four firms. note however, that the two different risk measures give different signals. the stocks (relative to pension funds) had lower betas with higher standard deviations. panel c of table 2 presents these metrics for net core returns (arithmetic mean, standard deviation, and beta). comparing the net core returns (table 2, panel c) with stock returns (table 2, panel a) measures the impact of pension asset performance. the comparison is summarized in table 3. the surprising result is the consistency across firms. each of the firms’ returns and standard deviations were reduced by the pension fund’s performance. these results are somewhat understandable since oil and gas companies have had a strong performance over the time horizon examined. the reduction in volatility follows since pension funds are generally well diversified, as demonstrated in the low standard deviation of pension stock returns. the most surprising metric was beta. being that pension assets typically include bonds, one would expect the pension fund to reduce beta; however, removing the pension returns reduced the beta for each of the firms. the averages shown in table 3 indicate that pension funds reduced stock returns by an average of 22 basis points, with a reductionof the standard deviation from 15.6% to 14.9% and increased beta from 0.36 to 0.39. table 3: impact of the pension fund performance on stock returns (%) hes cvx xom cop average return -0.02 -0.32 -0.13 -0.42 -0.22 vol (sd) -1.02 -0.76 -0.21 -0.84 -0.71 beta 3.31 3.11 4.52 2.45 3.35 5. correlations and the risk of pension assets the conflict between risk measures stems from the inter-company correlations of pension returns and net core returns. even though volatility assesses the returns of an asset as an independent set of numbers, many analysts find that a metric measuring how the security relates to a broader portfolio is more relevant. another view, especially relevant from a portfolio manager’s perspective, concerns the volatility of a representation of the industry as a whole. we construct an “integrated oil and gas” portfolio with equal weights in each security, and then compute the portfolio return.we then analyze the portfolio, pension, and net core return. the return for this portfolio is the arithmetic average of the four returns 96 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 for each year. table 4 provides the returns for an equally weighted portfolio for each year. the “diversification reduction” column compares the average standard deviation for the four oil companies ((sd hes +sd cvx +sd xom +sd cop )/4) with the standard deviation of the portfolio. the percentage difference stems from the portfolio returns being less than perfectly correlated. because the pension funds are typically well diversified, the relative comparison of the resultswith the diversification impact of combining pension returns being much smaller, is logical. but the degree of difference is surprising. pension fund diversification only reduces the standard deviation of 10.90% for the average of the four pension standard deviations to 10.11% for the portfolio of four pension funds, and a reduction in volatility of 7.25% ([10.90% 10.11%/10.90% = 7.25%). on the other hand, the portfolio reduction for the net core returns is over 26%, with the average net core’s standard deviation being 15.60% and the portfolio of net core return’s having a standard deviation of only 11.52 percent, where (15.60%-11.52%)/15.60% = 26.16 percent. the reduction in standard deviation presented above stems from the degree of diversification, serving as a reminder to the basic principles of modern portfolio theory. as table 5 demonstrates, the correlations among pension returns are much higher than the correlation among net core assets, explaining the diversification difference between the two portfolio combinations. table 5 presents the correlations across all components. table 4: portfolio risk panel a: stock returns (%) year stock returns portfolio average diversification reduction hes cvx xom cop 1992 1.48 6.97 5.94 11.09 6.37 1993 1.13 28.23 7.94 20.77 14.52 1994 4.62 8.02 1.99 18.63 8.31 1995 17.72 21.21 34.08 9.26 20.57 1996 11.60 25.95 23.42 31.34 23.08 1997 -2.01 21.22 26.20 13.78 14.80 1998 -5.87 12.56 21.81 -7.83 5.17 1999 16.30 10.15 14.21 16.03 14.17 2000 33.04 5.36 11.26 30.12 19.94 2001 -9.97 10.18 -6.90 10.83 1.04 2002 -2.78 -23.85 -7.18 -17.00 -12.70 2003 2.13 32.41 19.93 35.76 22.56 2004 49.49 23.59 25.16 31.98 32.55 average 8.99 14.00 13.68 15.75 13.11 vol (sd) 16.78 14.45 12.91 15.46 14.90 sd of portfolio 11.16 25.10 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 97 panel b: pension returns (%) pension returns portfolio average diversification reductionyear hes cvx xom cop 1992 7.01 6.86 5.40 1.70 5.24 1993 12.41 12.11 17.51 6.63 12.16 1994 -3.02 1.62 -0.43 0.00 -0.46 1995 23.18 20.08 19.52 24.06 21.71 1996 10.31 12.47 14.28 10.00 11.77 1997 14.82 16.74 16.01 20.63 17.05 1998 11.27 15.15 14.21 13.71 13.59 1999 13.27 15.19 35.16 12.91 19.13 2000 -2.44 2.35 1.18 -0.57 0.13 2001 -7.18 -7.36 -7.35 -10.03 -7.98 2002 -8.48 -7.11 -11.48 -14.29 -10.34 2003 21.36 18.80 21.50 15.97 19.41 2004 11.82 12.44 12.45 11.83 12.14 average 8.03 9.18 10.61 7.12 8.74 vol (sd) 10.28 9.21 12.77 11.36 10.90 sd of portfolio 10.11 7.24 panel c: net core returns (%) net core returns portfolio average diversification reductionyear hes cvx xom cop 1992 1.23 6.98 5.97 11.45 6.41 1993 0.56 29.78 7.30 21.24 14.72 1994 4.97 8.59 2.15 19.37 8.77 1995 17.42 21.30 34.91 8.49 20.53 1996 11.67 26.87 23.88 32.36 23.69 1997 -3.13 21.49 26.63 13.41 14.60 1998 -7.14 12.40 22.10 -9.46 4.47 1999 16.51 9.87 13.37 16.25 14.00 2000 35.08 5.52 11.59 31.71 20.97 2001 -10.14 10.92 -6.89 11.50 1.35 2002 -2.37 -25.67 -7.03 -17.29 -13.09 2003 0.29 33.90 19.86 37.57 22.90 2004 52.11 24.19 25.65 33.58 33.88 average 9.00 14.32 13.81 16.17 13.32 vol (sd) 17.80 15.21 13.11 16.30 15.60 sd portfolio 11.52 26.19 (continued) 98 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 table 5: correlations between components of firm returns (%) stock returns pension returns net core returns hes & cvx 15.81 97.39 15.32 hes & xom 36.86 86.84 39.54 hes & cop 60.68 94.25 53.04 cvx & xom 66.89 89.76 64.09 cvx & cop 67.19 97.26 69.97 xom & cop 33.87 84.06 31.76 average 46.88 91.59 45.62 within-firm correlations net core vs. pension stock vs. pension hes 13.99 17.80 cvx 73.00 74.59 xom 68.61 70.78 cop 29.69 32.54 average 46.32 48.93 the net core returns have an average correlation among all possible combinations of the four firms of only 45.6 percent, whereas the correlation among pension assets is 91.6 percent. even though pension assets have no direct industry affiliation, their performance appears to increase the similarity in the equity market behavior across firms. in other words, the pension assets increase the correlations among the firms from 45.62 percent to 46.88 percent. pension assets exaggerate the apparent industry affiliation, even though pension assets, in fact, diminish any industry-specific interrelationship. the bottom panel in table 5 includes the correlations between the pension returns and the net core returns, as well as the correlations between the pension returns and the stock returns. note that the correlations are always greater for the pension to stock relationship because the stock return includes the pension return. authors like haugen(1989) and jin, merton, & bodie (2006) suggest that firms may integrate pension asset allocation into overall firm risk management. the advantage to minimizing the correlation between net core returns and pension returns is that the firm is less likely to experience poor performance from operations when the pension assets are performing poorly. hes, by far, has the lowest correlation. if the relationship from the past holds into the future, hes would have less of a chance in facing binding constraints from being forced to make up for poor pension performance. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 99 6. conclusion while the demonstration was applied to healthy firms with the assumption that shareholders were the sole beneficiaries of pension performance, the model could be widely applied. neither creditors (for firms at risk of default) nor employees (who would increase claims on retirement income if the pension performed well, jin, merton, & bodie (1996)) were assumed to play a role in this analysis. for firms with some prospect of other claimants on pension assets, the model would be modified to estimate the likelihood of diminished impact of pension returns on stock returns. as discussed, pension assets may confuse the understanding of industryspecific impacts, by increasing the inter-firm correlation, but not for industry-specific events. such insight is likely to prove useful for external analysts in their attempts to understand the firm’s competitive position going forward. more specifically, a clear understanding of the role of pension returns would help industry-based longshort decisions for hedge funds to interpret the net result of changes in oil prices. for internal analysts, extracting pension returns could help in the assessment of the market’s interpretation of the effectiveness of competitive strategy. knowledge of the net core assets’ risk factors should enable management to select pension assets that have low sensitivity to firm/industry-specific or firmspecific risk factors (haugen (1989); jin, merton, and bodie (2006)). managing pension assets to minimize correlation with the net core assets would diminish the probability that the firm would experience a “double-down” or poor performance from both asset groups simultaneously. evidence of such management should appear in low correlation between pension returns and net core returns (table 5). in addition, pension plans can be readily compared within the industry. for example, xom’s pension has clearly outperformed its competitors in the industry, but only by assuming greater risk, both in terms of beta and standard deviation (table 2). comparing pension performance in the context of competitors may allow a rethinking of how the firm wishes to address asset allocation and funding levels. for human resource compensation decisions, compensation plans based on stock performance are clearly impacted by the presence, size, and performance of pension funds. extracting the pension results from the stock returns would have improved the compensation basis for cob by 42 basis points and cvx by 32 basis points (table 3). net core results may or may not have generated greater returns. however, it seems likely that employee options would tend to have more value if based on the more volatile net core returns instead of the sponsor’s stock return. author statement: brooks marshall (e-mail: marshasb@jmu.edu) is the submitting author. he is a professor of finance at the james madison university. references ballester, m., d. fried, and j. livnat (1998). pension plan cash flows: estimation and motives, journal of financial statement analysis 3(3): 41-50. 100 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 87-100 barth, m. e., w. beaver, and w. landsman, (1992). the market valuation implications of net periodic pension cost components, journal of accounting and economics 15: 2762. bodie, z., (1996). what the pension benefit corporation can learn from the federal savings and loan insurance corporation, journal of financial services research 10 (1): 83-100. black, f., (1980). the tax consequences of long run pension policy, financial analysts journal, (july-august 1980): 21-30. blitzer, david, howard silverblatt, and dave guarino, (2005). s&p 500 2004 pension status report, standard & poor’s. bulow, j., r. morck, and l. summers (1987). how does the market value unfunded pension liabilities? pension economics. coronado, j.l., and s.a. sharpe, (2003). did pension plan accounting contribute to a stock market ‘bubble’? brookings papers on economic activity (march). daley, l.a., (1984). the valuation of reported pension measures for firms sponsoring defined benefit plans, the accounting review 59 (april): 17798. feldstein, m., and s. seligman (1981) pension funding, share prices and national saving, the journal of finance (september 1981), 801-824. financial accounting standards board. (2006, september.) statement of financial accounting standards no. 158: employer’s accounting for defined benefit pension and other postretirement plans. retrieved february 13, 2007 from http://fasb.org/pdf/fas158.pdf franzoni, francesco, and j. marin (2006). pension plan funding and stock market efficiency, journal of finance, (2006), 61, 921-956. friedman, b., (1984). pension funding, pension asset allocation and corporate finance: evidence from individual company data,” financial aspects of the united states pension system (university of chicago press, chicago, il), 92-106, (november). haugen, robert, (1989). pension management in the context of corporate risk management, journal of portfolio management (fall). jin, li, robert c. merton, and zvi bodie (2006). do a firm’s equity returns reflect the risk of its pension plan? journal of financial economics 81 (1): (july). landsman, w., (1986). an empirical investigation of pension fund property rights, the accounting review 61: 662-691. oldfield, g., (1977). financial aspects of the private pension system, journal of money, credit and banking 9(1): 48-54. zion, d., carache, b. (2002). the magic of pension accounting, credit suisse first boston. international journal of banking and finance 3-1-2008 how defined, benefit pension assets affect the returns and volatility of the sponsor's stock brooks marshall timothy b. michael david m. maloney faramarz damanpour recommended citation portfolio preferences across markets: evidence from mutual fund ownership 48 the international journal of banking and finance, volume 9 (number 4) 2012: pages 48-73 portfolio preferences across markets: evidence from mutual fund ownership wen-hsiu chou florida international university, united states of america ____________________________________________________________ abstract this paper is about evaluating and comparing the portfolio preferences of domestic and foreign mutual funds in developed and emerging markets over the period 1998–2007. we find that foreign and domestic mutual funds have some different preferences toward firm characteristics and firms’ information environments, and economic development affects the preferences for both types of funds. a country’s characteristics and institutions also influence mutual fund investment decisions when fund managers form their portfolio holdings. results further show that foreign and domestic mutual funds play a monitoring role in their portfolio firms, but foreign mutual funds cannot monitor firms effectively in emerging markets. key words: portfolio preference, financial institutions, mutual funds, international asset allocation, developed vs merging markets jel classification: g11, g15, g23 _____________________________________________ 1. introduction to literature on ownership capital markets have become even more integrated and globalized over the last two decades as revealed by the mutual dependence of markets during and after the global financial crisis. this globalization reduces investment costs and constraints, which allows investors to reallocate assets easily across markets. it also enables companies to finance their projects through world capital markets. this pool of foreign capital is especially important for emerging markets because foreign capital not only increases market value, it also improves liquidity while reducing cost of capital (ferreira and matos, 2008; aggarwal et al., 2011). despite the rapid liberalization of capital markets in both developed and emerging economies, foreign financing and access to 49 foreign capital still vary considerably across firms and countries (ferreira and matos, 2008; leuz and warnock, 2009). previous studies in general focused on one developed market or some emerging markets as a group to reveal portfolio preferences of foreign investors; some studies focused on all national markets (for example, see dahlquist and robertsson, 2001; aggarwal, klapper and wysocki, 2005; ferreira and matos, 2008). it is still not clear whether foreign investors have different preferences for firms in developed and firms in emerging markets. fan, wei, and xu (2011) argue that firms in emerging markets are structurally different from firms in developed markets. therefore, some factors discovered in the previous literature may have different effects on investor’s choice of firms in developed versus emerging markets when attracting foreign investors. it is essential to understand what the differences are so firms in emerging markets may grow faster and more efficiently with the competition provided by foreign capital in order to catch up and compete with firms from developed markets. this study focuses on one type of institutional investor namely mutual funds to investigate whether foreign investors have different preferences for firms in developed and emerging markets. it could document whether factors in previous studies show different effects on foreign and domestic ownership in developed and emerging markets. mutual fund holdings are grouped into foreign and domestic portfolios. mutual funds from locations domiciled in countries other than the country where the firm is located are classified as foreign investors. if mutual funds and firms are both located in the same country, the mutual fund ownership is classified as domestic ownership. second, we explore how firms’ information environments affect mutual fund holding decisions. previous studies usually document that foreign investors suffer from more information disadvantages than domestic investors (bohn and tesar, 1996). thus, foreign investors in japan and sweden like to hold large stocks because large stocks are more visible and have more information than small stocks (kang and stulz, 1997; dahlquist and robertsson, 2001). we investigate whether foreign and domestic mutual fund ownership varies across different firmlevel information measures. three dimensions of measure for a firm’s information environment are used to proxy for the level of information asymmetry in different aspects. they are accounting standards (gaap and ias adoption), earnings management (accruals, earnings international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 50 smoothing, and earnings correlation) and analyst variables (number of analysts covering a firm, analyst forecast errors, and analyst forecast dispersion). results generally confirm the conjectures but with some variations. accounting standards and analysts play a major role on both foreign and domestic mutual fund ownership. foreign ownership increases with good accounting standards, better analyst coverage, and lower analyst forecast dispersion. however, the effects of good accounting standards and lower analyst forecast dispersion reduce when foreign mutual funds invest in emerging markets but remain strong for domestic mutual funds in both developed and emerging markets. earnings management seems to have little effect on the investment decisions for both foreign and domestic mutual fund holdings. we also explore country characteristics as to whether these have an asymmetric impact on foreign and domestic mutual fund ownership. foreign fund preference toward country characteristics is similar in both emerging and developed countries. higher gdp and larger size of stock market have a negative effect on foreign ownership. these results are consistent with previous literature that international investors prefer small markets to diversify their portfolio holdings (ferreira and matos, 2008). foreign trades, however, have a large positive impact on foreign mutual funds. this effect is indicative of the importance of visibility and familiarity for a country to attract foreign capital (chan, covrig and ng, 2005). contrary to foreign investors, a larger stock market has a positive effect but greater foreign trades have a negative impact on domestic mutual fund ownership. this negative effect of international trades is driven by emerging markets. in the developed markets, domestic fund ownership is positively associated with foreign trades. finally, legal environment plays important role on foreign mutual fund investment decisions. foreign mutual funds can only effectively monitor their holding firms in developed markets. previous literature documents that country’s laws and disclosure quality affect foreign investors’ intention to enter a market or to hold a firm’s stock (gelos and wei, 2005; ferreira and matos, 2008; leuz et al., 2009). here, security regulations, disclosure requirements, accounting standard, and anti-self-dealing are used to proxy for the enforcement of security laws, a country’s disclosure quality, the firms’ financial disclosure quality, and the quality of corporate governance in the country, respectively. 51 these factors are supposed to be positively related to foreign mutual fund ownership. results show that security regulations, disclosure requirements, accounting standard, and antiself-dealing are all positively related to foreign ownership in both developed and emerging markets. however, domestic mutual funds in general care more about disclosure requirements and anti-self-dealing but pay less attention to security regulations. when domestic mutual funds invest in emerging markets, all of these laws become important. with regard to firm characteristics, existing results show that foreign mutual funds in general prefer firms with large size, lower dividend yields, higher operating performance, more cash holding, and lower closely-held ownership. they also prefer younger firms, growth stocks, member firms of msci index, and firms with american depositary receipt (adr) issuance. these preferences are driven by developed markets because a relatively large number of listed companies worldwide are from developed markets. contrary to developed markets, foreign mutual funds in emerging markets focus more on firms’ operating performance, such as roe and sales growth. they also do not care about dividend yields. although foreign mutual funds still prefer msci members, they have less of a preference for firms with adr issuance. aggarwal et al. (2005) explain that this may be due to low cross-listing in emerging markets. when pooling developed and emerging markets together, domestic mutual funds generally have preferences similar to foreign mutual funds, but they do not care about firm size and dividend yields. in addition, msci membership and adr issuance actually are not important to domestic mutual fund ownership. the portfolio preferences are similar in developed markets. when comparing portfolio preferences in developed and emerging markets for domestic mutual funds, domestic mutual funds in emerging markets place more of an emphasis on dividend yields, roe, and sales growth. thus, the investment strategy that domestic investors adopt in emerging markets is generally firm-performance oriented. this study therefore aims to make several contributions to the debate on ownership preferences across the world. the empirical results that link firm characteristics, information environments, and country institutions to both foreign and domestic investors expand the limited evidence on international portfolio allocation. in the literature, this study is most closely related to dahlquist and robertsson (2001), aggarwal et al. (2005), and ferreira and matos (2008), which are examples of similar line of inquiries but with single aims. dahlquist and robertsson international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 52 use firms in sweden from 1991 to 1997 to investigate foreign and domestic investors’ portfolio preferences. aggarwal et al. use u.s. mutual funds and focus on emerging markets with one year of data. ferreira and matos use overall institutional investors from 2000 to 2005 to test domestic and foreign investors’ portfolio preferences across all countries. this study takes a first look at foreign and domestic mutual funds’ holdings across 28 countries from 1998 to 2007 by comparing portfolio preferences of foreign and domestic mutual funds in both developed and emerging markets. in addition, data contain mutual funds from 62 countries, not just from the u.s. 1 second, this study adds to a growing body of papers that examine international portfolio allocation. foreign investors have different portfolio preferences from domestic investors. in addition, both foreign and domestic mutual funds have different preferences toward firms in developed and emerging markets. results also provide some additional information for researchers to solve the puzzle of home or foreign bias. the country’s institutions play important role on mutual fund holding decisions. finally, this study is one of the few studies that investigate the effective monitoring role of institutional investors in the international markets. only domestic mutual funds play an effective monitoring role on firms in both developed and emerging markets. the reason may be that foreign investors face more information disadvantages in emerging markets than in developed markets. the rest of the paper is organized as follows. section 2 describes the sample and variables employed in the empirical tests, along with summary statistics. section 3 presents the firm-level empirical results while in section 4 we evaluate the impact of country-level institutions on foreign and domestic mutual fund portfolio decisions. the final section concludes the paper. 2. data and sample description in this section, we briefly describe the construction of sample and report summary statistics of key variables. the main sample comprises the following data sources: worldwide mutual fund holdings from thomson reuters; stock returns and financial variables from datastream and 1 u.s.-based foreign institutions hold more than us$ 2 trillion, non-u.s. foreign institutions hold us$ 1.7 trillion, and domestic institutions hold us$ 1.5 trillion in non-u.s. stocks. ferreira and matos document that u.s.-based foreign institutions, non-u.s. foreign institutions, and domestic institutions hold roughly the same level of assets in non-u.s. stocks. 53 worldscope; accounting standards from compustat global; and analyst information from i/b/e/s. the thomson reuters database contains mutual fund equity holdings information at the firm level for over 32,031 mutual funds worldwide. the portfolio holdings data include the identifications of funds and securities, and the number of shares held in a firm by each fund.2 we require sample firms to have an assigned sedol code from compustat global from 1998 to 2007. there are 92,747 firm-year observations. however, the sample sizes in the multivariate analyses vary and are lower due to missing values for some variables. to classify fund holdings into domestic and foreign holdings, we check whether the funds and the holding firms are from the same countries. if both fund and firm are from the same country, the fund holdings are classified as domestic ownership; otherwise, foreign ownership applies. we further split national markets into developed and emerging markets. there are 28 countries, 28,602 foreign mutual funds, 23,373 domestic mutual funds, and hence 92,747 firmyears in our sample over 1998 to 2007. the large number of worldwide mutual funds offers a common platform to analyze the investment decisions of a homogenous group of institutional investors who share the same investment savvy. their investment decisions should offer the subtle yet important attitude that mutual funds have towards firms in developed and emerging markets. we first investigate the overall portfolio preference of domestic and foreign mutual funds, followed by the portfolio preference of domestic and foreign mutual funds for each of developed and emerging markets. portfolio preferences include firm characteristics, firms’ information environments, and country-level institutions. we also test whether foreign and domestic investors play an effective monitoring role on firms. analyses provide several new and interesting results. drawing from the existing literature, firm characteristics in our regressions include: firm size (size), book-to-market ratio (bm), stock return (ret12), stock return volatility (σret), dividend yield (dy), return on equity (roe), a dummy variable (msci) that takes the value of one if the firm is included in the msci index for a given year, debt to asset ratio (da), cash holdings (cash), closely-held ownership (chold), sales growth (gsales), a dummy indicator (adr) that 2 see chan, covrig, and ng (2005) for a detailed description of the thomson reuters’ mutual fund holdings database. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 54 equals one if a firm has issued american depositary receipts, and firm age (age). size is the log of market capitalization at year-end. bm is the log of book-to-market equity ratio in june of year t. ret12 is the average of monthly stock returns over year t-1. σret is the standard deviation of monthly stock returns over year t-1. da is the ratio of long-term debt to total assets. cash is the ratio of cash and short-term investments to total assets. chold is the fraction of shares closely held by insiders and controlling shareholders in year t. gsales is the average of sales growth for years t-2 and t-1. to assess the effect of firm-level information environment on foreign mutual fund ownership, firms’ information measures include accounting standards (accstd), analyst coverage (nanalysts), forecast dispersion (disp), forecast errors (ferror), accruals (accrual), earnings smoothing (smooth), and earnings correlation (corr). these variables evaluate three dimensions of information disclosure quality at the firm level. accstd evaluates accounting standards; nanlysts, disp, and ferror are related to the information disclosed by analysts. accrual, smooth, and corr measure the earnings management. accstd is a dummy variable set to one if the firm uses gaap or ias as its accounting standard. nanalysts is the number of financial analysts covering a firm in year t. disp is the standard deviation of analyst forecasts scaled by the mean of analyst forecasts in year t. ferror is the absolute value of the difference between announced earnings and the mean of estimated earnings scaled by the mean of analyst forecasts in year t. accrual is the magnitude of accruals. we follow lai, ng and zhang (2012) to calculate the variable of accrual. smooth is the smoothness of the accounting reports and is the ratio of the standard deviation of operating income to the standard deviation of cash flows over the last five years. corr is the correlation coefficient between change of accruals and change of cash flows over the last five years. when calculating smooth and corr, a minimum of three years is required. country characteristics included in the analyses are gdp per capita (gdp), stock market capitalization to gdp (mcap), gdp growth (ggdp), trade to gdp ratio (tgdp), gdp growth volatility (σgdp), and stock market volatility (σmkt). all variables except stock market volatility are taken from the world development indicators. to access country-level information on disclosure or institutions, we implement indexes from la porta et al. (1998, 2006). these indicators include security regulations, the accounting standard index, and the anti-self-dealing index. the index for disclosure requirements is from world economic forum. median of the 55 index scores across all countries is used to classify countries. countries with score above the median are considered strongly regulated countries; otherwise they are considered weakly regulated countries. table 1: country-level sample distribution and descriptive statistics this table reports the time-series averages of country characteristics for 28 countries from 1998 to 2007, together with the number of firm-year observations (n(fyear)) covered in the sample and the fraction of shares held by foreign mutual funds (mf) in %. country characteristics include gdp per capita (gdp), stock market capitalization to gdp (mcap), gdp growth (ggdp) in year t-1, trade to gdp ratio (tgdp) in year t, gdp growth volatility (σgdp) over the last five years, and daily stock market volatility (σmkt) over year t-1. country n(fyear) mf gdp mcap ggdp tgdp σgdp σmkt argentina 475 0.92 8.94 0.47 0.02 0.33 0.06 0.29 australia 5,165 2.60 10.01 1.12 0.03 0.41 0.01 0.17 austria 741 5.05 10.11 0.30 0.02 0.97 0.01 0.15 belgium 1,443 2.24 10.05 0.77 0.02 1.63 0.01 0.17 brazil 2,230 2.17 8.24 0.42 0.03 0.24 0.02 0.31 denmark 1,209 2.66 10.33 0.64 0.02 0.89 0.01 0.17 finland 1,185 6.20 10.12 1.46 0.04 0.74 0.01 0.31 france 5,287 3.33 10.04 0.90 0.02 0.53 0.01 0.19 germany 5,422 3.96 10.06 0.52 0.02 0.71 0.01 0.19 greece 1,688 1.46 9.61 0.62 0.04 0.47 0.01 0.22 hong kong 1,838 2.29 10.20 4.93 0.04 3.23 0.04 0.21 india 3,660 2.51 6.27 0.56 0.07 0.37 0.02 0.26 italy 2,756 2.79 9.87 0.51 0.01 0.52 0.01 0.18 japan 20,984 1.67 10.54 0.84 0.01 0.24 0.01 0.21 mexico 1,039 2.47 8.69 0.26 0.03 0.61 0.03 0.23 malaysia 3,487 1.56 8.31 1.50 0.05 2.16 0.04 0.20 netherlands 1,801 5.79 10.10 1.22 0.02 1.29 0.01 0.18 norway 1,259 4.06 10.57 0.54 0.03 0.73 0.01 0.22 new zealand 691 3.07 9.58 0.41 0.03 0.61 0.02 0.16 portugal 487 2.24 9.30 0.47 0.02 0.67 0.01 0.16 south africa 1,625 2.01 8.08 1.98 0.04 0.56 0.01 0.25 singapore 2,427 2.13 10.08 2.04 0.06 4.28 0.04 0.19 spain 1,345 2.95 9.61 0.84 0.04 0.57 0.01 0.19 sweden 2,553 3.82 10.26 1.20 0.03 0.86 0.01 0.25 switzerland 2,459 4.61 10.45 2.66 0.02 0.84 0.01 0.16 thailand 3,548 1.86 7.73 0.56 0.04 1.30 0.05 0.29 taiwan 4,367 2.36 9.58 1.24 0.03 1.14 0.06 0.25 united kingdom 11,576 2.42 10.16 1.53 0.03 0.57 0.01 0.17 international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 56 table 1 is a summary on 28 countries, their characteristics, and the number of firm-years covered in each country. the number of firm-year observations varies substantially across countries with 20,984 firm-years in japan, 11,576 firm-years in the u.k., 5,422 firm-years in germany, and only 475 firm-years in argentina. average foreign mutual fund holdings are generally higher for developed markets than for emerging markets, ranging from 0.92 per cent for argentina to 6.20 per cent for finland. among the 28 countries in the sample, norway, hong kong, india, singapore, taiwan, and brazil have the greatest value of gdp, stock market size, gdp growth rate, trade to gdp ratio, gdp growth volatility, and market return volatility, respectively. however, india, mexico, italy, japan, and austria have the lowest value of gdp, stock market size, gdp growth rate, trade to gdp ratio, and stock market return volatility, respectively. table 2: firm-level sample istribution and descriptive statistics this table reports the time-series averages of firm characteristics from 1998 to 2007, together with the number of firm-year observations (n(fyear)) covered in the sample. firm characteristics include the fraction of shares held by foreign mutual funds (mf) in %, log of market capitalization (size), log of book-to-market ratio (bm), average of past one-year monthly stock returns (ret12), standard deviation of past one year monthly stock returns (σret), dividend yield (dy), return on equity ratio (roe), msci membership (msci), debt to asset ratio (da), cash holdings (cash), closely-held ownership (chold), sales growth (gsales), adr issuance (adr), and firm age (age). mean n(fyear) median std. dev. mf 2.57 90,006 0.55 5.52 size 5.64 73,529 5.51 1.90 bm -0.48 70,311 -0.41 0.88 ret12 0.01 73,503 0.01 0.05 σret 0.13 73,463 0.11 0.09 dy 0.02 69,622 0.02 0.03 roe 0.06 67,848 0.08 0.25 msci 0.46 73,907 0.00 0.50 da 0.23 69,392 0.17 0.23 cash 0.15 65,889 0.10 0.16 chold 0.43 63,301 0.42 0.23 gsales 1.15 68,406 1.07 0.47 adr 0.04 73,907 0.00 0.19 age 13.28 73,630 11.00 9.80 57 table 2 reports the distribution of variables at the firm level across all countries employed in the analyses. foreign mutual funds hold an average of 2.57 per cent of shares for the 90,006 total firm-year observations. around 4 per cent of the firms have issued adr, and around 46 per cent of the firms are members of msci index. closely-held ownership applies to an average of 43 per cent of the shares. firms in the sample generally have a negative bm and a positive past return. they also pay dividends and have a sales growth rate of 15 per cent. in an unreported table, the correlations between variables in the sample are moderately low, suggesting that the problem of multicollinearity should be minor. 3. determinants of mutual fund holdings at the firm level previous literature shows that u.s.-based foreign investors usually prefer firms with better information disclosure or firms that are more visible to foreigners such as large firms, firms with adr issuance, and msci member countries. one explanation is that foreign investors suffer from an information disadvantage. therefore, foreign and domestic investors should have some different portfolio preferences if foreign and domestic investors possess different levels of private information about local firms. in this section, we report findings on firm-level determinants of portfolio preferences across all countries for foreign and domestic mutual funds. following this are the firm characteristics for each developed and emerging market for both foreign and domestic mutual funds. determinants include firm characteristics and information environment. 3.1 what attracts foreign and domestic mutual funds? we first present results on which firm characteristics mutual fund investors care about the most. in this test, we control for country fixed effects and year fixed effects, and all associated tstatistics are calculated based on standard errors adjusted for heteroskedasticity and firm-level clustering. therefore, firm characteristics across countries in the regressions should be comparable. table 3 is a summary of results. in the first column of left panel, the foreign mutual fund international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 58 table 3: tobit regression results of foreign and domestic mutual fund ownership on firm characteristics this table reports the regression results of mutual fund ownership on various firm characteristics with t-statistics of the coefficient estimates in parentheses. in the three models on the left side, the dependent variable is foreign mutual fund ownership. results are first reported by including all countries, followed by developed countries, and lastly emerging countries. in the three models on the right, the dependent variable is domestic mutual fund ownership. firm characteristics include variables described in table 2. the sample period is from 1998 to 2007. year and country dummies are added to control for year and country effects, and clustered standard errors are corrected at the firm level. n(fyear) is the number of firm-year observations. foreign mutual funds domestic mutual funds variables all developed emerging all developed emerging size 0.83 0.82 0.96 -0.03 -0.03 -0.11 (25.48) (23.19) (9.94) (-1.09) (-0.73) (-1.45) bm -0.30 -0.21 -0.56 -0.62 -0.58 -0.80 (-5.70) (-3.45) (-4.98) (-9.80) (-8.32) (-5.20) ret12 -5.41 -5.37 -4.95 -2.19 -4.89 6.85 (-8.14) (-6.86) (-3.75) (-3.01) (-5.66) (5.13) σret 2.39 3.02 0.22 2.29 2.27 -0.40 (4.88) (5.28) (0.24) (3.82) (3.01) (-0.45) dy -2.82 -3.65 -1.94 0.96 -1.32 5.06 (-2.97) (-3.28) (-1.00) (0.78) (-0.92) (2.20) roe 0.54 0.39 1.33 1.52 1.43 2.06 (3.64) (2.41) (3.73) (9.00) (7.50) (6.37) msci 0.55 0.54 0.55 -0.47 -0.35 -0.76 (5.98) (5.24) (2.54) (-4.40) (-2.80) (-3.64) da 0.07 0.07 0.34 0.23 0.17 0.60 (0.39) (0.34) (0.71) (1.06) (0.65) (1.61) cash 1.38 1.23 1.60 -0.60 -0.74 -0.08 (5.01) (4.02) (2.44) (-2.19) (-2.41) (-0.13) chold -3.70 -3.79 -3.67 -3.47 -3.31 -3.95 (-18.21) (-16.19) (-8.49) (-18.05) (-14.89) (-10.50) gsales x100 -0.01 -0.02 0.17 0.02 0.01 0.12 (-1.00) (-1.09) (4.90) (0.60) (0.53) (2.61) adr 0.56 0.50 0.76 -1.36 -1.25 -1.68 (1.99) (1.78) (0.92) (-7.01) (-5.62) (-4.47) age -0.05 -0.04 -0.06 -0.04 -0.04 -0.08 (-9.93) (-9.27) (-3.29) (-7.45) (-6.62) (-4.84) n(fyear) 52,830 42,876 9,954 50,858 40,320 10,538 pseudo r2 4.21% 4.55% 3.60% 3.64% 3.83% 2.92% 59 holdings are mainly positive and statistically significantly related to firm size, roe, msci membership, cash holdings, and adr issuance. results indicate that firms with more information disclosure and better visibility attract more foreign investors, for example, large firms, firms with msci membership, and firms with adr issuance. the results further demonstrate that foreign ownership is negatively associated with book-to-market ratio, dividend yields, closely-held shares, and firm age. thus, foreign investors prefer growth stocks, low dividend payouts, less closely-held shares, and younger firms. consistent with ferreira and matos (2008), domestic mutual funds (the first column of right panel) generally show a lower preference toward large firms, firms with adr issuance, or firms listed in the msci index. they also prefer firms with lower cash holdings. since domestic investors have more information about local firms, they rely less on firm size to make their investment decisions. similar to foreign investors, domestic mutual funds show preference toward growth stocks, better roe, less closely-held shares, and younger firms. when countries are separated into developed and emerging markets, foreign investors’ portfolio preferences in developed markets are generally consistent with their overall portfolio preferences across all countries. contrary to the portfolio preferences for foreign investors in developed markets, foreign investors in emerging markets focus more on operating performance, such as strong roe and sales growth. dividend yield also plays no role in foreign mutual fund ownership. similar to foreign mutual funds, domestic mutual funds have some significantly different portfolio preferences for firms in developed and emerging markets. for example, domestic mutual funds in the emerging markets do not care about cash holdings but prefer firms that pay more dividends. it is interesting to note that adr issuance has almost no effect on foreign fund holdings in the emerging markets. there are two possible explanations: (i) the number of adr listings may be low in the emerging markets (aggarwal et al., 2005), and (ii) the data include all mutual funds worldwide, and adrs have less effect on foreign mutual funds from countries other than the u.s. 3.2 information disclosure quality and mutual fund holding preference next set of results are on disclosure affects in both foreign and domestic mutual fund holdings in developed and emerging markets. in this test, three dimensions of information international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 60 disclosure are used – accounting standards, analyst variables, and earnings management – to measure a firm’s information environment. to conduct the test, a dummy is set to one if the firm uses ias or gaap as its accounting standard. in the accounting literature, both accounting standards and auditor quality matter most for the quality of financial reports and foreign investors (see bradshaw, bushee, and miller, 2004; francis, 2004; covrig, defond, and hung, 2007). however, as francis and wang (2008) document, compustat global does not give proper auditor identification for japan, south korea, india, and pakistan. in addition, there is no precise way to determine the auditors’ identification in these four countries. therefore, this study focuses on accounting standards to measure the reliability, consistency and comparability of accounting information for a firm. to measure the information delivered by financial analysts for a firm, we implement the number of analysts covering the firm, analyst forecast dispersion, and analyst forecast errors. the magnitude of accruals, earnings smoothing, and the correlation between change in accruals and change in operating cash flows are the measures for earnings management. to test the relation between information disclosure and mutual fund holdings, and to see what measures have the most significant effects, each category of information measures (accounting standards, financial analysts, and earnings management) is included one by one in order to avoid the problems of reduced observations and collinearity between any two categories of measures. table 4 reports the results. it is obvious that accounting standards and information provided by analysts both play a major role on mutual funds investment decisions. when a firm adopts better accounting standards, foreign and domestic mutual funds will increase their holdings in the firm by 0.76 per cent and 0.54 per cent, respectively. this is economically significant. from table 2, the average foreign mutual fund holding is 2.57 per cent. thus, an increase of 0.76 per cent for foreign shares is representative of 30 per cent increase in ownership. the effect of better accounting standards on foreign mutual fund ownership is stronger in developed markets but reduces in emerging markets. this indicates that emerging markets lack enforcement, which weakens the link between better accounting standards and foreign holdings (francis and wang, 2008). however, better accounting standards have a more significant impact on domestic mutual fund investment decisions in emerging markets than in developed markets. this may imply that domestic investors have an information advantage over foreign investors. 61 table 4: information disclosure quality and mutual fund holding preference this table reports the regression results of mutual fund ownership on various measures for firmlevel information disclosure. measures of a firm’s information environment are categorized in the following three ways. (i) accounting standards: a dummy for the firm’s accounting standards (accstd) that equals one if the firm uses ias or gaap as its accounting standard; (ii) analyst variables, the number of analysts covering the firm (nanalysts), analyst forecast dispersion (disp), and analyst forecast errors (ferror); (iii) degree of earnings management: the magnitude of accruals (accrual), earnings smoothing (smooth), and correlation between the change in accruals and change in operating cash flows (corr). control variables include variables used in table 3. t-statistics reported in parentheses are calculated based on clustered firm-level heteroskedasticity-adjusted standard errors. n(fyear) is the number of firm-year observations. the sample period is from 1998 to 2007. year and country dummies are added to control for year and country effects. all countries developed market emerging market variables (1) (2) (3) (4) (5) (6) (7) (8) (9) panel a: foreign mutual funds nanalysts 0.09 0.08 0.19 (7.36) (5.82) (5.16) disp -0.41 -0.38 -0.60 (-2.50) (-2.18) (-1.25) ferror -0.01 -0.01 -0.04 (-0.25) (-0.14) (-0.47) accstd 0.76 0.82 0.78 (5.18) (5.30) (1.71) accrual -0.03 -0.06 0.08 (-0.68) (-1.34) (0.75) smooth 0.02 0.06 -0.22 (0.28) (0.62) (-1.42) corr -0.17 -0.20 -0.11 (-1.64) (-1.65) (-0.49) n(fyear) 41,766 33,313 43,670 34,436 27,629 35,588 7,330 5,684 8,082 pseudo r2 4.40% 4.63% 4.43% 4.67% 4.81% 4.71% 4.08% 4.28% 3.99% all countries developed market emerging market variables (1) (2) (3) (4) (5) (6) (7) (8) (9) panel b: domestic mutual funds nanalysts 0.05 0.04 0.09 (5.87) (4.50) (4.17) disp -0.87 -0.79 -0.94 (-5.30) (-4.37) (-2.49) ferror 0.00 0.00 -0.05 (-0.06) (-0.07) (-0.54) international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 62 accstd 0.54 0.41 1.07 (3.75) (2.58) (3.20) accrual -0.07 -0.09 0.00 (-1.58) (-1.63) (0.04) smooth 0.09 0.04 0.25 (1.11) (0.43) (1.50) corr -0.02 -0.06 0.22 (-0.15) (-0.44) (1.09) n(fyear) 40,108 32,136 41,252 32,199 26,684 33,199 7,909 5,452 8,053 pseudo r2 3.91% 4.16% 3.85% 4.16% 4.39% 4.06% 2.85% 3.37% 2.73% therefore, the lack of enforcement does not affect domestic mutual funds. among analyst variables, the number of analysts and analyst forecast dispersion are positively and negatively associated with the investment decisions of both foreign and domestic mutual funds, respectively. these two variables are significant at conventional levels. however, the significance of analyst forecast dispersion in emerging markets reduces for foreign investors. analysts’ forecast errors have the least effect on investment decisions for domestic and foreign mutual funds. this may show that visibility is more important than information quality for mutual funds because higher analyst coverage indicates that the firm is more visible and provides more information to investors. it is also possible that investors have difficulty judging the accuracy of information so the correlation between fund ownership and forecast errors is not significant. among all three earnings management measures, none of them has a significant effect on foreign or domestic holdings. these results are consistent between foreign and domestic mutual funds. 4. determinants of mutual fund holdings at the country level this section is on how country characteristics affect foreign and domestic mutual fund ownership. previous literature documents that a country’s information disclosure quality and its institutions both affect foreign investors’ intention to enter the stock market. for example, leuz, nanda, and wysocki (2003) show that information quality is poor for firms in countries with weak investor protection. gelos and wei (2005) and leuz et al. (2009) show that low country information 63 transparency or information disclosure quality reduces foreign investment and is likely to exacerbate the degree of information asymmetry faced by foreign investors. therefore, foreign mutual funds’ ownership varies with the countries’ institutions and information disclosure quality. however, it is not clear whether such country characteristics affect domestic mutual funds’ investment decisions. this study first examines how a country’s characteristics affect both foreign and domestic mutual fund ownership. then, we test how mutual funds’ ownership varies with a country’s information disclosure. lastly, we test whether foreign or domestic mutual funds play an effective role on monitoring. 4.1 country characteristics and mutual fund holdings here, country characteristics included in the test are a dummy variable to indicate emerging market, gdp, stock market capitalization to gdp, gdp growth, ratio of trade (exports plus imports) to gdp, standard deviation of annual gdp growth over the past five years, and the annualized standard deviation of daily stock market returns over the past one year. the dependent variable is mutual fund ownership in percent at the firm level. control variables include all firm-level characteristics used in table 3. t-statistics are reported with clustered firmlevel heteroskedasticity-adjusted standard errors. table 5 reports the results. from the left panel of table 5, it is clear that foreign mutual funds invest less in emerging markets. if all conditions are equal, foreign investors reduce their ownership by an average of 1.40% in emerging markets when compared with developed markets. all variables except trade to gdp and stock market volatility are negatively related to foreign ownership. thus, foreign investors actually invest less in big countries and this is caused by the diversification needs for foreign investors (ferreira and matos, 2008). international trading seems to be an important factor for attracting foreign capital. if a country has a stronger connection with other countries, foreign investors tend to invest more in this country. the portfolio preference is similar for foreign investors in both developed and emerging markets except when considering gdp, gdp growth volatility, and stock market volatility. the coefficients of gdp, gdp growth volatility, and stock market volatility become insignificant in emerging markets for foreign investors. for domestic mutual funds (right panel of table 5), there is no difference in fund ownership between developed and emerging markets. contrary to foreign investors, domestic mutual funds’ international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 64 ownership is positively related to market size and negatively related to trade to table 5: tobit regression results on mutual fund ownership on country characteristics this table reports regression results of foreign and domestic mutual fund ownership on various country characteristics. country variables include emed that is equal to one if the country is an emerging country, gdp per capita (gdp), stock market capitalization to gdp (mcap), gdp growth, ratio of exports plus imports to gdp (tgdp), standard deviation of annual gdp growth over past five years (σgdp), and the annualized standard deviation of daily stock market returns over year t-1 (σmkt). control variables include firm-level variables used in table 3. t-statistics reported in parentheses are calculated based on clustered firm-level heteroskedasticity-adjusted standard errors. n(fyear) is the number of firm-year observations. the sample period is from 1998 to 2007. year dummies are added to control for year effect. foreign mutual funds domestic mutual funds all developed emerging all developed emerging variables (1) (2) (3) (4) (5) (6) emed -1.40 -0.38 (-6.51) (-1.67) gdp -0.38 -1.67 -0.09 -0.44 -0.59 -0.27 (-4.44) (-7.02) (-0.92) (-5.14) (-2.41) (-2.76) mcap -0.25 -0.16 -0.40 0.81 0.55 1.56 (-3.53) (-2.09) (-1.91) (11.78) (8.36) (5.89) ggdp -9.80 -19.85 -5.09 2.99 15.54 -4.96 (-5.48) (-6.36) (-2.40) (1.62) (4.66) (-1.86) tgdp 0.76 0.91 0.59 -0.37 0.73 -1.95 (9.85) (8.28) (3.29) (-5.63) (7.25) (-13.58) σgdp -9.79 -31.37 -3.51 -51.10 -175.56 -8.91 (-3.64) (-5.15) (-0.96) (-15.74) (-22.05) (-2.52) σmkt 1.83 5.83 -0.83 1.01 1.65 0.52 (2.50) (4.73) (-0.96) (1.48) (1.35) (0.52) n(fyear) 52,527 42,573 9,954 50,637 40,099 10,538 pseudo r2 3.44% 3.67% 3.40% 2.19% 2.53% 2.64% gdp. it is reasonable that the size of equity market is positively related to domestic fund ownership because market size in general is associated with local people’s wealth. since foreign ownership is positively related to trade to gdp, the negative correlation between domestic ownership and trade to gdp indicates that well-known countries attract more foreign investors and this causes domestic investors to have a small fraction of ownership. the finding from emerging markets supports this argument because firms in emerging markets on average are 65 smaller than firms in developed markets. an increased foreign ownership in emerging markets will lead to a significant drop in domestic ownership. 4.2 country institutions and mutual fund holdings this subsection provides results on how country-level institutions affect investors’ decision making when they form their portfolio holdings. previous literature documents that countries’ laws and disclosure quality affect foreign investors’ intention to enter a market or to hold a firm’s stock (gelos and wei, 2005; ferreira and matos, 2008; leuz et al., 2009). here, we use security regulation to proxy for the enforcement of security laws, disclosure requirement to proxy for a country’s disclosure quality, accounting standard to proxy for the countries’ financial disclosure quality, and anti-self-dealing to proxy for the quality of corporate governance in the country. these factors are supposed to be positively related to foreign mutual fund ownership. however, the effect of these factors on mutual fund ownership is still not clear when markets are further split into developed and emerging markets, and funds are grouped into foreign and domestic mutual funds. table 6 first reports the results by using foreign mutual funds as the sample in panel a and then by using domestic mutual funds as the sample in panel b. it is obvious that a country’s institutions matter to foreign investors in both developed and emerging markets. however, the effect of institutions seems to be stronger in developed markets than in emerging markets except for security regulations. one possible explanation for this is that developed countries have better abilities to fairly enforce laws. among all institutional variables, disclosure requirements have the greatest effect on foreign holdings, while security regulations, accounting standard and antiself-dealing are all important to foreign investors. thus, to attract foreign capital, emerging countries have to improve their institutions to protect investors and provide quality information disclosure. contrary to foreign investors, domestic mutual funds have different behaviors in both developed and emerging markets. when all countries are pooled together, domestic mutual funds seem to not care about security regulations and accounting standard. however, disclosure requirements and anti-self-dealing are still important for domestic mutual funds. since domestic mutual funds know local firms better, information asymmetry is less of a problem when they international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 66 table 6: country institutions and mutual fund holding preference this table reports regression results of mutual fund ownership on various measures for country-level information disclosure. the dependent variable is the foreign or domestic mutual fund ownership in all markets, developed markets, and emerging markets. the measures for information disclosure quality at the country level include dummies for security regulations, disclosure requirements, accounting standards index, and anti-self-dealing index. see la porta et al. (2006) for the description of the indexes. the dummy for each country disclosure measure is equal to one if the country has an index score above the median of all countries. control variables include country variables used in table 5 and firm variables used in table 3. t-statistics reported in parentheses are calculated based on clustered firm-level heteroskedasticity-adjusted standard errors. n(fyear) is the number of firm-year observations. the sample period is from 1998 to 2007. year and country dummies are added to control for year and country effects. all countries developed markets emerging markets (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) panel a: foreign mutual funds security regulation 0.44 0.83 1.05 (3.59) (5.55) (2.52) disclosure requirement 2.06 2.23 1.05 (14.35 ) (13.29 ) (2.52) accounting standard 0.93 0.93 0.57 (6.48) (4.71) (1.84) anti-self-dealing 1.93 2.06 1.05 (15.18 ) (14.21 ) (2.52) emerging market -1.44 -1.90 -1.41 -1.83 (6.66) (-8.47) (6.52) (-8.27) n(fyear) 52,52 7 52,527 52,52 7 52,527 42,57 3 42,573 42,57 3 42,573 9,954 9,954 9,954 9,954 pseudo r2 3.44% 3.68% 3.54% 3.72% 3.72% 4.02% 3.71% 4.04% 3.42 % 3.42 % 3.50 % 3.42 % 67 table 6 (continued) all countries developed markets emerging markets (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) panel b: domestic mutual funds security regulation -1.42 -1.29 4.25 (11.29) (9.19) (9.58) disclosure requirement 1.51 0.73 4.25 (11.34 ) (4.55) (9.58) accounting standard -0.06 -1.56 4.01 (0.50) (9.64) (14.43 ) anti-self-dealing 0.57 -0.08 4.25 (4.74) (0.60) (9.58) emerging market -0.20 -0.71 -0.39 -0.47 (-0.87) (-3.19) (1.68) (2.06) n(fyear) 50,637 50,637 50,63 7 50,63 7 40,09 9 40,09 9 40,09 9 40,09 9 10,53 8 10,53 8 10,538 10,53 8 pseudo r2 2.32% 2.32% 2.19% 2.21% 2.62% 2.56% 2.61% 2.53% 3.04% 3.04% 2.98% 3.04% international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 68 make investment decisions. when countries are separated into developed and emerging markets, domestic investors demonstrate different weights on the country’s institutions. in developed countries, domestic investors care less about accounting standard and security regulations. they also pay less attention to anti-self-dealing. it is worth noting that all institutions play a significant role on the domestic investors’ decision making in the emerging markets. although domestic investors possess more information of local firms, they still value the quality of laws and information disclosure at the country level. therefore, good institutions not only attract more foreign capital, but also stimulate local people to invest in emerging market. 4.3 governance role of mutual funds ferreira and matos (2008) document that mutual funds and independent investment advisors are likely to play a governance role because they have fewer connections with their holding companies and therefore the monitoring costs will be lower. from previous sections, foreign investors and domestic investors have different portfolio preferences in developed and emerging markets. hence, it is interesting to see whether foreign and domestic investors can monitor firms effectively. in this subsection, we use return on assets (roa) instead of firm value to test whether mutual funds play a monitoring role. as ferreira and matos document, firm value and foreign ownership may be endogenously determined because higher firm value may attract more foreign investors and foreign investors push up the price, which causes a high firm value. using roa as a proxy for firm performance is a better way to disentangle the monitoring effect of institutional investors. table 7 reports the results by regressing firms’ roa on foreign ownership or domestic ownership with control variables. control variables include the entire firm characteristics used in table 3. year and country dummies are also added to control for year and country effects. from the results, it is obvious that both foreign and domestic mutual funds play a monitoring role on firms. firm operating performance is positively associated with fund ownership. particularly, domestic mutual funds seem to be a more effective monitor. for foreign mutual funds, funds’ ownership is related to higher roa in developed countries but the relation becomes insignificant in emerging countries. this indicates that foreign investors have difficulties to play an effective monitoring role for firms in emerging markets. however, the coefficients of domestic mutual 69 table 7: operating performance and mutual fund ownership this table reports the regression results of return on assets (roa) on foreign and domestic mutual fund ownership in all markets, developed markets, and emerging markets. control variables include variables used in table 3. t-statistics reported in parentheses are calculated based on clustered firm-level heteroskedasticity-adjusted standard errors. n(fyear) is the number of firm-year observations. the sample period is from 1998 to 2007. year and country dummies are added to control for year and country effects. foreign mutual funds domestic mutual funds variables all developed emerging all developed emerging mf 0.03 0.03 0.01 0.05 0.05 0.09 (2.80) (3.14) (0.46) (5.14) (3.97) (4.68) size 0.32 0.29 0.22 0.33 0.31 0.30 (6.60) (5.29) (2.23) (6.82) (5.66) (2.76) bm -0.89 -0.83 -1.70 -0.84 -0.77 -1.66 (-6.92) (-5.52) (-8.08) (-6.30) (-4.84) (-8.42) ret12 18.62 20.71 12.70 17.12 18.67 10.99 (10.06) (9.03) (6.24) (8.98) (7.74) (5.29) σret -16.65 -21.68 -1.49 -16.62 -20.59 -3.87 (-9.49) (-10.44) (-0.86) (-8.98) (-8.58) (-1.87) dy 26.83 25.70 31.55 25.54 25.06 31.06 (10.29) (8.21) (7.07) (9.70) (7.60) (7.32) roe 28.71 29.44 22.08 31.12 32.00 23.12 (40.22) (36.85) (18.38) (43.05) (39.40) (21.31) msci -0.25 -0.13 -0.02 -0.48 -0.33 -0.25 (-1.96) (-0.90) (-0.07) (-3.56) (-2.12) (-0.93) da -0.69 -0.61 -2.36 -1.09 -1.06 -2.64 (-2.42) (-1.86) (-4.44) (-3.69) (-3.08) (-5.40) cash -3.25 -4.29 6.21 -3.77 -5.01 6.36 (-4.68) (-5.57) (6.02) (-5.42) (-6.43) (6.77) chold 1.36 1.85 -0.02 1.12 1.60 0.24 (5.03) (5.93) (-0.04) (3.76) (4.54) (0.52) gsales x100 0.03 0.03 -0.08 0.02 0.02 -0.10 (2.74) (2.85) (-2.82) (1.27) (1.63) (-3.52) adr -0.80 -0.76 -0.45 -0.35 -0.22 -0.42 (-2.66) (-2.29) (-0.73) (-1.19) (-0.67) (-0.69) age -0.01 -0.01 -0.08 0.01 0.00 -0.07 (-0.91) (-1.38) (-3.04) (1.07) (0.37) (-2.96) n(fyear) 52,792 42,844 9,948 50,821 40,292 10,529 adjusted r2 0.51 0.51 0.51 0.54 0.53 0.53 international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 70 fund ownership are all significant in both developed and emerging markets. this monitoring effect of domestic mutual funds is especially strong in emerging markets. among all control variables, it is worth noting that size, past performance, dividend yields, and roe are all positively associated with roa. however, closely-held ownership has a positive effect on the roa in developed countries, but the effect disappears in emerging markets. results show that higher insider ownership does improve operating performance, but this relation only exists in developed countries. in the emerging markets, there is a lack of law enforcement to protect investors, resulting in higher monitoring costs for shareholders. because the penalties from being caught by laws or sued by shareholders are relatively minor, firm managers in emerging markets with higher ownership may actually use their power to extract private benefits from shareholders and to deter shareholders from playing monitoring role. it is also widely documented that good corporate governance is enforced through lower cash holdings (see e.g., yun, 2009). if this is the case, cash holdings should be negatively related to firm performance. as shown in table 7, the coefficient of cash holdings is negative, except emerging markets. in the emerging markets, cash holdings have a positive effect on firm performance, implying that cash holdings may play a different role in developed and emerging markets. 5. conclusions this study investigates the portfolio preferences of foreign and domestic mutual funds in developed and emerging markets. the sample covers 28 countries, 28,602 foreign mutual funds, 23,373 domestic mutual funds, and 92,747 firm-years from 1998 to 2007. previous studies in general only use the sample from developed or emerging markets with one year of data. they also focus on foreign investors to explore the preferences of asset allocation. it is still unclear whether previous findings apply to both domestic and foreign investors and remain evident in either developed or emerging markets. the maintained hypothesis is that firms in different markets have different structures. since foreign investors have information disadvantage in local markets, foreign and domestic investors should have different preferences when they form their portfolio holdings. this study uses both domestic and foreign mutual funds in developed and emerging markets to first investigate the overall portfolio preferences and then their portfolio 71 preferences in each market. results show that foreign and domestic investors have some different portfolio preferences, and these preferences vary between developed and emerging markets. foreign mutual funds in the developed markets prefer firms with large size, lower dividend yields, higher roe, more cash holdings, and fewer closely-held shares. they also like young growth stocks, firms listed in the msci index, and firms that have adr issuance. in emerging markets, they focus more on firms’ operating performance, such as roe and sales growth, but less on adr issuance. contrary to foreign mutual funds, domestic mutual funds overall do not care about msci membership and adr issuance. firm size and dividend yields also play a small role in domestic mutual funds’ portfolio holdings. such findings are consistent in developed markets. in emerging markets, domestic mutual funds place a greater emphasis on dividend yields, roe, and sales growth. firms’ information environments also affect both foreign and domestic mutual funds’ portfolio preferences. three dimensions of measure – accounting standards, earnings management, and analyst variables – are utilized to proxy for the level of the firms’ information asymmetry. results indicate that accounting standards and analyst-related measures are significantly related to fund ownership for both foreign and domestic mutual funds. however, the positive effect of strong accounting standards and lower analyst forecast dispersion reduces for foreign mutual funds in emerging markets but remains significant for domestic mutual funds. among all measures, the number of analysts covering a firm is the most important. results also would suggests that foreign mutual funds obviously invest less in emerging markets. higher gdp, larger equity market size, and higher gdp growth exhibit negative effects and foreign trades have a positive effect on foreign mutual funds. for domestic mutual funds, ownership is positively related to stock market size and negatively related to foreign trades. to explore whether institutions affect fund managers’ intentions to invest, four measures are used: security regulation, disclosure requirement, accounting standard, and anti-self-dealing. results show that all measures for a country’s institutions affect foreign mutual funds’ intentions to invest in firms in the country. the findings are consistent for both developed and emerging markets. however, domestic mutual funds only care about disclosure requirements in developed markets while all four measures matter in emerging markets. further results show that both international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 72 foreign and domestic mutual funds are able to play a monitoring role for firms, but foreign mutual funds can only monitor firms effectively in developed markets. author information: wen-hsiu chou is from the department of finance and real estate, college of business administration, florida international university, miami, fl 33199. author’s contact information: wchou@fiu.edu, (305) 348-0533. references aggarwal, r., i. erel, m. ferreira, and p. matos, (2011). does governance travel around the world? evidence from institutional investors, journal of financial economics, 100: 154-181. aggarwal, r., l. klapper, and p.d. wysocki, 2005, portfolio preferences of foreign institutional investors, journal of banking & finance, 29: 2919-2949. bohn, h. and l.l. tesar, (1996). u.s. portfolio investment in foreign markets: portfolio rebalancing or return chasing? american economic review, 86: 77-81. bradshaw, m., b. bushee, and g. miller, (2004). accounting choice, home bias, and u.s. investment in non-u.s. firms, journal of accounting research, 42: 795-841. chan, k., v. covrig, and l. ng, (2005). what determines the domestic bias and foreign bias? evidence from mutual fund equity allocations worldwide, journal of finance, 60: 14951534. covrig, v., m. defond, and m. hung, (2007). home bias, foreign mutual fund holdings, and the voluntary adoption of international accounting standards, journal of accounting research, 45: 51-70. dahlquist, m. and g. robertsson, (2001). direct foreign ownership, institutional investors, and firm characteristics, journal of financial economics, 59: 413-440. daske, h., l. hail, c. leuz, and r. verdi, (2011). adopting a label: heterogeneity in the economic consequences around voluntary ias adoptions, working paper at university of mannheim. fan, j., j. wei, and x. xu, (2011). corporate finance and governance in emerging markets: a selective review and an agenda for future research, journal of corporate finance, 17: 207214. ferreira, m.a. and p. matos, (2008). the colors of investors’ money: the role of institutional investors around the world, journal of financial economics, 88: 499-533. francis, j.r., (2004). what do we know about audit quality? british accounting review, 36: 345368. francis, j.r. and d. wang, (2008). the joint effect of investor protection and big 4 audits on earnings quality around the world, contemporary accounting research, 25: 157-191. gelos, r.g. and s. wei, (2005). transparency and international portfolio holdings, journal of finance, 60: 2987-3020. kang, j.k. and r. stulz, 1997, why is there a home bias? an analysis of foreign portfolio equity ownership in japan, journal of financial economics, 46: 3-28. la porta, r., f. lopez-de-silanes, and a. shleifer, 2006, what works in securities laws?, journal of finance, 61: 1-32. 73 la porta, r., f. lopez-de-silanes, a. shleifer and r. w. vishny, (1998). law and finance, journal of political economy, 106: 1113-1155. leuz, c., k.v. lins, and f.e. warnock, 2009, do foreigners invest less in poorly governed firms? review of financial studies, 22: 3245-3285. leuz, c., d. nanda, and p.d. wysocki, (2003). earnings management and investor protection: an international comparison, journal of financial economics, 69: 505-527. lai, s., l. ng, and b. zhang, (2012). does pin vary and affect asset prices around the world?, working paper. yun, h., (2009). the choice of corporate liquidity and corporate governance, review of financial studies. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 5 financial transaction and fiduciary obligation: ethics, economics or commingled commitment? 1 the international journal of banking and finance, volume 9 (number 4) 2012: pages 1-27 financial transaction and fiduciary obligation: ethics, economics or commingled commitment? s. romila palliam, lee g. caldwell and dilip k. ghosh gulf university of science and technology, kuwait ____________________________________________________________ abstract financial transactions and fiduciary obligations are simply intertwined. fiduciaries are subject to the principle of fidelity. it appears, at times at least, public trust in fiduciary commitments is declining as a result of fiduciaries’ selective reporting of financial events and the existence of conflicts when fiduciaries have selfish motives: motives being not always to maximize the trusting party’s value. it is the agency problem. this work attempts to enunciate that commitments and fiduciary obligations emanating from initial financial transactions are not to be violated or ignored as a matter of policy or practice. the questions that arise are: should a fiduciary be obliged to guarantee a certain outcome for the counter-party, and should a fiduciary be held accountable to a certain type of outcome? we examine what the guidelines are or should be put in place. initially, under the garb of some socio-religions edicts-cum-dicta, and then under the well-known economic analytics, we make our points and move the view to the forefront. key words: financial contracts, fiduciary obligation, agency problem, stewardship, fidelity jel classification: k22, m14, z12 _____________________________________________ 1. introduction governance encompassing ethical issues that entail a display of meaningful loyalty in all forms of political and corporate conducts has been a subject of intense debate over many years, and over the past decade, in particular, since enron, worldcom, pharmalat, onetel, inforsys, dubai world and other financial disasters. performing in the professionally responsible and loyal fashion with respect to other people’s resources or firm’s commitment of funds ought to characterize the conduct of investment managers and or fiduciary agents. 2 technically, it is the issue of profit creation vis-a-vi shareholders’, that is, investors’ wealth build-up. often, however, investors are significantly deprived of the fruits their investment funds spawn. stone notes (1934), “….when the history of the financial era which has just drawn to a close comes to be written, most of its mistakes and its major faults will be ascribed to the failure to observe the fiduciary principle, the precept as old as holy writ, that 'a man cannot serve two masters’.” more than a century ago, equity gave a hospitable reception to that principle and the common law was not slow to follow in giving it recognition. no thinking man can believe that an economy built upon a business foundation can permanently endure without some loyalty to that principle. regrettably, over the past few decades, fiduciary obligation became a judicial feat. in essence, less time and thought have been afforded to the concept of loyalty or fiduciary obligation, which is often associated with trust and trust-like relationships in which conflicts of interest and duty tend to arise (flannigan, 1989). consistent with these views, demott (1988) contends that it may very well be the lack of discussions about what fiduciary entails that have created the current investigations into several improprieties that relate public’s lack of faith in the reporting practices of corporations and the functioning of capital markets. any breach of public trust amongst stakeholders requires one to revisit the moral and ethical obligations which include accountability, honesty, and transparency of each stakeholder in a trust relationship. should governments and corporations wish to be virtuous and free, the obligations of moral leadership associated with the vocation can in no way be relinquished. there is a long list of studies that relate religious and moral teachings on fiduciary obligation in government and business management. islamic scholars have undertaken a thorough examination of relevant verses from the holy qur’an and the sunnah: ariff and iqbal (2011). they have established the basic principles that govern the rights and obligations of participants in the financial transactions. a dishonest dealing by a fiduciary is presented in the parable of the dishonest steward in the holy bible (luke, chapter 16) where the unjust steward was accused by his master for squandering the master’s wealth. while this parable is cited many a times in christian communities, no evidence exists whether this is widespread in current business dealings. it would suffice to say that the frequency with which the media are reporting issues relating to fraudulent conduct, poor stewardship and errors of omission and commission of international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 3 deceptions is increasing at an exponential rate. inherent and systemic corruptions in governments have also been witnessed more recently. consequently, the causes of the catastrophic collapse of governments and corporations require deeper meaning and understanding of fiduciary obligation. from an islamic perspective, beekun and badawi (2005) argue that the link between islam and business management has been sparse. trust and benevolence are variables that are factored into an islamic business model together with justice and balance. in islam there are two primary sources, namely the holy qur’an and the sunnah or hadith, both of which recommend unequivocally what fiduciary obligation is. essentially, the primary sources have implications for both individual and collective economic conducts. moreover, economic as well as non-economic values, norms and codes are inscribed in the shari’ah islamiyah, commonly referred to as islamic law that draws extensively and exclusively from the two primary sources. according to kamali (1989), the shari’ah “refers to commands, prohibitions, guidance and the principles that god has addressed to mankind pertaining to their conduct in this world and [for] salvation in the next”. the findings in the literature suggest that compliance to the revealed sources of knowledge (holy qur’an and sunnah) and compliance to shari’ah ought to be given the highest priority as sources of fiduciary power. any potential for objectionable behavior ought to be thwarted by a conscious sense of internalized islamic values. the western viewpoint demott (1988) in her seminal work contends that fiduciary obligation is one of the most elusive concepts in anglo-american law. notwithstanding the issues surrounding fiduciary obligation, an important clarification was made by canada's supreme court by mclachlin (2003) where a majority of the court agreed with mclachlin when she stated that fiduciary obligations were not obligations to guarantee a certain outcome for the vulnerable party, regardless of the fault. they do not hold the fiduciary to a certain type of outcome, exposing the fiduciary to liability whenever the vulnerable party is harmed by one of the fiduciary’s employees. rather, they hold the fiduciary to a certain type of conduct. this suggests that a fiduciary is not a guarantor of a certain outcome and a fiduciary does not breach his or her duties by simply failing to obtain the best result for the beneficiary. perhaps the widespread prevalence of these decisions in western courts may have coincided with the actions of individual muslims and therefore a sense of incongruence surrounding fiduciary obligation have crept into islamic business conduct. the very nature of the terms 4 “fiduciary” and “obligation” renders them to a sense of discrepancy resulting in some divergence between islamic philosophy and practice in economic life. the latter term, namely “obligation” is used in a variety of personal situations and embodies responsibility and accountability the extent of which is immeasurable. the former entails the use of the latin word fiducia meaning trust and fidere meaning to trust. the issue that further exacerbates the definition is whether fiduciary obligation is intrinsic or extrinsic. in this regard the british columbia court of appeal (1997) wrote of fiduciary as follows: every servant and every agent owes to his master or principal duties of good faith and fidelity (intrinsic). (s)he owes those duties because the law imports those duties as a contractual term or because the contract contains an express term to that effect (extrinsic) or because a court of equity would have imposed an obligation of good faith and fidelity as a matter of conscience (extrinsic and intrinsic). as a result the complex relationship existing in a servant-master and agent-principal scenario, a legal test for fiduciary duty is paramount. servants and agents together with masters and principals have an intrinsic fiduciary obligation to safeguard their own interpersonal interests. the applicability of fiduciary and obligation in variety of contexts presents an opportunity to determine situation-specific fiduciary obligation. resolving conflicts of obligations in fiduciary relationships requires an understanding of several subjective issues related to trust, faith and honesty. from an islamic perspective, khan and thaut (2007) in their seminal work on an islamic perspective on fair trade, maintains that economic exchange in the form of trade and commerce played a vital role in the expansion of islam and islamic values of kinship, tradition and communal relationship. scholars have identified mecca as the birthplace of islam and mecca was a market and center for commerce. khan and thaut (2007) suggest that the early muslims merchants were not only engaged in trade but they went to distant lands in connection with business. moreover, they identify the prophet muhammad as a successful trader with integrity for which he was conferred the title ‘the trustworthy’. this perhaps presented an opportunity to preach islam and therefore a literature review suggests that fiduciary obligation in early islam became intrinsic through the teaching of the prophet muhammad. the golden rule appears in the following statements attributed as revelation to muhammad, “woe to those . . . who, when they have to receive by measure from men, exact full measure, but when they have to give by measure or weight to men, give less than due”: qur’an (surah 83: verse 1). international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 5 the rich value system of trust, honor and duty in islam ought to be gradually passed on from one generation to another. moreover, wealth should be circulated widely and not held or concentrated in the hands of a few. to encourage circulation of wealth, the shari’ah encourages responsible trade. the shari’ah also promotes commutative sales contracts and thereby the attainment of a monetized economy rather than barter. on the redistributive side, the shari’ah also institutes zakat, which is a religious tax based on wealth or income depending on the category of wealth held by muslims. this is obligatory on all muslims who possess wealth beyond a fixed minimum and hold it for at least one calendar year. the islamic state collects it, in the absence of which individuals are obliged to pay it voluntarily on their own to the beneficiaries designated by the qur’an (poor, indigent, travelers, bankrupt, zakat collectors), to liberate slaves or oppressed people and “in the way of god” especially in the month of fasting. the foundations of the social system of islam rest on the belief that all human beings are equal and constitute one single fraternity. in this regard a fiduciary is expected to act thoughtfully and expediently. in the west courts impose a fiduciary obligation on the trusted party whereas in islam one’s conscience and one’s submission to shari’ah play a more meaningful role. good conscience requires one to act at all times in the interest of the trusting party. 1.1 identifying trust and obligation in fiduciary obligation since a fiduciary is one who is in a position of trust and him or her becoming the locus of control, it is the very essential characteristics of trust and control that present major problems in a fiduciary’s obligations. typically, a person or groups of interested persons get to trust one another for particular objectives or to accomplish certain tasks. a manifestation of trust becomes essential for commercial and economic activities to be sustained. trust can be abused by the trusted parties when they do not act with care and diligence. the trusted party may intentionally divert value away from the party who has reposed the trust. a literature survey would identify at least three types of trust in corporate relationships. these include deterrence-based trust; knowledge-based trust and identification-based trust. in this regard flannigan (1989) argues that only two types trust give rise to fiduciary obligations. he contends that in the first instance a person may trust another in an unconstrained and uninhibited personal way. this form of trust is generally a result of feelings of intimacy or security brought about by the affection or concern for another and may also arise out of past trust experiences or conduct 6 created over time. at times this form of trust may come up instantly as a result of the office occupied by the trusted person or as a result of his or her knowledge, experience and expertise. flannigan (1989) considers this as a 'deferential' kind of trust in that the trusting person will defer to the judgment of the trusted person. this deference may be total, or it may be only partial or situational and it is accompanied, in some cases, by elements of necessity, dependence or submission. in other cases there is no demonstrated vulnerability. however, what is paramount in relation to fiduciary obligation is that the trusted person is secured in the knowledge that his or her judgment is being relied on in the circumstances. in the second form of trust identified by flannigan (1989) there is typically no deference or vulnerability involved and an example given is the kind of trust that employers have in their agents and servants. in the flannigan paradigm then employers expect their agents and servants to act faithfully although they do not have unreserved faith in them. they remain wary and, in extreme circumstances, even skeptical of them. employers are 'vigilant' in the sense of remaining aware. they trust their agents and employees, but remain observant for signs that their trust is misplaced. the expectation underlying this kind of trust is that of fair dealing between competent and capable persons. bogle (2009) cites adam smith who contented already in the 18th century that managers of other people’s money rarely watch over it with the same anxious vigilance with which . . . they watch over their own . . . they very easily give themselves a dispensation, and bogle (2009) concludes that management negligence prevails to a point of an almost complete disregard of their duty and responsibility to their principals. following this trend of thought, one would argue that a dishonest steward could certainly guilty of a clear breach of fiduciary duty by failing to administer the property solely in the interest of the beneficiary and continuingto serve his own interest. 1.2 relations in fiduciary obligation fiduciary obligation is a necessary feature of an agency relationship. in this regard flannigan (1989) posits that an understanding of the relationship and obligation that renders “the overall legal regime in which the interests of the various parties are accommodated or balanced”. when someone has undertaken to act for and on behalf of another in a particular matter, this gives rise to a relationship of trust and confidence. some fiduciary relations are more intense than others. the relationships could include, among others, beneficiaries and trustees; wards and international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 7 guardians; principals and agents; clients and lawyers; administrators of estates and executors; and next of kin and legatees. the board of directors and the various officers are in a fiduciary relationship to the corporation and to the shareholders. moreover, each partner in a partnership is in a fiduciary relationship to the other partners. the partner has the duty and power to consider and take care of the interests of the other partners, too. following the same trend of thought, scott (1949) writing in the california law review contends that the greater the independent authority to be exercised by the fiduciary, the greater the scope of his fiduciary duty. “a trustee is under a stricter duty of loyalty than is an agent upon whom limited authority is conferred or a corporate director who can act only as a member of the board of directors or a promoter acting for investors in a new corporation.” 1.3 from agency theory to stewardship theory schoorman, mayer and davis (2007) identify trust as an important aspect of a business relationship. risk is at the core of business undertakings, and therefore it would not be surprising to conclude that trust would lead to risk taking in business relationships. the idea that “willingness to take risk” is embedded in trust. it is the view expounded by sitkin and george (2005). this, however, spawns the scope for opportunistic and self-serving conduct of agents which presents challenges in a fiduciary relationship. as a result, the notion that an agent may not act in accordance with the interest of the principal, commonly referred to as agency conflict, and may subject the principal to several agency costs. these costs according to flannigan (1989) are not created solely by those in the legal class of agents but other intermediaries and coparticipants in ventures, too. the function of the fiduciary obligation within this context would be to discourage any conduct that results in such intermediary or agency costs. jensen and meckling (1976), in their classic work on the theory of the firm, managerial behavior, agency costs and ownership structure, identify corporate managers as agents who are engaged to maximize the returns to principals, namely the shareholders. they assert that as agents do not own the corporation’s resources, they may commit certain ‘moral hazards’ merely to enhance their own personal wealth at the cost of their principals when they contract with executives to manage their firms for them. 8 in this regard davis, schoorman and donaldson (1997) contention that the agency theory provides a useful way of explaining relationships where the parties' interests are at odds and can be brought more into alignment through proper monitoring and a well-planned compensation system stands valid. however, a more comprehensive theory such as the stewardship theory needs to be considered for further knowledge and understanding on the issue. a distinguishing feature of the two different theories namely, agency theory from stewardship theory is the use of trust versus control systems to manage risk. the stewardship theory defines human behavior conduct beyond economic perspectives of perquisites. the literature dealing with stewardship is somewhat limited. on the surface, stewardship seems to be a simple concept, but the associated obligations are difficult to actualize. stakeholders are held accountable for their decisions that ought to lead to sustainable institutions and overall care of the various capitals entrusted to them. these include resources in the form of human capital, financial capital, natural resources, social capital and cultural capital. as davis, schoorman and donaldson (1997) see it, the stewardship theory has its roots in psychology and sociology; it examines situations in which executives as stewards are motivated to act in the best interests of their principals where the model of man is based on a steward whose conduct is ordered such that pro-organizational, collectivistic behaviors have higher utility than individualistic, selfserving behaviors. this is reflective essentially of general partner vis-à-vis limited partner in partnership model of business venture. it is argued then that given a choice between self-serving behavior and pro-organizational behavior, “a steward's behavior will not depart from the interests of his or her organization”. “a steward will not substitute or trade self-serving behaviors for cooperative behaviors. thus, even where the interests of the steward and the principal are not aligned, the steward places higher value on cooperation than defection …. because the steward perceives greater utility in cooperative behavior and behaves accordingly, his or her behavior can be considered rational”. if one considers the principal’s point of view that a fiduciary needs to protect and provide guidance and support to an organization together with the singular hope of ensuring the resources are handed over to new stewards in a condition that is as good or better, then the stewardship theory seems superior. religious teachings manifest stewardship theory in many forms. one can see this in the parable of the talents in the bible (luke, chapter 19) “well done, good and faithful slave! international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 9 you have been faithful in a few things. i will put you in charge of many things. enter into the joy of your master.” from an islamic viewpoint, khalifa is the arabic term for stewardship or fiduciary defined as a person or persons who hold(s) assets in trust for a beneficiary. khalifa relates to the nature of a legal trust (holding of something in trust for another). this entails a fiduciary contract in a fiduciary capacity with fiducial power. it is illegal for a fiduciary to misappropriate resources for personal gain. one of the well known hadith of prophet muhammad “kullu kum ra’in wa’ kullu ra’in mas’ool an rai’yatay-hi…” meaning that “every one of you is a caretaker, and every caretaker is responsible for what he is caretaker of’” defines the nature of stewardship in islam. it signifies and requires that fiduciary obligations must be honored always. 2. business, ethics and profits in section 1, we have discussed and adumbrated the moralistic approach, rooted in religious edicts and principles, that fiduciary obligations must be met because of the overarching importance of trusts and commitments underlying the financial transactions or any commitments. we have noted that fraudulent behavior often emerges and commitments are not met. here, within the framework of economic principles, we bring out the same issue and examine the root or underlying calculation of the violation of commitments. in the area of economics, ethics appears more in the name of reality and corporate governance than in the name of didactic dictums. in a piece (2010), ghosh, ghosh and zaher have raised the issue: business, ethics, and profits: are they compatible under corporate governance in our global economy? they have noted and observed the following: “….the traditional belief that profits and ethics are at odds with each other in the world of business. corporate governance appears to be a hindrance or a drag on profit maximization. (they) show that moral codes, public interest and social values pose no threat to profit maximization of any firm. it is demonstrated with the illustration of transfer pricing and public goods-based economy that profits and ethics are quite compatible within the strait jacket of societal norms and corporate goals.” in the section i, we have indicated that agents who are the custodians of people funds may engage in practices that are not consistent the investors’ interest. to make the point louder we bring out a sketchy income statement of a firm as follows, and then point to the reality that 10 counter-productive conduct of fiduciary agents may go against their own snatch-it-all, snatch-itaway short-term goal of maximizing fiduciary interest: table 1: income statement of xyz corporation ___________________________________________________________________________ 1. gross revenue ( xpr .= ) $100,000.000 2. minus fixed costs, including agents’ salary and bonus (f) 20,000,000 3. minus variable costs, e.g., wages, utilities, etc. (v) 25,000,000 4. earnings before interest and taxes (ebit) 55,000,000 5. minus interest expenses, bondholders, banks, etc. (i) 15,000,000 6. taxable income or profit before taxes (pbt) 40,000,000 7. minus taxes @25% (t) 10,000,000 8. net income (ny) 30,000,000 9. minus dividends to preferred stock holders (pd) 10,000,000 10. earnings available to common stock holders (eac) 20,000,000 _________________________________________________________________________ eac (in line 10) is the shareholders net wealth or net profit. if the agents of xyz corporation increase their emolument to make line 2 to be more (say, $30,000,000), ceteres paribus, line 10 drops to $12,500,000. fiduciary (that is, the top management of the firm or any individual trustee) can violate the embedded trust and rob the investor(s). note: repsdpseps n eac +≡= . here n is the number of shares outstanding, ,, dpseps and interest on loan are well-known and well-accepted fact of economic calculus that the discounted value of dps , that is, ∑ = + t t t t r dps 1 )1( measures the value of a share of the investor (where tdps is the dividend per share at time 1 and the time horizon for dividend flows is t periods from now. it is evident now that if eac is reduced by fiduciary by his/her grab-all or grab-most policy in line 2, eps drops and the value of the share drops, fiduciary’s compensation (fixed salary and bonus) has to go down, and in extreme situations the fiduciary can be given the pink slip. any prudent and far-sighted manager must therefore be induced to keep the obligation intact to the international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 11 extent possible for the reality of economic dynamics, even if ethics and moral pressure are not persuasive enough. note obligation is a constraint. economics is the science of constrained optimization (that is, either maximization or minimization). wealth maximization or profit maximization is never unconstrained. in the world of business we encounter two facets of economic reality simultaneously in general equilibrium: profit maximization (by producers-cum-sellers) and utility maximization (by consumers). when business ownership is construed within the straitjacket of corporate structure, profit maximization turns into what we have already labeled as wealth maximization of shareholders. since corporations are managed by agents such as the board of directors, ceos and cfos and so on, we know, as already noted, the agency problem, – the conflict between agents and owners of firms (that is, shareholders) rears its head, and this creates the failure by the agents to keep their commitment or obligation as self-interest dominates in their operations and calculations. corporate governance has come into existence to reduce or eliminate this conflict of interest. the existing literature waxes eloquently on various facets on this and related aspects and concerns. following the classic work of jensen, and meckling (1976), followed by fama (1980), and a series of papers (e.g., fama and jensen (1983), demsetz and lehn (1985), shleifer and vishny (1986), morck, shleifer and vishny (1988), jensen and murphy (1990), agrawal and knoeber (1996), han and suk (1998), bryan, hwang and lilien (2000), to wit a few) has examined how business, ethics, and profits can be compatible under the rules of corporate governance. it is should be noted, however, that most people think business firms and contractual deals are driven solely by profit motives. baumol (1959, 1962, 1970), williamson (1964), hicks [1954], and others, however, have raised questions as to the adequacy, tenability and unqualified validity of that view, particularly within the framework of long-term maximization principle. to frank knight (1888), “… perhaps no term or concept in economic discussion is used with a more bewildering variety of well-established meanings than profits”. building further on hawley’s view, knight considers profits as rewards for taking risks and facing uncertainty, notion alluded independently in section 1. 12 no matter how we construe profits, we recognize that business and profits are intimately linked and financial transactions and fiduciary obligations are intertwined. the issue at hand is: what has ethics to do with either business or profits? it is normally perceived that ethics is a set of moral codes that usually stands to inhibit wanton pursuit of selfish interest, and thus it puts a brake on the profits a corporation seeks to make and enlarge (and agents often try to divert them away from equity holders). very aptly, profits and ethics in this view are at odds with each other. the governments federal, state and local in most societies watch over the business practices of corporations and other profit-seeking entities in an effort to resolve any conflict between the two ends. the anti-trust division of the u.s. justice department regularly checks out and forestalls the creation of unwanted monopoly power in the united states. the securities and exchange commission (sec) monitors trading and brokerage houses, and makes insider trading outright illegal. the internal revenue service (irs) makes sure that every business organization fulfills its tax liabilities, provides the correct compensation to its employees and social security benefits, and so on. tax evasion is a felony punishable by laws in any land. in this view, it appears that ethics is antithetic to a positive business climate. empirical evidence can be cited to make this point clear in real life. in a more comprehensive work we can document claims made that ethics and profit are mutually exclusive. here, in this paper, we show that profits and ethics are not necessarily at loggerheads with each other. fiduciaries should realize that and should avoid non-fulfillment of fiduciary obligation. indeed, efforts to link ethics as a management tool to increase productivity and ultimately profits have been verified by burke and black (1990). it is possible to identify many claims that “…when we do what we believe is proper [i.e. ethical], the company gains.” some practitioners who have strongly incorporated social responsibility have noted that their gross yields are comparable to anyone in money market funds. while, anita roddick of the body show, a company committed to ethical decision-making, stresses that “for more than 10 of 15 years that it has been in business, its sales have grown an average of 50 percent a year. william ouchi (1981) sums up that “….profits are the reward to the firm as it continues to provide true value to customers, to help its employees to grow, and to behave responsibly as a corporate citizen.” international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 13 on the other hand, it may appear at times that relaxation of ethical standards may enhance profitability, whereas rigorous enforcement of the existing ethical standards and/or introduction of additional ones will attenuate profits as well as the growth of business in our societies. our conclusion is that such a perception is faulty, and often misleading and deceptive. it may be contended, and a point is always made by the opponents of ethics in business, that laissez faire means the absence of any impediment, and any ethical codes are noting but impairment of business. 2.1 what is the goal of business: social responsibility and/or profitability? milton friedman (1970) article in the new york times magazines, “the social responsibility of business is to increase its profits,” begins the debate which continues today. labeling the concept and any application of social responsibility as a “fundamentally subversive doctrine,” friedman has emphasized that the doctrine “does harm the foundations of a free society.” further, he stresses that the only responsibility of business is “to use its resources and engage in activities designed to increase profits.” arguments immediately from many scholars, stone (1975), bell (1973), and goodpaster, and matthews (1982) counter the milton doctrine. almost two decades later, grant (1991), in “friedman fallacies” calls friedman’s argument ‘erroneous’ because it rests on assumptions of an “apolitical political base, altruistic agents of a selfishness and good derived from greed .” o’neil and pienta (1994) argue that “profitability is a legitimate end of a business, but must be moderated by ethical considerations.” the authors note that schools of business are one place where selfishness is considered a virtue legitimatized by the profit maximization model. few have tested the ethics/profit proposition and where empirical evidence does exist when they are conflicting. owen and scherer (1993) cite almost an equal number of studies which supports as it reject a positive relationship between corporate social responsibility and firm financial performance. the major findings note that actions related to environmental pollution, corporate philanthropy and information disclosure have the greatest effect on market share. the evidence is modest, but the argument for including ethical decision making in firms as a responsibility is overwhelming. one of the clearest statements of the multiple responsibilities, beyond making a profit, of business is by angelidis and ibrahim (1991) who state that the diverse expectations of managers include “maximizing profits for owners and shareholders and for operating within the legal framework. they are also 14 expected to support their employees’ quality of work life, to demonstrate their concern for the communities within which their business operates, to minimize the impact of various hazards on the global environment, and to engage in purely social or philanthropic endeavors.” 2.2 a multinational corporation (mnc) and its potential behavior it is often observed or otherwise one can easily visualize that if a multinational corporation is not prohibited from being engaged in transfer pricing, it will undoubtedly enter into it because of its gains in way of tax savings can be made and the resultant increase in corporate income in the consolidated income statements of the company can be realized. scholar such as fowler (1978), eden (1998), and many others have examined transfer pricings and corporate income extensively. it is and can be contended that if capital is raised from all markets where the mnc operates, it can be shown that increase in consolidated income of the corporation via transfer pricing is not necessarily beneficial to the enterprise. to make the point clear, examine the following hypothetical income statements of the corporations and its subsidiaries or affiliates. let a and b be two subsidiaries (or affiliates) of the company, and c be the parent (combined) company. table 2: income statement a b c sales (r) $100,000 $200,000 $200,000 cogs (c) (60,000) (100,000) (60,000) ebit 40,000 100,000 140,000 interest (i) (4,000) (4,000) (8,000) income (ni) 21,600 57,600 79,200 table 3: modified income statement a b c sales (r) $150,000 $200,000 $200,000 cogs (c) (60,000) (150,000) (60,000) ebit 90,000 50,000 140,000 interest (i) (4,000) (4,000) (8,000) income (ni) 51,600 27,600 79,200 international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 15 the first situation is a depiction of the corporation’s low mark-up policy, and the modified statement describes the high mark-up policy under a constant marginal corporate income tax rate of the 40 percent. manipulative behavior, doctoring the subsidiary-level mark up and transfer pricing do neither change the total liability nor the total corporate income. but, if the tax rates are different for a and b, then the transfer pricing underlying this statement structure yields a different picture. let us assume that a faces 25 percent tax rate and b faces the same old 40 percent rate of taxation. under this modification, the income statements presented earlier appear as follows: table 4: income statement a b c sales (r) $100,000 $200,000 $200,000 cogs (c) (60,000) (100,000) (60,000) ebit 40,000 100,000 140,000 interest (i) (4,000) (4,000) (8,000) pbt 36,000 96,000 132,000 taxes (t) (9,400) (38,400) (47,400) income (ni) 27,000 57,600 84,600 table 5: modified income statement a b c sales (r*) $150,000 $200,000 $200,000 cogs (c) (60,000) (150,000) (60,000) ebit 90,000 50,000 140,000 interest (i) (4,000) (4,000) (8,000) pbt 86,000 46,000 132,000 taxes (t) (21,500) (18,400) (39,900) income (ni) 64,500 27,600 92,100 here, transfer pricing in combination with high mark-up at a generates tax savings of $7,500 (= $47,400 $39,900), and thus increases the corporate earnings of $7,500 (= $92,100 $84,600). because of this extra gain under differential tax structure, the corporation is seemingly induced to enter into unethical behavior of not pursuing the “arm’s length” transactions. 16 let us go beyond the numerical framework and introduce a theoretical structure around those paradigms presented earlier. let us use the following notations: ir = sales proceeds of company i ( cbai ,,= ); c = costs of goods sold (cogs); ebit = earnings before interest and taxes (= gross profits); i = interest expense or debt charge, assumed constant for both a and b; pbt i= profits before taxes of i (= taxable income); it = corporate tax for i; ini = net income of i. from definition, one can derive the following: }){1( icrtni aaa −−−= (1) }){1( icrtni bbb −−−= (2) and }){1( icrtni ccc −−−= (3) it is obvious that if ttt ba == (say, 40%), then transfer pricing markup differential is inconsequential. however, if ttt ba =< (say 40%), the company will attempt to change the mark-up policy and engage in transfer pricing. the question is: is the mnc really gaining by unethical practice in this case? the answer is not affirmative in clear terms. first, in corporate environment where reward of the agents are performance based, a and b are made to go through financial facelifts. a is made to look superior and b inferior, as the last line of table 4 exhibits. this may indeed create uncomfortable positions in two subsidiaries and diminution of actual as well as potential incentives towards growth and stability. intercompany enmity may do more damage than the potential gain unethically the company can generate in such data manipulation. the section 482 of the u.s. internal revenue code prohibits this practice of transfer pricing. 17 the irs regulations provide three ways and means to establish arm’s length price: (i) comparable uncontrolled prices, (ii) cost plus, and (iii) resale prices. the oecd committee on fiscal affairs did make the same recommendation for european countries. any attempt to get around these regulatory codes may be deliberately leaked because of the rivalry created through transfer pricing mechanism, and that may lead to disastrous consequences involving a much larger loss than the original profit created through contrivance. secondly, distortion in the reallocation process will affect the capital market position for both a and b. b will have reduced earnings per share, and the net worth of shareholders will go down. a will have an opposite situation: earnings per share will be up, and net worth will rise. but loss at one end and gain at the other end do not give the clear picture for market participants as a whole. it simply raises the inter-subsidiary and inter-personal income redistribution, and the final result as to whether it is better or worse for the society is not ascertainable. 2.3 the profit-maximizing behavior of a firm the traditional view that each firm optimizes the following maximand: )(())(( gxcgxr −=π (4) here ,π r, c, x, and g stand for profit level, sales revenue, cost for the firm, output level, and public good, respectively. assuming “well, behaved” revenue and cost functions, one obtains that when the following conditions hold: (5) ∂2π / ∂x2=∂2r/∂x2 ∂2c/∂x2 < 0 (6) profit is maximized. under this exercise, g is ignored as a ceteris paribus condition. before we bring this out, first let us point out the inadequacy of the results derived from (5) and (6). we all know that (5) brings out the equality of marginal revenue and marginal cost as the necessary condition for profit maximization, and (6) defining the sufficiency condition spells out that the rate of change of marginal revenue must be less than that of the marginal cost. note here the amoroso-robinson result: 18 marginal revenue = price ) 1 1( ε −× , where ε is the price elasticity of demand. since, it is a mere mathematical necessity that: marginal revenue > 0 only when ε > 1; marginal revenue = 0 only when ε = 1; marginal revenue < 0 only when ε < 1, one can immediately conclude that the producers of goods and services with inelastic demand (that is, with ε < 1) cannot reach a profit-maximizing situation when each additional output entails additional cost (signifying that marginal cost is positive). baumol [5], brown and revankar [7], and ghosh [15] have pointed out that profit-maximization alone cannot explain a firm’s behavior; some other objective(s) may be necessary to obtain the optimum. as already pointed out, g – the provision of public good has been ignored thus far. with its introduction, (5) and (6) are modified as (5*) and (6*): (5*) and ∂2π / ∂x2={∂2r/∂x2 ∂2c/∂x2} (∂2x/∂g2) < 0 (6*) here we postulate that 0)(),( >′= xfxfg , and hence, by jacobi’s theorem, x is invertible, that is, .0)(),()(1 >′≡= − gfgfgfx we do not know, however, if ∂2x/∂g2 is greater than or equal to or less than 0, and hence (6*) is not necessarily satisfied for the firm unconditionally. if the firm chooses the sales level that it thinks is profit-maximizing may, in turn, may be profitminimizing. if ∂2x/∂g2 = 0, it is a point of inflection, and if ∂2x/∂g2 < 0, the firm is in its most suboptimal profit situation. the traditional belief holds if and only if ∂2x/∂g2 > 0. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 19 it is a narrow view to assume that social good is a free good, and the firm has nothing to relate to it. a public road is a public good, and if it is not there, as president barack obama of the united states remarked recently on his campaign trail, the firm has to incur additional cost to transport its outputs and inputs. without proper adequate infrastructure, business can hardly survive, and much less to thrive on. free-rider mind-set may give rise to the belief that ethics or regulation is inimical to business and profits. a well-ordered environment, protected by laws that promote the harmony between consumers and producers, and ethics that stands as the terra firma of the laws can sustain a society with customers and vendors enjoying optimum position, side by side. not too long ago, under the clinton administration the universal health care coverage, costs, and employer mandates had touched off a debate on the advisability of the plan and the survivability of small businesses. the obama administration has passed it into law. the point is often made that insurance premium on the employer will raise the costs, lower the profits to the vanishing point, causing hereupon the demise of business. one point is ignored in this logic is that worry-free and healthy worker is a more efficient worker, and efficient worker means a higher output per input. it is like the harrod-neutral (labor-augmenting) technical progress, which essentially contributes to the growth of the corporation. unless the efficiency elasticity of cost is estimated, it is futile to think even that insurance policy as being mandated is profitreducing. thus, reduction in profit owing to insurance premium on employer is not a valid argument, particularly in the long-run context. if employers are what hicks (1954) calls “snatchers” (of short-term profits) and not “stickers” (for long-term profits), universal health care coverage may not win; but a more dynamic, long-range view of business, ethics and profits in an international economic structure in which most competing countries have medical guarantee for workers can justify the clinton proposal and the obama’s affordable health care law. 2.4 analytical structure: portfolio manager knows investor’s utility function next, take a step forward and move into an analytical development of portfolio selection, revision and comparative statics. following the works of royama and hamada (1967), ghosh (1988), and morishima (1993), who have done some of these exercises on the effects of parametric variations, we proceed by assuming that the current asset prices are normalized, and 20 each current price is then equal to 1; prices at the end of the period are expected to be ........,,, 21 nppp it means that 1+ ri = pi (7) i i i spm ∑= where pi’s is the price of the i-th asset, which is a stochastic variable. the investor’s expected values and variance co-variance terms are as follows: e(pi) = µi, and e(pi µi)( pj µj) = σij i, j = 1, 2,…, n. let µ = e(m) = i n i i s∑ =1 µ , and σ = ji n i n j ij ss∑∑ = =1 1 σ . assume that the investor maximizes his von neumann-morgenstern utility function: u = u(µ, σ) subject to: 0 1 msi n i i =∑ = π , and so his maximand is the following lagrangean: l = u(µ, σ) + (λ )0 1 msi n i i −∑ = π (8) where λ is the lagrangean multiplier, and (8) is the expression of constrained maximization. obviously, the first-order condition of the lagrangean optimand results in the following expression: .,....,2,12 nisuu ijiji ==+ ∑ λπσµ σµ (9) partial differentiation of (9) with respect to πk yields the following: international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 21       + − =             ∂ ∂ ∂ ∂ −       jjk k k j k jij i m s sm λδ π π λ π π0 , where ))((4222 α α αα α ασσσα α ασµα α αµσµµ πσπσσπσµσµµµ ∑∑∑∑ ++++= ijjijiijijji uuusuum an d +−= jkj um δµ( k k jkkj ds d ssusu µ σµ α α ασµµµ )2 ∑+ , jkδ being the kronecker delta. now one can immediately deduce the following: k k jr j jkjr j jkkrkrrk k r d d zssuzsuzuzzs s π µ σµλ π αα ασµµµµ })(2)({ ∑ ∑∑ ++−+−=∂ ∂ (10) here rz and krz are the ratios of cofactors of rπ and krg of the following determinant: d ≡ nnnn n n mm mm ... ...... ...... ...... ... ...0 1 1111 1 π π ππ is the value of the determinant d. with a bit of algebraic manipulation now it easy to show that: jr k k jkjk j krrk k r z d d s b buzzs s π µ µ µ δλ π σ )( ∂ ∂ +++−= ∂ ∂ ∑ (11) 22 = jr k k k j j krrk zd db uzzs π µ µ λ σ ∂ ∂ ++− ∑ (12) where σu u b u−= and σ µ u u b juj −= . here it is noted that 0=∑ jr j j zπ . consider a more well-accepted utility function à la von neumann-morgenstern as follows: u(m) = m – 1/2am2 a > 0 and m < 1/a (13) where m = i i i sp )1(∑ + . in this case then one can derive the following: jrjkkr k r z b ubzu s ∑∂ ∂ += ∂ ∂ µµ µµ σσ . (14) these terms on the right-hand side are the standard hicksian substitution and income effects. however, a change in kµ affects the individual's demand through two channels: through the change in the relativity of kµ to other iiµ ’s, and through the change in the absolute level of i i i s∑= µµ . in order to obtain the pure effect of kµ on rs , we introduce an imaginary change in µ so as to leave the individual on-the same level of µ , as before. such a compensated purely relative change in kµ gives rise to a change in the is ’s since, as (9) shows, the marginal rates of substitution between is ’s depend on their relative expected real values. this effect, which we may call the relative desire effect, is represented by the first term in (10). on the other hand, the change in the absolute level of µ , induced by the change in kµ , affects the marginal rate of substitution between µ andσ . if µ∂ ∂b > 0, then the individual will be prepared to bear more risk than before when the absolute level of µ is raised. such a change in the individual's risk aversion brings about a change in the holding of assets, which is represented by the second term of (10); it may be called the absolute desire effect. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 23 evidently, a similar analysis may be made if a change in a current price pi induces a change in the variance of the real value of an asset or the covariance of the real values of some two assets. optimum solution and comparative statics show that system works, and any irritations in the parameters such as manipulations by fiduciary will create chaos and instability, already noted through value creation of investors’ initial funds. 3. some concluding thoughts a profit versus ethics is an old debate resurfacing time and time again. it is an issue that involves individual gain and social welfare. a vast and penetrating literature, following the path-breaking analysis of arrow (1951), collard’s (1988) moral sentiments, gapinski’s (1988) right angle of non-profit theater, garfield’s (1999) profits and social responsibility mix, and so on have surfaced to highlight that profits and ethics are not inimical to each other. at times it may appear that social optimum is inconsistent with individual optimum choice, and at times these two ends seem to be mutually complementary. in case of conflict between these two objectives, it is an “impossibility” to find the optimum. a value judgment has to be made at this time, and one optimum has to be chosen in preference to the other. if maximum good of the many is deemed better than the maximum good to one, maximum benefits to many must be the superior choice. marris’s (1964) economic theory of “managerial” capitalism, svensson’s (1989) fairness, the veil of ignorance, and social choice, rawl’s (1971) theory of justice, schokkaet and overlaaet (1989) on moral intuition and economic models of distributive justice, ginzberg and solow’s work (1974) should lead us to recognize that ethics and profits can and should co-exist on rational grounds. social optimum is the choice in the society of many interests and individuals. the very acceptance of the government in our lives, and the “social contract” amongst the members of the society indicate that individual interest is subordinate to social welfare. a macroeconomic structure with its full interactive feedback mechanism and mutatis mutandis provisions should justify profits with ethics than without more strongly and cogently. in formal way one may visualize the social welfare (w ) as a function of agent’s utility ( 1u ), business entities’ utility ( 2u ), and then we have the following: 24 ),( 21 uuww = (15) where welfare function is quasi-concave. try to maximize this welfare function, subject to the following “well behaved” transformation function: 0),( 21 =uut so the problem is: max ),( 21 uuww = (16) subject to 0),( 21 =uut (17) this constrained maximization is simply a lagrangean )(l optimization: )},({),( 2121 uutuuwl λ+= (18) the canonical conditions on this optimization are as follows: ∂l/∂u1 = 011 =+ tw λ (19) ∂l/∂u2= 022 =+ tw λ (20) ),(/ 21 uutl =∂∂ λ (21) where 0/,0/,0/,0/ 22112211 >∂∂≡>∂∂≡>∂∂≡>∂∂≡ uttuttuwwuww from (19) and (20) we obtain: w1/w2 = t1/t2 (22) optimization yield the optimum social welfare: *1u and * 2u define this optimum constellation. if )( 21 uhu = , and 0)(/ 221 ≤′= uhdudu for 21 uu ≤ and international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 25 0)(/ 221 >′= uhdudu for 21 uu > the optimum is ascertainable. hence, the impression that ethics is a profit-reducing instruments or a drag on business is a misconception. with ethics, a business enterprise operates and optimizes in the world rationality. author information: s. romila palliam, jd, instructor in business law, college of business, gulf university for science and technology, kuwai. phone: +(965)-2530-7424, fax: +(965)2530-7030, e-mail: palliam.s@gust.edu.kw. lee g. caldwell is professor and dean of the college of business, gulf university for science and technology, kuwait. phone:+(965)-25307333, fax: +(965)-2530-7030, e-mail: caldwell.l@gust.edu.kw. dilip k. ghosh is professor of finance, college of business, gulf university for science and technology, kuwait. phone: +(965)-2530-7176, e-mail: ghosh.d @gust.edu.kw. their common postal address is: gulf university for science and technology, misref, kuwait. references ia compliance summit (2009). building a fiduciary society. washington, dc march 13. agrawal, anup and charles r. knoeber, (1996). firm performance and mechanisms to control agency problems between s and shareholders. journal of financial and quantitative analysis, 31 (september): 377-397. angelidis, john p., and nabil a. ibrahim. (1993). social demand and corporate supply: a corporate social responsibility model, review of business, vol. 15, no. 1, summer/fall. ariff, m., and iqbal, m., (2011). foundation of islamic banking: theory, practice and education. edward elgar publishing uk and usa. arrow, kenneth, (1951). social choice and individual values, john wiley & son, new york. baumol, william j., (1959). business behavior, value and growth, macmillan. ------------------------, (1962). the theory of the expansion of the firm, the american economic review, december, vol. lii. -------------------------, (1970).the firm with inelastic demand, in induction, growth and trade: essays in honor of sir roy harrod, eds. (1970). ellis, w. a. and m.g. scott, and j. n. wolfe, oxford clarendon press.. beekun and badawi (2005). balancing ethical responsibility among multiple stakeholders: the islamic perspective. journal of business ethics, no. 60. bell, daniel, (1973). the coming of post-industrial society: a venture in social forecasting, basic book. mailto:palliam.s@gust.edu.kw mailto:caldwell.l@gust.edu.kw 26 bogle, j.c., (2005). the battle for the soul of capitalism. yale university press, 2005. british columbia court of appeal (1997). in joostens canada v gibsons studios (42 british columbia law review 3d 149). brown, murray and revankar, n. (1971). a generalized theory of the firm: an integration of the sales and profit maximization hypotheses, kyklos, 1971. bryan, s., hwang l.s. and lilien s., (2000). ceo stock-based compensation: an empirical analysis of incentive-intensity, relative mix, and economic determinants, journal of business 73,661-693. burke j., r. grances, and a. black, (1990). improving organizational productivity: add ethics public productivity and management review, vol. xiv, 1, winter. clark, john b., (1939) distribution of wealth, macmillan. collard, d. a., (1988). new moral sentiments? a review [well-being: its meaning, measurement and moral importance]. oxford economic papers, n. s., june 1988, vol. 40, no. 2. davis, j. h., schoorman, f. d., & donaldson, l. (1997). “toward a stewardship theory of management,” academy of management review, vol. 22. demott, d.a., (1988), an analysis of fiduciary obligation, duke law journal, no. 5. demsetz h. and kenneth lehn (1985). the structure of corporate ownership: causes and consequences, journal of political economy, 93: 1155-1177. eden, lorraine, (1998). taxing multinationals: transfer pricing and corporate income taxation in north america, university of toronto press, toronto. fama, e. (1980). agency problems and the theory of the firm. journal of political economy, 88: 288-307. fama, e. f. and m. c. jensen (1983). separation of ownership and control. journal of law and economics 26: 301-325. flannigan, r.,(1989), the fiduciary obligation. oxford journal of legal studies, vol 9, no. 3, autumn. folwer, d. k., (1978). transfer prices and profit maximization in multinational enterprise operations, journal of international business studies, winter. friedman, m., (1970). the social responsibility of business is to increase its profits, the new york times magazine, september 13, 1970. gapinski, j. h., (1988). the economic right triangle of nonprofit theater, social science quarterly, september, vol. 69, no. 3. garfield, charles a., (1999). do profits & social responsibility mix? executive excellence, vol. 9, no. 3, march. ghosh, dilip k., (1978). market mechanism and the maximization and the maximization principle, the indian journal of economics, july vol. lix, part i, no. 232. ginzberg eli and robert m. solow eds., (1974). the great society, basic books, inc. goodpaster, kenneth e. and johns b. matthews, jr., (1982). can a corporation have a conscience? harvard business review, january-february. grant, colin, (1991). friedman fallacies, journal of business ethics, 10: 907-914. international journal of banking and finance, vol. 9, iss. 4 [2012], art. 2 27 han, k.c., and david y. suk (1998). the effect of ownership structure on firm performance: additional evidence. review of financial economics, vol. 7, no. 2:143–155. hicks, john r.,(1954). the process of imperfect competition, the oxford economic papers. jensen, m., meckling, w., (1976). theory of the firm: management behavior, agency costs, and ownership structure. journal of financial economics, 3, pp. 305-360. jensen, m., murphy k., (1990). performance pay and top-management incentives, journal of political economy 98, 225-264. kamali (1989). source, nature and objectives of shariah, the islamic quarterly, vol 33, no. 4. khan, a. a., and thaut, l., (2007), ‘islamic perspective on fair trade’, accessed august, 2012. www.islamicrelief.com/indepth/downloads/islam_and_fairtrade.pdf. knight, frank h., (1988) risk, uncertainly and profit, oxford university press. marris, r., (1964). the economic theory of “managerial” capitalism, the free press of glencoe. mclachlin (2003). canada's supreme court in e.d.g. v hammer 2003 scc 52, s.c.r. 459. morck, randall, andrei shleifer and robert w. vishny (1988). management ownership and market valuation: an empirical analysis, journal of financial economics, 20: 293-315 o’neil, robert f. and darlene a. pienta (1994), economic criteria versus ethical criteria toward resolving a basic dilemma in business, journal of business ethics, vol. 13, 1, january. ouchi william g., (1981). theory z: how american business can meet the japanese challenge, addison-wesley. owen, crystal l. and robert f. scherer (1993). social responsibility and market share, review of business, vol. 15, 1, summer/fall. rawls, john, (1971). a theory of justice, harvard university press, cambridge, massachusetts. schumpeter, j. a., (1927). the theory of economic development, cambridge university press. svensson, l., -g., (1989). fairness, the veil of ignorance and social choice, social choice and welfare, january, vol. 6, no. 1. schokkaet, e. and overlaet, b., (1989). moral intuitions and economic models of distributive justice, social choice and welfare, january, vol. 6, no. 1. shleifer, a., and r. vishny (1986). large shareholders and corporate control. journal of political economy, vol. 94, no. 3(june): 461–488. snyder, j. m., and g. h. kramer, (1988). fairness, self-interest, and the politics of the progressive income tax, journal of public economics, july, vol., 36, no. 2. stone, christopher d., (1975). why shouldn’t corporations be socially responsible?, where the law ends, new york: harper and row. stone, h.f., (1934). address to the university of michigan school of law, june 15, reprinted in harvard law review. williamson, oliver, e., (1964). the economics of discretionary behavior: managerial objectives in a theory of the firm, prentice-hall, inc. http://www.islamicrelief.com/ ijbf7-marina.indd the international journal of banking and finance, vol. 7. number 1: 2010: 99-118 99 ijbf exchange rate volatility and purchasing power parity: does euro make any difference? meher manzur and felix chan curtin university of technology, australia _____________________________________________________ abstract this paper provides a new test of the purchasing power parity (ppp) and its relevance for the euro. principal component analysis (pca) is employed to construct a pooled measure of infl ation for 12 euro-currency countries. this measure is used to test the ppp for euro against three major currencies, namely, those of the japan, uk and usa. the test results are then used to measure the speed of adjustment of the deviations from parity using rolling and recursive regressions procedures. finally, the forecasting accuracy of the ppp-based euro exchange rates is compared with those given by the random walk model, and the synthetic euro series provided by the european central bank. in general, the results are supportive of ppp. keywords: purchasing power parity, principal component analysis, pooled infl ation, random walk. jel classifi cation: f31, g15. _____________________________________________________ 1. introduction exchange rates have been notoriously volatile since the switch to fl oating by the major currencies in the early 1970s (manzur, 2003). over the recent decades, many countries adopted a number of different approaches to manage exchange rates. the best example comes from europe: 15 european union member countries have now abandoned their national currencies and adopted a single currency (euro). a number of emerging economies are at odds with variants of managed exchange rate regimes. smaller countries like panama and ecuador have gone for dollarisation, and several other high infl ation-countries like zimbabwe, mexico and argentina are viewed as suspects to this extreme option. thus, exchange rate volatility dominates policy debate and multinational business fi nance. ht tp :// ijb f.u um .e du .m y 100 the international journal of banking and finance, vol. 7. number 1: 2010: 99-118 dornbusch’s (1976) ‘overshooting’ model was among the fi rst to provide useful insights to explain exchange rate volatility. this model views goods prices and wages as ‘sticky’ and slow to adjust, while asset market clears instantaneously. consequently, the domestic monetary policy becomes impotent, rendering exchange rates to absorb shocks. this idea makes sense in theory; unfortunately, formal empirical tests do not provide evidence to support this hypothesis (rogoff, 2002). the infl uential ‘productivity bias’ hypothesis, proposed by balassa (1964) and samuelson (1964), predicts that the currencies of more affl uent countries are systematically overvalued due to higher relative prices of non-traded goods in those countries. however, further research has raised issues with the validity of balassa-samuelson thesis (bahmani-oskooee and nasir, 2005). despite meese and rogoff’s (1983) infl uential result that the random walk model performs at least as well as structural and time series (both univariate and multivariate) models of exchange rates, the idea of ‘disconnect’ between exchange rates and the real economy has been subjected to a growing scepticism (devereux and engel, 2002). fortunately, the conclusions concerning the validity of purchasing power parity (ppp) are becoming increasingly reliable. this theory states that prices should be internationally arbitraged so that they are the same in all locations when expressed in a common currency (absolute ppp). a less strict version of this theory states that the change in the exchange rate is equal to the infl ation differential (relative ppp). historically, ppp has been the fulcrum in the study of exchange rate volatility. its relevance has increased signifi cantly, following the introduction of euro in europe. ppp has profound implications for several issues surrounding the euro. it was argued that the nominal exchange rate volatility has now disappeared for the euro zone, but the difference in the underlying infl ation across countries and other cross-country differences may imply signifi cant shocks to real exchange rates affecting trade fl ows, price level convergence, asset pricing and portfolio management (koedijk, ben and mathijs 2004). interest is also mounting in investigating if the economic integration through adoption of a single currency has accelerated the convergence towards ppp in the euro zone. this paper provides a new test of ppp and its relevance for the euro. we introduced the method of principal component analysis (pca) to construct a pooled measure of infl ation among the 12 euro countries. given the overall infl ation of the 12 euro countries is not observable, a pca on the infl ation of 12 euro countries is expected to provide a consistent measure of pooled infl ation for the euro zone.1 this measure of infl ation is used to test the ppp condition for 1a related approach is the factor analysis (fa), which decomposes the observed data into a set of unobserved latent factors with different factor loadings, that is, the observed data is expressed as a (linear) combination of unobserved latent factors, whereas pca computes the appropriate linear combination of the observed data to generate the unobserved variable. since we intend to generate the unobserved measure of pooled infl ation based on a linear combination of the observed infl ation rates from the 12 countries, pca is a more appropriate method. ht tp :// ijb f.u um .e du .m y exchange rate volatility and purchasing power parity: does euro make any difference?: 99-118 101 the euro against the currencies of usa, uk and japan. the test results are then used to measure the speed of adjustment of the deviations from ppp using rolling and recursive regressions procedures. finally, the accuracy of the pca-based ppp-euro exchange rates is compared with those given by simple forecasting rules such as no-change extrapolation, and the synthetic euro series provided by the european central bank. in general, the results are supportive of ppp. the paper makes several contributions. firstly, as just mentioned above, the test of ppp for euro has useful policy and business implications. secondly, most, if not all, ppp studies used quarterly or annual data. this paper used monthly observations, thereby adding a new dimension to the literature. thirdly, our approach is simple, but innovative. while the method of pca is widely used for constructing unobserved latent factors in many areas, especially in medical research, marketing, psychology and panel data econometrics, its application is relatively new in macroeconomics and international fi nance. similarly, measuring the speed of convergence toward ppp using rolling and recursive regressions procedures is insightful. finally, our approach enables us to provide a test for the predictive power of ppp. the results indicated that ppp tends to outperform the random walk model – a result that will probably be infl uential in understanding modern exchange rate dynamics. the paper is organised as follows. a brief digression on the speed of adjustment of ppp deviations is given in the next section. section 3 introduces the concept of pca and the construction of the pooled infl ation measure. section 4 describes the data and preliminary analysis, followed by the empirical results in sections 5 and 6. the last section concludes the paper. 2. the speed of adjustment for ppp: a digression ppp is perhaps the oldest and most controversial theory of exchange rate determination. it is also one of the most-heavily researched topics in international fi nance (lan, 2003). manzur (1990) was among the fi rst to provide evidence to indicate that ppp is a long-run phenomenon. he was also among the fi rst to provide a measure of the long run: his results identifi ed fi ve years as being a broad measure of the length of the long run in so far as ppp is concerned. there is now a consensus in the literature that ppp is a long-run phenomenon, but the issue of speed of convergence toward long-run ppp is puzzling. several approaches have been adopted to explore the notion of ppp deviations and their speed of adjustment to the long run. a measure of persistence is what is now known in the literature as the half-life of ppp deviations, based on a concept originally from physics. ht tp :// ijb f.u um .e du .m y 102 the international journal of banking and finance, vol. 7. number 1: 2010: 99-118 table 1 estimates of ppp half-livese author(s) half-life (years) data frankel (1990) abuaf and jorion (1990) manzur (1990,1993) fung and lo (1992) wei and parsley (1995) frankel and rose (1996) cumby (1996) (i) (ii) 4.6 3.3 5 6.5 4.25 4.75 4 1 dollar-pound ten industrial countries seven industrial countries six industrial countries european monetary system non-european monetary system 150 countries big mac currencies lothian and taylor (1996) (i) (ii) 2.8 5.9 franc-pound dollar-pound papell (1997) (i) (ii) 1.9 2.8 ec (the european community) ems higgins and zakrajsek (1999) (i) (ii) (iii) (iv) 5 3 2.5 11.5 europe, cpi europe, wpi oecd, wpi open economies, cpi cheung and lai (2000) (i) (ii) (2-5) (under 3) industrial countries developing countries median mean standard error of mean 4 4.1 2.3 notes: a: where a study contains more than one estimate of half-life, we use (i), (ii), (iii) and so on to distinguish different estimates, with additional information provided in the fi nal column of the corresponding row. b: where a study does not report the half-life directly, we compute it from the speed-of adjustment estimate (on an annual basis) using c: in those cases where the underlying data are not annual and the parameter estimated is, we computed the speed of adjustment per annum as , where n is the number of periods per year. d: cheung and lai (2000) report the range of half-life estimates for two groups of countries. to compute the mean and its standard error in the last two rows of this table, we used 3.5 and 2 years as the respective point estimates. e: reproduced from lan (2003, p.25) with permission from the author and publisher. 1 ln 2 / (ln ) ln 2 / [ln )]. k jj h or h ht tp :// ijb f.u um .e du .m y exchange rate volatility and purchasing power parity: does euro make any difference?: 99-118 103 according to this measure, the half-life is the time taken by a given amount of ppp deviations to decay to half its magnitude (froot and rogoff, 1995, among others, for more discussion on ppp half-lives). a standard approach is to adopt a measure of the half-life for a general ar(1) process that allows asymptotic approximations around a unit root (abuaf and jorion, 1990, among others). andrews (1993), and andrews and chen (1994) discussed half-life measures for general ar(p) processes, but do not provide a precise measure of the half-life for such processes. another set of studies addresses this issue by constructing confi dence intervals through estimating impulse response functions with various methods (cheung and lai, 2000). table 1 (reproduced from lan, 2003, p.25) gives a summary of estimates of ppp half-lives from a set of selected studies in the literature. as can be seen, the observed half-life estimates are fairly diverse, and they range from as low as 1 year (crumby, 1996) to 6.5 years (fung and lo, 1992). in addition to crumby (1996), four other studies (namely, abuaf and jorion, 1990; lothian and taylor, 1996; papell, 1997; and higgins and zakrajsek, 1999) report a half-life estimate which is 3.3 years or below. the median and mean half-lives are 4 and 4.1 years, respectively. these average fi gures indicate that it takes about 4 to 5 years for the ppp deviations to disappear, which is broadly consistent with the survey results reported by froot and rogoff (1995). however, the issue is still unresolved and far from fully understood. as rogoff (1996) observes, it is puzzling to reconcile between the excessively high volatility of real exchange rates in the short run on the one hand, and the relatively longer half-life of ppp deviations (that is, 4 to 5 years) on the other. that is, the point estimates imply that ppp deviations are more persistent than what the conventional wisdom (sticky-price argument) would suggest (rossi, 2005). consequently, more work is needed to address this issue. the single currency in the euro zone provides useful test grounds for this and related issues to explore. 3. testing ppp with principal component we start with some notations before introducing the new test for ppp using principal component analysis (pca). let be a m x n matrix then a .j denotes the jth column of a, a j . denotes the jth row of a, and a' denotes the transpose of a. let x and y be two random variables then denotes the mean of x, var (x) denotes the variance of x and cov (x,y) denotes the covariance between x and y. if x is a m x 1 random vector then cov (x) denotes the m x m variance-covariance matrix of x . now, consider the relative ppp condition, (1)1 log d ft t t t s s ht tp :// ijb f.u um .e du .m y 104 the international journal of banking and finance, vol. 7. number 1: 2010: 99-118 where s t denotes the spot exchange rate in direct quote at time t, π t d and π t f are the infl ation rate for domestic and foreign country at time t, respectively. the fi rst step of testing ppp with euro is to construct π t d as a function of the infl ation from the 12 euro countries. an obvious approach is to construct a set of weights such that: (2) where w j denotes the weight given to country j, and π jt denotes the infl ation rate for country j at time t for j = 1,...,12. a typical approach to determine the appropriate weights is based on the size of the economy for each country. for example, manzur (1990), among others, used the gross domestic products as weights for the sample countries. however, this approach can be sensitive to the variables or methods used for determining the economy size, and consequently, bias the results. to avoid this problem, this paper used pca to determine the weights. an approach based on pca is more appropriate for this purpose for at least two reasons: (i) the dependent variable, π t d , is unobservable, and (ii) pca provides a set of robust weights as it is a function of infl ation rates rather than the subjective measure of the economy size.2 as the application of pca is relatively new to macroeconomics and international fi nance, we provide a brief outline of this procedure below. let x = (x 1 ,x 2 ,...,x m )' and y = (y 1 ,..., y m )' be two random vectors with m variates, then the idea of principal component is to fi nd a set of linear combinations, so that: y = wx where is a m x m matrix and: note that: and therefore: cov(y) = wcov(x)w' since cov(y) is a diagonal matrix, the weight matrix, w, is the matrix that diagonalises cov(x). moreover, cov(x) is a symmetric positive semi-defi nite matrix by defi nition, therefore w is the matrix consists of the eigenvectors of cov(x). specifi cally, w. j is jth eigenvector of cov(x) that corresponds to the jth eigenvalue of cov(x) with w'=w-1. this construction implies that the eigenvector associated with the largest eigenvalue is the linear combination that produces the 2for example, germany is the largest economy among the 12 euro-countries, but it is the most low-infl ation country in the group. if the size of the economy is used as weights to calculate the pooled infl ation for these 12 countries, then germany would attract the largest weight. this would render the pooled measure to be biased and inconsistent. 12 1 d t j jt j w ( )y y w x x ( , ) 0i jcov y y i j , ( ) ( )i jvar y var y i j ht tp :// ijb f.u um .e du .m y exchange rate volatility and purchasing power parity: does euro make any difference?: 99-118 105 0 1 : 0, 1 : 0 or 1 h h 3due to issues with missing data, this is the most recent and complete time series data we could obtain at the time of writing this paper. 4these estimates are likely to be unaffected by the fact that greece was given accession to adopt euro from 1 january, 2001. maximum variance given the random vector x, and the eigenvector associated with the second largest eigenvalue is the linear combination that produces the maximum remaining variance given the random vector x. y i is said to be the ith principal component of x. in the case of euro countries, let π t = (π 1t ,...,π 12t ) be a vector containing the infl ation rates for the 12 countries, then the fi rst principal component of π t , π t pc1, can be used as a measure of pooled infl ation, that is, a proxy for π t d. given the pooled infl ation measure for the 12 countries, a simple regression analysis based on equation (1) can now be applied to test ppp as follows: (3) testing equation (1) in the form of equation (3) is equivalent to testing: 4. data and empirical analysis the data used in this analysis are monthly observations, obtained through the datastream. these include the consumer price index (cpi) for the 12 euro countries, japan, united kingdom, and the usa, as well as the euro/us exchange rate, euro/yen exchange rate and the euro/gbp exchange rate. the sample began in december 1998 and ended in july 2007,3 with a total of 104 observations. the infl ation, π it , for the usa and the 12 euro countries were calculated as the log-difference of the corresponding cpi. all data are contained in a separate appendix, available on request. unless otherwise stated, the analysis in this paper was conducted using rv2.4.1. the pca on the infl ation rates from the 12 euro countries produced 12 eigenvalues. these are 1.714, 0.684, 0.279, 0.203, 0.144, 0.075, 0.044, 0.040, 0.037, 0.029, 0.015 and 0.013. table 2 contains the factor loadings (weights) for the fi rst principal components. as can be seen in this table, the total loading is being standardised to unity, with greece, ireland, luxembourg and spain accounting for 60% of the total loading.4 the pooled infl ation rate for the 12 countries was now calculated by using equation (2) with setting equal to the set of loadings as given in table 2. 1 log ( )d ft t t t t s s ht tp :// ijb f.u um .e du .m y 106 the international journal of banking and finance, vol. 7. number 1: 2010: 99-118 table 2 factor loadings from the first principal component figure 1 contains the us/euro, gbp/euro and yen/euro exchange rates (domestic currency cost of one euro) and their corresponding growth rates for the sample period. as shown in figure 1, the us dollar started off strong against the euro before 2002 and then it experienced a series of continuous downturns for the remainder of the sample period. this is also supported by the growth rate where the size and number of positive growth exceeded that of the negative growth. interestingly, the dynamics of pound/euro and yen/euro resemble a very similar story to that of us/euro. these currencies started off strong against the euro but became weaker after 2002. countries austria belgium finland france germany greece ireland italy luxembourg netherlands portugal spain loadings 0.043 0.029 0.073 0.062 0.019 0.273 0.110 0.026 0.110 0.084 0.065 0.106 figure 1. exchange rates and their growth rates (december 1998-july 2007). ht tp :// ijb f.u um .e du .m y exchange rate volatility and purchasing power parity: does euro make any difference?: 99-118 107 figure 2. infl ation rates: us, uk, japan and pooled euro countries (december 1998-jul 2007). figure 3. infl ation differentials between euro pooled infl ation and the three countries. figure 2 contains the plots for the usa, uk and japan infl ation rates. this fi gure also includes a plot for the ‘pooled’ infl ation (fi rst principal component) for the 12 euro countries. as can be seen, the usa and uk infl ation rates were mostly positive in the sample period and there does not appear to be any obvious cyclical movement in the infl ation rates in all four cases. a similar picture obtained in figure 3 which plots the differentials between euro ‘pooled’ infl ation and that in the other three countries. (the infl ation rates for the 12 euro countries are individually given in figure a.1 in the appendix. interestingly, this fi gure indicates that greece, portugal, netherlands and luxembourg have obvious cyclical movement in their infl ation rates, whereas the cyclical patterns are not quite as obvious in the other countries. also note that the factor loadings as reported in table 2 are consistent with this phenomenon in figure a.1). ht tp :// ijb f.u um .e du .m y 108 the international journal of banking and finance, vol. 7. number 1: 2010: 99-118 table 3 unit root tests for exchange rates, december 1998-jully 2007 adf test statistics asymptotic p-values usd/euro (level) -3.003 0.137 usd/euro (growth rate) -9.919 1.165e-12 gbp/euro (level) -2.420 0.369 gbp/ euro (growth rate) -6.363 1.407e-7 yen/euro (level) -3.089 0.109 yen/euro (growth rate) -10.161 3.996e-14 table 4 unit root tests for the regressors adf test statistics asymptotic p-values (π t pc π t us) -9.649 4.391e-18 (π t pc π t japan) -7.963 6.409e-13 (π t pc π t uk) -6.029 1.022e-7 next, we employed standard stationarity tests on the data. as can be seen in table 3, the augmented dicky-fuller (adf) test suggests that the three exchange rates are non-stationary (in levels), while their growth rates are stationary, as expected. we also employ adf tests on the regressors (infl ation differentials) and residuals of equation (3), and the results are reported in table 4 for the regressors and table 5 for the residuals. the results in these two tables indicate that the adf tests reject the null of non-stationarity in all cases. note that employing adf on the residuals in equation (3) is equivalent to testing for the presence of unit roots in real exchange rates (effi cient market approach to testing ppp). consequently, the results in table 5 (rejection of null of nonstationarity) are encouraging for ppp. table 5 unit root tests for the residuals residuals adf test statistics asymptotic p-values us -9.764 1.773e-13 japan -10.121 5.038e-14 uk -6.160 4.883e-8 ht tp :// ijb f.u um .e du .m y exchange rate volatility and purchasing power parity: does euro make any difference?: 99-118 109 as a further check, we estimated equation (3) using ordinary least squares, and the results are presented in table 6 for the three exchange rates. as can be seen, the results have passed the most common diagnostic tests for all three currencies. these tests include lm tests for serial correlation, heteroscedasticity and functional form, as reported in table 6. based on these diagnostics, the estimates of the parameters and their standard errors are now consistent, and they can be used for our hypothesis testing. as shown in table 6, the estimates of are signifi cant at 10% level for us/euro and gbp/euro, but not signifi cant for yen/euro. the coeffi cient estimates for β are not signifi cantly different from unity. unlike the unit-root test results (in table 5), the regression results provide a somewhat weak support for ppp in the case of us/euro and gbp/euro exchange rates. however, it is important to note that this seemingly weak support for ppp is fairly encouraging. the reasons for this are twofold. firstly the data are of monthly frequency which do not favour ppp (as it is more of a long-run phenomenon, discussed in section 2). secondly, the sample includes introductory periods of the euro, which are likely to be contaminated by exogenous elements. nevertheless, the results indicate that (relative) ppp did not hold all that tightly in the beginning of the euro introduction, but it seems reasonable to postulate that ppp might hold in the long run. in what follows, we turn to investigate this issue further. for the long-run analysis, we employ rolling and recursive estimation procedures. using rolling and recursive windows, estimates were obtained for the three currencies. the dynamic paths of these estimates would provide insight into the evolution of the β estimates and their associated t-ratios (in table 6) for all three cases. the details of the estimation methods are outlined below: step 1: select the sub-sample by choosing the fi rst k observations. step 2: construct the pooled measure of infl ation using pca as mentioned above with the k observations step 3: estimate equation (3) and save the coeffi cient estimates using the subsample. step 4: in the case of recursive windows, repeat step 1 to step 4 by setting k = k + 1; and in the case of rolling windows, repeat step 2 to step 4 by deleting the fi rst observation of the current sub-sample and setting k = k + 1. ht tp :// ijb f.u um .e du .m y 110 the international journal of banking and finance, vol. 7. number 1: 2010: 99-118 table 6 regression results for relative ppp usd vs. euro estimate standard error t-statistics p-values  0.1567 0.284 0.552 0.582 β 1.1001 0.657 1.676 0.097 r2 0.0271 serial correlation 0.111[0.739] f-statistics 2.808[0.097] heteroscedasticity 1.272[0.259] durbin-watson 2.021 functional form 0.808[0.369] yen vs. euro  0.176 0.350 0.503 0.616 β -0.04615 0.770 -0.060 0.352 r2 3.557e-05 serial correlation 0.041[0.840] f-statistics 0.004[0.952] heteroscedasticity 0.017[0.895] durbin-watson 1.999 functional form 0.041[0.840] gbp vs. euro  0.024 0.179 0.132 0.896 β 1.063 0.549 1.937 0.056 r2 0.0358 serial correlation 0.608[0.436] f-statistics 3.75[0.056] heteroscedasticity 0.119[0.731] durbin-watson 1.980 functional form 3.841[0.309] note: the numbers in the square brackets are the corresponding p-values. figures 4 and 5 contain the dynamic paths of the β estimates and their associated t-ratios using rolling windows, and the dynamic paths of the β estimates and their associated t-ratio using recursive windows, for all three exchange rates, respectively. rolling estimates rolling t-ratios figure 4. dynamic paths of and their t-ratios using rolling windows. ht tp :// ijb f.u um .e du .m y exchange rate volatility and purchasing power parity: does euro make any difference?: 99-118 111 both rolling and recursive windows start from the fi rst observation of our sample. as can be seen, for both usd/euro and gbp/euro, the dynamic paths of the β estimates exhibited a downward trend toward 1. this tends to be consistent with our results reported above, that is, relative ppp did not hold in the early stage of the euro introduction, but it seems to be valid over time for us and uk. however, the dynamic path of the β estimates is very unique in the case of yen/euro. as shown in figure 4, the rolling estimates of β suggests that relative ppp did not hold for much of the sample period, which is also supported by the dynamic path of the recursive estimates as shown in figure 5. however, the sharp and dramatic increase in the β estimates toward the end of the sample seems to suggest the tendency toward relative ppp. these results tend to suggest that (i) relative ppp did not hold in the initial period of euro introduction for us, uk and japan; and (ii) there is a tendency towards relative ppp for the three currencies over time. recursive estimates recursive t-ratios based on the results above, it is now straightforward to estimate the rate of convergence in fi gures. this is achieved by analysing the dynamic of the recursive β estimates as provided above. a possible model to consider is: (5) ˆfigure 5. dynamic paths of and their t-ratios using recursive windows. 0 ˆ exp( ) ~ (0, )t t t ua t u u iid ht tp :// ijb f.u um .e du .m y 112 the international journal of banking and finance, vol. 7. number 1: 2010: 99-118 t̂ ˆ 1 ˆ t t d dt ˆ ˆ( )t 0 ˆlog( ) log( )t ta t u where denotes the β estimate at the tth recursive window and thus  represent the rate of decay which can be interpreted as the rate of convergence in this case. to see this, differentiate equation (5) with respect to t and rearranging gives: which implies that  measures the percentage changes in due to a small change in t. the parameter in equation (5) can be estimated using ordinary least squares by taking the logarithmic transformation, which yields: (6) table 7 contains the  estimates for us/euro, gbp/euro and yen/euro. as can be seen, the estimates are highly signifi cant in all three cases. surprisingly, although the β estimates exhibited seemingly different dynamic over time, the rate of convergence is very similar in absolute terms. this result is signifi cant, as it indicates that the tendency towards ppp underlying each currency is very similar in the long run, even though the ppp deviations are persistent in the short run. table 7 estimates of the rate of convergence to ppp us/euro gbp/euro yen/euro -0.013 -0.012 0.015 -17.12 -19.57 4.452 the necessary condition for ppp to hold is that β = 1, therefore given the estimates of log (a 0 ) and , it is possible to calculate the expected time for each currency to converge toward ppp. this can be done by substituting in equation (6) and solving for t. table 8 contains the results for the three exchange rates. note that as the observations are monthly, is measured in terms of months and hence, dividing by 12 will give the estimated year of convergence toward longrun ppp for each currency. table 8 estimates of time for convergence to ppp us/euro gbp/euro yen/euro t (in months) 59.627 54.247 38.770 year 4.969 4.687 3.230 t̂ ht tp :// ijb f.u um .e du .m y exchange rate volatility and purchasing power parity: does euro make any difference?: 99-118 113 as shown in table 8, the length of the long run for ppp for both us/ euro and gbp/euro is around fi ve years, consistent with several other results discussed in section 2. interestingly, while the earlier results were not supportive of ppp for yen/euro (as shown in table 6), the adjustment towards ppp for this exchange rate is faster, just a little over three years. it is, however, to be noted that that yen/euro rates had the highest volatility in terms of the dynamic of in figures 4 and 5 (also see figure 1). 5. does ppp outperform the random walk model? a major business implication of ppp is that it provides a simple prediction model for exchange rates. if an exchange rate (in level) is overor undervalued relative to that implied by ppp, it is straightforward to form an expectation about its future movements. how accurate are the predictions based on ppp? in what follows, we provide a brief analysis of this issue. based on ppp, equation (3) can be used to forecast exchange rates given previous exchange rates and the rates of infl ation. the pooled infl ation rate for the 12 euro countries were calculated based on the fi rst principle component of their infl ation rates using the fi rst 50 observations and the remaining 54 observations will be used to compare the out-of-sample forecasts. the performance of this model was compared with the unit root model (no-change extrapolation) and the synthetic euro series provided by european central bank against the currencies of the usa, uk and japan. the three conventional measures of forecast errors, namely, mean square errors (mse), mean absolute errors (mae) and mean absolute percentage errors (mape), are used for comparison purposes. these measures are defi ned as follows: mean squared error (mse) = mean absolute error (mae) = mean absolute percentage error (mape) = where y t and denote the observed exchange rate and the predicted exchange rate at time t, respectively and h is the forecast horizon which equals to 54 in this case. the mse, mae and mape for the three currencies under the three models are presented in table 9. as can be seen, the forecast performance of the ppp model is very similar to those of the other two models. for us/euro, ppp clearly outperforms the synthetic euros of ecb, and performs at least as well as the random walk model. the results for yen/euro and gbp/euro are marginally mixed, but qualitatively (and statistically) very similar. overall, the result is highly encouraging for ppp. 21 1 ˆ h t t t h y y 1 1 ˆ h t t t h y y 1 1 ˆ / h t t t t h y y yht tp :// ijb f.u um .e du .m y 114 the international journal of banking and finance, vol. 7. number 1: 2010: 99-118 table 9 forecast performance: ppp, random walk and synthetic euro us/euro unit root ppp euro using pca synthetic euro mse 0.001 0.001 0.007 mae 0.025 0.026 0.070 mape 0.023 0.023 0.063 yen/euro mse 6.587e-8 6.746e-8 1.035e-9 mae 1.768e-4 1.844e-4 2.283e-5 mape 0.021 0.022 0.003 gbp/euro mse 8.187e-4 7.899e-4 9.356e-9 mae 0.022 0.022 7.355e-5 mape 0.014 0.014 4.738e-5 6. conclusions this paper has provided a new test of ppp and its relevance for the euro. we introduce the method of principal component analysis (pca) to construct a ‘pooled’ measure of infl ation among the 12 euro countries. as the overall infl ation of the 12 euro countries is not observable, we maintain that a pca on the infl ation of 12 euro countries provides a consistent measure of ‘pooled infl ation’ for the euro zone. the results indicate that the high-infl ation countries in the euro zone (such as greece and ireland) receive larger weights in the pool. this measure of infl ation is used to test the ppp condition for the euro against the currencies of usa, uk and japan. the results provide a somewhat weak support for ppp in the case of us/euro and gbp/euro exchange rates, and a rejection of ppp for yen/euro. however, this seemingly weak support for ppp is fairly encouraging, for at least two reasons. firstly, the data are of monthly frequency which do not favour ppp (as it is more of a long-run phenomenon, discussed in section 2). secondly, the sample includes introductory periods of the euro, which are likely to be contaminated by exogenous elements. the test results were then used to measure the speed of adjustment of the deviations from ppp using rolling and recursive regressions procedures. the results tend to suggest that (i) relative ppp did not hold in the initial (introductory) periods of euro for the three currencies under this study; and (ii) there is a tendency towards relative ppp for all three currencies over time. the ppp half-lives are found to be about fi ve years for us/euro and gbp/euro exchange rates, but just a little over three years for yen/euro. the accuracy of the ppp-based euro exchange rates is compared with those given by simple forecasting rules such as no-change extrapolation, and the synthetic euro series provided by the european central bank. the results indicated that the forecast ht tp :// ijb f.u um .e du .m y exchange rate volatility and purchasing power parity: does euro make any difference?: 99-118 115 performance of the ppp model is very similar to those of the other two models. for us/euro, ppp clearly outperforms the synthetic euros of ecb, and performs at least as well as the random walk model. the results for yen/euro and gbp/ euro were slightly mixed, but qualitatively (and statistically) very similar. this result is highly encouraging for ppp. the empirical results of this paper are expected to provoke further work on this issue at least in two directions. the fi rst is the issue of appropriate statistical procedures to employ for estimating the ppp half-lives (rossi, 2005). however, the half-life estimate for yen/euro (compared to those for us/euro and gbp/ euro) in this paper indicates that appropriate estimation procedures may be less of an issue in this case. consequently, further research will concentrate more towards the second direction: the role of economic fundamentals in explaining the persistence of ppp deviations. note that in estimating the ‘pooled’ infl ation in this paper, the differences in infl ation in 12 european countries (having one currency) are allowed for, but we have not explained the differences. thus, it may be possible in future research to examine these cross-country infl ation differences within the euro-zone more closely in explaining ppp deviations. author information: submitting author, assoc professor meher manzur, curting business school, gpo box u1987, perth, wa 6845, australia. e-mail; meher.manzur@cbs.curtin.edu.au. the authors are thankful to ariful hoque and muammer wali for excellent research assistance. financial support from curtin business school is gratefully acknowledged. appendix (continued) figure a1. infl ation rates: 12 euro countries (december 1998-july 2007). ht tp :// ijb f.u um .e du .m y 116 the international journal of banking and finance, vol. 7. number 1: 2010: 99-118 appendix figure a1. infl ation rates: 12 euro countries (december 1998-july 2007). references abuaf, n. and jorion, p. (1990). purchasing power parity in the long run. journal of finance, 45,157-174. andrews, d. (1993). exactly median-unbiased estimation of autoregressive/unit root models. econometrica, 61, 139-166. ht tp :// ijb f.u um .e du .m y exchange rate volatility and purchasing power parity: does euro make any difference?: 99-118 117 andrews, d. and chen, h-y. (1994). approximately median-unbiased estimation of autoregressive models. journal of business and economic statistics, 12, 187-204. bahmani-oskooee, m., and nasir abm. (2005). productivity bias hypothesis and the purchasing power parity: a review article. journal of economic surveys, 19, 671-696. balassa, b. (1964). the purchasing power parity doctrine: a reappraisal. journal of political economy, 72, 584-596. cheung, y-k. and lai, k. (2000). on purchasing power parity puzzle. journal of international economics, 52, 321-330. crumby, r. (1996). forecasting exchange rates and relative prices with the hamburger standard: is what you want what you get with mcparity? nber research paper 5675. massachusetts: national bureau of economic research. fung, h. and lo, w. (1992). deviations from purchasing power parity. the financial review, 27, 553-570. devereux, m., and engel, c. (2002). exchange rate pass-through, exchange rate volatility, and exchange rate disconnect. journal of monetary economics, 49, 913-940. dornbusch, r. (1976). expectations and exchange rate dynamics. journal of political economy, 84, 1161-1176. frankel, j. and rose, a. (1996). a panel project on purchasing power parity: mean reversion and between countries. journal of international economics, 40, 209-224. froot, k. and rogoff, k. (1995). perspectives on ppp and long run real exchange rates. in g. grossman and k. rogoff (eds), handbook of international economics (pp. 1647-1688). amsterdam: north-holland press. higgins, m. and zakrajsek, (1999). purchasing power parity: three stakes through the heart of the unit root null. staff paper no. 80. federal reserve bank of new york. koedijk, k., ben, t. and mathijs, a. (2004). purchasing power parity and the euro area. journal of international money and finance, 23, 1081-1107. lothian, j. and taylor, m. (1996). rela exchange rate behaviour: the recent fl oat from the perspective of the past two centuries. journal of political economy, 104, 488-510. lan, y. (2003). the explosion of purchasing power parity. in manzur m. (ed.), exchange rates interest rates and commodity prices (2nd ed.)(pp. 1-8). cheltenham, edward elgar. manzur, m. (1990). an international comparison of prices and excahange rates: a new test of purchasing power parity. journal of international money and finance, 9, 75-91. manzur, m. (2003). exchange rates interest rates and commodity prices: an introduction. in manzur m. (ed.), exchange rates interest rates and commodity prices (pp. 1-8). edward elgar, ht tp :// ijb f.u um .e du .m y french banks amid the global financial crisis the international journal of banking and finance, vol. 8, number 1, march 2011: 1-19 1 ijbf french banks amid the global financial crisis yingbin xiao international monetary fund usa ________________________________________________________________ abstract this paper runs the gamut of qualitative and quantitative analyses to examine the performance of french banks during 2006-2008 and the fi nancial support measures taken by the french government. french banks were not immune but proved relatively resilient to the global fi nancial crisis refl ecting their business and supervisory features. an event study of the impact of government measures on cds, debt, and equity markets points to the reduction of credit risk and fi nancing cost as well as the redistribution of resources. with the crisis still unfolding, uncertainties remain and challenges lie ahead, calling for continued vigil ance and enhanced risk management. keywords: french banks, financial crisis, recapitalization, debt guarantee, government support, refi nancing jel classifi cation: g01, g12, g15, g21, g28 ________________________________________________________________ 1. introduction the ongoing global fi nancial crisis has posed great challenges to fi nancial systems and governments to manage the economies around the globe. that is also the case in france. the unprecedented nature of the crisis and government bailouts highlight the importance of having a sound system and an appropriate policy response. it also points to the value of having a good grasp of the fi nancial sector performance and government support measures of systemic importance. this paper attempts to analyze the performance of french banks and the fi nancial support measures taken by the french government. france has a large and sophisticated fi nancial system, which accounts for ten percent of the global banking system and fi ve percent of the global capital markets. besides, it hosts the second largest mutual fund industry. the impact of the crisis on the french fi nancial system, especially on the banks, and the policies adopted after the crisis matter for the global economy and fi nancial stability. the rest of the paper is organized as follows. section 2 presents an overview of the french banking and supervision structure. section 3 conducts international comparative analyses of profi tability, asset quality, capital 1 xiao: french banks published by epublications@bond, 2011 2 the international journal of banking and finance, vol. 8. number 1, march 2011: 1-19 adequacy, leverage, quality of capital, funding profi le, and liquidity of banks. section 4 analyses business lines, potential spillovers, writedowns and losses, new capital raised, and market perceptions of risk. section 5 analyses the government support plan of banks and carries out an event study of the market impact of the measures. section 6 concludes. 2. banking and supervisory structure the french banking system features a small number of large universal banks spanning a wide range of business functions. the banks are largely organized along commercial, mutual, and cooperative lines. private commercial banks dominate the system with an asset value of about fi ve times that of mutual and cooperative banks. the close ties among banks, life insurance companies, and asset management companies, sometimes underpinned by cross-shareholding, tend to blur divisions among different types of institutions. the french banking sector is an integral part of the european and international banking system. home banks’ external claims well exceed host banks’ external claims. foreign banks have made few inroads into the french mainstream banking, the exception being the hsbc. although domestic banks dominate retail banking, the foreign banks are free to compete and have gained a strong market presence in wholesale banking and securities trading. however, cross border mergers and acquisitions may blur the line between domestic and foreign banks as leading french banks become partly foreign owned, as in the case of the belgian government’s ownership of bnp paribas after its recent acquisition of fortis bank. french banking is highly consolidated. the nine largest banks account for 75 percent of the total banking assets. the top fi ve banks represent half of the total deposits and more than a half of the total lending. banking concentration has largely stabilized over the last three years, but will rise with the merger between groupe caisse d’épargne (gce) and groupe banque populaire (gbp) this year. the fi nancial supervision structure is based on a functional approach with some twin peaks elements. the oversight of fi nancial services was reformed in 2003 with a view to enhancing regulatory effi ciency. hence, the fi nancial supervisory framework was reorganized and substantially simplifi ed. the monetary and financial code does not distinguish between commercial banks and investment banks. the prudential supervision of both banks and investment fi rms falls under the commission bancaire (cb), which ensures consistent coverage of all credit institutions. the proximity between the banque de france (bdf) and cb facilitates timely information-sharing as the governor of the bdf serves as the chairman of the cb, and bdf provides budget, human resources, and other support to the cb. the board of fi nancial sector authorities offers a domestic coordination framework. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 1 french banks amid the global financial crisis: 1-19 3 3. international comparative analyses the analyses employ selected soundness indicators of french banks and comparisons during 2006-08. the study period covers the pre-crisis period, the beginning, and the middle of the global fi nancial crisis. given the fi nancial integration and the global nature of the french banking business, a bank-based, rather than a country-based, comparison group is constructed. it consists of 48 large banks in advanced europe. to ensure comparability, only banks adopting the international financial reporting system (ifrs) are included. bank data are from bankscope and thomson financial. composite indicators weighted by size for the comparison group and nine major french banks respectively are constructed to facilitate aggregate comparisons. selected soundness indicators focus on bank profi tability, asset quality, capital adequacy, leverage, capital quality, funding profi le, and liquidity. profi tability is measured by operating income on average assets (roaa) and return on average equity (roae), gauging both the operating performance and the bottom line. asset quality is measured by the non-performing loans (npl) ratio and coverage ratio, gauging npl and the suffi ciency of provisions, respectively. capital adequacy is measured by the tier i ratio, a widely used regulatory indicator. leverage is defi ned as assets over shareholders’ equity. capital quality is measured by the share of tier ii capital in total capital and the core tier 1 ratio in view of investors’ recent focus. the banks’ funding profi le is measured by the share of wholesale funding in total funding and the share of short-term wholesale funding in short-term funding, gauging the banks’ dependence on capital market funding. liquidity is measured by the depositto-loan ratio and liquid ratio defi ned by liquid assets over liquid liabilities, examining liquidity from an asset-liability management (alm) perspective. 3.1 profi tability french banks were less profi table than their european peers before the crisis, but were affected less hard by the crisis. the operating roaa and roae were below those of the peers in 2006, refl ecting narrower interest margins and less exposure to profi table, but risky non-traditional banking activities. with intense competition among banks, the net interest margin of french banks eroded to a level below their peers. both french banks and their european peers registered plummeting operating performance and net income in 2008, with the profi tability of european banks turning negative while that of french banks overall remained slightly positive. this is largely due to the fact that the majority of french banks still earned modest profi ts in 2008: see figure 1. 3.2 asset quality figure 2 provides information on this aspect. the quality of french banks’ loan portfolios was almost on par with that of their european peers, but the fall of provisions was less pronounced during the crisis. both french banks and their xiao: french banks 4 the international journal of banking and finance, vol. 8. number 1, march 2011: 1-19 european peers saw a declining npl ratio in 2007, but a reversal was evident in 2008, refl ecting the turn of the economic and credit cycle, as well as the ensuing rise in defaults. the coverage ratio of both the french banks and the european peers dropped appreciably in 2008, suggesting less provision to cover more problem loans. french banks had less coverage than their european peers initially, but with a less marked deterioration ended up with a higher coverage ratio in 2007-08. fig ure 1: operating performance of french vs european banks figure 2: asset quality of french vs european banks 3.3 capital adequacy as is evident from the information in figure 3, regulatory capital adequacy of french banks was stronger than that of their european peers before the crisis, but gradually lost its relative strength: see figure 3. in 2006, the tier i ratio of french operating roaa -0.20 -0.10 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 2006 2007 2008 france developed europe nonperforming loan ratio 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 2006 2007 2008 france developed europe roae -5 0 5 10 15 20 2006 2007 2008 france developed europe coverage ratio 55 60 65 70 75 80 2006 2007 2008 france developed europe international journal of banking and finance, vol. 8, iss. 1 [2011], art. 1 french banks amid the global financial crisis: 1-19 5 banks was about 100 basis points above that of their peers. the outbreak of the crisis in 2007 eroded the buffers for both the french banks and their european peers. with massive government recapitalization efforts in some countries in 2008, the edge of the capital positions of the french banks shrank and converged with the others. figure 3: capital adequacy of french vs european banks 3.4 leverage the level and evolution of the leverage of french banks mirrored those of their european peers: see figure 4. both groups showed no signs of deleveraging from their pre-crisis levels, an interesting phenomenon that contradicts the conventional perception that banks would be forced or inclined to reduce leverage because of the crisis. the annual report of 2008 noted that the banks continued to expand their intermediation business, although at a slower pace. the report examined leverage as measured by the ratio of loan volume over equity and loan volume weighted by asset quality over equity. it concluded that tier 1 ratio 2006 0 2 4 6 8 10 12 14 16 5.5 6.5 7.5 8.5 9.5 10.5 11.5 12.5 13.5 14.5 15.5 16.5 all banks distribution french banks distribution tier 1 ratio 2007 0 5 10 15 20 25 30 5.5 6.5 7.5 8.5 9.5 10.5 11.5 12.5 13.5 14.5 15.5 16.5 all banks distribution french banks distribution tier 1 ratio 2008 0 2 4 6 8 10 12 14 16 18 5.5 6.5 7.5 8.5 9.5 10.5 11.5 12.5 13.5 14.5 15.5 16.5 all banks distribution french banks distribution sh ar e o f b an ki n g se ct or ( % ) sh ar e o f b an ki n g se ct or ( % ) sh ar e o f b an ki n g se ct or ( % ) xiao: french banks 6 the international journal of banking and finance, vol. 8. number 1, march 2011: 1-19 leverage stabilized and french banks did not appear to choose an aggressive strategy of reducing outstanding loans or restricting the distribution of riskier loans. figure 4: leverage of french vs european banks 3.5 quality of capital information on quality of capital is provided in figure 5. the crisis has sparked intense discussions about the quality of capital, the signifi cance of which is highlighted in supervisory guidelines. for example, the basel committee banking supervision (bcbs) guidelines noted that core tier 1 capital should be a predominant part of tier 1. the turner review pointed out that “the fsa therefore believes that required capital ratios for such banks should be expressed entirely in terms of high quality capital—broadly speaking the current core tier 1 and tier 1 defi nitions—and should not count dated subordinated debt as providing relevant support. this is in line with the direction of the basel committee deliberations.” the us supervisory capital assessment program (scap) argued that “supervisors have long indicated that common equity should be the dominant component of tier 1 capital….”. french banks had a higher capital quality initially, but their lead was eroded following the raft of global recapitalizations across the industry: see figure 5. both the french banks and their european peers increased the share of tier ii capital in the capital structure in 2008, although the increase was somewhat smaller for french banks. with investors putting less emphasis on tier ii capital, several european banks, including some french banks, have considered or conducted liability management operations to buy back their lower tier ii capital to improve the quantity and quality of capital. the core tier i ratio of french banks was about 15 basis points above that of others in 2007, and stood on par with their peers after the wave of government recapitalizations in 2008. leverage 0 5 10 15 20 25 30 35 2006 2007 2008 france developed europe international journal of banking and finance, vol. 8, iss. 1 [2011], art. 1 french banks amid the global financial crisis: 1-19 7 figure 5: capital quality of french vs european banks 3.6 funding french banks appear to have a rising and higher-than-average reliance on wholesale funding: figure 6. a bank’s funding strategy at different maturities tends to affect the banks’ fragility and its sensitivity to a liquidity dry-up. deposit funding and wholesale funding may carry different risks in causing a potential liquidity crisis and bringing about changes in funding costs. french banks have increased their recourse to wholesale funding with the rapid expansion of their activities and a shift to high-fee generating products for funding. an examination of french banks’ average funding profi le during 2006-08 reveals that wholesale funding represents 58 percent of the total funding and short-term (st) wholesale funding accounts for about 35 percent of st funding in the case of french banks, but 47 percent and 28 percent, respectively, in the case of their european peers. though at fi rst appearance it seems at odds with the high savings ratio of france, this refl ects the role of the money market funds and life insurance products in diversifi ed bank funding. figure 6: funding of french vs european banks share of tier ii capital 0 5 10 15 20 25 30 35 80027002 france developed europe core tier 1 6.00 6.10 6.20 6.30 6.40 6.50 6.60 6.70 6.80 80027002 france developed europe share of wholesale funding 0 10 20 30 40 50 60 70 2006 2007 2008 france developed europe share of wholesale funding in st funding 0 5 10 15 20 25 30 35 40 45 2006 2007 2008 france developed europe xiao: french banks 8 the international journal of banking and finance, vol. 8. number 1, march 2011: 1-19 the liquidity profi le of the french banks as seen in figure 7 from the alm perspective appears to be slightly more favourable than that of their european peers, although this is becoming less over time. while the funding profi le focuses on the liability side, the alm perspective focuses on liquidity matching from both assets and liabilities. the deposit-to-loan ratio shows the extent to which domestic credit is funded by banks’ deposit liabilities. any shortfall, as refl ected in a ratio of less than 100 percent, has to be funded by bank borrowing from the non-bank private sector or from overseas, both of which have become constrained in the crisis. banks with a high coverage ratio of customer loans by deposits are less vulnerable and more resilient in the face of increased tension in liquidity markets. the deposit-to-loan ratio of french banks rose to 75 percent in 2007, but slid in 2008, although it still remained above the level of the european peers. the liquid ratio of french banks declined by half to reach a similar level as their peers in 2008. however, the liquidity profi les of french banks may improve with the generalization of livret a and livret bleu (special saving accounts in france) at the beginning of 2009, which has encouraged banks to attract more deposits. figure 7: liquidity measures of french vs european banks 4. analyses of business and exposures the analyses of business and exposures supplement the above soundness check to gain more insights about the relative performance of french banks. specifi cally, this section analyses business lines, international exposures, writedowns and losses, new capital raised, and market perceptions of risk. financial data come from bloomberg, the bis, datastream, thomson financial, and the fi nancial statements of banks. domestic and international retail banking activities contributed to the fi nancial results of french banks. french banks have adopted a domestic retaildominated universal banking model. an analysis of the business lines shows that deposit/loan ratio 64 66 68 70 72 74 76 2006 2007 2008 france developed europe share of wholesale funding in st funding 0 5 10 15 20 25 30 35 40 45 2006 2007 2008 france developed europe international journal of banking and finance, vol. 8, iss. 1 [2011], art. 1 french banks amid the global financial crisis: 1-19 9 retail banking activities, especially international banking, continued their brisk growth throughout the crisis. the share of domestic business climbed modestly to 46 percent, refl ecting the low risk and high saturation of domestic markets. the contribution of international banking to operating income rose signifi cantly from 18 percent in 2006 to 27 percent in 2008, refl ecting the french banks’ geographic diversifi cation and brisk expansion to fast-growing countries. however, corporate and investment banking (cib) and asset management activities detracted from the fi nancial strength of the french banks. cib business, the key profi t driver in the past, was hit hardest by the crisis. its contribution to operating income fell from the pre-crisis level of 24 percent to 10 percent in 2008, refl ecting plummeting revenues and even sizable losses for some banks. they suffered from impairment charges and fair value losses from toxic assets including sub-prime related residential mortgage-backed securities (rmbs), collaterized debt obligations (cdo), asset-backed securities (abs), and exposures to monoline and credit derivative product companies (cdpc). increased risk aversion shunned investors away from complex cib products and services, which materially compressed cib revenues and depressed its profi tability. the economic downturn and market volatility also weighed on other structured products such as commercial mortgage-backed securities (cmbs), leveraged buy-outs (lbo), collaterized loan obligations (clo) as well as trading and advisory services. signifi cant outfl ows arising from volatility, a loss of confi dence associated with a “breaking of the buck” in the u.s. moneymarket funds, problems with the “dynamic” money market funds, and poor performances of hedge funds dealt a blow to profi ts from asset management activities, although to a lesser degree than for cib activities. with the partly offsetting income from private banking, the contribution from the combined asset management and private banking (am & pb) fell back to its pre-crisis level. figure 8: business lines of french banks evolution of business line 0 20 40 60 80 100 2006 2007 2008 domestic international cib am&pb other xiao: french banks 10 the international journal of banking and finance, vol. 8. number 1, march 2011: 1-19 the french banks’ exposure to foreign markets is not excessive and tilts towards mature markets. foreign claims on an immediate borrower basis represented 34 of bank assets and 128 percent of gdp at the end of 2008 while foreign claims on an ultimate risk basis were slightly lower. the exposure in terms of the size of the national economy aligns with countries such as austria, germany, sweden, and the u.k. and lies between the very low level of the u.s. and the very high level of switzerland. banks’ exposure to mature market dominates, representing 86 percent of total foreign claims. exposure to the u.s., italy, the u.k., and germany accounts for more than half of the total exposure to mature markets. the mature market bias suggests that potential spillovers from these markets may have a material impact on french banks. the french banks’ exposure to offshore fi nancial centers and emerging markets is limited. exposure to emerging markets on either an immediate borrower basis or an ultimate risk basis represented about four percent of bank assets and 14 percent of gdp at the end of 2008, only higher than the u.s. and the u.k. exposure to offshore fi nancial centers is even smaller: see figure 9. figure 9: foreign exposure of french banks the exposure to emerging markets is geographically diversifi ed. exposure to central and eastern europe (cee) made up about 36 percent of the total exposure to emerging markets, followed by africa and the middle east (28 percent), and asia and the pacifi c (24 percent) at the end of 2008. among the exposure to cee, exposure to the czech republic, russia, and poland accounts for half of the total exposure. among the three major french banks having exposure to cee, only one bank has a sizable exposure. although pressure on french banks could be mounting with a worsening of the fi nancial situation in the cee region or emerging markets in general, the overall risks may be manageable: figure 10. g foreign exposure (% of gdp) 0 100 200 300 400 500 600 france austria germany sweden switzerland uk us mature markets emerging markets offshore centers international journal of banking and finance, vol. 8, iss. 1 [2011], art. 1 french banks amid the global financial crisis: 1-19 11 figure 10: emerging market exposure of french banks losses and writedowns of french banks have been signifi cant, but comparatively less than in the hardest-hit countries: see figure 11. french banks have written down a large proportion of assets relative to their initial value. their total losses and write-downs since the onset of the crisis account for about 3 percent of losses and write-downs around the globe, considerably less than those of banks in the u.s. (55 percent), the u.k. (12 percent), germany (9 percent), and switzerland (7 percent), and also less than the share of french banks in the global banking system. the international comparison of the cb on the direct cost of the crisis for large international banks shows that the direct cost for the french banks was about 18 percent of tier 1 capital, lower than that of germany (about 33 percent), the u.k. (about 37 percent), the u.s. (about 86 percent), and switzerland (about 87 percent).1 recapitalization has been able to cover the losses and write-downs already incurred. french banks have made progress in shoring up their balance sheets by attracting capital from fi nancial markets and accepting government capital injections. state capital injections represent about one-third of new capital and play an increasing important role over time. total capital raised by french banks accounts for about 4 percent of the aggregate capital raised around the globe, roughly equal to their share of losses and write-downs. the market perception of credit risk of french banks seems to be more favorable than that of their european peers, especially for the lower class of debt. refl ecting rising credit risk, credit default swaps (cds) spreads of french banks and other european banks have increased considerably. in the case of senior cds spreads of french banks, spreads have shot up from less than 10 basis points before the crisis to more than 100 basis points recently. throughout most of the crisis period, senior cds spreads of french banks have been lower than those of peers, as measured by the itraxx euro senior cds index. in the case of junior cds spreads of french banks, spreads have surged from less than 30 basis points before the crisis to more than 150 basis points recently. as in the distribution of em exposure (in percent of bank assets) 0 5 10 15 20 25 france austria germany sweden switzerland uk us africa & middle east asia & pacific cee latin america cee exposure (in percent of total foreign claims) 0.00 0.20 0.40 0.60 0.80 1.00 czech republic estonia hungary latvia lithuania poland romania russia slovakia turkey ukraine 1 for details, see commission bancaire (2009). xiao: french banks 12 the international journal of banking and finance, vol. 8. number 1, march 2011: 1-19 case of senior cds spreads, throughout most of the crisis period, junior cds spreads of french banks have been lower than those of peers measured by the itraxx euro junior cds index. moreover, the difference is more striking. for example, junior cds spreads of french banks were 36 percent below those of the itraxx euro junior cds index at the peak of the crisis in march 2009 and 24 percent lower in june. figure 11: write-downs, losses and capital raised by french banks senior cds spreads of french banks were 17 percent and 8 percent below those of the itraxx euro senior cds index in march and june 2009, respectively. the market perception of french banks as being less risky than their peers, even for the lower class of debt, may indicate that markets place a premium on national champions: figure 12. figure 12: cd spreads of french banks geographical distribution of write-downs and losses and of capital raised write-downs and losses 2% 3% 9% 1% 2% 12% 55% 7% 9% canada france germany italy japan uk us switzerland others capital raised 1% 4% 5% 1% 4% 13% 42% 5% 25% cds of senior debt of financials (basis points, 5y) 0 50 100 150 200 250 1/06 5/06 10/06 2/07 7/07 12/07 4/08 9/08 1/09 average of top 4 french f inancials 1/ itraxx euro cds of junior debt of financials (basis points, 5y) 0 50 100 150 200 250 300 350 400 450 1/06 5/06 10/06 2/07 7/07 12/07 4/08 9/08 1/09 average of top 4 french f inancials 1/ itraxx euro international journal of banking and finance, vol. 8, iss. 1 [2011], art. 1 french banks amid the global financial crisis: 1-19 13 4. an event study of the french banking sector before the deepening of the crisis in september 2008, the french government focused on increasing the effi ciency and competitiveness of the fi nancial sector domestically and advocated regulatory and supervisory reforms internationally. in particular, to liberalize and reform the fi nancial sector, the government rolled out a fl urry of measures under the paris-place financière initiative and the loi de modernisation de l’économie (lme). in the aftermath of the trading fraud at sociétié générale (sg), the government took actions to strengthen internal controls and operation risk management. during the 2008 french presidency of the eu, it proposed thirty recommendations for europe to combat the crisis. the bankruptcy of lehman brothers intensifi ed the crisis and fuelled the worst bout of fi nancial contagion since wwii. concerns about exposures to lehman sparked massive turmoil in global fi nancial markets with the freezing up of interbank, senior unsecured, covered bond and securitization markets. the fi nancial shock in the us reverberated internationally, fuelled the global “fl ightto-quality,” and resulted in surging spreads, collapsing equity prices, and spiking volatility. the deepening of the crisis prompted large-scale sovereign interventions. governments around the world took unprecedented support measures to recapitalize banks and unfreeze credit. global government measures included recapitalization, guarantees, deposit insurance, asset swaps, asset purchases, as well as direct lending and crisis liquidity facilities. measures in some countries came with strings attached such as conditions on dividends, salary restrictions, lending rules, code of ethics, and government appointed board members. as part of the global actions, the french government created two separate agencies to recapitalize banks and provide government guarantees for bank refi nancing. recapitalization was handled by the société de prise de participations de l’état (sppe), a fully state-owned agency. refi nancing operations were undertaken by the société de financement de l’économie française (sfef), owned jointly by the french state (34 percent) and seven leading french banks (66 percent). sppe provided solvency support through the availability of tier 1 capital instruments. it earmarked €40bn of funds. in addition to the injection to dexia along with other governments, €10.5bn of the fi rst tranche in the form of deeply subordinated debt securities were injected into six banks, boosting their tier 1 ratios by about 50 bps. sppe also supported the merger between gce and gbp with an additional €3bn. the second tranche in the same amount to the same banks was announced, but has not been taken by all the banks yet. there are several improvements to the new recapitalization scheme over the original one. for example, banks are given the option of issuing preference shares. preference shares are new core tier 1 instruments established by the french law. they have the following characteristics: non-cumulative dividend, xiao: french banks 14 the international journal of banking and finance, vol. 8. number 1, march 2011: 1-19 no voting rights, non-convertible, preferential but capped remuneration, limited dilution, and loss-absorbing capacity. the new scheme also offers incentives for banks to buy back securities as the redemption amount would increase over time. sfef provided liquidity support through the provision of governmentguaranteed refi nancing. it raised market fi nancing by issuing state-guaranteed bonds, the proceeds of which were then used to on-lend to banks in proportion to the market share of each bank in terms of customer loans and assets for a period of one to fi ve years. sfef issued up to €265bn of guaranteed term debt (maximum maturity fi ve years), which enjoyed ‘aaa’ rating as the french government. debt had to be issued before the end of 2009. before the creation of the sfef, the government also issued guarantees on dexia’s obligations (a maximum of €55bn) along with the rescue package arranged for the bank with other governments. sfef proved to be popular with investors as refl ected in the very tight spreads of its bonds. its issuance, second only to the u.s. and about one third denominated in u.s. dollars, represented 20 percent of the global issuance of government guaranteed bonds. its ability to attract a wide range of investors from various market segments may result from its skillful set-up. by pooling liquidity, the agency enhanced the visibility and reduced the liquidity premium of its bonds. in addition, by construction, investors are not exposed to bank risks. the guarantor, the french government, is perceived to be capable of standing behind its promises. with the recent market improvement, french banks have tapped markets by issuing bonds without government guarantees. there have been several theoretical studies of best government support schemes, but very few empirical studies. the crisis provides good natural experiments of various government support schemes, but there seems to be only one empirical study by veronesi & zingles (2008) to examine the impact of the u.s. plan. to bridge this gap in the literature, this paper uses a similar method to analyse the french plan. the event window is between october 10 and october 14 in 2008, one day before and after the announcement of the french schemes on october 13, which was also the day when the u.s. government announced its revised paulson plan and the u.k. government announced its own schemes. the study aims to explore the impact of the french fi nancial sector support plan by combining both the balance sheet and the market information. specifi cally, it intends to: (a) the market impact on debt using senior and subordinated cds spreads and linking these to the maturity structure of the bank debt, (b) gauge the market impact on equity using capital asset pricing model (capm), (c) take into account other events happening at the same time by measuring the relative impact besides the gross impact, and (d) reverse engineer the black-scholes-merton model, as in xiao (2008), to measure the proportion of equity injection transferred to debt holders. following veronesi & zingles (2008), the impact on debt is calculated as follows: international journal of banking and finance, vol. 8, iss. 1 [2011], art. 1 french banks amid the global financial crisis: 1-19 15 1. the default probability from the cds (rr is recovery rate) is backed out. (1) 2. the gross impact on debt is equal to the difference between the present value of debt before the plan and after the plan (rf is the risk-free rate, b and a indicate before and after, respectively). (2) 3. to control other things happening at the same time, an adjusted impact is measured by subtracting debt changes in scor group, a reinsurance fi rm not receiving government money but experiencing narrowing spreads. (3) the results show that the french support plan drove down banks’ credit risk signifi cantly: see figure 13. the debt value of banks appreciated across the board after the announcement of the plan, with the gross appreciation ranging from 21 percent to 38 percent. as expected, taking into account other events happening at the same time lowering the impact, the positive effects of a value increase of at least 9 percent were still pronounced. in addition, subordinated debt benefi ted more than senior debt. on average, the value of the banks’ subordinated debt went up by 35 percent on gross terms and by 25 percent on adjusted terms, while the value of the banks’ senior debt went up by 26 percent and 14 percent, respectively. figure 13: impact on debt of french banks gross impact on debt 0 5 10 15 20 25 30 35 40 45 bnp ca na sg senior debt subordinated debt adjusted impact on debt 0 5 10 15 20 25 30 35 40 45 bnp ca na sg senior debt subordinated debt 1 10000/ rr cdsdp t t rf ddp t t rf ddp t t acdst a t t bcdst bpv 0 )1( *)1( 0 )1( *)1( 1000010000 sc b sc pv pv bpvpvpvadjusted xiao: french banks 16 the international journal of banking and finance, vol. 8. number 1, march 2011: 1-19 the impact on stock values is calculated as follows. the gross impact is measured by raw stock returns and the adjusted impact is measured by using abnormal returns from the capm. beta is estimated from daily stock prices during the period 1/1/07-10/09/08. both market benchmarks, cac40 and sbf250, are used. adjustments are done with beta equal to one and the estimated betas. the results show that the plan had a mixed impact on equity: figure 14. the gross impact was positive, with the equity value of the banks experiencing a modest increase of 2-7. however, the adjusted impact was negative across the board, regardless of the benchmarks and the beta estimation methods used. the loss of the equity value of the banks ranged from one to 23 percent. with the exception of one bank, the estimated betas produced larger equity value reduction than assuming betas equal to one. the different impact on debt and equity suggests the possibility of equity injections being transferred to debt, as analysed in myers (1977). figure 14: impact on equity of french banks to measure the proportion of equity injections into debt, the blackscholes-merton model is reverse engineered.2 in the black-scholes-merton model, asset value follows a geometric brownian motion: (4) equity is a call option and debt is a put option on bank assets. specifi cally, (5) (6) where e is the market value of equity, a is bank assets, d is the market value of debt, impact on equity -25 -20 -15 -10 -5 0 5 10 bnp ca na sg cic gross adj. with beta=1 adj. with beta=est. da/a=μadt+ adw daafdbaeedbae db rt )()()0,max( db rt daafdbadeadbdbd )()()0,max( 2 for details, see xiao (2008). international journal of banking and finance, vol. 8, iss. 1 [2011], art. 1 french banks amid the global financial crisis: 1-19 17 db is the distress barrier, μ a is the expected rate of return of assets, t is the time to maturity on debt in years,  a is the standard deviation of assets, dw is the weiner process, and f(a) is the asset distribution function. to obtain the share of equity injections into debt, the following procedure is followed: (a) calibrating the black-scholes-merton model to market data of bank equity and volatility, (b) backing out the implied value and volatility of bank assets, (c) calculating ex-ante the market value of bank bonds and equities after the announcement of the plan, and (d) determining the share of equity injection into debt by the value difference between the preand post-plan market values relative to capital injections. the results demonstrate that the share of transfer varies with the riskiness of banks. for the three largest french banks, the proportion of the transfer ranges from 13 to 56 percent. moreover, the transfer dovetails with the credit risk of banks measured by the average of the cds spreads of the banks’ senior and subordinated debt. the riskier the debt, the higher is the share of capital injections into debt. it indicates that capital injections may benefi t debt holders at the expense of shareholders, arguing for the necessity of injections by the government. figure 15: share of capital injections in french banks 5. conclusion french banks have demonstrated relative resilience to the global fi nancial crisis so far. the global crisis has put the banks to an unprecedented test and they were not immune to the severe fallout of the crisis. however, the banking model featuring diversifi cation in business, funding, and geography has helped to contain the risk at a manageable level. in addition, benefi ting from the comprehensive supervision, proactive regulation, and timely informationsharing among regulatory authorities, the banks enjoyed better initial conditions and have withstood the crisis thus far relatively well. 0 20 40 60 80 100 120 140 bnp ca sg 0 10 20 30 40 50 60 share of transfer(right scale) average cds(left scale) xiao: french banks 18 the international journal of banking and finance, vol. 8. number 1, march 2011: 1-19 government measures are necessary and benefi cial, but challenges still lie ahead. recapitalization and refi nancing measures have helped stabilize the system by reducing the risk and decreasing the fi nancing cost considerably. going forward, banks’ earnings and profi tability may continue to be under pressure with subdued cib activities, potential losses and writedowns from risky assets, the still low interest margins, as well as the rising counterparty risk and the cost of risk. with the global crisis still unfolding and the international debate on capital adequacy still evolving, the impact of the banks’ reduced comparative advantage in capital buffers remains to be seen. rising to the challenge would call for continued vigilance and enhanced risk management. author information: yingbin xiao, ph.d., is a senior economist of the international monetary fund, washington dc. dr. xiao would like to thank erik de vrijer, ann-marie gulde-wolf, danièle nouy, muriel tiesset, an anonymous referee, and participants at the french ministry of economy, finance, and industry in paris and the american midwest finance association 2010 conference in las vegas for helpful comments and suggestions. assistance from anastasia guscina and judith rey are also acknowledged. the views expressed in this paper are those of the author and do not necessarily represent those of the imf or imf policy. contact email: yxiao2@imf.org. references bis. (2004). bcbs compendium of documents volume ii: advanced supervisory methods, basel. black, fischer & myron s. scholes. (1973). the pricing of options and corporate liabilities. journal of political economy, 81, 637-654. choulet, & céline. (2009). banques européennes: les plans de soutien à l’épreuve de la recession. conjoncture 51, bnp paribas. commission bancaire. (2009). rapport de la commission bancaire pour l’année 2008. paris. federal reserve. (2009). the supervisory capital assessment program: design and implementation. federal reserve: washington dc. fsa. (2009). the turner review. financial services authority: london. gapen, michael, t., dale, f. gray, cheng hoon lim & yingbin xiao. (2005). the contingent claims approach to corporate vulnerability analysis: estimating default risk and economy-wide risk transfer. in corporate restructuring: lessons from experience, michael pomerleano and william shaw (eds). world bank. gapen, michael, t., dale, f. gray, cheng hoon lim & yingbin xiao. (2008), measuring and analyzing sovereign risk with contingent claims. imf staff papers, 55, 109–148. imf. (2009). global fi nancial stability report 04/09 (washington: international monetary fund). international journal of banking and finance, vol. 8, iss. 1 [2011], art. 1 french banks amid the global financial crisis: 1-19 19 lagarde, & christine. (2008). le plan français pour assurer le fi nancement de l’économie et restaurer la confi ance. communiqué de presse du conseil des ministres, paris. merton, & robert, c. (1973). theory of rational option pricing. the bell journal of economics and management science, 4, 141-183. merton, & robert, c. (1974). on the pricing of corporate debt: the risk structure of interest rates. the journal of finance, 29, 449-470. merton, & robert, c. (1977). an analytic derivation of the cost of loan guarantees and deposit insurance: an application of modern option pricing theory. journal of banking & finance, 1, 3-11. myers, s. c. (1977). determinants of corporate borrowing. journal of financial economics, 5, 147-175. noyer, christian, & nouy danièle. (2009). audition devant la commission des fi nances du sénat comptes rendus de la commission des finances, paris. ricol, & rene. (2008). rapport sur la crise fi nancière mission confi ée par le président de la république dans le contexte de la présidence française de l’union européenne 2008. paris. veronesi, pietro, & luigi zingles. (2008). paulson’s gift. mimeo: university of chicago. yingbin xiao. (2008). financing and risks of french fi rms. sosial science research network working paper no. 1435088. available at ssrn: http://ssrn. com/abstract=1435088. xiao: french banks islamic finance: an attractive new way of financial intermediation islamic finance: an attractive new way of financial intermediation munawar iqbal king abdul aziz university, saudi arabia abstract financial intermediation is a value-enhancing service. banks are among the most important financial institutions in a modern economy for that service. conventional banks use rates of interest (charged to clients and paid to depositors) on both the assets and the liabilities sides. since interest is prohibited in islam, islamic financial experts have developed a number of financial instruments that avoid any involvement in interest. they take the form of either risk-and-reward sharing or trading in commodities/assets to price assets. in this paper we describe the basic features of the most important among these financial instruments. even though islamic banks emerged in response to market needs of muslim clients, they are not religious institutions. like other banks, these are profit seeking institutions, simply following a different model of financial intermediation. while it is the preferred way of banking for one fifth of humanity, it offers a wider choice of financial products to all by generating a number of benefits for the society. the successful operation of islamic financial institutions has proven that this new model of financial intermediation is not only viable, but in many aspects, it is rather superior to the conventional model. the appealing features of the islamic model have attracted world-wide attention. islamic financial industry which started as a niche market in early 1970s in the middle east has made a place for it in more than fifty countries around the globe and has grown into a multi-trillion dollars industry. this paper attempts to explain the basic features of this fascinating model. key words: financial intermediation, islamic finance, islamic financial products jel classification number: g2 ________________________________________________________________ 1. introduction in any economy, there is a need to transfer funds from savers to investors because people who save are frequently not the same people, who have the ability to ijbf ________________________________________________________________ iqbal: islamic finance 2 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 exploit the profitable investment opportunities, i.e., they are not entrepreneurs. this function is performed either directly through stock and bond markets or through the process of financial intermediation in the financial markets. since interest is prohibited in islam, a different model of financial intermediation was proposed in 1950s which envisaged providing modern financial services without dealing in interest. at that time, it was considered wishful thinking since interest was deeply entrenched in almost all financial dealings. however, serious research work over the next two decades proved that islamic banking is not only feasible and viable; it is an efficient and productive way of financial intermediation. in response, several islamic financial institutions (ifis) emerged under heterogeneous social and economic milieu. what started as a small rural banking experiment in the remote villages of egypt in early 1960s has now reached a level where many mega international banks are offering islamic banking products. and several western countries, following a lead from the uk, have made room for islamic financial products/institutions in their legal structures. the most important operators in the financial markets are commercial banks. since the passage of the so-called glass-steagall act in 1933, commercial banks’ securities activities and affiliations between commercial banks and securities firms were prohibited. however, the market largely ignored the provisions of glass-steagall act and they gradually faded away. they were formally repealed in 1999 through the gramm-leach-bliley act. with the passage of the gramm-leach-bliley act, commercial banks, investment banks, securities firms, and insurance companies were allowed to consolidate. since then commercial banks are also one of the biggest players in stock and bond markets. financial intermediation enhances the efficiency of the saving/ investment process by eliminating many mismatches inherent in the needs of surplus and deficit units of an economy (iqbal & llewellyn, 2002). the surplus units are often small households who save relatively small amounts and the deficit units are the firms who often need relatively large amounts of cash. financial intermediaries remove this size mismatch by collecting the small savings and packaging these to make them suitable to the needs of the users. in addition, users of funds in general need funds for relatively long-term deployment, which cannot be met by individual suppliers of funds. this creates the mismatch in the maturity and liquidity preferences of individual savers and users of funds. the intermediaries resolve the conflict again by pooling the small funds. moreover, the risk preferences of small suppliers and large users of funds are also different. it is often considered that small savers are risk averse and prefer safer placements whereas the fund users may deploy funds in risky projects. therefore, the funds cannot be directly supplied. the role of the intermediaries again becomes crucial. they can substantially reduce this risk through portfolio diversification. furthermore, small savers cannot efficiently gather information about investment opportunities. financial intermediaries are international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 3 in a much better position to collect such information, which is crucial for making the investment successful. the role and functions of banks outlined above are indeed highly useful and socially desirable, but interest plays a central role in each of these functions. islamic financial intermediation endeavors to replace interest by partnership between owners of capital and human resources (entrepreneurs) on the basis of risk/profit sharing as a basic form of co-operation between capital and entrepreneur. trading activities are also an integral component of islamic banking. 2. basic principles of islamic finance in islamic jurisprudence there are two kinds of rulings (arkām). the first ruling is called ibādāt (worshipping) which governs the relationship between a man and his god. the general principal in case of ibādāt is that nothing is permitted unless endorsed by authentic and explicit analogical permission by the law giver. in other words, an act is considered worship only if so ordained. the second ruling is of arkām, called muʿāmalāt (mutual dealings) which governs the relationship among mankind. in this case, the general principal is that of permissibility (ibādah) i.e., everything is permitted unless clearly prohibited by the law giver. the collection of rulings comprises of what is called as sharī’ah (islamic law). since we are discussing the second set of rulings, the general rule is that of permissibility. we call this the “doctrine of universal permissibility”. the doctrine permits the contracting parties to agree on any conditions as long as they do not violate any sharī’ah ruling. this is supported by a well-known authentic hadīth which states: “all conditions mutually agreed upon by the muslims are upheld, except any condition that would allow what is prohibited or prohibit what is lawful.” [sunan tirmidhi:1352] this principle gives a very wide scope in designing contracts. however, the qualification, “unless clearly prohibited by sharī’ah” must never be ignored. when one looks at the permissible acts and contracts and the non-permissible ones, one is struck with the very wide scope of running human relations within the boundaries of sharī’ah in all fields of life including financial transactions. there are very few kinds of activities that are prohibited. furthermore, when we reflect even on these prohibitions, one can easily ascertain that the purpose of such prohibitions is to provide a level playing field to protect the interests of weaker parties; to ensure justice and fairness; to ensure mutual benefit for the parties as well as the society at large; minimize chances of conflicts and promote social harmony. islamic law is so beneficent that it not only ensures protection iqbal: islamic finance 4 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 of interests of the two parties to the contract, it also protects the interests of third parties, future generation and even the en the prohibitions are very few. in the following paragraphs we discuss those prohibitions that are most relevant for constructing financial contracts. 2.1 prohibition of ribā islam prohibits all forms of ribā. but what does this term encompass? the word ribā as a noun literally means in arabic, an increase, and as a root, it means the process of increasing. ribā has been understood throughout muslim history as being equivalent to interest paid on a loan. the prohibition of ribā essentially implies that the fixing in advance of a positive return on a loan as a reward for waiting is not permitted by the sharī’ah. it makes no difference whether the return is big or small, fixed or variable or an absolute amount to be paid in advance or on maturity, or a gift or service to be received as a condition for the loan. it also makes no difference whether the loan was taken for consumption or business purposes. in its basic meaning ribā can be defined as “anything (big or small), pecuniary or non-pecuniary, in excess of the principal in a loan that must be paid by the borrower to the lender along with the principal as a condition,1 (stipulated or by custom), of the loan or for an extension in its maturity.” as mentioned in the definition of ribā given above, anything, big or small, stipulated in the contract of loan to be paid in addition to the principal is ribā. such additional payment in modern terminology is known as interest. thus ribā and interest are the same. the equivalence of ribā and interest has always been unanimously recognized in muslim history by all schools of thought. while the basic rationale for the prohibition/condemnation of ribā is moral, it creates many economic problems too. the ease of availability of credit on interest also creates inherent instability in the financial system. as nobel laureate maurice allais (1993) rightly points out: “the world economy as a whole rests today on gigantic pyramids of debts, buttressed one against another in a fragile equilibrium. never in the past had such a colossal accumulation of promises to pay been witnessed. undoubtedly, never will it be so difficult to master. be it speculation on currencies or speculation on stocks and shares, the world has become one big casino with gaming tables distributed along every latitude and longitude. the game and the bids, in which millions of players take part, never cease.” iqbal (2013) dwells on the dangers and consequences of unreasonable ease of access to credit and the resulting debt overhang which has caused havoc with global economy several times, the most recent example being the still raging 1 thus any excess given by the debtor out of his own accord, and without the existence of a custom or habit that obliges him to give such excess is not considered as ribā. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 5 european debt crisis. several other recent publications show that the root cause of these crises is excessive debt facilitated by interest-based contracts (reinhart, carmen & rogoff, 2009). it is pertinent to note here that the practice of interest has been condemned by many most prominent thinkers in human history and by all devine religions. mills and presley (1999) have traced the history of prohibition of interest in biblical texts. the basic rationale for the prohibition of interest is based on ethics, most importantly to ensure justice to all parties. in many cases, charging of interest is also demeaning. for example, if the loan is for procuring things necessary for survival, charging interest delivers a severe blow to the very basic fabrics of social life which requires cooperation, care and help of the needy, and protecting the weaker members of the society. 2.2 prohibition of gharar prophet (pbuh)2 prohibited any trading activity involving ‘gharar’. in jurisprudential literature, gharar has been variously defined. mainly there are three views which are summarized by al-darir (1997). first, gharar applies to cases of uncertainty, as in the case of not knowing whether something will take place or not. a second view holds that gharar applies trading of unknown. thus, according to ibn hazm gharar in sales occurs when the purchaser does not know what he has bought and the seller does not know what he has sold. third is a combination of the two opinions; gharar covers both the unknown and the doubtful, as exemplified by the definition proposed by al-sarakhsi which states that gharar obtains where consequences of a contract are not known. this is the view favoured by most jurists. gharar has been one of the most difficult islamic juristic terms to explain to non-specialist. some authors have dealt with it very comprehensively in long treaties.3 perhaps the best way to envision its true scope and import is to list some examples. in modern contracts, gharar appear in many ways. the most important categories are the following: i) ignorance of the genus. for example, saying “i sell you 1 kilogram of apples for five dollars.” it involves gharar because it is not clear what type of apples are the subject of sale. ii) ignorance of the species. for example, saying “i sell you my pet for 100 dollars.” it is not known what the pet is. iii) ignorance of the attributes. for example, saying “i sell you my car for 5000 dollars.” it is not known which car. (however, if the object is available and offered for inspection with known defects declared then there is no gharar.) iv) ignorance of the quantity of the object. for example, saying “i sell you a box of oranges for 20 dollars” without mentioning the type of oranges 2 short form for “may peace of allah be upon the soul of prophet muhammad” 3 the best reference, in view of this author is darir al-siddiq (2012) iqbal: islamic finance 6 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 and the weight. branding and standardization can remove this type of gharar. v) ignorance about price. for example, saying “i sell you this dress for a week’s salary.” vi) ignorance of the specific identity of the object. for example, saying “i sell you one flat in this building for fifty thousand dollars.” mentioning of flat number and allowing physical inspection is necessary. vii) ignorance of the time of payment in deferred sales. for example, saying “i buy this house from you for 100,000 dollars which i will pay when my farm is sold.” viii) inability/uncertainty to deliver the object. for example, saying “i sell you the bird sitting on that tree,” or “i sell you my bmw which has been stolen if it is recovered for only $1000.” ix) contracting on a non-existent object. for example, saying “i sell you the harvest of my farm from the next crop.” x) not being able to inspect the object. for example, saying “i sell you the contents of this carton for fifty dollars on as is basis.” xi) more than one option in a contract unless one is specifically chosen. for example saying “you can either take my car for ten thousand dollars,” or “my boat for fifteen thousand dollars.” the sale would become valid only after you exercise your option and specifically choose what you are buying. xii) contingent sale. “i sell you my house for $25000, if peter rented his house to me.” these examples are not exhaustive but they should be sufficient in giving a fairly good idea of what the prohibition of gharar implies. in essence, gharar refers to acts and conditions in exchange contracts, the full implications of which are not clearly known to the parties. this is something very similar to ‘asymmetric information’. lack of knowledge with respect to all implications of a contract vitiates against the principle of voluntary consent of all parties which is a necessary condition in all contracts of exchange according to islamic law. in other words, had the parties known the full implications of the contract, one (or both) of them may have preferred not to enter into that contract. in the presence of asymmetric information, the agreement of the parties cannot be considered as voluntary consent. the objective of prohibition of gharar is to minimize possibilities of post-facto misunderstandings and conflicts between the contracting parties. as such, any conditions in contracts that may give rise to ambiguities and/or misinterpretations and are preventable must be avoided. jurists make a distinction between two types of gharar: gharar fāḥish (substantial) and gharar yaseer (trivial). the first type is prohibited while the second is ‘tolerated’ since this may be unavoidable without causing considerable damage to one of the parties with the possibility of post-facto conflict. in many cases, it is simply not possible to reveal all information; not because the seller wants to hide anything, but because it is in the nature of the subject matter. the international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 7 buyer has to trust the seller. for example, the buyer of a built house has to take the word of the seller as to what kind of material is used in the foundations of the house. the seller obviously cannot dismantle the house to reveal the foundations to the buyer. his intention is not to hide essential information, but the nature of the subject matter of the contract is such that he cannot show it to the buyer. therefore, such lack of knowledge does not violate contracts. the principle in such cases is that the seller must act as a trustworthy person. penalties may be imposed ex-post if it is proved otherwise. in the literature, some other prohibited types of trade or conditions therein are also mentioned. for example, the following cases are mentioned: prohibition of selling something not owned; ghish (cheating, for example adulteration); two mutually inconsistent contracts with (conditions) bunched into one contract; ikrāh (coercion); ghaban (fraud); but on closer examination one finds that these are covered by the prohibition of gharar. in economic parlance the term is very close to “asymmetric information.” its prohibition necessitates ‘full disclosure’ by all parties of a contract. any intentional concealment renders the contract void. small, unintentional or unavoidable lack of disclosure (gharar yaseer) is tolerated (not to be considered allowed). post contract exposure is still subject to legal recourse for compensation. 2.3 prohibition of gambling gambling involves transfer of wealth without any value added. it is zero-sum game and does not make any addition to the national wealth. prohibition of gambling implies prohibition of all zero-sum games and taking risks that are self-created. as compared to this, taking risk which are normally involved in almost all business activities do not come under the purview of gambling. islamic financial contracts encourage sharing of these risks among the parties. 2.4 loan and trade contracts cannot be combined this is prohibited by an authentic hadith. the rationale for this prohibition can easily be seen. if allowed, a lender may advance the loan on an interest-free basis in the loan part of the deal but they can buy something at a cheaper price in the sale part of the deal which amounts to ribā. 2.5 two deals cannot be combined into one the prophet (pbuh) prohibited “two deals in one”. some scholars take this hadith as a general prohibition of combining contracts. however, actually it implies that if more than one option is offered, one must be chosen before the deal is finalised. for example, if a seller offers a product for sale such as: (i) $1000 on cash payment basis; (ii) $1100 on payment after six months; (iii) $1500 on payment after one year etc., the buyer must choose one of the options and close the deal. the rationale of such prohibition is that it amounts to gharar which, as iqbal: islamic finance 8 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 stated above, is prohibited in business contracts. the seller will not know when (and hence how much money) he will receive. 2.6 right to profit is contingent upon taking responsibility (for loss) this is in line with the principle of justice to all parties. in the field of financial contracts, any condition which gives a party assured benefit without him taking any risk is not allowed. for example, in an interest-based loan contract the lender’s capital is guaranteed, yet he gets a positive return. all risk of loss is borne by the borrower. that is not allowed. as compared to this, in a muḍārabah contract, if a loss occurs the financier bears the monetary loss and the entrepreneurs loses any return for his efforts. in other words, the capitalist loses part of the money capital and the entrepreneur loses ‘sweat capital’. thus both parties share profit in case of profit and loss in case of loss. 3. islamic banking and finance: a different way of financial intermediation islam fully recognizes the useful role that financial intermediation can play value. historically, the role of a financial intermediary in the islamic economy is found in the principle of al-muḍārib udārib; a practice which has existed in islamic history since early centuries. it can be expressed as, “the one who mobilizes funds, on profit-sharing basis, can extend these funds to the users on the same basis”. similarly, in leasing, the lessee who possesses the usufruct, may sell these against a higher price (rent), and create additional value. in the early islamic period, most caravan trades were financed by muḍārabah (trust financing) and money transfer was quite common among businessmen. islamic scholars consider the earning of profits from an intermediary role as a genuine occupation. it is however, noticeable that such financial intermediation is interwoven with the production and exchange of real goods and services. the functions of islamic banks and other financial intermediaries are similar to their conventional counter-parts. islamic finance experts have shown that there are alternative islamic modes and models through which these functions can be performed. in fact, they have shown that islamic models can perform these functions in a better way. an islamic bank, like other banks, is a company whose main business is to mobilize funds from savers and supply these funds to businessmen/entrepreneurs. it is organized as a joint stock company with the shareholders supplying the initial capital. while a conventional bank uses the rate of interest for both obtaining funds from savers and supplying these funds to businessmen, an islamic bank performs these functions using various financial modes compatible with the sharī’ah. these will be described later in the paper. first we would like to present the most basic features of islamic finance. while islamic banks perform mostly the same functions as conventional banks, they do this in distinctly different manners. some of the distinguishing features of islamic banking are given below: international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 9 3.1 risk sharing the most important feature of islamic banking is that it promotes risk-sharing between the provider of funds (investor) and the user of funds (entrepreneur). by contrast, under conventional banking, the investor is assured a predetermined rate of interest. since the nature of this world is uncertain, the results of any project are not known with certainty ex-ante. therefore, there is always some risk involved. in conventional banking, all this risk is borne by the entrepreneur. whether the project succeeds and produces a profit or fails and produces a loss, the owner of capital gets away with a predetermined return.4 in islam, this kind of unjust distribution is not allowed. in islamic banking both the investor and the entrepreneur share the results of the project in an equitable way. in the case of profit, both share this in pre-agreed proportions. in the case of loss, all financial loss is borne by the capitalist and the entrepreneur loses his labour. askari, iqbal, krichene and mirakhor (2012) discuss in detail on what benefits risk/profit sharing model of financial intermediation brings to an economy. 3.2 emphasis on productivity relative to credit worthiness under conventional banking, almost all that matters to a bank is that its loan and the interest thereupon are paid on time. therefore, in granting loans, the dominant consideration is the credit-worthiness of the borrower and/or his providing sufficient collateral as a cover. under profit and loss sharing (pls) banking, the bank will receive a return only if the project succeeds and produces a profit. therefore, an islamic bank will be more concerned with the soundness of the project and the business acumen and managerial competence of the entrepreneur. this feature has important implications for the distribution of credit as well as the stability of the system. chapra (2000) lists the limitations/ harms of a collateral-based banking. 3.3 moral dimension conventional banking is secular in its orientation. in contrast, in the islamic system all economic agents have to work within the moral value system of islam. islamic banks are no exception. as such, they cannot finance any project which conflicts with the moral value system of islam. for example, they will not finance a wine factory, a casino, a night club or any other activity which is prohibited by islam or is known to be harmful to society. in this respect islamic banks are somewhat similar to “ethical funds” which are now becoming popular in the western world. 4 except in the extreme case of bankruptcy. iqbal: islamic finance 10 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 3.4 close linkages between monetary and real sectors in islamic model of financial intermediation, there is a close link between the monetary and real sectors of the economy. the financial products are either asset-backed or trade-linked. in this way ‘pure’ speculation is eliminated. this feature provides a safeguard against forming of artificial economic bubbles that often lead to economic and financial instability 3.5 wider set of products an important point to be noted in the way islamic banking works is that it offers a wider choice of products. in addition to some fixed-return modes that can serve necessarily the same functions that interest serves in conventional banking, islamic banks can use a variety of innovative profit-sharing financing techniques. the addition of various profit-sharing modes, which can be designed to meet the needs of different classes of risk-takers, to the menu of available financial products, renders several advantages, which will be noted later in the paper. 4. islamic financial products in this section we briefly explain the most commonly used financial products being used by islamic financial institutions and lately by corporations and governments. 4.1 mushārakah (partnership) mushārakah literally means sharing. in the islamic finance literature it refers to a joint enterprise in which all the partners share the profit or loss of the joint venture. the financial term is derived from the islamic legal term “shirkah” with the same literal meaning but having a broader application. in the islamic fiqh literature shirkah is of two kinds: the first is shirkat-ul-milk which means joint ownership of two or more persons of a particular property/asset which may come into existence either through inheritance or joint purchase. the second kind of shrikah is shirkat-ul-ʿaqd, which means a partnership established through a contract. such contractual partnerships are usually established for commercial purposes and take several forms such as partnership in the capital of the enterprise, partnership in labour and management, common goodwill or a combination of these elements. mushārakah as a financial contract refers to an arrangement where two or more parties establish a joint commercial enterprise and all contribute capital as well as labour and management as a general rule. the profit of the enterprise is shared among the partners in agreed proportions while the loss will have to be shared in strict proportion of capital contributions. the basic rules governing the mushārakah contract include: ʿ international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 11 i) profit of the enterprise can be distributed in any proportion by mutual consent. however, it is not permissible to fix a lump sum profit for anyone. ii) in case of loss, it has to be shared strictly in proportion to the capital contributions. iii) as a general rule all partners contribute both capital and management. however, it is possible for any partner to be exempted from contributing labour/management. in that case, the share of profit of the sleeping partner has to be in strict proportion of his capital contribution. iv) the liability of all the partners is unlimited. as a mode of finance, an islamic bank can advance money to a client using the contract of mushārakah. normally the bank will use the option of being a sleeping partner. the contract can be more widely used by islamic funds whereby the unit holders can assume the role of sleeping partners. the contract can also be used in securitized assets. 4.2 mudārabah (passive partnership) mudārabah is a special type of partnership. this is a contract between two parties: 5 a capital owner (called rabb al-māl) and an investment manager (called mudārib). profit is distributed between the two parties in accordance with the ratio that they agree upon at the time of the contract. financial loss is borne by the capital owner; the loss to the manager being the opportunity cost of his own labour, which fails to generate any income. except in the case of a violation of the agreement or default, the investment manager does not guarantee either the capital extended to him or any profit generation. some other important features of the mudārabah contract include: i) while the provider of capital can impose certain mutually agreed conditions on the manager, he has no right to interfere in the day-to-day work of the manager. ii) mudārabah is one of the fiduciary contracts. mudārib is expected to act with utmost honesty, otherwise he is considered to have committed a grave sin (in addition to worldly penalties). this has important implications for the moral hazard problem. iii) the liability of the rabb al-māl is limited to the extent of his contribution to the capital. iv) the muḍārib is not allowed to commit the mudārabah business for any sum greater than the capital contributed by the rabb al-māl. v) all normal expenses related to mudārabah business, but not the personal expenses of the mudārib, can be charged to the mudārabah account6. 5 there could be more than two parties. the contract is explained using a two-party example only for simplicity. 6 however, in case of travelling outside the place of business, the muḍārib is entitled to travel and living expenses during the trip. iqbal: islamic finance 12 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 vi) the contract of mudārabah can be terminated at any time by either of the two parties on giving a reasonable notice. (this condition may create serious problems in the context of modern commercial enterprises. however, in the light of the ‘doctrine of universal permissibility’ mentioned above, the parties can agree on any conditions in the contract that will regulate the termination so as not to cause any damage to the enterprise). vii) no profit distribution can take place (except as an ad hoc arrangement, and subject to final settlement), unless all liabilities have been settled and the equity of the rabb al-māl restored. as a mode of finance applied by islamic banks, on the liabilities side, the depositors serve as rabb al-māl and the bank as the mudarib. mudārabah deposits can be either general, which enter into a common pool, or restricted to a certain project or line of business. on the assets side, the bank serves as the rabb al-māl and the businessman as the mudārib (manager). however the manager is often allowed to mix the mudārabah capital with his own funds. in this case profit may be distributed in accordance with any ratio agreed upon between the two parties, but the loss must be borne in proportion to the capital provided by each of them. 4.3 diminishing partnership this is a contract between a financier (e.g., a bank) and a beneficiary in which the two agree to enter into a partnership to own an asset, as described above, but on the condition that the financier will gradually sell his share to the beneficiary at an agreed price and in accordance with an agreed schedule. 4.4 bay almurābahah (sales contract at a profit mark-up) in the classical fiqh literature, there is a sales contract called bay al-mu’ajjal which refers to sale of goods or property against deferred payment (either in lump sum or instalments). bay al-mu’ajjal need not have any reference to the profit margin that the supplier may earn. its essential element that distinguishes it from cash sales is that the payment is deferred. strictly speaking, the deferred payment can be higher than, equal to or lower than the cash price. there is another sale contract known as bay al-murābaḥah, which refers to a sale in which the seller declares his actual cost and the parties agree on adding a specific profit margin. basically, this is a two party buying and selling contract. no financial intermediation is involved. the islamic banks have created a mode of finance by combining the concepts of bay al-mu’ajjal and bay al-murābahah. they use this contract as a mode of finance in the following manner. the client orders an islamic bank to purchase for him a certain commodity at a specific cash price, promising to purchase such commodity from the bank once it has been bought, but at a deferred price, which includes an agreed upon profit margin called mark-up in favour of the bank. thus, the international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 13 transaction involves an order accompanied by a promise to purchase and two sales contracts. the first contract is concluded between the islamic bank and the supplier of the commodity. the second is concluded between the bank and the client who placed the order, after the bank has possessed the commodity, but at a deferred price, that includes a mark-up. the deferred price may be paid as a lump sum or in instalments. in the contract between the islamic bank and the supplier, the bank often appoints the person placing the order (the ultimate purchaser) as its agent to receive the goods purchased by the bank. the basic rules governing the murābahah contract include: i) the subject of sale must exist at the time of sale. ii) the subject of sale must be in the ownership of the seller at the time of sale. iii) the subject of sale must be in the physical or constructive possession of the seller. iv) the delivery of the sold commodity to the buyer must be certain and should not depend on a contingency or chance. v) as in any sales contract the price must be specified and once specified it cannot be increased in case of default. vi) the time of delivery must be specified. vii) the payments schedule must be specified. 4.5 ijārah (leasing) in the simple lease contract the usufruct generated over time by an asset, such as machinery, airplanes, ships or trains is sold to the lessee at a predetermined price. this is called an operating lease, as opposed to a financial lease. the operating lease has a number of features that distinguish it from other forms of leasing. firstly, the lessor is himself the real owner of the leased asset and, therefore, bears all the risks and responsibilities of ownership. all defects, which prevent the use of the equipment by the lessee, are his responsibility, even though it is possible to make the lessee responsible for the day-to-day maintenance and normal repairs of the leased asset. secondly, the lease is not for the entire useful life of the leased asset but rather for a specified short-term period (for a month, a quarter, or a year) unless renewed by mutual consent of both the parties. 4.6 ijārah wa iqtina’ (lease ending in the purchase of the leased asset) since the entire risk is borne by the lessor in the operating lease, there is a danger of misuse of the leased asset by the lessee. the financial lease helps taking care of this problem by making the lease period long enough (usually the entire useful life of the leased asset), to enable the lessor to amortize the cost of the asset with profit. at the end of the lease period the lessee has the option to purchase the asset from the lessor at its market value at that time. the lease is not cancellable before the expiry of the lease period without the consent of both the parties. there is, therefore, little danger of misuse of the asset. iqbal: islamic finance 14 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 a financial lease has other advantages too. the leased asset serves as a security and, in case of default on the part of the lessee, the lessor can take possession of the equipment without court order. it also helps reduce the lessor’s tax liability due to the high depreciation allowances generally allowed by tax laws in many countries. the lessor can also sell the equipment during the lease period such that the lease payments accrue to the new buyer7. this enables the lessor to get cash when he needs liquidity. this is not possible in the case of a debt because, while the sharī’ah allows the sale of physical assets, it does not allow the sale of monetary debts except at their nominal value. some of the jurists have expressed doubts about the permissibility of financial leases. the rationale they give is that the long-term and non-cancellable nature of the lease contract shifts the entire risk to the lessee, particularly if the ‘residual’ value of the asset is also fixed in advance. the end result for the lessee may turn out to be worse than the outright purchase of the asset through an interest-bearing loan. a financial lease has thus the potential of becoming more exploitative than outright purchase. suppose the lease contract is for five years. the lessee would have to continue making lease payments even if he does not need the asset, say, after two years. in the case of a purchase through an interest-bearing loan, the purchaser can sell the asset in the market and repay the loan, thus reducing his loss. this cannot be done in a financial lease. if a party is unable to make lease payments, it may lose its stake in the asset even though part payment of the asset price has been made beyond the rental charge he would normally pay in an operating lease. however, there are jurists who consider financial leases to be permissible if certain conditions are satisfied. firstly, the lessor must bear the risks of leasing by being the real owner of the leased asset. the lessor cannot lease what he does not own and possess, and should be responsible for all the risks and responsibilities related to ownership. therefore, a leasing contract where the lessor acts only as an intermediary between the supplier and the lessee and plays the role of only a financier, with ownership of the asset being nothing more than a legal device to provide security for repayment of the loan and legal protection in case of default, is not allowed. in this case the lessor leases an asset before buying and taking possession and gets a reward without bearing any risk. secondly, lease payments cannot start until the lessee has actually received possession of the leased asset and can continue only as long as it remains usable by him. thirdly, all manufacturing defects and later damages which are beyond the control of the lessee, should be the lessor’s responsibility8. the lessee can, however, be made responsible for the proper upkeep and maintenance of the leased asset. as a form of financing used by islamic banks in practice, the contract takes the form of an order by a client to the bank, requesting the bank to purchase a piece of equipment, promising, at the same time, to lease it from the bank after 7 the new buyer has to agree to continue the lease on the conditions previously agreed unless the lessee willingly agrees to new conditions. 8 some of these can be insured against, but this has to be done by the lessor at his own cost. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 15 it has been purchased. rent instalments are calculated in such a manner as to include, in reality, recovery of the cost of the asset plus the desired profit margin. thus, this mode of financing includes a purchase order, a promise to lease, and a leasing contract with a provision to transfer ownership of the leased asset to the lessee at the end of the lease agreement. this transfer of ownership is made through a new contract, in which the leased asset is either given to the lessee as a gift or is sold to him at a nominal price at the end of the lease agreement. according to a decision of the international islamic fiqh academy, this second transfer-of-ownership contract should be signed only after termination of the lease term, on the basis of an advance promise to affect such a transfer of ownership to the lessee. 4.7 bay alsalam bay al-salam (usually written in short form as salam) is a sales contract in which the price is paid in advance at the time of contracting, against delivery of the purchased goods/services at a specified future date. not every commodity is suitable for a salam contract. it is usually applied only to fungible commodities. some basic rules governing the salam sale are given below: i) the price should be paid in full at the time of the contract. ii) goods whose quality or quantity cannot be determined by specification cannot be sold through the contract of salam. an example is precious stones. iii) goods can be sold only by specifying the attributes. they cannot be particularized to a given farm, factory or area. iv) the exact date and place of delivery must also be specified. islamic banks can provide financing by way of a salam contract by entering into two separate salam contracts, or one salam contract and an instalments sale contract. for example, the bank could buy a commodity by making an advance payment to the supplier and fixing the date of delivery as the date desired by its client. it can then sell the commodity to a third party either on a salam or instalments sale basis. if the two were salam contracts, the second contract would be for delivery of the same quantity, description, etc., as that constituting the subject-matter of the first salam contract. this second contract is often concluded after the first contract, as its price has to be paid immediately upon conclusion of the contract. to be valid from the sharī’ah point of view, the second contract must be independent, i.e., not linked to the delivery in the first contract. should the second contract consist of an instalments sale, its date should be subsequent to the date on which the bank would receive the commodity. 4.8 bay al-istisnah (contract to manufacture) and bay alistisnah al-tamwilī (financing by way of istisnah ) bay al-istisna (usually written in short form as istisna) is a contract in which a party orders another to manufacture and provide a commodity, the description of iqbal: islamic finance 16 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 which, delivery date, price and payment date are all set in the contract. any party can cancel the contract after giving a notice to the other before the manufacturing work starts. however, after the manufacturing work has started, the contract cannot be cancelled unilaterally. istisna is similar to salam in the sense that both are exceptions to some general conditions of sale which prohibit selling of something which is not owned and is not in the possession of the seller at the time of sale. however, there are some differences between the two which are summarized below: i) the subject of istisnah is always a thing which needs manufacturing, while salam can also be effected on things that do not involve manufacturing. ii) in the case of salam full payment of price is necessary whereas in case of istisnah the payment can be delayed. iii) the time of delivery in case of salam must be specified at the time of the contract. in the case of istisnah this is not necessary. al-istisnah al-tamwilī, which is used by islamic banks, consists of two separate istisnah contracts. the first is concluded between the beneficiary and the bank, in which the price is payable by the purchaser in future, in agreed instalments and the bank undertakes to deliver the requested manufactured commodity at an agreed time. the second istisnah contract is a subcontract concluded between the bank and a contractor to manufacture the product according to prescribed specifications. the bank would normally pay the price in advance or during the manufacturing process in instalments. the latter undertakes to deliver the product to the bank on the date prescribed in the contract, which is the same date as that stated in the first istisnah contract. the original purchaser (i.e., the bank’s client) may be authorized to receive the manufactured commodity directly from the manufacturer. 4.9 wakālah (agentship) wakālah is a contract whereby someone appoints another to act in/pursue some matter on his behalf. it is permissible to charge a fee for providing such services. in islamic finance, the contract is being used extensively in the area of fund management. under this contract, an islamic bank serves as an investment manager for some clients. usually the bank establishes a mutual fund in which clients buy units. the fund promoter (bank) is the wakīl (agent) of the unit holders and charges a fixed fee for its services. the profit or loss is passed on to the fund providers after deducting bank’s fees. it is worth noting that mutual funds are different from mudārabah accounts in the islamic banks. the basic difference between the two is that in the latter case the bank is mudārib and hence shares in the risk of investment. in case of investment funds, all risk (loss) is borne by the unit holders. for the same reason, the fund promoters have no share in profits, all of which after deducting fixed management fees of the promoters, is passed on to the unit holders. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 17 4.10 tawarruq in many cases the clients would need liquidity which cannot be linked to purchase of commodities. conventional financial system has the capability of providing cash for those who want it, provided they are willing to give back more than they borrow and are able to provide a collateral or guarantee. islamic financial system does not have this facility. recently, some banks have started using a new instrument called tawarruq. in the classic fiqh literature there is an instrument called tawarruq for obtaining cash in case of difficulties facing an individual. it works like this. suppose a asks b for a loan. b says that i do not have any cash but i can lend you one bushel of wheat. a takes that loan in kind, sells it in the market and gets the cash that he needed. at the time of repayment, a goes to the market buys a bushel of wheat of the same quality and returns it to a. in essence, this is an arrangement of loan in kind. since one of the conditions of loan is that the ownership must be transferred to the borrower, he can do whatever he wants with the borrowed commodity. thus his selling it for cash is legitimate. since what he returns after the period of the loan is exactly the quantity borrowed, there is no question of ribā in this kind of deal. hence, the arrangement is permissible. however, the way this instrument is being practiced by banks is a lot more different. it appears to be simply a ‘devious artifice’ (ilah) to go around prohibition of interest through an intermediate process, the end result being what was prohibited. say, a goes to the bank for obtaining cash on loan. on the request of the client, the bank buys, at a credit price, on client’s behalf some commodities on paper. in order to do that a signs a wakālah contract appointing the bank as his agent for this purchase. bank certifies that it has done so at, say a credit price of $1.0 million due after one year. a becomes owner of these commodities. he now signs another wakālah contract, this time appointing the bank as his agent to sell those commodities. bank certifies that it has done so at a cash price of $900,000 and pays this amount to a. all of these transactions are only on paper. it can easily be seen that it is money obtained now ($900,000) for more money to be paid later ($1.0 million). the real asset involved in the process is no more than a dummy. a single piece of real asset can form the basis of innumerable successive tawarruq deals. there are a few scholars who have allowed tawwaruq argued on the basis of the example given above in the classic fiqh literature. however, one can easily see that the two are far apart. therefore, even those scholars place some restrictions to be complied with, if this instrument is to be declared sharī’ah-compatible. basically, they require the banks to actually buy and take at least constructive possession of the commodities involved and then sell them. according to my information almost all of the banks using this instrument are not complying with those conditions. hence, i believe that it is only a subterfuge for a ribā transaction. iqbal: islamic finance 18 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 4.11 bay al-dayn (sale of debt) many islamic instruments, the most prominent example being bay al-murābahah, give rise to debt obligations. can such debt be sold at a discount in order to obtain immediate cash? according to a majority of islamic scholars, sale of debt is not permissible except at face value. the traditional muslim jurists are unanimous on the point that sale of debt is possible only on its face value. 9 discounting debt is not allowed in sharī’ah. the overwhelming majority of the contemporary muslim scholars are of the same view. however, some scholars of malaysia have allowed selling debt at a discounted price. they refer to a ruling of shafaī school wherein it is held that the sale of debt is allowed. they did not pay attention to the fact that the shafaī jurists have allowed it only in a case where the debt is sold at its par value. in fact, the prohibition of bay al-dayn is a logical consequence of the prohibition of ribā. a ‘debt’ receivable in monetary terms corresponds to money, and every transaction where money is exchanged for the same type of money, the exchange must be at par value. any increase or decrease from one side is tantamount to ribā. some scholars argue that the permissibility of bayal-dayn is restricted to a case where the debt is created through the sale of a commodity. in this case, they say, the debt represents the sold commodity and its sale may be taken as the sale of a commodity. such an argument is not correct. for, once the commodity is sold, its ownership is passed on to the purchaser and it is no longer owned by the seller. what the seller owns is nothing other than money due. therefore, if he sells the debt, it is no more than the sale of money and it cannot be termed as the sale of the commodity. that is why this view has not been accepted by the overwhelming majority of the contemporary scholars. the international islamic fiqh academy of jeddah, which is the largest representative body of sharī’ah scholars and has the representation of all the muslim countries, including malaysia, has unanimously upheld the prohibition of bay al-dayn at a discount.10 4.12 ju ālah ju ālah is a contract whereby one party (offerer) undertakes to pay a prescribed fee to the second party (offeree) for performing a given task. for example a may offer to b that if you build this wall, i will pay you $100. the contract of ju ālah could have a useful role in the area of issuing of letters of guarantee by islamic banks. however, the classical fiqh literature is almost unanimous to the point that the guarantee is a charitable transaction and no fee can be charged on a guarantee. the most the guarantor can do is to claim his actual secretarial expenses incurred in offering the guarantee. usmani (2004, pp.130-131) notes that some contemporary scholars are considering the problem from a different angle. they feel that guarantee has become a necessity, especially in international 9 see usmani (2004), p.217. 10 see the international islamic fiqh academy, jeddah, resolution no. 101 (4/11), 1998. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 19 (c on ti nu ed )trade where the sellers and the buyers do not know each other, and the payment of the price by the purchaser cannot be simultaneous with the supply of the goods. there has to be an intermediary who can guarantee the payment. it is utterly difficult to find the guarantors who can provide this service free of charge in required numbers. keeping these realities in view, some sharī’ah scholars of our time say that the prohibition of guarantee fee is not based on any specific injunction of the holy quran or the sunnah of the prophet. it has been deduced from the prohibition of ribā as one of its ancillary consequences. moreover, guarantees in the past were of simple nature. in today’s commercial activities, the guarantor has to bear some costs for giving the guarantee. therefore, they opine, that the prohibition of guarantee fee should be reviewed in this perspective. in practice, islamic banks are using this contract for providing letters of credit and letters of guarantee to their clients. 4.13 sukūk sakk (singular of sukūk) literally means cheque or promissory note for receivable. technically sukūk refer to financial instruments meant to mobilize resources from the market based on the strength of one’s balance sheet, credentials, track record, goodwill and prospects of the proposed project. they are meant to provide an islamic alternative to conventional bonds. sukūk can play a positive role in mobilization of savings on a vast scale. they benefit investors as well as those who have projects to finance that bear the promise of eventually generating sufficient revenue to meet the costs yet leave a surplus. their proliferation increases the efficiency of the financial system. also, they are capable of meeting credit needs of government and businesses in a manner that keeps credit supply linked with real assets. financial engineers have come up with fancy, names for various types11 where ‘engineering’ is more in names than in substance. the basic ideas are quite simple. sukūk are basically, certificates based on ownership of certain assets. generally, these certificates are negotiable in secondary markets. they represent ‘ownership’ in the assets underlying the issue. those with variable returns are based on mudārabah or mushārakah. there are also sukūk with predetermined, fixed income streams. the most popular of these is the one based on ijārah, (lease). there are sukūk based on salam or istisnah contracts. also there are hybrid issues whose underlying assets are mixtures of these. murābahah receivables being debt obligations are not considered fit for sukūk issue. but they have been accepted in such a mixture as long as they are in a minority. due to this last point, while sukūk offer a usefully potential mechanism for secondary market resource mobilization, they also open the way for sale of debt receivables (as minority share in a general sukūk issue). since, the sale of debt except at its face value is not generally acceptable by islamic jurists, the use of sukūk where debt receivable are a noticeable proportion, remain suspect from a sharī’ah point of view. 11 for various types of sukūk in vogue, please see the paper of meysam safari in this issue. iqbal: islamic finance 20 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 as ariff (2012) points out, sukūk have several advantages over conventional bonds, for which ṣ ukūk provide an alternative. these include: • first, large firms with a sure profit track are not afraid of sharing part of their profits with long term sukūk investors. these investors will not sell off at the first signs of bad news because of the enshrined risk-sharing design of contracts. these fundamental differences in the contract structures lead to a new capital market product called the sukūk certificate, which was first issued for public trading in 2000. • second, while there is transfer of assets to lenders, the assets are going to be returned to the borrower once debt is repaid, a strong incentive to limit debt. • third, because these investors want to avoid dealing in usury and interests (so it becomes an islamic finance product), they are known to hold on to their security until maturity much more than would common bond investors. studies also show that, over its 12-year history, the yield is higher by a margin of about 10-40 basis points in low risk cases (government debt) and about 110 basis points in high risk cases (corporate debt). profit sharing is only slightly more expensive than a one-sided contract that requires a no-default coupon payment with no regard to the outcome of investments! 5. combination of products in one contract most modern contracts are hybrid contracts in which more than one product and/ or their basic terms and conditions, are incorporated. it is, therefore, important to explain the rules governing such combinations. as standalone contracts, each component may be permissible but the combination as a whole may have the end result similar to a prohibited contract. in islamic law, it is not only the words of the contracts that are important. the spirit of the law must be reflected in the end result. many qurãnic verses and ahādīth heap disdain on attempts to avoid the spirit of the law through subterfuges or going around the bush. in general the following principles should be observed in designing any contract structure to be sharī’ah-compliant. 5.1 the combination must not contradict an explicit text if there is an explicit text in the primary sources of islamic law that certain types of contracts cannot be combined, for whatever reason, then any structure that involves such a combination becomes unacceptable. if analysed, the rationale of prohibition can also be found. some examples are: i) it is prohibited to combine a loan contract with a sales contract. rationale: if allowed, a lender may advance the loan on an interest-free basis in the loan part of the deal but they can buy something at a cheaper price in the sale part of the deal which amounts to ribā. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 21 ii) the prophet prohibited “two deals in one”. some scholars take this hādīth as a general prohibition of combining contracts. however, actually it implies that if more than one option is offered, one must be chosen before the deal is finalised. rationale: such combination amounts to gharar in the combined contract, which is prohibited in business contracts. 5.2 not trying to go around the bush a product structured on the basis of a combination of contracts should not be intended to circumvent impermissible transactions. such attempts are called ḥ iyal (legal artifices) and are not generally allowed. for example, the prophet (pbuh) prohibited bay ʿ al-inah, which is a sale and buy-back arrangement. for example, a sells his house to b for a cash price of us$500,000 and simultaneously buys it back from b at a credit price of us$600,000. it is easy to see that the end result is exchange of $500,000 now for us$600,000 later. this is nothing but ribā. the “selling” and “buying back” of the house is inconsequential. the recent ‘sukūk debacle’ was a result of similar commitment. 5.3 the combination must not involve contradictory contracts/conditions each type of contract has unique legal implications and obligations. contracts which are mutually contradictory cannot be combined. if the legal consequences do not conflict each other then there is no problem in combining them. for example, a collateral condition supports a loan contract and is hence acceptable. on the other hand, a condition in a marriage contract stipulating that the spouses will live apart, defeats the purpose of the contract and is hence not acceptable. 5.4 the combination must not involve contingent contracts the prophet (pbuh) has prohibited a sale that is circumscribed with a condition (bay wa shar). if the combination is such that the execution of one contract is contingent upon another contract, such combination is not permissible. for example, william saying to peter that i sell my house to you provided nancy rents her house to me. 6. summary and conclusions even though islamic banks emerged in response to the market needs of muslim clients, they are not religious institutions. like other banks, they are profit seeking institutions. while it is the preferred way of banking for one fifth of humanity, it offers a wider choice of financial products to all. several potential benefits can arise from the emergence of islamic banking. these include: i) the range of contracts available to savers and entrepreneurs is widened. the menu ranges from low risk trade-linked products to high risksharing contracts. iqbal: islamic finance 22 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 ii) the financial system is enriched by the establishment of financial institutions with different modus operandi. this diversity enhances the stability of the financial system because the behavioral characteristics of different types of banks are likely to vary. iii) competition among alternative banking models is expected to increase the efficiency of the financial system. iv) the financial needs of muslims can be met in accordance with their faith. since the public’s acceptance of the services provided by the industry play a vital role in creating stable and efficient markets, plurality and inclusiveness are important for the development of financial markets. v) the allocation of financial resources on the basis of profit-and-loss sharing (pls) gives maximum weight to the profitability of investment as compared with credit worthiness in the conventional system. such allocation of resources is expected to be more efficient than that on the basis of interest. vi) as a result of pls contracts, the liabilities side of the balance sheet tends to become symmetrical with the assets side. this helps making islamic banks less vulnerable to external shocks and insolvency. vii) the liability to share bank losses by investment depositors motivates them to be more vigilant about the operations of their banks and to demand greater transparency and more effective audit. banks are also under pressure to evaluate their clients’ projects more carefully and to monitor the risks more effectively. viii) since in the case of both profit-sharing and sale-based contracts, bank assets are created in response to investment opportunities in the real sector of the economy, and all financing is linked to commodities or assets, the real factors related to the production of goods and services (rather than speculative maneuvers) become the prime determinants of the rates of return. ix) debt creation in islamic finance is generally not possible without the backing of goods and services. monetary expansion would thus tend to take place in step with the growth of the real economy. this is expected to control inflationary pressures. x) destabilizing speculation would also be significantly curtailed as would the erratic and mass movement of short-term funds. that will provide a formidable protection against economic and financial instability. xi) the islamic alternative for interest-based bonds, ṣukūk, is a potent financial product that can meet the needs of the corporate and government sectors, but ensuring that their debts do not become unsustainable. xii) like ‘ethical funds’, islamic banks do not provide finance to projects considered socially undesirable. that introduces greater social responsibility. in brief, islamic banking should contribute to greater allocative efficiency, market discipline, financial stability and social responsibility. in a world beset with financial crises, these strengths inherent in islamic banking ṣ international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 islamic finance: an attractice new way of financial intermediation: 1-24 23 should offer a new ray of hope for achieving the cherished goal of systemic stability. the experience of the last 50 years has shown that islamic banking is a viable, dependable and well-supervised activity. it is just another way of performing the financial intermediation function. islamic banks are subject to the same regulatory standards and fall under the purview of the same supervisory authorities as conventional banks. in addition they have to comply with the islamic rules which mean that their overall legal and supervisory environment is more stringent. in brief, islamic banking adds a healthy dimension to the international financial system. author information: munawar iqbal, phd, is a professor at the institute of islamic economics, king abdul aziz university, saudi arabia. he is a founder of the international association of islamic economics as well as first editor of two professional journals, the review of islamic economics and the islamic economic studies. he has taught and researched on this topic for several decades, having published more than forty books/articles. he lectures and conducts training in islamic finance throughout the world. the author wishes to record his gratitude to the two anonymous referees and the editors of the journal for providing review comments on an earlier version of this paper. munawar may be contacted at: e-mail: munawariqbal@gmail.com. references ariff, m. (2012). the search for a better debt instrument. october, 24. retrieved from elgarblog.wordpress.com ariff, m., iqbal, m. & mohamed, s. (eds). (2012). the islamic debt market for sukuk securities: the theory and practice of profit sharing investment. new york, us and cheltenham, u.k: edward elgar. askari, h., iqbal, z., krichene, n. & mirakhor, a. (2012). risk sharing in finance: the islamic finance alternative. singapore: john wiley and sons. umer chapra, m. (2000). international financial stability: the role of islamic finance. policy perspectives, vol. 4, no. 2. darir alsiddiq. (2012). gharar: impact on contracts in islamic fiqh, jeddah: dallah al-barakah. iqbal, m. & llewellyn, d. t. (2002). islamic banking and finance: new perspectives on profit sharing and risk. new york, us and cheltenham, u.k: edward elgar. iqbal, m., & molyneux, p. (2005). thirty years of islamic banking: history, performance and prospects. houndsmill, uk and new york, usa: palgrave mcmillan. iqbal, m. (2013). determining the role of debt in the economy and a new approach for solving sovereign debt crises. international journal of banking and finance, 10(1), 99-130. iqbal: islamic finance 24 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 mills, p. s. & presley, j. r. (1999). islamic finance: theory and practice. london: macmillan. allais, m. (1993). the monetary conditions of an economy of markets: from the teachings of the past to the reforms of tomorrow. jeddah: islamic research and training institute. reinhart, c. m. & rogoff, k. s. (2009). this time is different: eight centuries of financial folly. princeton, nj: princeton university press. usmani, m. t. (2004). an introduction to islamic finance. karachi: maktaba ma’ariful quran. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 4 cross-border mergers and acquisitions between industrialized and developing countries: us and indian merger activity the international journal of banking and finance, vol. 8, number 1, march 2011: 35-58 35 cross-border mergers and acquisitions between industrialized and developing countries: us and indian merger activity gordon v. karels, edward lawrence and jin yu university of nebraska-lincoln, florida international university and st. cloud state university, us ________________________________________________________________ abstract in this paper we study the cross border mergers and acquisition between the us. and indian fi rms. our empirical work suggests that us fi rms realize signifi cant losses on the announcement of acquisitions of indian targets while indian targets realize signifi cant gains on the announcement of mergers with us acquirers. publicly-traded indian fi rms realize insignifi cant returns on their announcement of acquisitions of publicly-traded us fi rms but realize signifi cant positive returns on announcements of acquisitions of privately-held us fi rms and subsidiary fi rm targets. publicly-traded us targets realize insignifi cant gains when us acquired by indian fi rms. keywords: cross border mergers and acquisitions, acquirer’s abnormal returns, target’s abnormal returns, publicly traded acquirers, privately held targets jel classifi cation: g34, g14 ________________________________________________________________ 1. introduction much of the current research on cross-border mergers focuses on analyzing the returns of industrialized fi rm’s acquisitions of emerging world targets.1 little attention has been given to acquisitions of developed country targets by emerging world country fi rms. also, researchers have focused primarily on the returns to the acquirers and none of the studies to date have investigated the effect of announcement of mergers and acquisitions on the target fi rms in the cross border ijbf 1 we use “emerging” and “developing” interchangeably and “industrialized” and “developed” interchangeably throughout the paper. karels et al.: cross-border mergers and acquisitions 36 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 mergers and acquisitions involving developed and developing world fi rms. we study both types of cross-border merger activities by examining the indian and us company returns when us fi rms are acquirers of indian fi rms and when indian fi rms are acquirers of us fi rms.2 we fi nd that there is no country effect when a publicly-traded fi rm acquires a publicly-traded target or subsidiary fi rm target but there is a signifi cant country effect for public-traded acquirers of privately-held targets us (acquirers earn signifi cantly lower returns than indian acquirers). our investigation for the target fi rms reveals that acquisition by any us fi rm (public, private or subsidiary fi rm) is benefi cial to the shareholders of the indian targets whereas the us public targets earn insignifi cant returns on the announcement of their acquisition by indian fi rms. we also fi nd that the abnormal returns for the targets of publicly-traded acquirers are consistently higher than the abnormal returns of targets of privately-held acquirers. figure 1 shows the dramatic increase over the last few years in mergers and acquisitions between the us and indian companies. prior to 1995 there were few cross-border mergers and acquisitions between the us and indian fi rms. post 1995, us acquisitions of indian targets reached more than 100 fi rms in 2000, then declined rapidly following the dot.com bubble burst. by 2006, us acquisitions of indian targets were again close to 100 fi rms. indian acquisitions of us fi rms followed a similar pattern with increases up to the year 2000 (but only about 50 transactions), then a decline following the dot.com bubble burst. by 2006, the number of transactions had risen above the year 2000 level. in figure 2, we separate the cross-border mergers and acquisitions by types of ownership structure – publicly-traded, privately-held or non-traded subsidiary. the fi gure shows that indian acquirers of us fi rms are largely publicly-traded fi rms whereas the us targets are typically privately-held fi rms. mergers and acquisitions between us acquirers and indian targets are spread across the publicly-traded, privately-held and subsidiary fi rms for both acquirers and targets. conn, cosh, guest and hughes (2005) assert that there are important theoretical reasons why the acquisition of domestic and cross-border targets may differ and why acquisition of private targets may differ from the acquisition of public targets.3 2 the economic environment and the business environment of countries play a crucial role in the decision to acquire or being acquired in cross-border merger and acquisitions (la porta, lopez, shleifer & vishny, 1998, 1999, 2000). we cannot get meaningful information by studying all the developed world and the developing world countries together. doing the merger and acquisition study for all the developing world countries separately in one paper would make the paper too voluminous. hence we limit the scope of this paper to mergers and acquisitions in india and between india and the us and leave the mergers and acquisition in the remaining developing countries as a topic for future research. 3 we refrain from the discussion as to why cross border mergers and acquisitions and why acquisition of private targets is different from the acquisition of public fi rms. those issues are discussed at length in section 2 of conn et al. (2005). international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 37 figure 1: number of mergers and acquisitions, 1995 to 2006 source: sdc mergers and acquisitions database figure 1 shows an increasing trend in the number of mergers and acquisitions between u.s. and indian fi rms. accordingly, we examine mergers and acquisitions between the us and indian fi rms by separating them into different pairings: publicly-traded acquirers and publicly-traded targets, publicly-traded acquirers and privately-held targets, publicly-traded acquirers and subsidiary fi rms, privately-held acquirers and publicly-traded targets and subsidiary fi rm acquirers and publicly-traded target fi rms. in the following fi gure we categorize the mergers and acquisitions by publicly-traded fi rms, by privately-held fi rms, by subsidiaries of large fi rms and by others (government fi rms, joint ventures etc.). indian acquirers of u.s. fi rms are largely public and private fi rms but the u.s. targets are largely privately held and subsidiaries of large fi rms. mergers between the u.s. acquirers and indian targets are spread across publicly-traded, privately-held and subsidiary fi rms for both acquirers and targets. there is substantial literature on the announcement effects of crossborder acquisitions on acquiring fi rms in developed countries such as the us and the uk but there is limited work that has been done examining mergers and acquisition between the acquirers and the targets of a developing country (such as india) and an industrialized country (such as the us). in cross-border mergers and acquisition studies involving acquirers, moeller and schlingemann (2005) have one target indian fi rm; francis, hasan and sun (2008) have 13 target indian fi rms; and chari, ouimet and tesar (2010) have 33 target indian fi rms in their samples. using a comprehensive sample of 248 fi rms, in this paper we study the announcement effect of mergers and acquisitions on the acquirers of indian 0 20 40 60 80 100 120 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 #s o f m er ge rs a nd a cq ui si tio ns us target & indian acquirer indian target & us acquirer karels et al.: cross-border mergers and acquisitions 38 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 targets. also, none of the studies done to date have studied the announcement effect of mergers and acquisitions on the target shareholders of the fi rms from the developing world. conn et al. (2005) stated that despite the increase in the acquisition of cross-border public and non-public targets, nearly all acquisition studies were limited to domestic targets which were publicly traded. they studied the acquisitions of both domestic and cross-border targets that were public and private but they limited their study to the effect of acquisition on uk acquirers only. utilizing merger and acquisition data from 1995 to 2007 we address the announcement effect on publicly-traded acquirers of publicly-traded, privatelyheld and subsidiary targets, and on publicly-traded targets by privately-held fi rms for us acquirers and indian targets and indian acquirers of us targets. figure 2: types of mergers and acquisitions source. sdc mergers and acquisitions database we fi nd that indian target fi rms when acquired by a publicly-traded us fi rm, realize positive abnormal returns around the announcement date whereas acquiring fi rms suffer a loss of market value around the announcement date. indian targets acquired by privately-held us fi rms enjoy signifi cant gains on the announcement of a merger or acquisition. in the case of an indian acquisition of a us target, the acquiring indian fi rm experiences positive abnormal returns around the announcement date. the returns to an indian acquirer of a publiclytraded us fi rm are statistically insignifi cant whereas indian acquirers realize signifi cant abnormal returns around the announcement date of the acquisition of privately-held us fi rms and the subsidiaries of us fi rms. us targets realize statistically insignifi cant returns on the announcement of a merger or acquisition by indian fi rms. 0 50 100 150 200 250 300 indian acquirer us target us acquirer indian target us target & indian acquirer indian target & us acquirer #s o f m er ge rs a nd a cq ui si tio ns private public subsidiary others international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 39 our results indicate that the shareholders of indian acquirers of privatelyheld us targets and us subsidiary fi rm targets gain from the announcement of an acquisition whereas the share prices of us acquirers of privately-held indian targets fall on the announcement of an acquisition.4 we investigated if this difference in cross-border mergers and acquisitions was due to fi rm characteristics or was due to the country environment. we examined the fi rm characteristics of the acquirers and found that indian and the us acquirers of privately-held targets have similar high tech status and similar diversifi cation levels. indian acquirers of us subsidiary targets have substantially greater high tech status and higher diversifi cation levels than the us acquirers of indian subsidiary targets. for all target categories, the relative size of targets is substantially smaller for us acquisitions. the market value of indian acquirers is also signifi cantly lower than the market value of us acquirers. both indian and us fi rms, acquire majority stakes in target fi rms but the average stake acquired is higher in private and subsidiary fi rm targets than in publicly-traded targets. to determine if the reaction to the news of an acquisition is due to fi rm specifi c characteristics, we regressed the three-day cumulative abnormal return around announcement (-1 day to +1 day after announcement) on fi rm specifi c characteristics and a dummy variable identifying the country of the acquirer. our results indicate that there is no infl uence of the country environment when a publicly-traded fi rm acquires a publiclytraded target or subsidiary fi rm target as the dummy variable for the country remains insignifi cant. the dummy variable is signifi cant for public-traded acquirers of privatelyheld targets indicating that the us acquirers earn signifi cantly lower returns than indian acquirers. our results differ from the recent research of chari, ouimet and tesar (2010) and gubbi, aulakh, ray, sarkar and chittor (2010). chari et al. (2010) studied the returns to the developed market acquirers of emerging market fi rms. they reported a positive and signifi cant abnormal return for the developed country acquirer of targets from emerging markets whereas we found a negative but insignifi cant abnormal return for the us acquirers of indian targets. they found that private targets were associated with signifi cantly higher announcement returns for acquirers from developed world countries whereas we found that us acquirers of private indian targets earned signifi cantly negative abnormal returns. gubbi et al. (2010) found that the indian acquirers of foreign targets (both developing and developed country fi rms) gained on the announcement of their mergers and acquisitions. however, when they regressed the abnormal returns on explanatory variables, the coeffi cient on private targets was positive and insignifi cant in one model (as in their table 4) and negative and insignifi cant in the other (in table 5). we found that indian acquirers of targets earned signifi cant, positive abnormal returns on the announcement of us acquisitions of targets but this gain was limited to the acquisition of private targets only. 4 our results are consistent with the fi ndings of moeller and schlingemann (2005) and denis, denis and yost (2002), who fi nd that cross-border m&as for acquirers decrease acquirers’ value. our results do not support the fi ndings of chari et al. (2010), francis, hasan and sun (2008), kiymaz (2004), doukas (1995), and doukas and travlos (1988), who fi nd that cross-border m&as are value enhancing for acquirers. karels et al.: cross-border mergers and acquisitions 40 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 the size of our sample allows us to avoid the grouping of targets and acquirers from different countries. whereas, the developed world countries are somewhat economically and culturally similar to each other there are large differences across the countries of emerging markets. studying cross-border mergers and acquisitions between representative developing and industrialized countries should provide less noisy information. the rest of the paper is organized as follows: section 2 reviews related literature and section 3 discusses the data sources and methodology. section 4 presents the empirical results. section 5 provides some of the probable cases for the gains/losses of the shareholders of the acquirers and the targets, and section 6 concludes. 2. literature review we summarize the literature on cross-border mergers and acquisitions in table 1; us targets gain signifi cantly when acquired by foreign fi rms, foreign acquirers of us fi rms gain signifi cantly whereas the us acquirers of foreign fi rms show mixed results. the fi rms in these studies are predominantly from industrialized countries. markides and ittner (1994) examined 276 u.s. international acquisitions made from 1975 to 1988 and found the two-day cumulative abnormal return [car (-1, 0)] for acquiring fi rms to be 0.32% (statistically signifi cant). most of the acquisitions in markides and ittner’s study came from the manufacturing and fi nance, and insurance industries and are from predominantly developed countries. markides and oyon (1998) used a sample of 236 us acquisitions consisting of 47 canadian targets and 189 european targets (32 from france, 81 from uk, 27 from germany, 13 from italy, 14 from spain, 15 from holland and 7 from belgium) and found that us acquisitions in europe generated signifi cant returns while returns for us acquisitions in uk and canada were not signifi cant. seth, song and pettit (2002) investigated 100 cross-border acquisitions between foreign acquirers and us targets during the time period 1981-1990. they found that the cumulative abnormal return [car (-10, 10)]: 10 days before and 10 days after the fi rst bid by the ultimately successful bidders) to the foreign bidder was 0.11 per cent.5 the acquirers of the 100 cross-border acquisitions in this study were mainly from industrialized countries great britain (52), japan (10), canada (10), australia (8), west germany (3), and switzerland (2). conn et al. (2005) studied 4,000 uk domestic and cross-border public and private acquisitions. for the acquirers of publicly traded targets they found signifi cant losses for domestic mergers and acquisitions but insignifi cant losses for cross-border mergers and acquisitions. they found signifi cant gains for acquirers of privately-held fi rms for both domestic and cross-border targets. 5 seth et al. (2002) did not study the returns for the targets. also, the statistical signifi cance of the car is not reported in their paper. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 41 t ab le 1 : s um m ar y of l it er at ur e on c ro ss -b or de r m er ge rs a nd a cq ui si ti on s c r o ss -b o r d e r s tu d y d a ta t im e p e r io d e v e n t w in d o w t a r g e t s h a r e h o ld e r s a c q u ir e r s h a r e h o ld e r s 1 d a y p ri o r to t h e a n n o u n c e m e n t d a te n o t s tu d ie d 0 .3 2 % g a in ( si g n if ic a n t) 1 0 d a y s p ri o r to 1 0 d a y s a ft e r a n n o u n c e m e n t d a te n o t s tu d ie d 0 .2 9 % g a in ( n o t si g n if ic a n t) s e th e t a l. ( 2 0 0 2 ) 1 0 0 , b e tw e e n f o re ig n a c q u ir e rs a n d u s t a rg e ts 1 9 8 1 t o 1 9 9 0 1 0 d a y s p ri o r to 1 0 d a y s a ft e r a n n o u n c e m e n t d a te 3 8 .3 % g a in ( 1 % ) 0 .1 1 % g a in ( in si g n if ic a n t) e u n , k o lo d n y a n d s c h e ra g a ( 1 9 9 6 ) 2 2 5 f o re ig n a c q u is it io n o f u s f ir m s 1 9 7 9 t o 1 9 9 0 5 d a y s b e fo re t o 5 d a y s a ft e r a n n o u n c e m e n t d a te 3 7 .0 2 % g a in (1 % ) 1 d a y p ri o r to a n n o u n c e m e n t d a te n o t s tu d ie d 0 .6 3 % g a in ( 1 % ) fo r fo re ig n a c q u ir e rs a n d 0 .3 6 % l o ss ( in si g n if ic a n t) f o r u s a c q u ir e rs 1 0 d a y s p ri o r to 1 0 d a y s a ft e r a n n o u n c e m e n t d a te n o t s tu d ie d 1 .9 6 % g a in ( 1 % ) fo r fo re ig n a c q u ir e rs a n d 0 .2 5 % l o ss ( in si g n if ic a n t) f o r u s a c q u ir e rs m a rk id e s a n d o y o n ( 1 9 9 8 ) 2 3 6 a c q u is it io n s b y u s c o m p a n ie s in e u ro p e a n d c a n a d a 1 9 7 5 t o 1 9 9 8 1 d a y p ri o r to a n n o u n c e m e n t d a te 0 .3 8 % g a in ( 5 % ) e c k b o a n d t h o rb u rn ( 2 0 0 0 ) 3 9 0 a c q u is it io n s o f c a n a d ia n t a rg e ts b y u s fi rm s 1 9 6 2 t o 1 9 8 3 d u ri n g t h e m o n th o f m e rg e r a n n o u n c e m e n t 3 .5 9 % g a in ( 1 % ) 0 .1 9 % l o ss ( in si g n if ic a n t) n o t s tu d ie d f o r p u b li c t a rg e ts : 0 .8 2 % l o ss (1 % ) fo r a ll t a rg e ts , 0 .9 9 % l o ss ( 1 % ) fo r d o m e st ic , 0 .0 9 % l o ss (i n si g n if ic a n t) f o r c ro ss b o rd e r n o t s tu d ie d f o r p ri v a te t a rg e ts : .8 6 % g a in ( 1 % ) fo r a ll t a rg e ts , 1 .0 5 % g a in ( 1 % ) fo r d o m e st ic a n d 0 .3 8 % g a in ( 5 % ) fo r c ro ss -b o rd e r ta rg e ts m o e ll e r a n d s c h li n g e m a n n ( 2 0 0 5 ) 4 4 3 0 d o m e st ic a n d c ro ss b o rd e r a c q u is it io n s b y u s a c q u ir e rs 1 9 8 5 t o 1 9 9 5 1 d a y p ri o r to 1 d a y a ft e r a n n o u n c e m e n t d a te n o t s tu d ie d f o r u s a c q u ir e rs o f th e c ro ss -b o rd e r sa m p le : 0 .3 0 7 % ( in si g n if ic a n t) ; fo r u s a c q u ir e rs o f th e d o m e st ic s a m p le 1 .1 7 3 % (1 % ) f ra n c is , h a sa n a n d s u n ( 2 0 0 8 ) 1 4 9 1 u s a c q u is it io n s o f fo re ig n f ir m s a n d 7 6 1 2 u s d o m e st ic a c q u is it io n s 1 9 9 0 t o 2 0 0 3 1 d a y p ri o r to 1 d a y a ft e r a n n o u n c e m e n t d a te n o t s tu d ie d f o r u s a c q u ir e rs o f th e c ro ss -b o rd e r sa m p le : 0 .9 6 % ( 1 % ); f o r u s a c q u ir e rs o f th e d o m e st ic s a m p le : 1 .4 9 % ( 1 % ) c h a ri , o u im e t a n d t e sa r (2 0 1 0 ) 5 9 4 d e v e lo p e d -m a rk e t a c q u is it io n s o f fi rm s in e m e rg in g m a rk e ts a n d 1 6 2 4 a c q u is it io n s o f fi rm s in d e v e lo p e d m a rk e ts . 1 9 8 6 t o 2 0 0 6 1 d a y p ri o r to 1 d a y a ft e r a n n o u n c e m e n t d a te n o t s tu d ie d f o r d e v e lo p e d -m a rk e t a c q u ir e rs o f fi rm s in d e v e lo p in g m a rk e ts : 1 .1 6 % (s ig n if ic a n t) ; in si g n if ic a n t re tu rn f o r d e v e lo p e d -m a rk e t a c q u ir e rs o f fi rm s in d e v e lo p e d m a rk e ts g u b b i, a u la k h , r a y , s a rk a r a n d c h it to r (2 0 1 0 ) 4 2 5 c ro ss -b o rd e r m e rg e rs a n d a c q u is it io n s b y i n d ia n fi rm s 2 0 0 0 t o 2 0 0 7 5 d a y s p ri o r to 5 d a y s a ft e r a n n o u n c e m e n t d a te n o t s tu d ie d 2 .5 8 % g a in ( 5 % ) m a rk in d e s a n d i tt n e r (1 9 9 4 ) 2 7 6 u s i n te rn a ti o n a l a c q u is it io n s 1 9 7 5 t o 1 9 8 8 c o n n e t a l. ( 2 0 0 5 ) 4 0 0 0 u k d o m e st ic a n d c ro ss b o rd e r p u b li c a n d p ri v a te t a rg e ts 1 9 8 4 t o 1 9 9 8 1 d a y p ri o r to 1 d a y a ft e r a n n o u n c e m e n t d a te c a k ic i, h e ss e l a n d t a n d o n ( 1 9 9 6 ) 1 9 5 f o re ig n f ir m s th a t a c q u ir e d u s f ir m s a n d 1 1 2 u s a c q u is it io n s o f fo re ig n fi rm s 1 9 8 3 t o 1 9 9 2 karels et al.: cross-border mergers and acquisitions 42 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 moeller and schlingemann (2005) studied the announcement effect on us acquirers for a sample of 383 cross-border transactions and 4047 domestic takeover transactions during 1985 to 1995. they found a three-day (-1, +1) market adjusted return of 0.307 per cent for cross-border acquirers and 1.173 per cent for domestic acquirers. in the moeller and schlingemann (2005) study, the uk is the most frequent target country (31%), followed by canada (21%), france (9%) and germany (9%). in their sample there is only one target fi rm from india. in a recent study, francis, hasan and sun (2008) used a sample of 1,491 foreign acquisitions by us fi rms and 7,692 us domestic acquisitions. they found that over the full sample period of 1990-2003, acquirers in domestic m&as experienced an average abnormal stock return of 1.49 per cent whereas the acquirers in cross-border m&as experienced an average abnormal return of 0.96%. among their 1,491 transactions, 1,275 (85.50%) were from integrated fi nancial markets, with the remaining 215 (14.50%) from segmented fi nancial markets. there were only 13 indian fi rms acquired by us fi rms in this sample. most merger and acquisition studies have focused on publicly-traded fi rms within the domestic us market or those between the us and other industrialized countries. little research has examined the returns to shareholders from crossborder mergers and acquisitions between the us and developing country fi rms.6 chari, ouimet and tesar (2010) fi lled this gap and examined the returns to shareholders of developed country fi rms that undertook acquisitions in emerging markets. they found that when developed country acquirers gained control of emerging-market targets, they experienced positive and signifi cant abnormal returns of 1.16 per cent, on average, over a three-day event window. though chari et al. (2010) studied the acquisition of targets from india by developed world fi rms there were only 33 target fi rms from india. in this paper we have a sample of 248 fi rms from the acquiring targets in india. chari et al. (2010) did not study the effect of announcement of mergers and acquisitions on target fi rms. also, they did not study the acquirers of targets from developed world fi rms (like us) by developing world fi rms (like india). research by gubbi, aulakh, ray, sarkar and chittor (2010) studied 425 foreign mergers and acquisitions by indian fi rms from january 2000 to december 2007. they found positive abnormal returns for the acquiring fi rm shareholders when combining all target fi rm countries (developing and emerging) into one group. they investigated if the abnormal returns for the acquirers of developed world targets were different than those for emerging world target and found that acquirers had statistically higher abnormal returns when the target fi rms were located in advanced economies. they also investigated if the returns for the acquirers of private targets were different from the returns for public targets but 6 francis, hasan and sun (2008) have only 14 per cent of their cross-border sample from developing countries and the moeller and schlingemann (2005) study has only 5.2 per cent of targets from developing countries. rossi and volpin (2004) examined the determinants of cross-board mergers and acquisitions from 1990 to 1999 with few of their target fi rms from developing countries. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 43 found no signifi cant results. in their table 4, they performed a cross-sectional regression on the abnormal return for the acquirers around announcement date and reported a negative and insignifi cant ‘private’ variable which was one for private targets and zero for public targets. in their table 5, gubbi et al. (2010) reported a positive an insignifi cant ‘private’ variable for a different cross sectional regression model. hence their fi ndings on the abnormal returns to the acquirers for different public/private stat were ambiguous. also, gubbi et al. (2010) did not study the announcement effect of mergers and acquisitions on the target fi rms. cross-border mergers and acquisitions can be benefi cial for the acquirers resulting in positive abnormal returns on the announcement date; they can also result in losses for the acquirers. researchers have given several reasons for the gains from cross-border mergers and acquisitions. caves (1971, 1998), morck and yeung (1991, 1992) and williamson (1979) point to the internalization benefi ts in the cross-border mergers and acquisitions. firms extract above normal returns from cross-border investments by internalizing the host country market imperfections when their fi rm specifi c assets cannot fi nd comparable values elsewhere. ayban and ficici (2009) note that the resulting rents derived from internalization are expected to be capitalized into a higher value of the fi rm. baldwin and caves (1991) point out that cross-border mergers and acquisitions may result in gains from diversifi cations when businesses seek synergies arising from intangible and information-based assets like brand names, technical knowledge and r&d expertise. according to kogut (1983) cross-border acquisitions may increase the operational fl exibility of the fi rm by giving it the opportunity to exploit market conditions. conn et al. (2005) state that in crossborder merger, geographical diversifi cation by direct investments in overseas subsidiary permits fi rms to expand the boundary of the fi rm. this will result in an increase in revenues for the fi rm. researchers give several reasons for the loss from cross-border mergers and acquisitions. conn et al. (2005) argue that overseas targets are more diffi cult to value accurately because of imperfect information. they point out that there are diffi culties of managing the post-merger process when cultural differences make integration and acculturation, a diffi cult, time-consuming and expensive process. the bigger the cultural gap, the bigger the relative size of the target the worse the problem may be. aybar and ficici (2009) point out that differences in natural culture, customer preferences, business practices and institutional forces may jeopardize the potential gains of cross-border mergers and acquisitions. hitt, hoskissons and ireland (2001), hitt, ireland, camp and sexton (2001) and kissin and herrera (1990) point out that complication in target assessments, misidentifi cation of asset complementarities, informational asymmetries and high premiums paid for target may have adverse effects on the value of acquiring fi rms. the above factors infl uence the developed and developing country fi rms differently. furthermore, when a fi rm from a developed country acquires a developing country fi rm, it gains from the cheap labour thus reducing its karels et al.: cross-border mergers and acquisitions 44 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 operational cost whereas when a fi rm from a developing country acquires a developed world fi rm, it accesses the technological know-how thus increasing the growth potential for the fi rm. in their studies, chari et al. (2010) have grouped countries into developed markets and emerging markets and gubbi et al. (2010) have clubbed developed and developing world target countries together. whereas, the developed world countries are economically and culturally similar to each other and can be grouped together, there exist noticeable differences across the countries of emerging markets. while the countries in the developed world are all democracies, there are different government structures in the emerging world countries. whereas the countries in the developed world predominantly follow the same religion and have similar cultures, there are differences in the cultural and religious practices of the emerging world population. whereas the judicial system in the developed world countries is strong, it is not so strong in all the emerging world countries. the fi nancial markets in all the developed world countries can be considered to be at least weak form effi cient; the fi nancial markets for many emerging world countries are not even weak form effi cient. due to the difference within the emerging world countries, a fundamental factor may play an important role in one emerging world country but may not be important for mergers in other emerging world countries. it may also be that because of the differences across the emerging world countries effect due to fundamental factors cancel out when we club the data from different countries together. studying the cross-border mergers and acquisitions between two countries at a time especially when the merger and acquisition is between the developed world and the emerging world countries would hence provide more meaningful information. also combining all targets in one group may cancel the effects of their different organizational form hence one should be cautious in deriving any inference from the papers studying all targets together and more meaningful information can be derived by splitting the targets into public and private. in this paper, we examine the gains and losses to shareholders of both targets and acquirers, from mergers and acquisitions between the us and the developing country of india by separating the targets into their organizational forms (public, private and subsidiary fi rms). the overall purpose of our analysis is to bring new evidence to view on shareholder wealth from the cross-border (especially when the acquirer is from an emerging world country) merger and acquisition activity. 3. data, variables and methodology this study focuses on cross-border acquisitions between the us and india announced over the period january 1995 – august 2007.7 we extract our merger and acquisition sample from sdc’s (securities data corporation, a database 7 we start the data in 1995 as there were very few mergers and acquisitions between indian and us fi rms before that time. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 45 from thomson financial) mergers and acquisitions database during the sample period of january 1995 to august 2007. we use crsp for daily returns and daily index returns for us fi rms and datastream for the stock prices of indian fi rms. we use the bse200 (bombay stock exchange) and the bse500 from datastream for the indian market index. the bse 500 is a more compensative index than bse200, but it was fi rst introduced on february 1, 1999. hence, we use bse200 as the indian market index during 1995-1998 and bse500 as the indian market index after 1998. table 2: number of acquiring and target fi rms in the sample this table shows the number of acquiring and target fi rms in our initial sample and the fi nal sample for the mergers and acquisitions between the us and the indian fi rms. the initial data set consists of 676 us acquisitions of indian fi rms and 230 indian acquisitions of us fi rms. this sample includes all the bidding and targets fi rms irrespective of whether or not the merger was successful. firms without complete daily return or stock price information are excluded from the sample. as reported in table 2, this reduces our initial sample to 248 us acquiring fi rms and 98 indian target fi rms, and 16 us target fi rms and 128 indian acquiring fi rms. in our sample all of the subsidiary targets are non-traded fi rms. for the cross-sectional regression we collect the data on fi rm characteristics from sdc. there are several indian fi rms with incomplete or missing data on sdc. we use datastream to obtain the data on these fi rms. excluding relative size data and the data on percentage owned after transaction, we have complete data on 99 indian acquirers of us targets and 129 us acquirers of indian targets. when we include the percentage owned after transaction our sample has 62 indian acquirers of targets and 69 acquirers of indian targets. when we include type of merger and acquisition total m&a data available for us public firms data available for indian public firms total m&a data available for us public firms data available for indian public firms private acquirer & private target 09 42 private acquirer & public target 4 4 71 34 private acquirer & subsidiary 9 53 public acquirer & private target 110 81 88 83 public acquirer & public target 16 11 11 81 63 38 public acquirer & subsidiary 45 34 66 63 subsidiary & private target 55 11 subsidiary & public target 2 0 61 26 subsidiary & subsidiary 6 34 others 3 1 2 77 39 032latot 16 128 676 248 98 us target & indian acquirer indian target & us acquirer karels et al.: cross-border mergers and acquisitions 46 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 the relative size variable our sample size falls to 46 indian acquirers of us targets and 41 us acquirers of indian targets. we collect the data on the exchange rate from the website of the federal reserve bank of st. louis. we compute the announcement date abnormal return on a three-day window of -1 day to 1 day after the announcement. we use the standard eventstudy methodology to analyse the impact of the acquisitions announcement on shareholders’ wealth for both the acquiring and the target fi rms. we set the event date as the announcement date of the acquisition as reported in the sdc platinum database. the abnormal return for stock j on day t ( ) is computed using the marketadjusted returns model: (1) where is defi ned as the raw return of the common stock of the jth fi rm on day t, and is the expected return on stock j. to calculate the expected return on stock j, we use the market model: (2) where is the daily market return using a value-weighted index, and for these fi rms are computed through a regression of the fi rm returns on the market returns during the time period of 120 days to 5 days before announcement. for us fi rms, the daily value-weighted return from crsp is used as the proxy for the us market return; while for indian fi rms, the bse200 and bse500 indices from datastream are used as the proxy for the indian market returns.8 the average abnormal return for day t is calculated as: (3) where n is the number of fi rms in the sample. over an interval of three days beginning with day -1 and ending with +1, the cumulative average abnormal return is: (4) )( jtjtjt rerar )()( mtjjjt rre ,1 n ar aar n j jt t . 1 1 1,1 t taarcaar 8 bse500 started in 1999 so we use bse200 as a proxy for market index before 1999. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 47 to test the null hypothesis that cumulative average abnormal returns from -1 day to +1 day, is zero, we compute the t statistic (tstat) as: (5) where is the average abnormal return on day t, is the sample variance of the average abnormal return for day t, and n is the number of fi rms in the sample. our objective is to test if the cumulative average abnormal returns for acquirers or targets are signifi cantly different from zero. we perform the following hypothesis test: where is the cumulative average abnormal return. we use the t-test statistics of equation (5) to determine the statistical signifi cance of the results. 4. results in this section, we analyse the shareholder wealth effects for both the acquiring and the target fi rms. we report the cumulative abnormal returns for acquirers and targets during the three-day event window of -1 day to 1 day after the announcement. 4.1 us acquirers and indian targets in panel a of table 3 we present the caars for us acquirers of indian publiclytraded, privately-held and non-traded subsidiary fi rms. though the sign on the cumulative abnormal returns around the announcement date is negative, the cumulative abnormal returns for acquiring us fi rms are not signifi cant. when we split the sample9 into the us acquirers of publicly-traded, privately-held and subsidiary fi rm indian target companies, we fi nd the abnormal returns on the announcements of us acquisitions of publicly-traded and subsidiary fi rm targets are statistically insignifi cant. shareholders of us fi rms who acquire privatelyheld indian fi rms realize statistically signifi cant abnormal returns of -1.1 per cent on the three day window (-1 to +1). ns caar tstat t t 1 1 2 1,1 0: 10h 9 we do not study separately the acquirers of government fi rm targets and joint ventures. these targets are included in the overall sample. for this reason the sum of the acquirers of public targets (63 fi rms), private targets (83 fi rms) and subsidiary fi rm targets (63 fi rms) is less than the overall sample (248 fi rms). karels et al.: cross-border mergers and acquisitions 48 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 table 3: abnormal returns of acquiring and indian target fi rms panel a: american acquiring fi rms u.s. publiclytraded fi rms acquire indian targets (248 fi rms) u.s. publiclytraded fi rms acquire indian publiclytraded fi rms (63 fi rms) u.s. publiclytraded fi rms acquire indian privatelyheld fi rms (83 fi rms) u.s. publiclytraded fi rms acquire subsidiaries of indian fi rms (63fi rms) caar[-1,1] -0.0041 -0.0038 -0.0110* 0.0025 panel b: indian target fi rms indian targets acquired by u.s fi rms (98 fi rms) indian targets acquired by u.s. publicly-traded fi rms (38 fi rms) indian targets acquired by u.s. privately-held fi rms (34 fi rms) indian targets acquired by subsidiaries of u.s. fi rms (26 fi rms) caar[-1,1] 0.0408*** 0.0449*** 0.0355** 0.0407** ***signifi cant at the 1% level **signifi cant at the 5% level *signifi cant at the 10% level panel a presents the cumulative average abnormal returns for publiclytraded us fi rms that announced the acquisition of indian target fi rms. the overall results in column 2 shows negative abnormal returns for the shareholders of the acquiring fi rms. when we separate them into different groups the results show losses for the acquirers of publiclytraded, privately-held and subsidiary fi rm targets. in panel b we present the cumulative average abnormal returns around the announcement date for indian target fi rms. overall results in column 2 show that shareholders of target fi rms earn signifi cant returns around the announcements of mergers and acquisitions. these results hold even when we separate these targets into being acquired by publicly-traded (public), privatelyheld (private) or subsidiary fi rms. in panel b of table 3 we present the caars for publicly-traded indian targets acquired by publicly-traded us fi rms, privately-held us fi rms and subsidiaries of us fi rms. we fi nd a signifi cant gain of 4.08 per cent for the set of indian targets as a whole. when we look at the abnormal returns by the type of us acquirer – publicly-traded, privately-held or us subsidiary fi rms we fi nd that the indian targets realize signifi cant gains of approximately 4 per cent across all types of acquirers for the three-day event window. the above results indicate that the acquisition by any us fi rm (public, private or subsidiary fi rm) is benefi cial to the shareholders of the indian targets. the shareholders of us acquirers of indian targets suffer losses around the announcement of mergers and acquisitions. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 49 4.2 indian acquirers and us targets in panel a of table 4 we show the caars for indian acquirers of us targets. the acquisition announcement by indian fi rms is associated with a 2.71 per cent stock price increases for the indian acquiring fi rms. when we decompose the us targets into publicly-traded and privately-held fi rms, we fi nd that the indian fi rms do not realize statistically signifi cant gains for acquisitions of publiclytraded us targets. indian fi rms realize a statistically signifi cant gain of 3.09 per cent for the acquisition of privately-held us fi rms and a gain of 2.26 per cent for their acquisition of us subsidiary fi rms. in panel b of table 4 we report the caars of the us targets in the crossborder acquisitions. our sample size is very small. for the sixteen publiclytraded us fi rms acquired by all types of indian fi rms, the us targets experienced large positive abnormal returns but these were statistically insignifi cant. when we decompose the indian acquirers into privately-held and publicly-traded companies, we fi nd large cumulative abnormal returns for the us targets but these are not statistically signifi cant. these results indicate that indian acquiring fi rms realize gains on the announcement of their acquisition of us fi rms although the gains are concentrated in acquisitions of privately-held or subsidiaries of us fi rms. no signifi cant gains are earned in the acquisition of publicly-traded us companies. table 4: abnormal returns for indian acquiring and target fi rms panel a: american acquiring fi rms indian publiclytraded fi rms acquire u.s. targets (128 fi rms) indian publiclytraded fi rms acquire u.s. publicly-traded fi rms (11 fi rms) indian publiclytraded fi rms acquire u.s. privately-held fi rms (81 fi rms) indian publiclytraded fi rms acquire subsidiaries of u.s. fi rms (34 fi rms) caar [-1, 1] 0.0271*** 0.0054 0.0309*** 0.0226** panel b: us targets u.s. targets acquired by indian fi rms (16 fi rms) u.s. targets acquired by indian publiclytraded fi rms (11 fi rms) u.s. targets acquired by indian privately-held fi rms (4 fi rms) caar [-1, 1] 0.1812 0.1850 0.0389 ***signifi cant at the 1% level **signifi cant at the 5% level *signifi cant at the 10% level karels et al.: cross-border mergers and acquisitions 50 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 panel a presents the cumulative average abnormal returns for publiclytraded indian fi rms which announced the acquisition of u.s. targets. the overall results in column 2 show positive abnormal returns for the shareholders of the acquiring fi rms. when we separate them into different groups, publiclytraded (public), privately-held (private) and subsidiary fi rms the results indicate substantial gains for the acquirers of the privately-held targets and subsidiary fi rm targets. panel b presents the cumulative average abnormal returns around the announcement date for the us target fi rms. the shareholders of the target fi rms gain from the announcement of mergers and acquisitions but this gain is statistically insignifi cant. 5. explanations of the empirical results 5.1 public, private and subsidiary fi rm acquirers and public targets obviously, the shareholders of public targets will agree to sell their perpetual dividend stream and future capital gains when they believe that the future prospects of the company’s growth are bleaker than the buyer’s beliefs. they would also agree to sell if they get a very lucrative offer from a buyer and there are possibilities for further negotiations of the prize. in either of these cases the merger is good news for the shareholders and should result in positive abnormal returns. bargeron, schlingemann, stulz and zutter (2008) fi nd that the premium paid for acquisitions of public targets is signifi cantly lower when the acquirer is a private fi rm instead of a public fi rm. they propose a managerial discretion theory of takeovers where managers may gain from the acquisitions that do not benefi t shareholders. due to the private benefi ts of acquisitions for managers, they pay more for target fi rms than shareholders would. bargeron et al. (2008) provide evidence consistent with the managerial discretion theory of takeovers by showing that the difference in target shareholder gains between acquisitions by privately-held and by publicly-traded fi rms fall as the proportion of managerial ownership of the publicly-traded bidder increases. our results are consistent with bargeron et al. (2007) in that we fi nd the abnormal returns for the targets of publicly-traded acquirers to be consistently higher than the abnormal returns of targets of privately-held acquirers; (a) the abnormal returns for the indian targets of publicly-traded us acquirers are 4.49 per cent (signifi cant at 1% level) and the abnormal returns for the indian targets of privately-held us acquirers are 3.55 per cent (signifi cant at 5% level), (b) the abnormal returns of the us targets of publicly-traded indian acquirers are 18.50 per cent and the abnormal returns of the us targets of privately-held indian acquirers are 3.89 per cent, though both of these are statistically insignifi cant. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 51 5.2 public-traded acquirer and publicly traded, private and subsidiary fi rm targets in section 4 we found that shareholders of us acquirers of indian targets suffered losses on and around the announcement of mergers and acquisitions of indian companies. however, shareholders of indian fi rms gained on the announcement of their acquisition of privately-held us fi rms and subsidiaries of us fi rms. we investigate if this opposite reaction is due to fi rm specifi c characteristics by regressing the three-day cumulative abnormal return around announcement (-1 day to +1 day after announcement) on the following fi rm specifi c characteristics10: • relative size = value of deal/market value of acquirer.11 moeller and stulz (2004) fi nd that for domestic acquisitions large fi rms loose when they acquire small fi rms whereas small fi rms gain when they acquire large fi rms. moeller and stulz (2004) fi nd that small acquirers acquiring large fi rms have 2 per cent higher announcement returns. • acquirer size = market value of acquirer. mitchell and stafford (2000) fi nd that announcement returns are positively related to the size of the acquirer for domestic acquisitions. • high-tech = “1” if both target and acquirer are defi ned as high tech by the sdc thomson data source and “0” otherwise. conn et al. (2005) fi nd that fi rms (both acquiring and target) in the same high-tech industry have a positive infl uence on the returns of the acquirer. • related = “1” if target and acquirer have the same four-digit sic code and “0” otherwise. this variable measures if the merger is within the industry or if the fi rms are diversifying. megginson, morgan and nail (2004) fi nd that for domestic acquisitions gains are higher in related acquisitions. • value = “1” if acquirer’s market to book value of equity is in quintile one and “0” otherwise. • glamour = “1” if acquirer’s market to book value of equity is in quintile fi ve and “0” otherwise. these defi nitions of value and glamour variables are from conn et al. (2005). • tq = tobin’s q of the acquirer. we estimate tobin’s q as the market value of fi rm divided by the book value. • downer= dummy equals “1” if the majority stake is acquired (greater than 50%) and “0” otherwise. chari et al. (2010) fi nd that acquirers from the developed world gain when they announce acquisition of majority stake in fi rms from the emerging world. • daqc = “1” if the acquirer is from a and “0” if the acquirer is from india. 10 we defi ne these characteristics similar to conn et al. (2005). due to the limitation of data available on the sdc thompson data source, we are unable to study all the fi rm characteristics discussed in conn et al. (2005). 11 where the data on value of deal is unavailable we use the defi nition of relative size given by cakici, hassel and tandon (1996) who defi ne relative size as the value of outstanding equity of targets/equity of bidder. karels et al.: cross-border mergers and acquisitions 52 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 we provide the summary statistics by cross-border relationship for the above variables in table 5. we fi nd that the indian and the us acquirers of private targets have similar high-tech status levels. the indian acquirers of publiclytraded us targets and us subsidiary fi rm targets are more often classifi ed as high-tech status as compared to us acquirers of publicly-traded indian fi rms and indian subsidiary fi rm targets. irrespective of the target fi rm status, indian acquirers appear to engage in company diversifi cation more than us acquirers. the relative size of the targets is substantially smaller for us acquisitions. the market value of indian acquirers of us targets is also signifi cantly lower than the market value of us acquirers of indian targets. though the average percentage owned after the transaction by indian acquiring fi rms is higher than the percentage owned after the transaction by acquirers, both indian and acquirers, acquire majority stakes in the target fi rms. for both the us and indian acquirers, the average stake acquired in the private and subsidiary fi rm targets is higher than the average stake acquired in publicly-traded targets. table 5: summary statistik of the us and indian acquiring and target fi rms panel a: indian acquirer of us target publicly traded target privately held target subsidiary targe high tech 0.583 0.532 0.500 related 0.250 0.274 0.308 market value of acquirer ($million) 2238 835 1185 market value/ book value of equity 10.739 4.464 4.980 acquirers book value of common equity ($million) 171 210 260 acquirer total assets ($million) 292 377 508 # of fi rms above data is available for 11 62 26 average % owned after transaction (oat) 75.75 88.82 99.71 oat data available for (# of firms) 4 41 17 relative size 0.362 0.110 0.481 relative size data available for (# of firms) 9 27 10 total number of fi rms 16 110 45 international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 53 panel b: us acquirer of indian target publicly traded target privately held target subsidiary targe high tech 0.264 0.558 0.273 related 0.415 0.349 0.394 market value of acquirer ($million) 52523 6143 32489 market value/ book value of equity 7.349 7.214 4.465 acquirers book value of common equity ($million) 14066 2093 6965 acquirer total assets ($million) 135979 5381 34739 # of fi rms above data is available for 53 43 33 average % owned after transaction (oat) 50.3 84.24 79-87 oat data available for (# of firms) 21 28 20 relative size 0.016 0.027 0.006 relative size data available for (# of firms) 19 12 10 total number of fi rms 81 88 66 in panel a we present the summary statistics for the indian acquirers of us targets and in panel b we present the summary statistics for the us acquirers of indian targets. a comparison of panel a and panel b shows that the relative size of us targets is substantially higher than the relative size of indian targets and the us acquirers have higher market value than the indian acquirers. the indian acquirers of the us targets diversify their business more than the us acquirers of the indian targets. we regress the caar for the acquirers of public, private and subsidiary targets on these variables to investigate whether there are abnormal return differences to the acquirer’s home country after controlling for fi rm characteristics. the data for relative size and percentage owned after transaction is available for fewer fi rms hence for each of the target types we fi rst run the regression without the variables, ‘relative size’ and ‘downer’ as the lack of data on these variables substantially reduces the sample size. we use the following three models to test if the caars are country dependent: karels et al.: cross-border mergers and acquisitions 54 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 mode (6) model ii: (7) model iii: (8) where lmv is the log of market value of the acquirer. exc in the above equation is the exchange rate variable12 which measures the effect of changes in the exchange rate on mergers and acquisitions. we follow cakici et al. (1996) who estimate the effects of exchange rate variable on bidding shareholders’ wealth as a result of foreign acquisitions of the us fi rms, and compare these to a control sample of foreign acquisitions by us fi rms. we pool the country data and run the above regression models for the acquirers of publicly-traded, privately-held and subsidiary targets separately. the results are reported in table 6. for the acquirers of publicly-traded and subsidiary targets we fi nd an insignifi cant daqc variable suggesting that there is no country difference in the abnormal returns of the indian and the us acquirers of the publicly-traded and subsidiary targets. in the following table we present the regression on the caar (-1 day to +1 day after the announcement) as a dependent variable for the acquirers of public, private and subsidiary targets. after controlling for fi rm characteristics we fi nd that the dummy variable for country (daqc=1 for us acquirers and daqc=0 for indian acquirers) is negative and signifi cant for the acquirers of private targets suggesting a lower abnormal return for the us acquirers as compared to the indian acquirers of private targets. the coeffi cient for the sample of publicly-traded acquirers of privatelyheld targets is negative and signifi cant suggesting a lower caar for the us acquirers of privately-held indian targets as compared to the indian acquirers of privately-held us targets. 12 we compute the exchange rate variable using the method adopted by cakici et al. (1996). we take the indian currency’s average exchange rate (per dollar) for the sample period 1995-2007 and subtract the indian currency’s exchange rate for the year of acquisition. then we divide this difference by the average exchange rate. iiiiii iiii excdaqcglamourvaluetq lmvlatedrehightechcaar 87654 321 ii iiiii iiii downer excdaqcglamourvaluetq lmvlatedrehightechcaar 9 87654 321 iiiiii iiiii lativesizereexcdaqcglamourvalue tqlmvlatedrehightechcaar 98765 4321 international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 55 table 6: regression results of cross-border mergers and acquisitions 6. conclusion prior empirical research on mergers and acquisitions is predominantly related to the developed markets of the us and europe. there is limited work on mergers and acquisitions when targets from the developed world are acquired by companies from the developing world. in the cross-border mergers and acquisitions between developed and developing world fi rms, researchers have studied the effect of the announcement of mergers and acquisitions on the acquirer fi rms only. none of the studies to date have investigated the effect of announcement of mergers and acquisitions on the target fi rms. in this paper we studied the cross-border mergers and acquisitions between the us and india for both acquirers as well as target fi rms. we used data on mergers and acquisitions from january 1995 to august 2007 and found that the mergers and acquisitions with us acquirers and indian targets resulted in signifi cant losses for the acquirers and signifi cant gains for the targets. mergers and acquisitions between indian acquirers and us targets resulted in signifi cant gains for the acquirers and insignifi cant gains for the targets. we further examined the abnormal returns by decomposing the sample of fi rms into publicly-traded, privately-held and non-traded subsidiary fi rms. we found that the us acquirers of publicly-traded indian fi rms realize insignifi cant losses while publicly-traded indian targets acquired by us fi rms earned signifi cant returns on the announcement of a merger or acquisition. indian acquirers of publicly-traded us fi rms earned insignifi cant gains/losses and us targets of publicly-traded indian acquirers earned insignifi cant positive abnormal returns on the announcement of mergers and acquisitions. these results are similar to the results we found in the existing literature. variable i ii ii i ii ii i ii ii intercept -0.030 -0.017 -0.020 0.037* 0.041 0.067 0.017 0.064 -0.017 high tech 0.009 0.026 0.020 0.009 0.002 -0.001 -0.002 0.001 0.063*** related -0.001 -0.013 -0.034* 0.003 0.014 0.005 -0.003 0.008 -0.008 lmv 0.005 0.007 0.004 -0.001 -0.002 -0.002 0.002 -0.001 0.004 tq 0.001 -0.005 0.000 -0.001 0.001 -0.002 0.000 0.000 -0.005 value 0.022 -0.008 0.071** -0.001 0.000 -0.006 0.021 0.019 0.018 glamour -0.004 -0.010 0.014 -0.006 -0.034* 0.019 -0.036 -0.011 -0.003 exc 0.045 0.088 0.038 0.058 -0.121 0.181 0.070 -0.011 0.197 daqc -0.019 -0.006 -0.014 -0.044*** -0.053*** -0.044* -0.020 -0.024 -0.014 relative size -0.105* -0.024 0.028*** # of firms 65 25 28 104 69 38 59 37 20 rsquare 0.095 0.266 0.401 0.196 0.197 0.197 0.146 0.232 0.786 public target private target subidiary target karels et al.: cross-border mergers and acquisitions 56 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 in the cross-border acquisition of private targets, publicly-traded us acquirers suffered losses whereas publicly-traded indian acquirers realized gains. the indian acquirers of us subsidiary fi rm targets realized signifi cant gains while the us acquirers of indian subsidiary fi rm targets realized insignifi cant gains/losses. we investigated these cross-border merger and acquisition results to determine if the results were due to fi rm characteristics or country dependent by regressing the cumulative abnormal returns on the fi rm characteristics and a dummy variable for the country of the acquirer. for the private targets, we found the country dummy variable to be signifi cant after controlling for fi rm characteristics indicating a lower abnormal return for us acquirers of privately-held indian fi rms as compared to the indian acquirers of privately-held us fi rms. the dummy variable is not signifi cant for the acquirers of publicly-traded fi rms and for subsidiary fi rm targets indicating that fi rm characteristics and not the country environments govern the returns on these mergers and acquisitions. author information: professors gordon karels, edward lawrence and jin yu are staff member of the fi nance faculty respectively at the university of nebraska-lincoln, florida international university and st. cloud state university, all in the usa. they may be contacted at: gkarels@unl.edu; edward. lawrence@fi u.edu; iyu@stcloundstate.edu. references aybar b. & ficici a. (2009). cross-border acquisitions and fi rm value: an analysis of emerging-market multinationals. journal of international business studies, 40, 1317-1338. baldwin j. r. & caves, r. e. (1991). foreign multinational enterprises and merger activity in canada. in l. waverman (ed.). corporate globalization through mergers and acquisitions. calgary: university of calgary press, 89-122. bargeron, l., schlingemann, f. p., stulz, r. m. & zutter, c. j. (2008). why do private acquirers pay so little compared to public acquirers? journal of financial economics, 83, 375-390. cakici n., hassel, c. & tandon, k. (1996). foreign acquisitions in the united states: effect on shareholder wealth of foreign acquiring fi rms. journal of banking and finance, 20, 307-329. caves, r. e. (1971). international corporations: the industrial economics of foreign investment. economica, 38, 1-27. caves, r. e. (1998). research on international business: problems and prospects. journal of international business studies, 29, 5-19. chari, a., ouimet, p. & tesar, l. (2010). the value of control in emerging markets. review of financial studies, forthcoming. conn, r. l., cosh, a., guest, p. m. & hughes, a. (2005). impact on uk acquirers of domestic, cross-border, public and private acquisitions. journal of business, finance and accounting, 32, 815-870. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 cross-border mergers and acquisitions between industrialized and developing countries: 35-58 57 denis, d. j., denis, d. k. & yost, k. (2002). global diversifi cation, industrial diversifi cation and fi rm value. the journal of finance, 57, 1951-1979. doukas, j., (1995). overinvestment, tobin’s q and gains from foreign acquisitions. journal of banking and finance, 19, 1285-1303. doukas, j. & travlos, n. g. (1988). the effect of corporate multinationalism on shareholders’ wealth: evidence from international acquisitions. journal of finance, 43, 1161-1175. eckbo, b. e. & thorburn, k. s. (2000). gains to bidder fi rms revisited: domestic and foreign acquisitions in the us. journal of financial and quantitative analysis, 35, 1-24. eun, c. s., kolodny, r. & scherga, c. (1996). cross-border acquisitions and shareholder wealth: tests of the synergy and internalization hypothesis. journal of banking and finance, 20, 1559-1582. francis, b. b., hasan, i. & sun, x. (2008). financial market integration and the value of global diversifi cation: evidence for acquirers in cross-border mergers and acquisitions. journal of banking & finance, 32, 1522-1540. gubbi, s. r., aulakh, p. s., ray, s., sarkar, m. b. & chittor, r. (2010). do international acquisitions by emerging economy forms create shareholder value? the case of indian fi rms. forthcoming in the international journal of business studies. hitt, m. a., hoskisson, r. e. & ireland, d.r. (2001). a mid-range theory of the interactive effects of international and product diversifi cation on innovation and performance. journal of management, 20, 297-326. hitt, m. a., ireland, r. d., camp, s. m. & sexton, d. l. (2001). strategic entrepreneurship: entrepreneurial strategies for wealth creation. strategic management journal, 22, 479-491. kissin, w. d. & herrera, j. (1990). international mergers and acquisitions. journal of business strategy, 11, 51-54. kiymaz, h. (2004). cross-border acquisitions of fi nancial institutions: impact of macroeconomic factors. journal of banking and finance, 28, 1413-1439. kogut, b. (1983). managing political risk assessment. sloan management review, 24, 71-73. la porta, r., lopez-de-silanes, f., shleifer, a. & vishny, r. (1998). law and fi nance. journal of political economy, 106, 1113-1155. la porta, r., lopez-de-silanes, f., shleifer, a. & vishny, r. (1999). corporate ownership around the world. journal of finance, 54, 471-517. la porta, r., lopez-de-silanes, f., shleifer, a. & vishny, r. (2000). investor protection and corporate governance. journal of financial economics, 58, 3-27. markides, c. c. & ittner, c. d. (1994). shareholder benefi ts from corporate international diversifi cation: evidence from u.s. international acquisitions. journal of international business studies, 25, 343-366. markides, c. & oyon , d. (1998). international acquisitions: do they create value for shareholders? european management journal ,16, 125-135. karels et al.: cross-border mergers and acquisitions 58 the international journal of banking and finance, vol. 8. number 1, march 2011: 35-58 megginson, w. l., morgan, a. & nail, l. (2004). the determinants of positive long-term performance in strategic mergers: corporate focus and cash. journal of banking and finance, 28, 523-52. mitchell, m. l. & stafford, e. (2000). managerial decisions and long-term stock price performance. journal of business, 73, 287-329. moeller, s. b. & stulz, r. m. (2004). firm size and the gains from acquisitions. journal of financial economics, 73, 201-228. moeller, s. b. & schlingemann, f. p. (2005). global diversifi cation and bidder gains: a comparison between cross-border and domestic acquisitions. journal of banking and finance, 29, 533-564. morck, r. a. & yeung, b. (1991). why investors value multinationality. journal of business, 64, 165-187. morck, r. a. & yeung, b. (1992). internalization: an event-study test. journal of international economics, 33, 41-56. mulherin, j. h. & audra l. b. (2000). comparing acquisitions and divestitures. journal of corporate finance, 6, 117-139. rossi, s. & volpin p. f. (2004). cross-country determinants of mergers and acquisitions. journal of financial economics, 74, 277-304. seth, a., song, k. p. & pettit, r. (2002). value creation and destruction in crossborder acquisitions: an empirical analysis of foreign acquisitions of u.s. fi rms. strategic management journal, 23, 921-940. williamson, o. e. (1979). transaction-cost economics: the governance of contractual relations. journal of law and economics, 22, 233-261. international journal of banking and finance, vol. 8, iss. 1 [2011], art. 3 79 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 how to cite this article: enow, s. t. (2023). volatility persistence in international financial markets in the post covid-19 era. international journal of banking and finance, 18(2), 79-96. https:// doi.org/10.32890/ ijbf2023.18.2.4 volatility persistence in international financial markets in the post covid-19 era samuel tabot enow the independent institute of education vega school, south africa enowtabot@gmail.com received: 12/8/2022 revised: 8/11/2022 accepted: 30/11/2022 published: 25/6/2023 abstract the long-term behaviour of stock markets are of significant importance to asset managers and financial experts due to its direct link with security price valuation. volatility persistence has a significant impact on the returns of security prices due to its time varying properties. however, there is no real meaningful effect of current volatility on future security prices and returns if the volatility is transitory and not persistent. the aim of this study was to explore conditional volatility properties and determine whether the current volatile environment would persist in the jse, s&p 500, nasdaq index, sse, cac 40 and dax markets. using a garch 1.1 model and a markov switching model, the findings revealed that volatility would persist in the jse, s&p 500, nasdaq index, sse, cac 40, and the dax from their arch and garch coefficients, as well as the delay parameters. in addition, the effects of past volatility in the nasdaq, cac 40, and dax would remain in the forecast of variance. a diversified and broader investment approach should be used in the jse, s&p 500, nasdaq https://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 80 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 index, sse, cac 40, and dax indexes to mitigate risk, and portfolio formation should not concentrate on any sector or asset classes. keywords: volatility, financial markets, covid-19, garch coefficient, arch term, conditional variance. jel code: g1, g2, g4. introduction modelling volatility has always been an interesting phenomenon in finance and it is certainly an interesting time to investigate volatility persistence in financial markets because of the recent covid-19 pandemic, which has had a mirage effect around the globe. due to the introduction of the vaccine, financial markets are recovering from market disruptions caused by the covid-19 pandemic (shang, 2021). the spread of fear and uncertainty has created massive swings in market prices in almost all financial markets around the world. this was evident in stock markets like the s&p 500 index that had plummeted by over 30 percent at the beginning of 2020 with the spread of the pandemic (li, 2020). the french stock market index (cac 40) and the german blue chip companies trading on the frankfurt stock exchange (dax) fell by over 12 percent, which contributed to an overall 11 percent decrease in the value of european stocks at the beginning of february 2020, before increasing to a record high (hathorn, 2021). prior to the widespread impact of the virus, there was a significant jump in oil prices, breaking through the $1700 per ounce ceiling for the first time since 2012. however, the oil price tumbled to $250 per ounce in march 2020, losing about 60 percent of its value in the first quarter of 2020 due to the mismatch of demand fears and supply concerns (camp et al., 2020). a similar situation was seen in the bond market, where nearly $4 trillion of municipal bonds experienced unprecedented volatility as investors sold off their positions amid concerns related to covid-19 (liang, 2020). the crisis produced unrivalled government responses, which included the introduction of extraordinary stimulus packages. there was a relaxation of banking regulations to ensure capital buffers were not impeding banks from supporting and stabilising economies. these actions were partly to curb financial market volatility and help ensure price stability. 81 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 the concept of volatility persistence refers to how today’s volatility affects the conditional variance of future volatility (wang & yang, 2017). therefore, today’s unconditional volatility variance may be infinite and will continue in the future. applying this definition in financial markets means that large or small volatility changes will tend to follow the same pattern in the future with unpredictable features and successive disturbances. the study of karanasos et al. (2014) reveals that volatility has time varying properties with structural breaks. muguto and muzindutsi (2022) contend that volatility should not persist, because positive and negative news will induce an equal change. there has always been a compelling need for forecasting volatility persistence in financial markets as it determines the gains and losses from the erratic behaviour of financial markets. it can also be used to determine the level of risk involved in holding a security. the aim of this study was to investigate the conditional volatility properties of the major financial markets around the world post the covid-19 era. this study is significant to investment practitioners and market participants as it explores the extent to which the current volatile environment will persist, which has important implications for risk management and portfolio management. more specifically, knowledge of volatility persistence is important because the value of financial assets is directly linked to the level of volatility prevailing in the market; it denudes the linkage between some underlying risk factors and security price movements (christiansen et al., 2012). moreover, establishing the amount of risk to take should be based on the knowledge of the extent of volatility persistence in financial markets (engle, 1982). literature review theoretical perspective heightened financial market volatility is a direct consequence of macro-economic uncertainty and a lack of liquidity (kundu & paul, 2022). these two economic forces are the main drivers of price fluctuations in financial markets. macro-economic uncertainty makes it very difficult to price an asset, while a lack of liquidity causes fire sales where assets are traded at a lower price (dow & han, 2018). a clear distinction between realised and implied volatility should be made in the analysis and discussion of market volatility. realised 82 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 volatility is associated with technical analysis in which it is concerned with the past, and provides a vivid picture of historical asset price movements within a particular time frame (paraschiv, 2020). conversely, implied volatility describes expectations which are often used in option pricing and financial market trading (mayhew, 1995). there exists a volatility gauge called the vix index which is used by market participants and investors to access the level of risk and uncertainty in the market. in essence, the vix is used to assess volatility expectations over a short period. high levels of uncertainty in the vix index often results in fewer trading activities, a drop in liquidity, and a negative feedback loop. the three largest volatility spikes recorded in the vix index were in 1987, 2008, and 2020, which was followed by reduced investor holdings. although volatility is usually analogous to bear markets, heightened financial market volatilities are also experienced in bull markets (elgammal et al., 2021). this was evident in the 1990 dotcom boom where market volatility rose considerably alongside the tech stocks. the growth expectations that were placed on these untested tech stocks with the accompanying excitement gave rise to uncertainties, which had led to an unsustainable bubble growth. however, market volatility does find stability when market shock subsides and when market participants get a better understanding of the economic environment (degiannakis et al., 2014). prior literature (krichene, 2003; bobeică & bojeşteanu, 2008; oh et al., 2008; thupayagale, 2012; gyamfi et al., 2016) contend that volatility tends to have a long-term memory due to recurring macroeconomic cycles. these long-term memories are justified by hysteresis and repetitive irrational behaviour, which are contrary to the efficient market hypothesis (emh). irrational behaviours are captured in the heteroskedastic variance of the financial market (maheu & mccurdy, 2000). accordingly, modelling the heteroscedasticity behaviour and understanding the unconditional mean and variance of a security index is therefore, necessary for asset pricing, risk management, and portfolio optimisation. however, there are three underlying challenges in forecasting conditional volatility: (1) inferring a latent time series from a noisy observation and modelling non-linear temporal dynamics; (2) defining a positive symmetric covariance matrix; and (3) computing maximum likelihood estimations (bauwens et al., 2006). due to the aforementioned challenges, successfully quantifying the realised volatility does not necessarily lead to forecasting the implied 83 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 volatility. hence, the need arises for a more sophisticated model to capture the arbitrage effect between the realised and implied volatility. the covid-19 pandemic has caused volatility spikes in the global financial markets. this certainly calls for concern as investors, market participants, and the general public are sceptical about the increasing risk as a result of volatility spikes. although there are some risks worth taking, most risks are detrimental and should be avoided. for example, there is a risk when investing in a market where the economy is experiencing recession. moreover, the question of the amount of risk to be taken should be analysed in conjunction with the concept of current and future volatilities. volatility of idiosyncratic moves in stock markets will not be rewarded, as long as volatility persists (visaltanachoti & pukthuanthong-le, 2009). there are mainly two sources of volatility in stock markets, i.e., the amount of new information in the market, and macro-economic uncertainty. new information about certain events has a significant impact on stock prices. however, some information is important while others are not. new important information causes investors to change their expectations, which in turn affect the market price (bookstaber & pomerantz, 1989). this new information arrives in clusters and alters the way investors perceive the future. macro-economic uncertainty is another factor that influences financial market volatility. concerns about the macro-economic outlook contribute to financial market volatility because it is very difficult to price an asset in an uncertain environment. that is why investors are very interested in the macroeconomic events associated with high volatility. according to engle and rangel (2006), the macro-economic factors that play an important role in driving financial market volatility include the following: high inflation, slow upward growth, recession and changes in short-term interest rates. volatility in the financial markets is still staggering because of the russian-ukraine crisis. currently, financial markets are dominated by the russian-ukraine crisis, which has increased volatility and uncertainty in the financial system. the spill over effect is a decreased liquidity in certain markets, which prompt investment practitioners to shift their attention to active management. commodity-based countries and countries with large amounts of united states denominated debt are experiencing tighter financial conditions. russia is the third largest oil producer, accounting for 11 percent of the world’s total supply (carpenter, 2022). that is why the disruption caused by the russian84 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 ukraine war has significant consequences in the global supply of oil, as well as natural gas exports to the european union. considering that financial markets are largely driven by market sentiments and are also emotionally structured, fear and panic tend to be reflected in stock prices (ackert et al., 2003). speculative and high growth stocks tend to be significantly affected by market volatility because fundamental drivers are driven by market sentiments. future returns and the growth of speculative stocks take a downward turn with an increase in uncertainty, and investors tend to shift their sentiments toward risk-half. this means that they are less willing to take risks in the long and short term. volatility in financial markets during covid-19 has been extensively investigated during the pandemic. a summary of these studies are highlighted. table 1 evidence of market volatility during the covid-19 pandemic study (author & year of study) model period country findings topcu et al. (2021) lag augmented vector auto regression 3 january to 15 october, 2020 united states (us) the fear and uncertainty of the pandemic triggered volatility across financial markets. endri et al. (2021) garch model 2 march 2020 to 16 march 2020 indonesia evidence of high volatility which had a negative impact on stock price returns on the indonesian stock exchange. gherghina et al. (2021) garch model january 2020 to april 2021 romania notable evidence of market volatility in the bucharest exchange trading index which was similar to that of the 2008-2009 financial crisis. (continued) 85 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 study (author & year of study) model period country findings ibrahim et al. (2020) continuous wavelet transformation analysis and plots and gjr-garch analysis 15 february to 30 may 2020 asia-pacific region financial markets in china, japan, south korea, malaysia and philippines experienced high volatilities during the covid-19 pandemic, although government interventions help curb some of the volatility. mishra & mishra (2020) fixed effect and garch model 2 july 2019 to 12 june 2020 china, hong kong, india, indonesia, israel, japan, malaysia, philippines, singapore, south korea, thailand, and taiwan the covid-19 pandemic amplified market volatility due to the impact of widespread fear. rahman et al. (2021) canonical correlation analysis 22 january 2020 to 31 december 2020 nasdaq 100 options index, the s&p 500 and dow jones industrial average, dax, cac 40, and the euro stock 50 index the announcement of new positive and death cases from the pandemic significantly increased market volatility. table 1 provides evidence of significant volatility during the covid-19 pandemic. as has been well documented in the study of muguto and muzindutsi (2022), volatility can normalise in the future and cease to persist. therefore, this study fills the gap in the literature by investigating the extent to which the current volatile environment will persist and in so doing, forecasts the implied volatility for selected financial markets. the next section highlights the blueprint used in the data analysis. 86 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 methodology to investigate volatility persistence, this study used a generalised autoregressive conditional heterosedasticity (garch) (1,1) model as proposed by bollerslev (1986), and the markov switching model developed by hamilton (1989). in the garch (1,1) model, conditional volatility at time t is an autoregressive moving average (arma) relying on past volatilities and lagged values of the error term (bauwens et al., 2006). in this respect, the garch model is very useful in investigating volatility persistence with lagged shocks together with its momentum. this model provides a parsimonious alternative to the higher autoregressive conditional heteroscedasticity (arch) model (ruilova & morettin, 2020). the garch (1,1) model has the following two important parameters; the arch term and garch coefficient. the arch term captures the extent to which the volatility changes over time due to the previous lag of autoregressive conditions, while the garch coefficient reveals the level of volatility symmetry in the market (bollerslev, 1986). the sum of the arch and garch coefficients reveals the extent to which volatility will persist in the financial market. volatility persistence is evident when the sum of the arch and garch term is closer than, or equal to 1 and vice versa (nelson, 1990). the garch (1,1) model (bollerslev, 1986) is given by the formula (1): (1) where conditional variance α = error term ϕ = arch term = lag value of conditional variance β = garch coefficient = lag square error term despite the model’s relevance, malik et al. (2005) contend that in the absence of a regime shift, the garch (1, 1) model may overestimate volatility persistence. in order to have a robust finding, a markov switching model has also been used to supplement the garch (1, 1) model. a markov switching model is very useful in modelling the time-varying behaviour of volatility in financial markets (mike et al., 1998). these time-varying behaviours of volatility may be subject to market shocks that create structural breaks (ndako, 2012). in the context of this study, the covid-19 pandemic may have caused ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c = ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c 87 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 market shocks that will affect the long memory in financial markets. the switching mechanism in the markov model captures the complex volatility pattern by dating the breaking points, which is not possible in other models. failure to incorporate these structural breaks can result in misspecifications. a markov switching model for a given parameter is given in (2): (2) where are mean zero random variables (kuan, 2012). most importantly the parameters are the delay parameters in which their probabilities will also indicate whether the current volatility will persist, hence complementing the garch (1, 1) model. the study used a sample of six major international financial markets, namely the johannesburg stock exchange (jse), the standard and poor 500 index (s&p 500), the nasdaq index, shanghai stock exchange (sse), the french stock market index (cac 40) and the german blue chip companies trading on the frankfurt stock exchange (dax). this is because these markets are among the largest financial markets in each continent around the world. the sample period was from january 1, 2020 to december 31, 2021, which was the crux of intense volatility in financial markets due to the covid-19 pandemic. results tables 2 and 3 present the results of the descriptive statistics and heteroscedasticity test, respectively. the r square values in table 2 range from 1 percent to 19 percent in the financial markets under consideration. most importantly, the durbin-watson statistics values are between 1.96 to 2.13, indicating the absence of autocorrelation (kenton, 2021). the f-statistics p-values for the jse, s&p 500, nasdaq, cac 40, and dax, are significant at 5 percent, which indicate that the variance of the returns is not constant. this may signal the presence of volatility clustering. from table 4, it can be seen that all conditions for the stability test are satisfied because the coefficients of the conditional variance are between 0 and 1, and the sum of the arch and garch coefficients is less than 1 (bera & higgins, 1993). the average return of the s&p 500 and nasdaq index ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c 88 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 is positive and statistically significant at 5 percent. the past value of the nasdaq is significant at 5 percent, meaning that the past returns in the nasdaq can be used as a gauge for future returns. these results are also evident using the cac 40 and dax index with significant past value returns. the past value returns of the nasdaq, cac 40, and dax have a very strong predictive ability because of the coefficients of their past value returns, and the arch term and garch term are all significant at 5 percent, which is congruent with the study of nguyen et al. (2020). table 2 descriptive statistics r square adjusted r square mean dependent variable durbin-watson statistics jse 0.01 0.01 0.032% 2.02 s&p 500 0.08 0.08 0.025% 2.13 nasdaq 0.19 0.19 0.029% 2.08 sse 0.01 0.01 0.025% 1.96 cac 40 0.04 0.04 0.025% 2.13 dax 0.02 0.02 0.026% 2.07 table 3 heteroscedasticity test: arch f-statistics p-value (f-statistics) p-value (chi square-statistics) jse 3.05 0.0482* 0.0482* s&p 500 22.95 0.000* 0.000* nasdaq 57.76 0.000* 0.000* sse 2.7 0.1006 0.1002 cac 40 10.85 0.000* 0.000* dax 4.96 0.0073* 0.0075* the results from table 4 also indicate that volatility will persist in the jse, s&p 500, nasdaq index, sse, cac 40 and the dax, as the sum 89 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 of their arch and garch coefficients are significant at 5 percent and close to 1. table 4 also indicates that the decaying rate of volatility in the jse, s&p 500, nasdaq, sse, cac 40 and dax are 0.02, 0.04, 0.04, 0.09, 0.05, and 0.03 respectively. furthermore, the garch coefficients are greater than the arch coefficients, confirming that volatility will persist in all financial markets under consideration. table 4 garch (1.1) results average return value of past average return arch coefficient garch coefficient sum of arch and garch coefficient decaying rate of volatility (1-sum of arch & garch coefficient) jse -0.0003 (0.71) -0.028 (0.52) 0.05 (0.0001)* 0.93 (0.000)* 0.98 0.02 s&p 500 0.001 (0.00)* 0.02 (0.41) 0.27 (0.00)* 0.69 (0.00)* 0.96 0.04 nasdaq 0.002 (0.007)* -0.12 (0.02)* 0.19 (0.00)* 0.77 (0.00)* 0.96 0.04 sse 0.01 (0.327) 0.01 (0.86) 0.19 (0.00)* 0.72 (0.00)* 0.91 0.09 cac 40 0.001 (0.088) -0.099 (0.04)* 0.19 (0.00)* 0.76 (0.00)* 0.95 0.05 dax 0.001 (0.0815) -0.109 (0.05)* 0.16 (0.00)* 0.81 (0.00)* 0.97 0.03 note. *significant at 5% regarding the markov switching output in table 5, sigma is significant in all financial markets for both regime 1 and regime 2. moreover, the p-values of the delay parameters () are all significant at 5 percent, confirming the garch (1, 1) output results. the two robust findings confirm the persistence of volatility in the sample financial markets. as already documented in the studies by ibrahim et al. (2020); gherghina et al. (2021); endri et al. (2021); enow (2023); rahman et al. (2021); and topcu et al. (2021), markets will remain volatile with the arrival of positive and negative news. 90 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 table 5 markov switching model results jse variable coefficient standard error z-statistics p-value regime 1 c 0.001487 0.002638 0.563682 0.5730 log(sigma) -3.577146 0.115888 -30.86714 0.0000* regime 2 c -0.000538 0.000772 -0.696638 0.4860 log(sigma) -4.439066 0.156048 -28.44678 0.0000* transition matrix parameters 1.441298 0.660442 2.182323 0.0291* -2.498119 1.065777 -2.343942 0.0191* s&p 500 variable coefficient standard error z-statistics p-value regime 1 c 0.001591 0.000444 3.579874 0.0003 log(sigma) -4.694589 0.039875 -117.7320 0.0000* regime 2 c -0.011364 0.008663 -1.311830 0.1896 log(sigma) -2.794207 0.106898 -26.13894 0.0000* transition matrix parameters 4.560793 0.532895 8.558519 0.0000* -2.475898 0.594259 -4.166362 0.0000* nasdaq variable coefficient standard error z-statistics p-value regime 1 c -0.002542 0.004119 -0.617204 0.5371 log(sigma) -3.309863 0.090075 -36.74569 0.0000* regime 2 c 0.001989 0.000535 3.720936 0.0002 log(sigma) -4.545140 0.039892 -113.9360 0.0000* transition matrix parameters 2.466291 0.472654 5.217963 0.0000* -4.212685 0.476894 -8.833580 0.0000* ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c (continued) 91 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 sse variable coefficient standard error z-statistics p-value regime 1 c -0.002288 0.002817 -0.812123 0.4167 log(sigma) -3.842527 0.100703 -38.15694 0.0000* regime 2 c 0.000825 0.000429 1.925522 0.0542 log(sigma) -4.793461 0.039309 -121.9440 0.0000* transition matrix parameters 2.258948 0.506996 4.455559 0.0000* -4.201432 0.509416 -8.247548 0.0000* cac 40 variable coefficient standard error z-statistics p-value regime 1 c -0.001 0.003 -0.359 0.720 log(sigma) -3.517 0.074 -47.325 0.000* regime 2 c 0.001 0.000 1.911 0.056 log(sigma) -4.737 0.041 -115.533 0.000* transition matrix parameters 3.106 0.539 5.768 0.000* -4.550 0.562 -8.103 0.000* dax variable coefficient standard error z-statistics p-value regime 1 c -0.000763 0.002638 -0.289369 0.7723 log(sigma) -3.553676 0.072844 -48.78471 0.0000* regime 2 c 0.000854 0.000463 1.843283 0.0653 log(sigma) -4.737073 0.052580 -90.09305 0.0000* transition matrix parameters 2.708401 0.529242 5.117515 0.0000* -3.958786 0.577152 -6.859174 0.0000* note. *significant at 5% ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c ℎ𝑡𝑡𝑡𝑡 = α + ϕℎ𝑡𝑡𝑡𝑡−1 + β𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 (1) bollerslev (1986) ℎ𝑡𝑡𝑡𝑡 α = error term ϕ = arch term ℎ𝑡𝑡𝑡𝑡−1 = lag value of conditional variance β = garch coefficient 𝜇𝜇𝜇𝜇𝑡𝑡𝑡𝑡−12 = lag square error term r 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 𝑍𝑍𝑍𝑍𝑡𝑡𝑡𝑡 = � 𝛼𝛼𝛼𝛼0 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼0 + 𝛼𝛼𝛼𝛼1 + 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡−1 + 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 � 𝛽𝛽𝛽𝛽𝑧𝑧𝑧𝑧𝑡𝑡𝑡𝑡 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜀𝜀𝜀𝜀𝑡𝑡𝑡𝑡 𝜌𝜌𝜌𝜌11 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜌𝜌𝜌𝜌21 𝜌𝜌𝜌𝜌11-c 𝜌𝜌𝜌𝜌21-c 92 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 conclusion using the garch model, the aim of this study was to investigate the extent to which volatility will persist in the international financial world after the covid-19 pandemic. the results indicate that volatility will persist in the financial markets under consideration. these findings are reliable because they appear to corroborate the results presented in nguyen et al. (2022), the proposition that financial markets in developed countries have stronger long-term memory than less developed financial markets. in addition, the effects of past volatility in the nasdaq, cac 40, and dax will remain in the forecast of variance due to the significant positive arch and garch coefficients. moreover, a small number of market participants in the nasdaq, cac 40 and dax may influence the stock price movements in either direction within a short period. as alluded to by pereira and zhang (2010), persistent volatility will affect the market volatility in the bond and equity markets. more specifically, one would expect to see a decrease in demand in the order-driven markets accompanied by wider bid-ask spreads. there might also be a decline in market depth in the sovereign bond markets. the number of corporate credit instruments may not match the corresponding increase in trading volume, which will increase the cost of providing liquidity. investors and market participants should focus on a diversified investment approach in this market so that risks are not concentrated on any one sector or asset class. additionally, a broader strategy of executing trade is highly recommended because financial markets will experience an increase in irrational behaviour because of a lack of confidence in the market. in short, financial markets are experiencing metastasis and therefore, market participants and investors should expect agitations, as well as persistent and volatile financial markets (mohamed, 2022). acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. references ackert, l. f., church, b. k., & deaves, r. (2003). emotion and financial markets. economic review (federal reserve bank of atlanta), 88(q2), 33-41. 93 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 gherghina, s. c., armeanu, d. s., & joldes, c. c. (2021). covid-19 pandemic and romanian stock market volatility: a garch approach. journal of risk and financial management 14(341), 1-29. bauwens, l., laurent, s., & rombouts, j. v. k. (2006). multivariate garch models: a survey. journal of applied econometrics, 21(1), 79–109. bera, a., & higgins, m. (1993). arch models: properties, estimation and testing. journal of economics survey, 7, 305-366. bobeică, g., & bojeşteanu, e. (2008). long memory in volatility. an investigation on the central and eastern european exchange rates. european research studies, 6(4), 8-18. bollerslev, t. (1986). generalized autoregressive conditional heterosedasticity. journal of econometrics, 31, 307-327. bookstaber, r. m., & pomerantz, s. (1989). an information-based model of market volatility. financial analysts journal, 45(6), 37-46. camp, k. m., mead, d., reed, s. b., sitter, c., & wasilewski, d. (2020). from the barrel to the pump: the impact of the covid-19 pandemic on prices for petroleum products. [online] https://www.bls.gov/opub/mlr/2020/article/from-the-barrel-tothe-pump.htm carpenter, j. w. (2022). the 5 biggest russian oil companies. [online] https://www.investopedia.com/articles/markets/100515/5biggest-russian-oil-companies.asp. christiansen, c., schmeling, m., & schrimpf, a. (2012). a comprehensive look at financial volatility prediction by economic variables. bank of international settlement working paper 374. [online]. https://www.bis.org/publ/work374.pdf degiannakis, s. a., filis, g., & kizys, r. (2014). the effects of oil price shocks on stock market volatility: evidence from european data. the energy journal, 35(1), 35-56. dow, j., & han, j. (2018). the paradox of financial fire sales: the role of arbitrage capital in determining liquidity. the journal of finance, 73(1), 229-274. http://www.jstor.org/ stable/26653275. elgammal, m. m., ahmed, w., & alshami, a. (2021). price and volatility spill overs between global equity, gold, and energy markets prior to and during the covid-19 pandemic. resources policy, 74, 102334. https://doi.org/10.1016/j. resourpol.2021.102334. 94 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 engle, r. (1982). autoregressive conditional heteroscedasticity with estimates of the variance of united kingdom inflation. econometrica, 50(1), 987-1007. endri, e., aipama, w., razak, a., sari, l., & septiano, r. (2021). stock price volatility during the covid-19 pandemic: the garch model. investment management and financial innovations, 18(4), 12-20. https://doi.org/10.21511/imfi.18(4).2021.02. engle, r., & rangel, g. (2006). the spline garch model for unconditional volatility and its global macroeconomic causes. review of financial studies, 21(3), 1-42. enow, s. t. (2023). forecasting volatility in international financial markets. international journal of research in business and social science, 12(2), 197-203. gherghina, s. c, armeanu, d. s., & joldes, c. c. (2021). covid-19 pandemic and romanian stock market volatility: a garch approach. journal of risk and financial management, 14(341), 1-29. https://doi.org/10.3390/jrfm14080341. gyamfi, e. n., kyei, k. a., & gill, r. (2016). long memory in asset returns and volatility: evidence from west africa. investment management and financial innovations, 13(2), 24-28. hathorn, d. s. (2021). ranking the biggest factors affecting the performance of the cac40. [online].https://www.dailyfx.com/ cac-40/factors-affecting-the-performance-of-the-cac40.html ibrahim, i., kamaludin, k., & sundarasen, s. (2020). covid-19, government response, and market volatility: evidence from the asia-pacific developed and developing markets. economies, 8(105), 1-22. karanasos, m., paraskevopoulos, a. g., ali, f. m., karoglou, m., & yanti, s. (2014). modelling stock volatilities during financial crises: a time varying coefficient approach. journal of empirical finance, 29, 113–28. kenton, w. (2021). durbin watson statistic definition. [online]. https:// www.investopedia .com/terms/d/durbin-watson-statistic.asp krichene, n. (2003). modelling stochastic volatility with application to stock returns. imf working papers 03(125). kuan, c. m. (2002). lecture on the markov switching model. institute of economics academia sinica, 8(15), 1-30. kundu, s., & paul, a. (2022). effect of economic policy uncertainty on stock market return and volatility under heterogeneous market characteristics. international review of economics & finance, 80, 597-612. 95 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 liang, j. n. (2020). corporate bond market dysfunction during covid-19 and lessons from the fed’s response. [online] https://www.brookings.edu/research/corporate-bond-marketdysfunction-during-covid-19-and-lessons-from-the-fedsresponse/ li, y. (2020). this was the fastest 30% sell-off ever, exceeding the pace of declines during the great depression. [online] https:// www.cnbc.com/2020/03/23/this-was-the-fastest-30percentstock-market-decline-ever.html maheu, j. m., & mccurdy, t. h. (2000). identifying bull and bear markets in stock returns. journal of business & economic statistics, 18(1), 100-112. https://www.jstor.org/stable/1392140 malik, f., ewing, b. t., & payne, j. e. (2005). measuring volatility persistence in the presence of sudden changes in the variance of canadian stock returns. the canadian journal of economics/ revue canadienne d’economique, 38(3), 1037-1056. http:// www.jstor.org/stable/3696071. mayhew, s. (1995). implied volatility. financial analysts journal, 51(4), 8-20. http://www.jstor.org/stable/4479853. mike k. p. so., lam, k., & li, w. k. (1998). a stochastic volatility model with markov switching. journal of business & economic statistics, 16(2), 244-253. https://doi.org/10.2307/1392580. mohamed, t. (2022). 6 experts warn markets are at breaking point – and the financial system may be starting to crack. [online] https://www.businessinsider.co.za/stocks-bonds-dollarmarket-experts-financial-system-credit-suisse-banks-2022-10. mishra, p. k., & mishra, s. k. (2020). corona pandemic and stock market behaviour: empirical insights from selected asian countries. millennial asia, 11(3), 341-365, muguto, l., & muzindutsi, p. f. (2022). a comparative analysis of the nature of stock return volatility in brics and g7 markets. journal of risk and financial management, 15(2), 1-27 https:// doi.org/10.3390/jrfm15020085. ndako, u. b. (2012). financial liberalization, structural breaks and stock market volatility: evidence from south africa. applied financial economics, 22(15), 1259-1273. nelson, d. b. (1990). stationarity and persistence in the garch (1, 1) model. econometric theory, 6(3), 318-334. nguyen, d. b. b., prokopczuk, m., & sibbertsen, p. (2020). the memory of stock return volatility: asset pricing implications. journal of financial markets, 47, 100487. 96 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 79–96 oh, g., kim, s., & eom, c. (2008). long-term memory and volatility clustering in high-frequency price changes. physica a: statistical mechanics and its applications, 387(5-6), 12471254. paraschiv, f., reese, s. m., & ringkjøb skjelstad, m. (2020). portfolio stress testing applied to commodity futures. computer management sciences, 17, 203–240. pereira, j. p., & zhang, h. h. (2010). stock returns and the volatility of liquidity. the journal of financial and quantitative analysis, 45(4), 1077–1110. rahman, m. m., guotai, c., gupta, a. d., hossain, m., & abedin, m. z. (2021): impact of early covid-19 pandemic on the us and european stock markets and volatility forecasting. economic research-ekonomska istraživanja, 1-19. https://doi.org/10.108 0/1331677x.2021.1997626. ruilova, j. c., & morettin, p. a. (2020). parsimonious heterogeneous arch models for high frequency modeling. journal of risk and financial management, 13(38), 1-19. shang, y., li, h., & zhang, r. (2021). effects of pandemic outbreak on economies: evidence from business history context. front. public health, 9(632043), 1-12. thupayagale, p. (2012). long memory in the volatility of local currency bond markets: evidence from hong kong, mexico and south africa. in (ed.), risk management current issues and challenges. intechopen. https://doi.org/10.5772/50888. topcu, m., yagl, i., & emirmahmutoglu, f. (2021). covid-19 and stock market volatility: a time-varying perspective. economic bulletin, 41(3), 1-11. visaltanachoti, n., & pukthuanthong-le, k. (2009). idiosyncratic volatility and stock returns: a cross country analysis. applied financial economics, 19(16), 1269-1281. wang, j., & yang, m. (2017). conditional volatility persistence. [online] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3080693 ijbf7-marina.indd the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 139 ijbf institutional and macroeconomic determinants of stock market development in mena region: new results from a panel data analysis mondher cherif and kaouthar gazdar rheims university euromed marscille cemm and rheims university abstract this paper provides new evidence on the infl uence of macroeconomic environment and institutional quality on stock market development, using data from 14 mena countries over the period of 1990-2007. using both panel data and instrumental variable techniques, we found that income level, saving rate, stock market liquidity, and interest rate infl uence stock market development with the expected theoretical signs. our results also showed that the banking and the stock market sectors are complementary instead of being substitutes. we found that the institutional environment as captured by a composite policy risk index does not appear to be a driving force for the stock market capitalisation in the region. our last results are robust to different specifi cations and empirical techniques. keywords: ethnic heterogeneity, institutions, instrumental variable technique, mena countries, political risk, panel data. jel classifi cations: g20, g28, o55. 1. introduction as an integral part of fi nancial development, stock markets have received over the last decade, a great deal of attention as a source of economic growth. the theoretical argument for linking fi nancial development to growth is that a well-developed fi nancial system performs several critical functions to enhance the effi ciency of intermediation by reducing information, transaction, and monitoring costs. indeed, several previous studies on fi nance and development highlight that countries with better developed fi nancial systems would experience faster economic growth. then, the question of what determines stock market development becomes important and is the subject of a large and still growing research literature from which some general conclusions can be drawn. in brief, ht tp :// ijb f.u um .e du .m y 140 the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 there is agreement that countries should adopt appropriate macroeconomic policies, encourage competition within the fi nancial sector, and develop a strong and transparent institutional and legal framework for fi nancial sector activities. the development of equity markets is crucial toward further development of the fi nancial system. the stock market increases fl exibility in the fi nancial intermediation process, as it provides investors with a clear exit strategy. further more, the stock market provides an important indicator for information sharing among investors, company valuation, and the prospect of macroeconomic fundamentals. indeed, banks dominate fi nancing in several developing and even in most developed countries, and stock markets remain a small part of the overall fi nancial system. early research emphasized the role of the banking sector in economic growth. king and levine (1993a, b) show that bank development affects economic growth in a sample of more than 80 countries. levine (1997) (levine zervos, 1998), beck, demirguc and levine (2000) and levine, loayza and beck (2000) have confi rmed this fi nding. however, these studies have omitted the role of the stock market and recent research has therefore begun to focus on the linkages between the stock markets and economic development. with the growing importance of stock markets in the context of fi nancial liberalization and global integration, a number of theoretical and empirical studies have focused on stock market indicators and economic growth. their fi ndings showed how stock market development might boost economic growth. levine and zervos (1998), for instance, found that stock market development plays an important role in predicting future economic growth. the results of beck and levine (2004) confi rmed these fi ndings. using a panel data set of 40 countries and applying the generalized method of moment technique, they found that stock markets and banks positively infl uence economic growth. while the question of whether stock markets promote economic growth has gained considerable attention in academic and policy works, there are little theoretical and empirical studies on the determinants of stock market development in mena countries. mena countries have embarked on economic reform and structural adjustment programs. however, they have not yet emerged as economics powers, which may explain the lack of research on mena capital markets (ben naceur, 4 (2007)). from a geographical perspective, relatively little of the recent research in fi nancial development has been directed at the mena region. hence, the purpose of this paper is to fi ll the void in the literature and make an in-depth analysis of the mena fi nancial sector in order to identify their main determinants. to better understand what drives the stock market development in this region, we explored in this paper both macroeconomic and potential institutional determinants. the scope of our study covered 14 mena countries during the 19902007 periods. we employed two different econometric methodologies. firstly, we use a panel data analysis under the fi xed and random effects specifi cations. our results show that while macroeconomic factors such as saving rate, fi nancial intermediary development, and stock market liquidity are important determinants of stock market development in the mena region, institutional environment is ht tp :// ijb f.u um .e du .m y institutional and macroeconomic determinants of stock market development in mena region: 139-159 141 1aghion, alesina and trebbi (2004) argue that political institutions infl uence economic policy, but they are themselves endogenous since they are chosen, in some way, by members of the polity. not a good predictor of stock market capitalization. thus, the political risk index has no signifi cant effect on stock market development. secondly, building on the institution and development literature, we applied the instrumental variable (iv) techniques that besides endogeneity1 accounts for measurement error in the institutional quality proxies. in doing so, we avoided the shortcomings of the existing literature in this area. the results of instrumental variable estimations show that our fi ndings were unaltered. in fact, while macroeconomic factors are important determinant' of stock market capitalization, institutional environment has no effect on stock market development. the remainder of the paper is structured as follows. section 2 provides a brief literature review of macroeconomic and institutional driving forces for stock market development. section 3 presents the evolution of the stock market in the mena countries. section 4 presents the data and the adopted econometric methodology. the empirical results are presented and interpreted in section 5. finally, section 6 offers some conclusions and policy recommendations. 2. determinants of stock market development: a brief literature review major prior studies related to the present paper include demirguç-kunt and levine (1996), levine and zervos (1998), garcia and liu (1999), boyd, levine and smith (2001), beck, demirguc-kunt and levine (2003), ben naceur et al. (2007) as well as yartey (2008). these studies focused on the effects of macroeconomic and institutional factors on stock market development in developed and developing countries. unlike the study of ben naceur et al. (2007) conducted with a special focus on the mena region, we added to the macroeconomic factors, the institutional determinants of the stock market development in this region. in studies on fi nance and development, a causal relationship between fi nancial development and economic growth has been suggested along three lines: fi rstly, fi nancial deepening promotes economic growth; secondly, economic growth stimulates fi nancial development; and thirdly, economic growth and fi nancial development infl uence each other (levine, 2005). against this background, analyzing what determines stock market development has become a prominent topic in recent years. in this literature, macroeconomic and institutional factors have been found to be the most important driving forces for the development of stock markets. a large body of evidence has documented the importance of macroeconomic conditions in determining fi nancial development. recent research shows that stock market development might boost economic growth, and empirical evidence tends to provide some support to this fi nding. demirguc-kunt and levine (1996) examine the relationship between stock market development and fi nancial intermediary in developing countries and fi nd that most stock market ht tp :// ijb f.u um .e du .m y 142 the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 2 information sharing institutions. indicators are highly correlated with fi nancial intermediary development. levine and zervos (1998) presented empirical evidence that stock market development plays an important role in predicting future economic growth. garcia and liu (1999) investigated empirically the macroeconomic determinants of stock market development. using pooled data from 15 industrial and developing countries from 1980 to 1995, they found that real income level and stock market liquidity are important predictors of market capitalization, while macroeconomic stability does not have any explaining power. they also concluded that banks and markets are complementary instead of being substitutes. one of the basic pre-requisites for fi nancial market development is macroeconomic stability. using a cross sectional regressions and a dynamicpanel generalized-method of moments (gmm) estimator for a sample of 65 countries over 1960-1995 period, boyd et al. (2001) provided evidence that there is a signifi cant and economically important negative relationship between infl ation and fi nancial development. building upon garcia and liu's (1999) work, ben naceur et al. (2007) examined the macroeconomic determinants of stock market development in the mena region. using an unbalanced panel data from 11 mena countries (over 1979-1999) and employing fi xed and random effects specifi cations, they found that saving rate, credit to private sector, the ratio of value traded to gdp and infl ation change are the important determinants of stock market development. similar to garcia and liu (1999), they also found that fi nancial intermediaries and stock markets are complements rather than substitutes in the growth process. there is also a substantial amount of research trying to identify the impact of the institutional quality on fi nancial development. this strand of research goes back to the seminal contributions of la porta, lopez. de-silanes, sheifl er and vishny (1997 and 1998) on how the legal rules covering protection of corporate shareholders and creditors, the origin of these rules, and the quality of their enforcement affect fi nancial development. applying the propositions of la porta et al. (1997 and 1998) to the transitions economies, pistor,raiser and stanislaw (2000) found that the effectiveness of legal institutions has a stronger impact on equity and credit market development. in an extension, beck et al. (2003) evaluated empirically the law and endowments theories of fi nancial development. using cross-country regressions on a sample of 70 former colonies, they have provided evidence for both theories. thus both legal systems and country’s initial endowments are important determinants of fi nancial development. using a new sample of 129 countries over 25 years, djankov, mcliesh and shleifer (2007) report that creditor protection through the legal system, and information sharing institutions are associated with higher ratios of private credit to gdp. their results also showed that improvements in creditor rights or the introduction of credit registries2 leads to an increase in the private credit to gdp ratio. in addition, they reported that legal origins are an important determinant of both creditor rights and information sharing institutions. ht tp :// ijb f.u um .e du .m y institutional and macroeconomic determinants of stock market development in mena region: 139-159 143 using arellano and bond dynamic panel gmm estimations and several data sets over 1980-2003 periods, baltagi, demetriades and law (2007) fi nd that institutions can explain a large part of the variation in fi nancial development across countries and over time. the predictions of the theoretical model developed by holder (2007) are consistent with the previous empirical works. his fi nding shows that better property rights institutions make fi nancial repression more costly for the elite and tend therefore to increase fi nancial development. their predictions also show that better contracting institutions lower the costs of fi nancial transactions, which have countervailing effects on equilibrium fi nancial development. girma and shortland (2008) contributed to this literature by evaluating the infl uence of the political system and legal origin in fi nancial development. using panel data on developed and developing countries from 1975-2000, their results show that the degree of democracy and political stability are signifi cant explanatory factors in determining the speed of fi nancial development. in a more recent study to examine the determinants of fi nancial development, law and habibullah (2009) used data from 27 economies during 1980-2001. their dynamic panel data analysis results demonstrated that institutional quality is a statistically signifi cant determinant of banking sector development and capital market development. besides the legal framework discussed above, informal institutions begin to gain some attention among economists. calderon et al. (2001) exploited the link between trust and both the structure and development of fi nancial system. examining the simple correlation analysis for a sample of 48 countries during 1980-1995, calderon et al. (2001) fi nd that trust is positively linked with both fi nancial development and effi cient fi nancial structure. when they use the ordinary least squares regressions they have found that trust and rule of law are strongly related to fi nancial system indicators. in the empirical literature of stock market development, an oftencited paper related to our study is yartey (2008). this study examines the macroeconomic and institutional determinants of stock market development. using a panel data of 42 countries over 1990 to 2004, he found that income level, gross domestic investment, banking sector development, private capital fl ows, and stock market liquidity are important determinants of stock market development. he also provides evidence that institutional factors such as law and order, political risk, and bureaucracy quality are important determinants of stock market development. 3. stock market evolution in the mena region the mena countries have reformed their fi nancial sectors over the past three decades. however, while they have made progress, their efforts have been eclipsed by faster reform and growth in other parts of the world. the stock markets in our sample of mena countries have seen considerable development since the 1990s. the market capitalization of mena countries has risen from ht tp :// ijb f.u um .e du .m y 144 the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 $124 billion in 1995 to more than $1635 billion in 2007 (world development indicators: wdi 2007). to understand the economic importance of the stock market capitalization in our sample, we examined the capitalization ratio. this ratio is defi ned as the value of domestic equities traded on the market relative to gdp. as we can observe from figure 1, this ratio has increased from 14% in 1990 to about 104% in 2007. the high growth of the capitalization ratio coincided also with an increase in the number of listed companies. the number of listed companies has more than doubled; growing from less than 1050 companies in 1990 to about 2400 companies in 2007. source: world development indicators (world bank, 2007) and author’s calculations. to examine the mena region stock markets depth, this paper measures the activity of stock market using the total value traded as share of gdp, which gives the value of stock transactions relative to the size of the economy. figure 2 shows that this measure increased from about 4 % in 1990 to roughly 56% in 2007. source: world development indicators (world bank, 2007) and author’s calculations. figure 1. stock market capitalization in the mena countries (1990-2007) figure 2. total value traded per gdp in mena countries (1990-2007) ht tp :// ijb f.u um .e du .m y institutional and macroeconomic determinants of stock market development in mena region: 139-159 145 3following demirguç-kunt and levine (1999), we consider the total value traded as a share of gdp as a measure of stock market activity. this measure is also used to gauge market liquidity because it measures trading relative to economic activity (levine and zervos, 1998). to clearly understand the liquidity picture, we examine the turnover ratio. the turnover ratio is defi ned as the ratio of the value of total shares traded to market capitalization. it measures the activity of the stock market transactions relative to its size. many analysts use the turnover ratio as a measure of transactions costs. high turnover ratio implies high transaction and, consequently, high effi ciency. in our sample of countries the turnover ratio has increased from under 33% in 1990 to about 49% in 2007 which can be interpreted as an effi ciency gain in the mena region stock markets. source: world development indicators (world bank, 2007) and author’s calculations. within the mena region, there is substantial variation in the degree of fi nancial development. some countries have advanced fi nancial sectors, while for others progress in this area has been limited. as we can observe from table 1, stock market development indicators exhibit a considerable variability across countries and over market capitalization and market activity. when considering the market capitalization, jordan, qatar, and israel seem to outperform other countries; while tunisia, iran, and lebanon are at the end of the list. in terms of activity,3 saudi arabia and kuwait have a relatively more active stock market, followed by jordan. however, tunisia, bahrain, malta and lebanon have the least active stock markets. this is partly as a result of the limited number of companies that are listed on these country exchanges. for example, in 2007 the numbers of listed companies were 50, 15 and 11 in tunisia, malta and lebanon respectively. finally, on the basis of the turnover ratio, we fi nd that saudi arabia has relatively the most liquid stock market followed by kuwait. figure 3. turnover ratio in the mena countries (1990-2007) ht tp :// ijb f.u um .e du .m y 146 the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 table 1 indicators of stock market development in the mena countries, 1990-2007 country number of listed companies market capitalization/ gdp value traded / gdp turnover ratio bahrain 43 0.992 0.0438 0.043 egypt 435 0.339 0.096 0.292 iran 329 0.143 0.022 0.162 israel 654 0.583 0.292 0.466 jordan 245 1.072 0.364 0.264 kuwait 181 0.822 0.523 0.664 lebanon 11 0.162 0.024 0.123 malta 15 0.352 0.022 0.056 morocco 74 0.312 0.056 0.175 oman 125 0.239 0.060 0.250 qatar 40 0.885 0.146 0.235 saudi arabia 111 0.616 0.745 1.012 tunisia 50 0.118 0.013 0.100 uae 90 0.468 0.194 0.483 source. world development indicators (world bank 2007) and author’s calculations. 4. data and econometric methodology the approach taken in this paper was to model the impact of macroeconomic and institutional factors on stock market development in mena region. our initial intention was to cover all countries in the mena region, but given that some countries have not yet established stock markets (for example djibouti, libya, syria, and yemen) and other countries have established stock markets recently (for example, algeria), the sample included are only 14 mena countries: bahrain, egypt, iran, israel, jordan, kuwait, lebanon, malta, morocco, oman, qatar, saudi arabia, tunisia and uae over the period of 1990-2007. a. the data data were extracted from various sources. employing the november 2008 beck et al. (2000) database on fi nancial development and structure, stock market and fi nancial system indicators from 1990 to 2007 are extracted. other information related to macroeconomic stability, saving and investment rates are collected from the world development indicators (world bank, 2007) database. the institutional indicators were collected from the international country risk guide (icrg) compiled by the political risk services (prs group). these indicators rely exclusively on polls of experts. the main advantages of these datasets are that they were available for a considerable time span; thus allowing to test the dynamics and relevance of institutions in affecting stock market development (daude and stein 2007). ht tp :// ijb f.u um .e du .m y institutional and macroeconomic determinants of stock market development in mena region: 139-159 147 4stock market development is a multi-dimensional concept, it is usually measured by the number of listed companies, stock market liquidity, etc. we focused on market capitalization as a percentage of gdp because it is less arbitrary than the other measures. the dependent variable of interest is stock market development. following yartey (2008), we measured stock market development using market capitalization as a proportion of gdp . this measure equals the value of listed shares divided by gdp.4 the independent variables consisted of a set of macro and institutional factors as defi ned below. b. the variables macroeconomic factors as macroeconomic factors, we adopt those as defi ned in (garcia and liu (1999)). income level: real income has been found to be highly correlated with the size of the stock market. we used the logarithm of the real gdp in us dollars to measure the income level. to avoid the causality problem, we simply use last year’s income level. we expected to have a positive impact on stock market development. savings: the saving rate is calculated as the ratio of gross saving to gdp. we expected a positive effect on the stock market size. to avoid the causality problem, we used last year’s saving. investment rate: investment is considered an important determinant of stock market capitalization as stock markets represent one way to intermediate saving to investment projects. we use the ratio of gross fi xed capital to gdp as a measure of investment. we expect it to be the important determinant of stock market capitalization. to avoid the causality problem, we used last year’s investment rate. financial intermediary development: since both bank and stock markets channel savings toward investment projects, they can be either complements or substitutes. we used two indicators of fi nancial intermediary development. the fi rst was the domestic credit to private sector. this indicator measures the role of banks in providing long-term fi nancing to private corporations. the second was the ratio of broad money supply m3 to gdp. this ratio is a measure of the size of the banking sector in relation to the economy as a whole. stock market liquidity: we used two indicators of stock market liquidity. the fi rst is the turnover ratio, defi ned as the total value of domestic shares traded divided by market capitalization. it indicates the trading volume of the stock market relative to its size. the second is the value traded ratio, which equals the total value of domestic equities traded on each country’s major stock exchanges as a percentage of gdp. we expected this measure to have a positive impact on stock market capitalization because a large amount of savings is channeled through stock markets. ht tp :// ijb f.u um .e du .m y 148 the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 macroeconomic stability: macroeconomic stability may be an important factor for the development of the stock market. to determine the impact of macroeconomic stability on market capitalization, we used two indicators of macroeconomic stability: real interest rate and current infl ation. we expected that there is a strong negative relationship between these two indicators and stock market development. institutional factors the study of the impact of the institutional environment on stock market development is related to the recent literature on the relationship between the legal institutional framework and corporate fi nance. we contribute to this literature by looking at the link between stock market development and political risk, a measure of the institutional quality that supports the viability of external fi nance. the paper looks also at the impact of institutional quality on stock market development because it is widely believed that the strengthening of institutions in a country could broaden appeal and confi dence in stock market investment. as mentioned by perotti and van oijen (2001), equity investment thus becomes gradually more attractive as political risk is resolved over time. yartey (2008) showed that the political risk and institutions are strongly associated with growth in stock market capitalization. thus, the development of good quality institutions (resolution of political risk) can be an important factor in the development of stock markets in emerging economies. therefore, the development of good quality institutions can improve the attractiveness of equity investment and lead to stock market development. to assess the role of public institutions in determining stock market development, we constructed a yearly composite political risk index5 using the icrg variables from the prs group. the composite index is the sum of the 12 indices of government stability, socioeconomic conditions, investment profi le, internal confl ict, external confl ict, corruption, military in politics, religion in politics, law and order, ethnic tensions, democratic accountability, and bureaucracy quality. this index ranges from zero to one hundred, with lower values suggesting poorly performing institutions. we expected countries with good quality of institutions and therefore low political risk to have welldeveloped stock markets. 5to capture the effect of institutional quality on the fl ows of capital from rich countries to poor countries, alfaro, kalemli-ozkan and volosovych (2008) constructed a yearly composite index using the icrg variables from the prs group (2001). to examine the institutional determinants of fi nancial fl ows, papaionnou (2009) also used a proxy of institutional quality, a composite indicator constructed by prs, namely, the icrg "political risk" rating. ht tp :// ijb f.u um .e du .m y institutional and macroeconomic determinants of stock market development in mena region: 139-159 149 6the four measures were chosen because of their importance in past results. yartey (2008) found that law and order, democratic accountability, and bureaucracy quality are important for stock market development. 7fixed effects model indicates that the individual effects are correlated with the explanatory variables. in accordance with yartey (2008), who argued that the political risk indicator has a problem in the sense that it imparts very little direction toward targeting a particular aspect of institutions, we considered the impact of four sub-indicators of the composite icrg index on stock market development: law and order, bureaucracy quality, democratic accountability, and corruption.6 quality of bureaucracy: [0-6] high points are given to countries where the bureaucracy has the strength and expertise to govern without drastic changes in policy or interruptions in government services. law and order: [0-6] index where higher scores indicate sound political institutions, a strong court system, and provision for an orderly succession of power. corruption: [0-6] index refers to corruption in the political system. countries that have low levels of corruption have high values of the index and vice versa. democratic accountability: [0-4] this is a measure of how responsive the government is to its people. c. econometric methodology the model to be estimated is the following: yi,t =i + macroi,t+  instit + μi,t , for i = 1, 2,…n, t = 1,2,…ti for i = 1, 2,…n, t = 1, 2,...ti (1) where y i,t the dependant variable, is defi ned as mcap, macro is a matrix of macroeconomic variables made up of income level, saving rate, investment rate, credit to private sector, m3 to gdp, stock market liquidity, and macroeconomic instability. the inst variable is the indicator of institutional quality, i is the unobserved country specifi c fi xed effect, and μ i,t is the error term for each observation. fixed effects as well as random effects models were considered in this study. we used the hausman test to select the appropriate estimator. if the hausman test rejects the null hypothesis that the individual effects are not correlated with the explanatory variables, the most suitable estimation would then be the fi xed-effects model.7 while the panel data techniques (fi xed effects and random effects specifi cations) account for time-invariant country characteristics and time trends that may infl uence stock market development, fi xed and random effects models are not a panacea, since reverse causation and measurement error might still plague the estimates when we consider the institutional quality. thus, when ht tp :// ijb f.u um .e du .m y 150 the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 we considered the institutional quality, we used the instrumental variable (iv) techniques in addition to the fi xed effects and random effects models. besides endogeneity, the instrumental variable (iv) estimates also accounted for measurement error in the institutional proxies. 5. the results the results were grouped and presented in three sub-sections: (a) macroeconomic determinants of stock market development, (b) institutional quality and stock market development, and (c) unbundling. as a prelude to these three categories, table 2 provides summary statistics on the variables. table 2 descriptive statistics variable mean std. dev. min max market capitalization .481 .479 .019 2.98 income level 10.407 .522 9.283 11.552 saving rate .212 .107 -.044 .419 investment rate .221 .054 .106 .392 value traded .181 .465 .0007 3.934 turnover ratio .288 .399 .008 2.884 domestic credit .570 .213 .206 1.157 m3/gdp .801 .416 .260 2.490 infl ation .060 .096 -.013 .807 interest rate .060 .084 -.362 .314 political risk index 67.183 10.021 10.333 87.583 a. macroeconomic determinants of stock market development table 3 summarizes the results of the fi xed and random effects models for the sample of the 14 mena countries from 1990 to 2007. in the fi rst column of table 2, we present results when we include only the last year’s income, last year’s saving rate, domestic credit to the private sector divided by gdp and last year’s value traded to gdp ratio in the regression (1). we used it as the basic regression. the results showed that last year’s saving rate, domestic credit to private sector and last year’s traded to gdp ratio have a positive and signifi cant effect on stock market capitalization. conversely, last year’s income has no signifi cant impact on market capitalization. when last year’s saving increases by ht tp :// ijb f.u um .e du .m y institutional and macroeconomic determinants of stock market development in mena region: 139-159 151 one percentage point, market capitalization increases by 1.58 percentage points, which implies that most of the increase in savings is channeled through the stock markets. if value traded to gdp ratio increases by one percentage point, market capitalization increases by 1.06 percentage points. therefore, stock market liquidity has a positive effect on market capitalization. financial intermediaries also promote stock market development. when domestic credit to private sector increases by one percentage point, stock market capitalization increases by 1.22 percentage points. to test the effect of the investment rate on stock market capitalization, regression (2) includes last year’s investment instead of savings rate. the result showed that investment rate is not a good predictor of market capitalization, since its coeffi cient is insignifi cant. however, contrary to regression (1), income level has a signifi cant and a positive effect on stock market capitalization. to test the effect of an alternative measure of fi nancial intermediary development on market capitalization, regression (3) includes m3 to gdp ratio instead of domestic credit to private sector. this confi rmed the positive impact of fi nancial intermediary growth on development of the stock market. comparing regressions (1) and (3), we recorded that domestic credit to the private sector seems to be a better measure of fi nancial intermediary and a better predictor of stock market development. if m3 to gdp ratio increases by one percentage point, market capitalization increases by 0.48 percentage points. this is consistent with our expectation. in order to test the incidence of another measure of stock market liquidity, regression (4) includes least year’s turnover ratio instead of the ratio of value traded to gdp. the turnover ratio has a positive and signifi cant effect on market capitalization. when last year’s turnover ratio increases by one percentage point, market capitalization increases by 0.87 percentage points. comparing regressions (1) and (4), we can see that the value traded to gdp is a better measure of stock market liquidity, and plays a more important role in determining stock market capitalization. to investigate the effect of macroeconomic stability, current infl ation is introduced in model (5). infl ation has a positive sign even though it is not statistically signifi cant. to examine the effect of real interest rate on stock market capitalization, we used real interest rate instead of infl ation in model (6). the results show that last’s year income, saving rate, and value traded are all positive and statistically signifi cant in explaining stock market capitalization. real interest rate is signifi cant and has the expected negative sign. finally to test the hypothesis that the investment rate is a better predictor of stock market capitalization than the savings rate, we compared regressions from (3)-(6) with regressions (7)-(10) when we employ the investment rate instead of savings rate. the results proved the predominance of savings rate as a best predictor of stock market capitalization. ht tp :// ijb f.u um .e du .m y 152 the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 r eg re ss io n (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) (9 ) (1 0) in co m e le ve l .0 69 (0 .4 7) 1. 00 2* (4 .6 9) .7 54 * (4 .2 2) 1. 61 6* (4 .1 0) .8 40 ** (2 .7 0) 1. 38 4* (3 .5 3) 1. 25 7* (7 .3 2) 1. 40 2* (5 .1 4) 1. 16 6* (4 .8 7) 1. 30 1* (5 .0 5) s av in gs r at e 1. 58 6* (4 .0 4) 1. 10 7* * (2 .1 6) -. 25 6 (0. 33 ) 1. 13 8* ** (1 .9 5) 2. 08 4* * (2 .4 9) in ve st m en t r at e -. 12 5 (0. 21 ) .5 70 (0 .9 0) -. 54 4 (0. 68 ) .0 29 (0 .0 5) -. 40 1 (0. 62 ) d om es ti c c re di t 1. 22 6* (5 .2 5) .7 85 * (2 .7 4) .9 23 (1 .5 0) .5 28 (1 .1 6) .4 54 (0 .6 9) 1. 14 3* (2 .9 3) 1. 04 1* (3 .3 5) -. 02 5 (0. 08 ) m 3/ g d p .4 82 ** (2 .5 8) .5 21 * (2 .6 9) v al ue t ra de d 1. 06 0* (9 .4 3) .2 93 * (5 .6 5) .9 48 * (8 .2 8) .9 23 * (6 .8 5) .8 21 * (5 .1 7) .2 64 * (5 .1 9) .2 61 * (4 .8 7) .7 06 * (6 .9 3) t ur no ve r r at io .8 73 * (3 .9 8) .3 79 * (4 .7 0) in fl at io n .6 79 (1 .1 1) 1. 53 6* (2 .6 3) in te re st r at e -2 .4 82 * (2. 72 ) -. 80 0* * (2. 43 ) r 2 0. 79 0. 51 0. 64 0. 53 0. 63 0. 69 0. 51 0. 47 0. 55 0. 64 h st at is ti c (h au sm an sp ec ifi c at io n) 0. 06 1 (r e ) 0. 00 (f e ) 0. 00 (f e ) 0. 00 (f e ) 0. 00 (f e ) 0. 00 (f e ) 0. 00 (f e ) 0. 00 (f e ) 0. 00 (f e ) 0. 00 (f e ) n o te : p an el e st im at io ns o f 14 m e n a c ou nt ri es . t he d ep en da nt v ar ia bl e is t he r at io o f m ar ke t ca pi ta li za ti on t o g d p . h -s ta ti st ic s co rr es po nd t o h au sm an t es t fo r co m pa ri so n be tw ee n fi xe d (f e ) or r an do m ( r e ) ef fe ct s sp ec ifi c at io ns . t -s ta ti st ic s fo r th e co ef fi ci en ts a re i n pa re nt he se s. * si gn ifi c an t at 1 % , * *s ig ni fi ca nt a t 5% , a nd * ** si gn ifi c an t at 1 0% . t ab le 3 m a cr o ec o n o m ic d et er m in a n ts o f s to ck m a rk et d ev el o p m en t ht tp :// ijb f.u um .e du .m y institutional and macroeconomic determinants of stock market development in mena region: 139-159 153 8regression (1) table (2). 9to instrument corruption mauro (1995) has used the ethno-linguistic fragmentation. 10 to instrument political stability and freedom daude and stein (2007) have used also the ethno-linguistic fragmentation. 11easterly and levine (1997) found empirical evidence to support their claim that the very high level of ethnic diversity of countries in africa is an important contributor to their poor economic performance. 12 the index of ethnic fractionalization is the probability that two randomly selected individuals from a given country will not belong to the same ethnic group. b. institutional quality and stock market development we now report the impact of institutions on stock market development. the results of the panel and instrumental variable estimations are presented in table 4. to test the effect of institutional quality on stock market development, we added the political risk index to our baseline regression.8 as we can see, last’s year income level, savings rate, and total value traded have a positive and signifi cant effect on stock market capitalization. the domestic credit to private sector was also positive but statistically insignifi cant. the political risk index has the expected positive sign even though it is not statistically signifi cant. this outcome indicates that institutional environment is not a good predictor of stock market development in mena countries. our second approach was to run iv regressions using instruments that are not subject to reverse causality and can account for the institutional variation. following the institutions and development theories (mauro 19959; daude and stein 2007)10, we use the indicators of ethnic diversity as instruments. la porta, loped-de-silanes, sheifl er and vishny (1999) pointed out that ethnic diversity leads to corruption and low effi ciency in governments that expropriate the ethnic losers. several authors have interpreted the fi nding of a positive relationship between ethnic diversity and poor economic performance to be a consequence of the high probability of confl ict associated with a highly fractionalized society11. for this reason, many papers use the ethno-linguistic fractionalization index as the indicator of ethnic heterogeneity. in a more recent study, aghion et al. (2004) argued that racial fragmentation and institutions are not independent from each other. building on this theory, we use in regression (2) an index of ethnolinguistic fragmentation from roeder (2001) as an instrument of political risk index. the fi rst interesting result is that in term of signifi cance, the results are analogous to the fi xed effect models. thus, political risk index has a negative sign even though it is not a signifi cant determinant of stock market development. in contrast, fearon (2003) pointed out that the index of fractionalization12 cannot capture important differences in ethnic structures. similarly, montalvo, and reynal-querol (2005) argued that the measure of ethnic heterogeneity ht tp :// ijb f.u um .e du .m y 154 the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 r eg re ss io n (1 ) p an el t ec hn iq ue s es ti m at io n (2 ) iv e st im at io n (i ns tr um en t = e th ni c fr ag m en ta ti on ) (3 ) iv e st im at io n (i ns tr um en t = e th ni c po la ri za ti on ) (4 ) iv e st im at io n (i ns tr um en ts = e th ni c fr ag m en ta ti on & e th ni c po la ri za ti on ) in co m e l ev el .7 45 * ( 2. 80 ) .0 21 ( 0. 26 ) .3 14 ( 0. 50 ) .1 33 (0 .5 1) s av in g 1 .1 91 ** ( 2. 23 ) 2. 50 6* ( 2. 54 ) 6 .9 23 ( 1. 01 ) 4 .0 37 (1 .5 3) d om es ti c c re di t .2 90 ( 0. 72 ) 1. 69 4* ( 2. 86 ) 6 .1 33 ( 0. 97 ) 3 .4 36 (1 .3 8) v al ue t ra de d .9 87 * ( 8. 32 ) 1. 26 8* ( 5. 30 ) 1 .1 39 * ( 3. 39 ) 1 .1 83 * (5 .4 5) p ol it ic al r is k in de x .0 02 ( 0. 53 ) .0 25 ( -1 .2 4) -. 22 8 ( -0 .8 0) .1 03 ( -0 .9 3) r 2 0 .6 2 0. 52 0 .6 0 0 .4 3 p -v al ue s (h au sm an t es t) 0 ,0 14 ( f e ) p -v al ue s (h an se n jte st ) 0 .1 n o te . p an el e st im at io n (m od el 1 ) an d in st ru m en ta l v ar ia bl es e st im at io n (m od el s 2, 3 a nd 4 ) of th e 14 m e n a c ou nt ri es . t he d ep en da nt v ar ia bl e is th e ra ti o of m ar ke t c ap it al iz at io n to g d p . h -s ta ti cs co rr es po nd s to h au sm an t es t fo r co m pa ri so n be tw ee n fi xe d (f e ) or r an do m ( r e ) ef fe ct s sp ec ifi c at io ns . jte st c or re sp on ds t o h an se n' s ov er id en ti fi ca ti on t es t. t he t -s ta ti st ic s fo r th e co ef fi ci en ts ar e in p ar en th es es . w e us ed e th ni c fr ac ti on al iz at io n as i ns tr um en t in m od el ( 2) , th e et hn ic p ol ar iz at io n in de x as i ns tr um en t in m od el ( 3) , an d bo th e th ni c fr ac ti on al iz at io n an d et hn ic p ol ar iz at io n as i ns tr um en ts i n m od el ( 4) . *s ig ni fi ca nt a t 1% , * *s ig ni fi ca nt a t 5% , a nd * ** si gn ifi c an t at 1 0% . t ab le 4 in st it u ti o n a l d et er m in a n ts o f s to ck s m a rk et d ev el o p m en t in m e n a c o u n tr ie s ht tp :// ijb f.u um .e du .m y institutional and macroeconomic determinants of stock market development in mena region: 139-159 155 13in the fractionalization index, the size of each group has no effect on the weight of the probabilities of two individuals belonging to different groups, whereas in the polarization index these probabilities are weighted by the relative size of each group. see montalvo and reynol (2005) for more explications. 14papaioannou. (2009) has used ethnic polarization to instrument institutional quality. 15the p-values (10%) exceed the conventional 5% signifi cance level. appropriate to capture potential confl ict should be a polarization measure13. in fact, in accordance with horowitz (1985), montalvo, and reynal-querol (2005) showed that the most severe confl icts arise in societies where a large ethnic minority faces ethnic majority. the index of ethnic fractionalization is not able to capture this idea appropriately. as the result, in regression (3) we used the ethnic polarization14 from montalvo and reymal-querol (2005) as an instrument of institutional quality. as with ethnolinguistic fractionalization, this instrumentation strategy also suggests that the index of political risk has no signifi cant effect on stock market capitalization. thus, the political risk index appears an insignifi cant determinant of stock market development. regression (4) reports iv models using the two indicators of ethnic diversity (ethnic fragmentation and ethnic polarization) as instruments for institutional development. this approach is the most effi cient since it helps obtain a stronger fi rst stage fi t and more properly isolate the exogenous components of institutions. it is also helpful, since having more than one instrument, one can test for instrument validity performing. we used hansen’s over identifi cation test (j-test) to check the null hypothesis of whether the instruments for institutions we chose are valid. p-values for the over identifi cation (10%)15 suggest that we cannot reject the null hypothesis of instrument validity. c. unbundling as mentioned earlier, the problem with the concept of political risk index is that it tells us very little about which aspect of institution policies should be targeted for change. to remedy this defi ciency, the paper studied the impact of some components within the index of political risk on stock market development. the results are reported in table 5. in regression (1), we used the protection against corruption index. the coeffi cient on corruption is statistically insignifi cant at standard confi dence levels. in regression (2), we looked at the effect of bureaucratic quality. good bureaucracy quality enhances the regulatory capacity of countries and therefore should be positively associated with stock market development. however, the result shows that bureaucracy quality does not appear a signifi cant determinant of stock market development. democratic accountability and law and order do not also appear a signifi cant determinant of market capitalization in mena countries. ht tp :// ijb f.u um .e du .m y 156 the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 table 5 institutional determinants (sub-indicators of political risk index) of stock market development in mena countries regression (1) (2) (3) (4) income level .008 (0.01) .779* (3.13) .069 (0.64) .029 (0.38) saving 1.714* (4.72) 1.269** (2.35) 1.556* (3.83) 1.672* (4.93) domestic credit 1.190* (5.05) .294 (0.73) 1.187* (4.81) 1.222* (5.49) value traded 1.129* (9.66) .983* (8.27) 1.056* (9.29) 1.120* (9.84) bureaucratic quality -.076 (-0.72) corruption -.019 (-0.61) accountability .011 (0.42) law and order .017 (0.94) r2 0.59 0.62 0.60 0.58 hstatistic (hausman specifi cation) 0.19 (re) 0.04 (fe) 0.24 (re) 0.14 (re) note: panel estimations of the 14 mena countries. the dependant variable is the ratio of market capitalization to gdp. h-statistics corresponds to hausman test for comparison between fi xed (fe) or random (re) effects specifi cations. the t-statistics for the coeffi cients are in parentheses. *signifi cant at 1%, **signifi cant at 5%, and ***signifi cant at 10%. 6. conclusions and policy recommendations stock market development is an integral part of fi nancial development, which is, in turn, associated with economic growth. in this paper, we have highlighted the role of selected macroeconomic and institutional variables in explaining stock market development in 14 mena countries for the period 1990-2007. according to previous studies, institutional factors as political risk, law and order, democratic accountability and bureaucratic quality are important determinants of stock market development in emerging markets. hence, the resolution of political risk can increase investor confi dence and propel the growth of the stock markets in emerging economies. the empirical analysis found three interesting results. firstly, we have found that while savings rate, fi nancial intermediary, stock market liquidity, interest rate and income are important determinants of stock market development, investment and infl ation do not prove to be signifi cant. secondly, the banking sector is a complement to the stock market in fi nancing investment and in the ht tp :// ijb f.u um .e du .m y institutional and macroeconomic determinants of stock market development in mena region: 139-159 157 growth process in mena region. these fi rst results are generally in agreement with the theoretical and empirical literature. lastly, when we considered the institutional environment, and in contrast to some recent research (yartey, 2008), we found that the index of political risk has no signifi cant effect on stock market capitalization. our results are robust to the use of different econometrics strategies. in fact, when we instrumented the political risk index by both ethnic fragmentation and ethnic polarization, the results were similar to those of panel data estimates. our fi ndings have important policy implications for mena countries. firstly, economic growth plays a crucial role in stock market development. policymakers in mena may initiate policies to foster growth and development as countries liberalize their fi nancial system. secondly, a well-developed banking sector is important for stock market development in the region and can play a driving force in promoting the stock market as demonstrated by the experiences of many east asian countries. thirdly, domestic savings is an important determinant of stock market development in the mena countries. thus, mena policymakers should encourage savings by appropriate policies. fourthly, stock market liquidity has a positive impact on stock market development. consequently, improving stock market liquidity can be another tool for promoting stock market development. finally, even if our empirical results do not show a positive effect on institutional quality on stock market development, mena countries should improve their institutional framework because good institutions reduce political risk which is an important factor in investment decision. author statement: mondher cherif would like to acknowledge chaire ag2r prémalliance “finance autrement: investissement solidarités responsabilité” for the support of this research. he is a professor at the rheims university: he also holds another position as shown earlier. email: mondherc@gmail.com. references aghion, p., alesina, a., and trebbi, f. (2004). endogenous political institutions. quarterly journal of economics, 119 (2), 565-613. alfaro, l., kalemli-ozkan, s., and volosovych, v. (2008). why dosen’t capital fl ow from rich to poor countries? an empirical iinvestigation. review of economics and statistic , 90 (2), 347-368. baltagi, b. h., demetriades, p. , and law, s.h. (2007). financial development, openness and institutions: evidence from panel data. center for policy research working papers no.7. beck, t., demirguç-kunt, a., and levine, r. (2000). a new database on fi nancial development and structure. world bank economic review, 14 (3), 597-605. ht tp :// ijb f.u um .e du .m y 158 the international journal of banking and finance, vol. 7. number 1: 2010: 139-159 beck, t., levine, r. , and loayza, n. (2000). finance and the sources of growth. journal of financial economics, 58 (1-2), 261-300. beck, t., and levine, r. (2004). stock markets, banks and growth: panel evidence. journal of banking and finance, 28 (3), 423-442. beck, t., demirguc-kunt, a., and levine, r. (2003). law, endowments, and fi nance. journal of financial economic , 70 (2), 137-181. ben naceur, s., ghazouani, s., and omrani, m. (2007). the determinants of stock market development in the mena region. managerial finance, 33 (7), 477-489. boyd, j.h., levine, r., and smith, b.d. (2001). the impact of infl ation on fi nancial sector performance. journal of monetary economics, 47 (2), 221-248. daude, c. and stein. e. (2007). the quality of institutions and foreign direct investment. economics & politic , 19 (3), 317-344. demirguç-kunt, a., and levine, r. (1996). stock markets, corporate fi nance, and economic growth: an overview. world bank economic review, 10 (2), 223-239. djankov, s., mcliesh, c., and shleifer, a. (2007). private credit in 129 countries. journal of financial economics, 84 (2), 299-329. easterly, w., and r. levine. (1997). africa’s growth tragedy: policies and ethnic divisions. quarterly journal of economic , 112 (4), 1203-1250. fearon, j.d. (2003). ethnic and cultural diversity by country. journal of economic growth, 8 (2), 195-222. holder, r. (2007). institutions, trade and the political economy of fi nancial development. nncr ttrade regulation, swiss national centre of competence in research, working paper no. 2007/27. horowitz, d.l. (1985). ethnic group in confl ict. berkeley: university of california press. garcia , v.f., and liu, l. (1999). macroeconomic determinants of stock market development. journal of applied economics, ii (1), 29-59. girma, s. and shortland , a. (2008). the political economy of fi nancial development. oxford economic papers, 60 (4), 567-596. king, r.g., and levine, r. (1993a). finance and growth: schumpeter might be right. journal of economics, 10 (3), 717-738. king, r.g., and levine, r. (1993b). finance, entrepreneurship, and growth: theory and evidence. journal of monetary economics, 32 (3), 513-542. la porta, r., lopez-de-silanes, f., sheifl er, a., and vishny, r.w. (1999). the quality of government. journal of law, economics and organization, 15 (1), 222-279. la porta, r., lopez-de-silanes, f., sheifl er, a., and vishny , r.w. (1998). law and fi nance. journal of political economy, 106 (6), 1131-1150. la porta, r., lopez-de-silanes, f., sheifl er, a., and vishny , r.w. (1997). legal determinants of external fi nance. journal of finance, 53 (1), 1131-1155. levine, r. (1997). financial development and economic growth: views and agenda. journal of economic literature, 35 (2), 688-726. ht tp :// ijb f.u um .e du .m y institutional and macroeconomic determinants of stock market development in mena region: 139-159 159 levine, r., loayza, n. and beck, t. (2000). financial intermediation and economic growth: causality and causes. journal of monetary economics, 46 (1), 31-77. levine, r. and zervos, s. (1998). stock markets, banks, and economic growth. american economic review, 88 (3), 537-558. levine, r. (2005). finance and growth: theory and evidence. in p. aghion and s. durlauf (eds.), handbook of economic growth. elsevier: amsterdam. law, s.h., and habibullah, m.s. (2009). the determinants of fi nancial development: institutions, openness and fi nancial liberalization. south african journal of economics, 77 (1), 45-58. mauro, p. (1995). corruption and growth. quarterly journal of economics, 110 (3), 681-712. montalvo, j.g., and reynal-querol, m.(2005). ethnic polarization, potential confl ict and civil wars. american economic review, 95 (3), 796-816. papaioannou, e. (2009). what drives international fi nancial fl ows? politics, institutions and other determinants. journal of development economics, 88(2), 269-281 pagano, m. (1993). financial markets and growth: an overview. european economic review , 37 (2-3), 613-622. perotti, c. e., and van oijen, p. (2001). privatisation, political risk and stock market development in emerging economies. journal of international money and finance, 20 (1), 43-69. pistor , k., raiser, m., and stanislaw, g. (2000). law and fi nance in transition economies. economics of transitio, 8 (2), 325-368. roeder, p.g. (2001). ethnolinguistic fractionalization (elf) indices, 1961 and 1985. retrieved from http//:weber.ucsd.edu\~proeder\elf.htm yartey, c. a. (2008). the determinants of stock market development in emerging economies: is south africa different? imf working paper no.08/32. zoli, e. (2007). financial development in emerging europe: the unfi nished agenda. imf working paper no. 07/245 . ht tp :// ijb f.u um .e du .m y ht tp :// ijb f.u um .e du .m y determinants of off-balance sheet business in the case of gcc banking sectors 46 the international journal of banking and finance, volume 9 (number 3), 2012: pages 46-68 determinants of off-balance sheet business in the case of gcc banking sectors mohammad elian gulf university for science and technology, kuwait __________________________________________________________ abstract this paper identifies the association between off balance sheet businesses and a number of determinants identified for the banking sectors of the gulf cooperation council countries. the fixed effects least squares dummy variable model is used to identify the determinants for a large sample of 64 banks over a recent fifteen-year period. the results reveal that bank-specific variables have important roles in influencing off balance businesses. as for the regulatory variable, capital items are less important, which is contrary to the long-held market discipline hypothesis, under which secure banks are predisposed to engage in more off balance businesses. the macroeconomic variable reveals that higher real gdp growth does not necessarily cause an increase in the off balance activities. however, its positive impact indicates that the off balance business actions follow business cycles, and the overall growth of economy. prudential regulators, as a policy matter, need to consider region-wide implications of these findings. this is important given the fact that regulating how off balance business is conducted in the region would influence costs and the scope of banks, hence also the monetary policy. key words: off-balance sheet, regulatory pressure, macroeconomic conditions, gcc. jel classification: c33, f65, g21 ___________________________________________________________ 1. introduction for the last three decades, banking products have gone deeply into transformations in the face of a wide range of financial innovations that are viewed as critical for economic growth. these innovations, in conjunction with securitization and financial derivatives, include a considerable expansion in range and scope of the fee-based forms, off-balance sheet (obs) businesses which causes a shift from traditional intermediary credit functions to more marketable credit 1 47 instruments. as a result, banks have offered a wide spectrum of obs forms by originating financial contracts that result in the acquisition of certain assets and liabilities with contingent conditions, which is enters as off balance sheet (obs) businesses. this paper investigates this phenomenon in the banking sectors in the gulf cooperation council (gcc) countries. for banks, the expansion toward obs businesses allow them to downsize their on-balance sheet items by generating additional fee income to offset size reduction of interest rate margins generated from banks’ granting loans. other motivational factors include global trade relations, hedging and reducing (or avoiding) regulatory restrictions i.e., reserve requirements, capital and deposit insurance premiums that are not imposed on obs businesses. on the contrary, other parties regulators, securities analysts and economists raise a major concern regarding the risk to the banking sector associated with such contingent claims-based business activities that magnifies leverage, causing more overall risks for banks: interest risks, credit risks and liquidity risks as pointed out in bennett (1986). further, given the current financial deregulation environment worldwide, banks have expanded their risk-taking levels, causing more failure cases in the marketplace. it is very obvious that obs businesses have both risk-increasing and risk-reducing attributes where increased obs earnings can potentially compensate for increased obs exposure. it may even actually reduce the probability of insolvency for most financial firms. nevertheless, obs activities have the potential to seriously harm the banks when major crisis occurs. recall the parallel banks that provided us$12 trillion capital to the us banking sector and the collateralized debt obligations as off balance items were the two root causes of the 207-8 global financial crisis (ariff et al., 2012). as empirical literature, several studies of obs business are available, though these are confined to developed banking markets in the us, europe and some economies in asia. in asia, the phenomenon is most prevalent even though a limited evidence is forthcoming for emerging banking markets (see jagtiani et al 1995a, b; fung and cheng, 2004 and nachane and saibal, 2007). therefore, while a systematic test of obs businesses in the context of emerging markets seems to be a reasonable justification, this paper bridges this gap in literature by testing within international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 48 the gcc countries. 1 in particular, it is an useful study since prudential regulators of this group of countries act in unison as region. these regulators may benefit from this dissemination of our findings. as for methodology, a bank level panel data estimation technique is used. this allows identifying and measuring effects that are simply not demonstrable in cross sections or timeseries analysis. the results of the panel data estimates reveal that bank-specific variables have an important role influencing obs businesses. the results show insignificant influence of the regulatory as well as the macroeconomic variables. the present study contributes to the above literature by exploring the usage of obs businesses and testing for its determinants within the gcc. the focus on the gcc banking industry is justified by the following attributes. first, with more deregulation, the market risk facing banking decisions are augmented in the light of the risky scheme of lending that causes growing integration need with most regional/global counterparties. simultaneously, the usage of hedging instruments has become crucial. second, given the global as well as the regional competitions and the reduction in interest income, gcc banks began to compete through the usage of non-interest sources of income. third, the policy initiatives that have been provided by the regulatory authorities following the reversal of speculative capital inflows that cause the banking industry to face severe squeezed liquidity conditions. such liquidity positions started to tighten in the beginning of the year 2008, reflecting an increasing reliance on external financing. hence, the record low liquidity ratio reflects relatively high asset-liability maturity mismatch in the gcc banking, calling for more off balance sheet items to be issued for hedging purposes. fourth, the fluctuation in oil prices creates unstable cash flows and, hence, volatile liquidity conditions: see hassan et al (2010). the rest of the paper is structured as follows: section 2 provides a brief review of the related literature. section 3 presents the methodology, specification of tests and data collection, while 1 the gulf cooperation council (gcc) includes saudi arabia, kuwait, qatar, united arab emirates, oman and bahrain. these countries share common culture, language, religion, and all are considered as oil producing countries and facing nearly similar deregulation and risks environment. 49 section 4 describes the data and its empirical findings. concluding remarks are given in section 5. 2. literature on innovation adoption and obs worldwide a downward trend in granting loans by traditional banking businesses occurred in recent decades, which is attributed to economic changes, financial innovations, and relaxation of restrictions as de-regulation took center stage. this trend led to an expansion of businesses toward non-interest services, specifically obs businesses. it seems that the growth of the obs businesses in the late 1970s and 1980s was a natural response to clients’ demand for credit guarantees and interest rate insurance at that time of high inflation and thus high interest rates. overall, the literature suggests a downward trend of traditional businesses and the expansion of the obs usage. boyd and gertler (1994) stress the financial innovations as a major cause of the downward trend of traditional banking businesses, while edwards and mishkin (1995) justify the drop in such businesses to the economic forces. they show the bank’s declining importance as a source of funds to non-financial borrowing clients, increased activities toward asset securitization, growth of commercial papers and junk bond markets, increase in financial services provided by non-bank intermediaries. the result was reduced bank spreads, costs and income advantages causing reasonable incentives to justify a bank to involve in the obs businesses. siems and clark (1997) reveal the emergence of new forms of intermediation. the expansion toward fee-based businesses have been stimulated by several factors, which include changes in fundamental economics, technological progress, along with changes in the surrounding political enforcements. angbazo (1997) argues that, due to primarily the increased volatility in both interest rate and foreign exchange rate markets, banks started to provide numerous risk management services that were tailored to specific needs of their clients. such practice was targeted in order to strengthen client relations in order to capture additional noninterest income through fee-based contingent claims securities trading. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 50 hassan and sackley (2002) identify, for the 1980s and early the 1990s, the banking industry faced numerous changes in the overall financial environment in highly volatile conditions 2 causing banks to expand their obs businesses as alternative sources to secure returns for banks. khambata and bagdi (2003) ascribe the growth in obs businesses to factors that include competition, market pressures and volatility in the market forces, movements toward deregulation, informational innovations and advancement in technology, lower levels of profitability from the traditional on-balance sheet banking businesses. they also include a desire to minimize the degree of risk linked with such businesses or with expansion of the investment opportunities to secure higher returns required by investors. on the issue of determinants of obs businesses, pavel and philis (1987) examine the determinants of loan sales of commercial banks. they conclude that diversification, capital, binding capital and reserve requirements have important influences on loan sales, suggesting that banks should and will start to sell loans when capital ratios are low and charge-offs are high. they conclude that banks with binding capital constraints are more likely to engage in obs business, swaps and loan securitization, than banks with excess capital. this result is supported by baer and pavel (1987), who assert that banks normally are involved in loan securitization and standby letter of credits (slcs) to avoid breaching regulatory requirements, namely reserve constraints and deposit insurance premium. however, benveniste and berger (1987) and koppenhaver (1989) argueed that binding capital requirements are not significant influential factors toward issuing slc guarantees. koppenhaver (1989) states that bank size, reserves, and loan losses are the main determinants of banks expanding obs business. similar results are reported by kaufman and mote (1994): banks may engage in their normal businesses more efficiently and may even grow more rapidly under alternative regulatory regimes, particularly those regimes that allow for expansion across state lines and do not restrict banks to commercial and investment businesses. 2 these conditions include changes in the financial market forces fluctuated economic conditions, highly competitive local and international pressures, movements toward deregulation, new capital adequacy requirements, progressed technology in communication and data processing, diminishing returns that normally generated from onbalance sheet businesses. 51 sinha (2005) identifies that obs businesses are positively related to operating profit ratio and negatively with the non-performing ratios. nachane and ghosh (2007) have us believe not only regulatory factors but also market forces are at work in the diffusion pattern of obs businesses. they show that non-performing assets and bank size are of prime concern for foreign as well as public sector banks. ahmad and hassan (2010) suggest there exists an informational economy of scope between loan and obs businesses that is profit driven. they conclude that banks participate in more obs businesses to reduce risk resulting from granting loans, while it is true political and economic events negatively affect banks obs businesses. 3. methodology specification and data collection 3.1. model estimation several econometric models are available and may be used as empirical models to form the basis for examining the obs businesses in the banking industry. as for gcc, the standard diffusion model developed by mansfield (1961) is appropriate. this model is the most commonly used model in diffusion studies: hannan and mcdowell (1984); jagtiani et al (1995a); bondt (1997); nachane and saibal (2007); and ahmad and hassan (2010). the basic proposition is that for the proportion of banks not adopting an obs item at time (t), their adoption begins at time (t + 1). this is a function of the proportion of those adopters at time (t). based on such a premise, the growth in the proportion of banks adopting the obs innovation over time may be expressed as the logistic function: xt+1 – xt =  (t – xt) xt/t (1) where xt is the total or cumulative number of banks adopting an obs item by time (t) and (t) is the total number of banks in the banking industry.  is the parameter which can measure the rate of the obs diffusion across banks. replacing discrete time by continuous time and integrating produces a logistic time cure describing obs diffusion: international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 52 xt/t )}exp (1{ 1 t   (2) while xt/t is the proportion of banks adopting the innovation (issuing an obs item) at time (t),  and  are the parameters which measure the rate of diffusion. 3 the expected value of  can be assumed to be a function of innovation (industry) characteristics. generating  through ols estimation is considered a standard practice. this can be achieved by equation (2) in its revised form and then by regressing the resulting estimates of  on the chosen firms-specific and innovation-specific variables as influences on the rate of diffusion: hannan and mcdowell (1984). overall, given a relevant transformation and the panel structure of the data, the basic equation to be estimated can be as follows: inobs itti itp itp in                   ))}(1( )( (3) pit (xt/t) is the proportion of bank (i) issuing an obs item at time (t) relative to the total asset size of bank (i) at time (t), including total of on-balance sheet items and off-balance sheet items of bank (i) at time (t). following jagtiani et al (1995) and bondt (1997), the justification for this is that we consider the scale on which banks issue an obs items is a linear relationship between the number of banks engaged in issuing obs as a proportion of all banks and the amount of obs items in terms of the total balance sheet size (on and off-balance businesses). in light of the aforesaid discussion and depending on the characteristics of the unobservable bank-specific variables, αi, the researcher needs a two way error component regression model, which can be outlined. more specifically, pooled regression data model presupposes the fact that differences across units can be captured in differences in constant term (as for the fixed effect 3 the s shaped diffusion curve of equation (2) expects that the proportion of banks adopting obs items will move upwards at an accelerating bound until a proportion of 50 per cent of total banks is attained at time t = ( )  . hence, obs adoption would move up at a decreasing bound, and 100 per cent obs diffusion is loomed asymptotically, see morrison (1996), bondt (1997) and nachane and saibal (2007). 53 modes) or alternatively, as individual specific constant terms are randomly distributed across cross-sectional units (as for the random effects models). the fixed effects model as a method for pooling time-series cross sectional data considers all individual differences as captured by differences in the intercept parameter. the intercepts (αi) were considered as fixed parameters that we could estimate directly using the least squares estimator. the fixed model assumes that the errors eit are independent with mean zero and constant variance σe 2 , for all individuals and in all time periods. following that, all behavioral differences between individual firms and over time are captured by the intercept. where the individual intercepts are there to control for the firm-specific differences. the resulting econometric model is: уit = αi + β1x1it + β2x2it + β3x3it + β4x4it + ………+ βnxnit + eit. (4) for the random effect model, even it assumes that all individual differences are captured by the intercept parameters, considers the individual differences as random rather than fixed. the random model considers αi as mutually independent random variables that are independent of the equation’s error term. thus, the constant term model αi is a separate constant term for each bank: αi = α1d1+ α2d2…... , with the dj’s bank-specific dummy variables. in other words, the random individual differences can be included in the econometric model by specifying the intercept parameters αi to consist of a fixed part that represent the population average and random individual differences from the population average, ui. so, in the random effects model αi is a firm-specific disturbance: αi = α + ui. the random individual differences ui which are called random effects, are analogous to random error terms that have zero mean, uncorrelated across individuals and have a constant variance σu 2 , so e(ui) = 0, cov (ui,uj) = 0 and var (ui) = σu 2 . for this research, while the data set exhausts the population, there is an earlier argument supporting the hypothesis that bank-wise variances in performance derived from divergences in initial terms/conditions such as scale of operations, indicates there are bank-specific constants. that in turn supports the choice of the fixed effect estimator: nachane and ghosh 2002). in other words, while we are not drawing n individuals (banks) randomly from a large population, the focus, in the present study, is on a specific set of firms (banks in our case) and the inference is international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 54 restricted to the behavior of these firms (banks), so the use of the fixed effects model is appropriate: bondt 1997 and baltagi (2008). 4 however, while there is no prior condition to suggest correlation between the cross-section specific error component (i) and the regressors (explanatory variables), hence panel data test could also be estimated as the random effects model in which substantial differences between the fem and rem is concluded. 5 the hausman test is used to compare the coefficient estimates from the two models, to make the decision as to which model is more appropriate. 6 to conclude, the used model in this article is an extension of the earlier basic model presented in equation (3) above. these variables are bank-specific variables (characteristics) in addition to regulatory and macroeconomic explanatory variables. accordingly, the empirical test is based on the following general pooled regression equation that typifies a modified version of equation (3): inobs itti itp itp in                   itit ))}(1( )( (5) i denotes the number of banks (i = 1, 2, 3, ………., 64) and t denotes the number of years (t = 1, 2, 3, ……………,15). inobsit (as logistic transformation of pit), is the dependent variable and represents the proportion of bank (i) issuing obs items at time (t) relative to the total asset size of bank (i) at time (t), including total of on-balance sheet items and off-balance sheet items of bank (i) at time (t). the time trend ( t ) accounts for the autonomous diffusion (deterministic trend), where its coefficient of indicates the autonomous speed of obs diffusion. it is a function defining bank (i) specific variables (creditworthiness variables) at time t.  it is a function defining bank (i) nonspecific variables at time t. it: is the residual term that is included to reflect all other influences on bank (i)'s probability to issue obs items. 4 in the case where n is large, a fixed effects model would lead to an enormous loss of degrees of freedom, causing the need to use the random effects model rather than the fixed effects model. 5 statistically, if it is assumed that i and the regressors are uncorrelated, rem may be recommended, but if they are correlated, then, the fem would be more appropriate. 6 the null hypotheses underlying the hausman test is that fem and rem do not differ substantially. 55 for our case, the bank-specific variables include market power, bank size, loan ratio, net loan ratio, net loan charge-offs and profitability. the bank nonspecific variables include the capital adequacy ratio and the growth in real gross domestic product in order to proxy regulatory and macroeconomic condition variables, respectively. 3.2. proxies of variables the dependent variable (obs) is measured by the natural logarithm of bank (i) obs items at time t. the time trend variable (time) indicates the number of years since the first year of the sample period. its coefficient reflects the autonomous speed of diffusion (  ), which depends on development in financial technology, learning factors and changing preferences regarding the obs produced items. larger  indicates more rapid adaptation of an obs product, bondt (1997). for market power (marp), it is expected that the more market power a bank has, the more the bank has ability to expand businesses toward on-balance sheet and off-balance sheet businesses. while market power increases, the creditworthiness of a bank is expected to have a positive impact on the probability of banks to underwrite an obs item. for measurement purposes, the level of market share (the percentage share of bank (i) deposits to total bank deposits) is used as a relevant statistic to provide a snapshot of the market power at a point in time. this ratio measures the absolute size of total deposits which is held by the i th bank relative to total size of the banking market as measured by total deposits. so, although the market power of bank (i) is assumed to be directly related to the bank share of the deposits market, the marp coefficient is expected to be positive, indicating that bank with more market power tends to engage in more obs businesses. for the bank size (bsiz), the impact on obs businesses can go either way (positive/negative). on the one hand, large size banks, in terms of total assets for example, will have more chance to diversify their banking businesses, employ high skilled and well trained employees, derive the benefits of the economies of scale, and be more efficient in providing higher quality services. large banks may also have well developed and widespread networks with access to large and sophisticated clients who are likely to be asking for obs products. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 56 further, large banks armed with specialized management skills will ask for and move toward more obs products engagement, thus meeting the demand for large and institutional clients who, on the other hand, will be in favor to of engaging large banks to handle their obs demand. in addition, greater market confidence is normally directed to large banks, causing less failure possibility. such indicators, along with others, will encourage banks to expand their businesses toward the obs items, keeping in mind that obs businesses may require that a bank be of a certain size in order to realize economies of scale or to use qualified management skills needed for obs products. on the other way around, it can be argued that larger banks in terms of total assets would have the probability to be more risk-diversified (less business risk) causing less usage of an obs product. in this study, a relevant proxy is the natural log of total assets for bsiz: the coefficient is expected to be positive/negative. the loan ratio (loar) is measured as loan to total assets. as reported by angbazo (1997), higher loan ratio normally increases interest rate risk, leading to more engagements in obs businesses for hedging purposes. another rational to support such expected impact, as mentioned by nachane and ghosh (2007), lies in the way it is used. approving clients’ loans in which banks access their clients’ investment information, will facilitate the decision on the offer of relevant obs risk management products. while the loar is directly related to obs businesses, its coefficient is expected to be positive, indicating that banks with more loan ratio tend to undertake more obs businesses. the net loan ratio (nlor), which is measured as gross loan minus loan loss reserves as a percent of total assets. this can be regarded as a measure of the maturity gap and maturity mismatch between a bank assets and liabilities. as net loan ratio widens, the maturity gap widens as well, providing a positive incentive to use obs products since the increase in risk exposure has to be hedged. banks with more willingness to make risky loans may be more willing to take more obs risk (bondt 1997). the coefficient is expected to be positive, indicating that bank with more net loan ratio tend to undertake more obs businesses. for the net charge-off ratio (nchr), the impact is ambiguous. on the one hand, considering the net charge-off as a proxy of non-performing loans, the impact is expected to be negative. an 57 increase in a bank’s non-performing loans would decrease its creditworthiness, leading to lower levels of obs practices. on the other hand, higher levels of a bank charge-offs indicate higher levels of loan portfolio risk, reflecting the increased possibility of issuing or supplying obs products for hedging purposes. this would also simultaneously generate substitute income (fee income) to compensate for bad debt loans. hence, an increase in the level of charge-off would have a positive impact on issuing an obs product. the charge-offs (net loan write-offs) which are based on the difference between loans actually written-off and recoveries on loans previously classified as uncollectable, to gross loan ratio is used as a proxy of the nchr variable. the impact of the profitability (prof), measured using net income from intermediation profitability, can be positive or negative. on the one hand, profitability cab be considered as a measure of a banks’ creditworthiness and indicates a clients’ trust, causing anticipated positive relationship between profitability and the issue of obs items. further, high profits increase cash flows, which can be retained to absorb future loses to reflect the chance to improve creditworthiness. banks with more creditworthiness would attract or would use more of obs businesses. this is supported by the argument that more profitability is an indication of superior management armed with optimal utilization of bank assets and of higher productivity that produces lower costs, reflecting an improvement in the interest spreads. on the other hand, banks with low net income would attempt to increase their underwriting in fee-based forms of income to offset reduced interest income generated from traditional credit businesses. hence such banks would supply more derivatives (obs) products. keeping in mind those banks with high net interest income would lock in their interest spreads by using derivatives to hedge. the moral hazard hypothesis suggests that banks with low net interest income (low profitability) would speculate using derivatives, while banks with high net interest income will not hedge, reflecting lower usage of obs products. as for non-bank-specific variables, the capital adequacy ratio and the growth in real gross domestic product are included as controls for regulatory and macroeconomic conditions respectively. for the banks’ capital adequacy ratio (capc), the impact is ambiguous and can be viewed differently (furlong and keeley, 1989; cited by bondt, 1997; nachane and ghosh, 2007; and ahmad and hassan, 2010). while the capital adequacy ratio (car) measures the banks’ international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 58 capital to protect its depositors and insure stability, obs businesses are expected to be larger for banks with higher car ratio since these banks are more creditworthy therefore clients would place more trust towards the performance of such banks. a high car with accompanying high credit rating would be an incentive for a bank to be an attractive provider/supplier of obs products. alternatively, a high car ratio reduces a banks’ marginal gain from increasing the risk in the asset portfolio. as bank capital increases, the ability to assume/bear risk increases as well, but the need for obs products to hedge the risk exposure may decrease. therefore, a bank with high car could also be expected to take on less obs risk and issue a lower volume of obs item/products. the capital adequacy ratio is used as proxy for capital. 7 variables included in the tested model, their expected sign, and economic rationale all are shown in table 1. table 1: variables in the empirical model ______________________________________________________________________________ variable code proxy expected economic rationale__________ sign of coefficient time trend time (t) + time↑⇒ obs technology diffusions↑⇒obs↑ bank-specific variables market marp (deposits/total deposits) + marp↑⇒ scale economies↑⇒obs↑ power bank size bsiz ln(total asset) ± bsiz↑⇒ scale economies↑⇒obs↑ bsiz↑⇒ bank risk↓⇒obs↓ loan ratio loar (loans/total assets) + loan↑⇒scope economies and risk↑⇒obs↑ net loan nlor (net loans/total assets) + nlor↑⇒scope economies and risk↑⇒obs↑ ratio net charge-offs nchr (net loan write-offs/ gross loan) ± nchr↑⇒scope economies and risk↑⇒obs↑ profitability prof (net profit/total assets) ± prof↑ ⇒creditworthiness↑⇒obs↑ prof↑ ⇒creditworthiness↑⇒obs↓ bank-non-specific variables: regulatory and macro-economic variables capital capc (primary capital/risky-assets) ± carh↑⇒ creditworthiness↓⇒obs↓ ratio carl↑⇒ creditworthiness↑⇒obs↑ real gdp gror g + gdp↑⇒economic activity↑⇒obs↑ growth _____________________________________________________________________________ 7 the emphasis is on the response of banks to the 8% total risk-based capital standards which signals to the degree of regulatory pressure for adequately capitalized banks, car more than 8 per cent. banks in the gcc region have acceptable leverage ratios and are well capitalized with capital adequacy ratios (car) above the minimum standard levels of the cars 59 the effect triggered by variations in the economic activity is captured by the real gross domestic product growth rate (gror) which is used to control for the macroeconomic conditions. under high growth rates of gdp, demand for obs products will respond positively taking into consideration the growth in the business cycles. the variable coefficient is expected to be positive, reflecting that with the higher growth rates of gdp and the probabilities of issuing obs items will increase. 3.3. data collection the published source of data in emerging markets rarely provides satisfactory data to commence panel data investigation. considerable efforts were needed to construct the targeted size of panel database using bank scope and the imf data sources. the growth in real gdp variable is obtained from the international financial statistics (ifs) online database. to prepare the sample, we include all banks over the sample period reflecting a total balanced panel of 64 banks drawn from bahrain, kuwait, oman, qatar, united arab emirates and saudi arabia. to avoid outliers, the filtering process eliminated short lived banks, foreign banks and central banks. overall, to estimate the empirical version of the model, the pooled data sample includes a total of 7,680 annual observations over 15 years. the period represents an era of significant challenge to the gcc banking industry in which banks have been attempting to improve and measure their performance while facing several changing economic conditions such as regulation, high inflation, volatility in oil prices, volatility in interest rates, high variability in structuring and pricing derivatives worldwide, volatility in foreign exchange rates against currencies other than the us dollar, and the reduction in demand for some exports, in addition to the effect of deregulation, competitive pressures and increased uncertainty about returns compared with the 1980s. one may add that this period is of particular interest given the debate between regulators about obs businesses and capital requirement. further, this period spans the years prior to and after the global financial crises. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 60 4. empirical findings before embarking on presenting the empirical findings, it is worth mentioning that the obs businesses display a wide variation over the period. this is not surprising since most banks are relatively new participants in obs items. given the standard classification of bank-scope data base, obs businesses include (a) managed securitized assets, (b) other obs exposure to securitization, (c) guarantees, (d) acceptances and documentary credits, (e) committed credit lines, (f) other contingent liabilities. 8 yet, 25 per cent of banks have reported their usage of guarantees, 18 per cent for acceptances and documentary credits, 19 per cent for committed credit lines, and 39 per cent for other contingent liabilities. managed securitized assets and other obs exposure to securitization are not reported by the gcc banks as obs items under usage. the limited usage of the obs businesses indicates that the banking sector in the gcc region still relies on the traditional banking businesses as the main source and use of funds. 9 so far, most banks assets portfolios are composed of loans, securities investments, and islamic finance products. given such asset structures, during the financial crises, gcc banks faced significant losses through mark-to-market valuations of their trading portfolios, keeping in mind that those banks in the gcc held 18 per cent of their portfolios in securities as at the end of 2008 while only 1 per cent was held in derivatives: hassan et al., 2010. overall, it seems that a reasonable levels of profitability was generated from on balance sheet activities, in conjunction with the low volatile conditions. both interest rate and foreign exchange rate markets caused less incentives for obs businesses as alternative sources of returns. 8 for categorization purposes, obs businesses can be categorized into two categories. first, those businesses that generate income or expenses without the creation or holding of underlying asset or liability (market related activities) i.e., when a bank acts as a broker (taking a fee for arranging for funds to be provided to borrowers without making loans or raising deposits), rather than as a dealer (making and holding loans and any other funding sources. another example is providing banking services, such as cash management, that generate income without requiring assets or liabilities. second, those businesses that involve the bank’s commitments and contingent claims, which might be grouped to the following: (1) financial guarantees i.e., standby letter of credit, line of credit, revolving loan agreement, securitization of asset with re-course, and note issuance facilities such as euro-notes, revolving underwriting facilities (rufs), and standby notes. (2) trade finance i.e., commercial letter of credit, acceptance participation, and investment activities that includes all derivative instruments such as forward commitments, financial futures, interest rate swaps, option (put, call, collars), and currency swaps. 9 for example, as source of funds, the contribution of the foreign liabilities is limited and bond financing is small (2 per cent of total liabilities), causing an aggravation of the maturity mismatch between assets and liabilities in the gcc banking industry. 61 table 2 is a summary of empirical findings. the fit (r 2 ) of the estimated equations of models 1, 2 and 3 is acceptable. the models explain almost 83 per cent of the variation in the dependent variable. the adjusted (r 2 ) is lower, but it exceeds 80 per cent for the three models. the hausman test value, with the p-value of the estimated chi-squire (χ 2 ) statistics is very low, helps to rejects the rem. this supports the choice of the fem, indicating that the random error terms (i) are probably correlated with one or more of the regressors. hence, in short, the fem is preferred to rem. as shown by the results from model 3, the autonomous speed of diffusion is significantly different than zero, while the speed of diffusion is relatively slow for all countries under consideration. the time variable has a significant and negative sign, indicating that the usage of obs businesses is decreasing over the test period suggesting lower adoption of obs items. it indicates obs businesses in the gcc region may be considered as risk increasing businesses rather than risk decreasing activities. this conclusion may be attributed to deficiency in the technological infrastructure and/or lower engagement in the speculation and hedging businesses within the market workplace. or it could simply be the results of withdrawals from this activity post financial crisis. at the bank-specific firm level, market power has a positive and significant impact, indicating that a bank with more market power would be categorized as more creditworthy, engendering such banks to underwrite an obs item. the bank size has positive and significant impact, reflecting that there are economies of scale influence resulting from obs businesses in the gcc region. this would help us to conclude the relative contribution of obs businesses within the overall balance sheet products. the loan ratio has a positive and significant impact on the obs businesses in the gcc banking industry indicating that loans and obs businesses are not substitutes. this indicates economies of scope between loans and obs businesses as well. the result suggest that banks would issue more obs products in order to minimize the resulting risk generated from loans. the net loan ratio has a positive and significant influence on obs businesses. this indicates a tendency to making more risky loans, and reflects widening maturity mismatches between assets and liabilities causing more risk exposure to be hedged, thus leading to issuing an obs products. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 62 table 2: estimated results from fixed effect least squares dummy variable model (lsdv) with cross sectional weights ___________________________________________________________________________________________________ inobsit = αi+ t + φit+it ……………………….………………………….model (1) inobsit = αi+ t + φit+χit+it ………………………..……………………..model (2) inobsit = αi+ t + φit+χit+ γit.+it ……………………………..…………...model (3) (i = 1, 2, 3, ………….., 64) (t = 1, 2, 3, ………….,15) variable model (1) model (2) model (3) constant 1939.21*** 1939.29*** 1939.91*** (1262.08) (1256.46) (1207.39) time -0.022*** -0.010*** -0.015** (-5.69) (-0.897) (-0.65) marp 515.71*** 523.56*** 549.75*** (8.39) (8.50) (8.74) bsiz 9.41*** 9.40*** 9.29*** (40.85) (40.73) (38.01) loar 3.36*** 3.29*** 3.77*** (2.75) (2.70) (2.95) nlor 2.66** 2.71** 2.18** (2.05) (2.08) (1.61) nchr 1.98*** 1.97*** 1.97*** (4.23) (4.24) (4.27) prof -0.55 -0.80 -1.04 (-0.20) (-0.28) (-0.36) capc 0.009 0.01 (0.71) (0.80) gror 1.30 r 2 0.832 0.832 0.829 adjusted r 2 0.819 0.819 0.815 f-statistic 62.955 61.932 59.70 p-value 0.0000 0.0000 0.0000 (f-statistic) hausman test ho: fem vs rem chi-squire (χ 2 ) statistics (6) 704.127 (7) 703.623 (8) 669.504 p-value 0.0000 0.0000 0.0000 count. 946 946 946 explanatory notes: dependent variable is inobs; t-statistics are reported in parentheses; ***, **and *indicate significance at the 1, 5, and 10 percent level, respectively. model (1) below presents the estimation of the banks’ specific variables (creditworthiness variables) captured by φit. model (2) incorporates additional regulatory proxy affecting the obs businesses captured by χit. model (3) incorporates additional macroeconomic proxy affecting the obs businesses captured by γit. the three models are extracted basically from our formal model, equations 3 and 4. the net charge-off ratio has a significant and positive sign. the positive sign indicates that banks are using obs businesses/products to manage loan portfolio risks, mainly risks that are normally result from bad debt. this suggests asset quality of the gcc banks has improved where banks are facing high credit growth rates and the non-performing loans to total loans have been 63 declining since. banks are asked by the prudential authorities to undertake substantial generous loan loss provisions against anticipation of continued increasing non-performing loans. the coverage ratio of provisions to non-performing loans across the gcc is very high by international standards: hassan et al., 2010. the profitability has a negative and insignificant impact on obs businesses, indicating that higher obs usage is not prompted by profitability considerations. banks are earning high net interest income reflecting the deep emphasis on granting loans as traditional banking businesses and less prominence is given to fee-based forms as sources of income. following the moral hazard hypothesis, the concentration risk of lending resulting from lending to few clients and/or lending to sectors that have high price exposure, creates the need to ask banks with high net interest income to lock in their interest spreads by using derivatives to hedge, while banks with low net interest income (low profitability) would use derivatives products to secure more profits. while the derivatives market in the region is still at its early stage of development, together with the fixed exchange rate regimes for the currencies of most gcc economies, low usage of derivatives to hedge or to speculate is obvious. as shown by the f-statistics of model 3, it seems that adding the two regulatory and macroeconomic variables does not significantly enhance the statistical fit of regression. the coefficients of capital adequacy and gdp growth are statistically insignificant, although both variables have positive signs as suggested by theory. it seems that banks with high capital ratio are less likely to engage in obs businesses in order to avoid regulatory constraints, which supports the capital avoidance hypothesis banks with low capital are more likely to engage in obs businesses. this also helps to confirm that low regulatory pressure has positive impact on the use of obs products. it indicates that low regulatory pressure increases creditworthiness of banks, and increases clients’ incentives to engage in the obs products offered by banks under consideration. 10 this would have us argue that regulators imposing binding restrictions on bank 10 in other words, low capital regulation, as implied by high car, signifies relevant capital position (high credit rating), reflecting more creditworthiness position of a bank, causing higher engagement in obs businesses (koppenhaver and stover, 1991). alternately, low capital regulation (low regulatory pressure) also reduces the marginal propensity to increase the risk of a banks’ asset portfolio (furlong and keeley 1989, cited by nachane and ghosh 2007 and khasawneh and kabir 2010).therefore, banks with high capital ratios (implying low regulatory pressure) can be expected to have lower obs risk, and hence, engage in lower obs businesses. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 64 capital will lead to the banks engaging in less obs usage: a supporting for the argument that obs businesses should be avoided as these businesses are risk-increasing. one possible reason for this observation is that the impact of the regulatory variable (the capital adequacy ratio) may actually arise from the ground reality that gcc countries have their currencies fixed since 2003 to the major currency for reason of exchange rate stability. the domestic interest rates are largely a function of us dollar interest rates with variation of up to 30 basis points from time to time as country risk or transaction costs. prudential authorities in the countries have concluded that this is an effective policy to deliver low price inflation, mainly for tradable commodities while it also enhances the credibility to the local currencies. while inflation positively correlated with the high oil prices that boosted incomes and local demand in these countries. however, it seems that the exchange rate pegging policy provides a level of stability to current prices and future price expectation makes such issue not to emerge as a problem: gray and blejer, (2006). the insignificant impact of the macroeconomic factor on the obs usage may indicate that higher real gdp growth does not necessarily cause an increase in the obs usage. the positive impact of the real gdp growth on the obs usage indicates that the obs businesses follow the overall economic growth and business cycles. more economic growth would create more obs usage, indicating the double role of the obs businesses as income generating and risk management techniques. in other words, an upturn in the economic activities would increase business risks and there is an incentive to supply more volume of obs businesses. 5. conclusions earlier empirical research pertaining to off balance sheet business has documented evidence on countries where the phenomenon is prevalent in the developed banking markets. this paper is on several countries described as gcc countries reliant on oil revenue with pegged currencies. these countries may be described as emerging with lot less obs activities, thus providing a converse environment to test some of the theories on what chases obs activities to increase or 65 retard in such emerging economies. the determinants included not only bank-specific variables but also regulation and macroeconomic conditions the results appear to endorse a conclusion that the speed of diffusion of adopting off balance businesses by banks is relatively slow. the negative sign on the coefficient for time indicates decreasing trend overtime. it suggests a declining adoption of obs businesses. also, the estimates suggests that bank-specific variables such as market power, bank size, loan ratio, net loan ratio and net charge-offs ratio, all have theory-suggested positive impact on the dependent variable (obs business activity). insignificant negative impact is found for profitability. results show insignificant influence from regulatory as well as the macroeconomic variables, although the impact is positive in both cases. the positive influence of the regulatory and macroeconomic variables supports the argument that prudential regulators, region-wide, need to be aware of the obs adoption as financial innovations. we also suggest that the same forces are in play in the region as in other more active obs places. for the gcc banking industry, the low adoption of most obs items can be justified. first, most bank businesses are directed towards on-balance sheet products in which the gcc banks would have significant concentration risk in the context of lending to few targeted clients and/or lending to sectors that have high price exposure (sectors that are subject to market price fluctuations and asset bubbles such as real estate and equities): hassan et al. (2010). second, on the level of capital markets, in spite of having an advanced infrastructure, most gcc capital markets have not yet started to provide derivative instrument/products such as stock futures, index futures, options, debt swaps, interest rate futures, and commodities futures. accordingly there is deficiency in the offering many obs products simply as a reflection of the (i) high profitability from normal on-balance activities and (ii) lack of willingness – perhaps because of the early stage of financial development or perhaps cultural norms towards risky obs – of expertise in obs structures. these things may change if the environment changes. add to these, the other facts of (iii) good interest spreads in banks, and (iv) pegged currencies, it would be international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 66 some time before obs activities could pick up as it did in the developed economies. 11 gray and blejer (2006). there are some policy relevant findings. given the heavy focus towards traditional banking activities, prudential authorities have to provide more focus towards banks asset/liability management practices, where banks need to re-evaluate their obs positions as part of their overall liquidity management strategies, not so much obs for securing fee incomes. this seems crucial given the tightening in the international liquidity positions in the current years. second, there is a need for education and incentives to develop the derivatives to manage the liquidity risk of the balance sheet in the gcc region. third, the influence of oil price effect on the banking profitability needs to be studied so as to see how obs activities could be beneficial to offset risk on the balance sheet through obs activities. author information: mohammad elian is an associate professor of finance, department of economics and finance, gulf university for science and technology (gust), kuwait. his address is gulf university for science and technology, mubarak al-abdullah area/west mishref, p.o. box 7207, hawali, kuwait 22093. he may be contacted via e-mail: elian.m@gust.edu.kw. references ahmad, k. and hassan, k. (2010). the determinants and policy implications of off-balance sheet activities in mena countries commercial banks. economic research forum, october, 1-31. angbazo, l. (1997). commercial banks net interest margins, default risk, interest rate risk and off-balance sheet banking. journal of banking and finance, 21, 55-87. ariff, m., farrar, j., and khalis, a., (2012). regulatory failure and the global financial crisis. edward elgar publishing, uk and usa. baer, h. and pavel, c. (1987). does regulation derive innovation? economic perspective, federal reserve bank of chicago, 3-16. bennett, b. (1986). off balance sheet risk in banking: the case of standby letters of credit. economic review, federal reserve bank of san francisco, issue win, pages 19-29. benveniste, l. and berger, a. (1987). securitization with recourse: an instrument that offers uninsured depositors sequential claims. journal of banking and finance, 11, 403-424. 11 for example, foreign exchange forward contracts against the us dollar are not used under the expectation that the exchange rate will not change in the future, however, foreign exchange forward contracts against the euro and other currencies are needed. further, other complex derivatives products such as futures and options can be used throughout the liquid us markets without facing exchange rate risks, reflecting an advantage of the monetary policy decision to peg the exchange rate by allowing an effective outsourcing. http://ideas.repec.org/a/fip/fedfer/y1986iwinp19-29.html http://ideas.repec.org/s/fip/fedfer.html 67 bondt, g. (1997). off-balance sheet diffusion: european evidence from bank-level data. onderzoeksrapport wo&e nr 528/9743, december, 1-14. boyd, j. and gertler, m. (1994). are banks dead? or are the reports greatly exaggerated? federal reserve bank of minneapolis quarterly review 18, 1-27. edwards, f. and mishkin, f. (1995). the decline of traditional banking: implications for financial stability and regulatory policy. nber working paper, w4993, 1-31. fung, m. and cheng, a. (2004). diffusion of off-balance sheet financial innovation: information complementary and market competition. pacific-basin finance journal, 12, 525-540. furlong, f. and keeley, m. (1989). capital regulation and bank risk-taking. journal of banking and finance, 35, 883-891. hannan, t. and mcdowell, j. (1984). rival precedence and the dynamics of technology adoption: an empirical analysis. journal of economics and business, 49, 533-547. jagtiani, j., saunders, a. and udell, g. (1995a). the effect of bank capital requirement on bank off-balance sheet financial innovation. journal of banking and finance, 19, 647658. jagtiani, j., nathan, a. and sick, g. (1995b). scale economies and cost complementarities in commercial banks: on and off-balance sheet activities. journal of banking and finance, 19, 1175-1189. hassan, k. and sackley, w. (2002). determinants of the us saving and loan off-balance sheet activities: an empirical investigation. finance india: the quarterly journal of indian institute of finance. 16, 3, p. 915-931 gray, s. and blejer, m. (2006). the gulf cooperation council region: financial market development, competitiveness, and economic growth. imf, 1.4, 41-51. al-hassan, a., khamis, m. and oulidi, n. (2010). the gcc banking sector: topography and analysis. imf working paper, 10/87, 1-37. kaufman, g. and mote, l. (1994). is banking a declining industry? a historical perspective. federal reserve bank of chicago, economic perspective, may/june, 2-21. khambata, d. and bagdi, r. (2003). off-balance sheet credit risk of the opt 20 japanese banks. journal of international business regulation, 5, 57-71. koppenhaver, g. (1989). the effects of regulation of bank participation in the guarantee market. research in financial services, 1, 165-180. koppenhaver, g. and stover, r. (1989). standby letter of credit and large bank capital: an empirical analysis s. journal of banking and finance, 15, 315-327. morrison, j. (1996). how to use diffusion models in new product forecasting. the journal of business forecasting, summer, 6-9. nachane, d. and ghosh, s. (2002). determinants of off-balance sheet activities: an empirical analysis of public sector banks. economic and political weekly, special issue in money, banking and finance, february, 421-428. nachane, d. and saibal, g. (2007). an empirical analysis of the off-balance sheet activities of indian banks. munich personal repec archive (mpra), paper no. 17304. online at http://mpra.ub.uni-muenchen.de/17304. pavel, c. and philis, d. (1987). why commercial banks sell loans: an empirical analysis. economic perspective, federal reserve bank of chicago, june, 3-14. international journal of banking and finance, vol. 9, iss. 3 [2012], art. 4 http://mpra.ub.uni-muenchen.de/17304 68 siems, t. and clark, j. (1997). rethinking bank efficiency and regulation: how off-balance sheet activities make a difference. financial industries studies, federal reserve bank of dallas. sinha, r. (2005). off balance sheet exposure of indian commercial banks: some empirical results. indian institute of capital markets, the 9 th capital markets conference. available at http://ssrn.com/abstract=977813. http://ssrn.com/abstract=977813 how to cite this article: ibrahim, h,. zulkafli, a. h., & jabeen, g. (2020). board education, growth and performance of family ceo listed firms in malaysia. international journal of banking and finance, 15(2), 25-46. https://doi.org/10.32890/ijbf2020.15.2.2 board education, growth and performance of family ceo listed firms in malaysia 1haslindar ibrahim 2abdul hadi zulkafli 3gul jabeen school of management 3universiti sains malaysia, malaysia 1corresponding author: haslindar@usm.my; 2hadi_zml@usm.my; 3guljabeen04@gmail.com a r t i c l e i n f o article history: received 29 november 2019 revised 3 february 2020 accepted 17 february 2020 published 31 july 2020 jel code is g32, g34 keywords: board professional qualification, board degree, growth, family ceo firms. a b s t r a c t this paper examines the relationship between board education, board size, growth, ownership and firm performance of family ceo and nonfamily ceo listed firms in malaysia. a sample of 37 firms and data were collected over a period of five years from 2012 to 2016. the 37 samples of family firms were subdivided into family ceo (21), and non-family ceo (16) firms. the independent variables were board education as measured by the proportion of board degrees (bdeg) and the proportion of board professional qualifications (bpro), board size (bsize), growth, and ownership. meanwhile, firm performance was measured by using return on equity (roe) and return on assets (roa). the findings showed that there was a significant difference the international journal of banking and finance, vol. 15. number 2, 2020: 25-46 25 26 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 between family ceo and non-family ceo firms at a five percent level for board professional qualifications confirming that altruism and nepotism were observed among family members which supported the argument of characteristics of nepotism such as granting jobs to family members regardless of merit. in addition, this study also found board professional qualifications as significant but negatively related to external firm performance in family ceo firms. this showed that board education has not really been emphasized among board members. besides, growth has significant influence on family firm performance which is evidently reflected in their contribution to the country’s gdp. 1. introduction family businesses are predominantly affluent companies that have flourished through wealth generation. asian literatures have shown that family firms in many countries, including australia, taiwan, hong kong, singapore and china, have performed well (filatotchev, lien & piesse, 2005; la porta, lopez-de-silances & shliefer, 1999). in addition, a study conducted by faccio, lang & young (2002) in 13 western european countries verified that at a 20 percent cut-off level, 44.3 percent are family owned firms. however, the percentage of family firms increases to 57.2 percent when they exclude ireland and the united kingdom in the sample. similarly, the study conducted by claessens, djankov and lang (2000) on nine east asian countries at the 20 percent cut off-level revealed that the percentage of family firms in the sample statistically increased from 38.29 percent to 58.68 percentwhen japan was excluded from the sample. apparently, japan has the largest share of widely held family firms at 79.8 percent based on the 20 percent cut-off level. in malaysia, about 70 percent of listed firms are family businesses and they are the major contributors to the country’s gross domestic product (gdp) (amran & ahmad, 2010). meanwhile, corporate governance emphasizes on the collective action taken when a firm faces problems with alternative stakeholders. in the process board education, growth and performance of family ceo listed firms in malaysia: 25-46 27 of addressing problems or conflicts, firms often rely on large shareholders who have significant ownership and control rights to influence decisions. in the global corporate world, large shareholders with significant controlling power are commonly found to be family firms. apparently, family controlled ownership in firms are typical and common in worldwide corporations. apart from the performance between family firms and non-family firms, there were numerous studies carried out to examine their comparison (anderson & reeb, 2003; miller & le breton-miller, 2006; villalonga & amit, 2006; ibrahim & samad, 2011; amran & ahmad 2010). even though family firms contribute significantly to the malaysian economy, they are found to lack independence and internal control. therefore, these problems need to be addressed in family firms as family businesses are governed by family traits, which do not exist in other businesses (mishra, randoy & jenssen, 2001). despite the importance of corporate governance on firm performance, research on malaysian family firms seems to lag behind. admittedly, there is too little empirical evidence to prove that malaysian family firms perform better than non-family firms. for instance, ibrahim and lau (2018) revealed from a study that board size, independent directors and duality do show a strong relationship with firm performance. thus, this study would like to focus more on the academic background or qualifications among board members especially in family firms. the board qualifications come entrusted with a wide range of observable or unobservable capabilities in this knowledge overflow modern period which appear to be a critical issue related to firm performance. to ensure high corporate governance quality of firms through identification and measurement of its capabilities, board members especially the chief executive officer (ceo) has to obtain educational qualifications which include a degree(s) or postgraduate recognition for better and reliable communication with the stakeholders (bhagat, bolton & subramanian, 2010). according to gottesman and morey (2006a), superior intelligence can be measured with educational qualifications where managers with a higher level of education often perform better than managers with a low level of education. furthermore, investigations by ujunwa, nwakoby, and ugbam (2012) found that firms with board members holding phd degrees have positive relationships and that such firms perform better. however, generally it is important to remember that superior managerial skills are not always obtained from a high level of academic qualifications. in terms of distinction, soft skills like entrepreneurial and leadership skills are often developed from non-academic related activities. in fact, there are also findings which show fast-growing and high-performing firms that are founded and managed by average-educated individuals. these contradicting results from previous studies complicate the importance of board educational qualifications, and furthermore, these studies are comparatively scarce within the literature. hence, it is vital to investigate whether or not the educational qualifications of the ceo and board members influence firm performance. 28 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 therefore, the study on family firms which have been managed by the family ceo and non-family ceo in malaysia, needs to be investigated to find out whether their board educational qualifications play an important role in sustaining their firm performance or otherwise. according to minichilli, corbetta, and macmillan (2010), their study revealed that the family ceo is beneficial for the performance of a family firm. on the contrary, burkart, panunzi & shleifer (2003) argued that a reliable non-family ceo is vital for the performance of a family firm. apart from the performance between family firms and non-family firms, there were numerous studies carried out to examine their comparison (ibrahim & lau, 2018; ibrahim & samad, 2011; miller & le breton-miller 2006; villalonga & amit 2006; anderson & reeb 2003). even though family firms contribute significantly to the malaysian economy and the importance of corporate governance mechanisms on firm performance, research on malaysian family firms seemed to lag behind. therefore, there needs to be further research to investigate the emphasis on family firms as family businesses which are governed by family traits, do not exist in other businesses (mishra et al., 2001). 2. literature review 2.1 altruism and nepotism in family firms altruism is a powerful force within family life and by extension, within the family firm. it ensures that parents are protective of their children, supports family members to be considerate towards each other and cultivates loyalty and commitment to the family and firm. altruism holds family members together with the firm belief that they have a residual claim on the family estate (stark & falk, 1998). according to agency theory, the agency cost can be reduced through fostering, monitoring and enforcing agreements when ownership goes through the process of aligning interests among family agents towards growth opportunities and risk. therefore, family agent performance is not monitored regularly. nevertheless, information asymmetries among family agents can be reduced meanwhile increasing informal agreements usage through increased communication and cooperation within the family firm with altruism (daily & dollinger, 1992). besides, altruism benefits the agency in a way that creates a heightened sense of interdependence among family agents. in fact, employment relates welfare directly to firm performance. lubatkin, schulze, and dino (2005) stated in their study that altruism is both a blessing and a curse for a family firm because it will give a better chance to enter the family members in the market, and also produce good founders and managers. board education, growth and performance of family ceo listed firms in malaysia: 25-46 29 furthermore, it is deduced that agency problems which grow from altruism and self-control are complicated when the ceo exercise discretion in placing control of the firm’s resources. intrinsically, this broadens the ceo’s capacity to make altruistic transfers such as employment, perquisites, and privileges to family members that they would not be receiving if they were employed elsewhere. henceforth, a variety of agency costs is created when these privileges and the sense of entitlement are evoked (gersick, davis, hampton, & lansberg, 1997). meanwhile, according to ford and mclaughlin (1986), nepotism is the act of compassion toward one’s family members or friends. in economic or employment terms, it can be granting jobs to friends and relatives, without regard to merit. apparently, such practices impact businesses negatively. in other words, they can erode the support of other employees, reduce the quality and creativity of management and belittle the importance of competence and high-level performance (zax & ichniowski, 1988). keles, ozkan, and bezirci (2011) conducted a study on turkish family firms which showed that nepotism, favouritism and cronyism negatively affected the organization’s trust in the family business. nepotism often provides a positive perspective in many smaller family firms. this is due to the practice of “succession” with alternative cheap source of labour. however, according to zax and ichniowski, (1988), the practice is misleading where the utmost priority for employment will be competence, only then accompanied by years of constant, superior performance that pave way to succession. nepotism is neither good nor bad, in and of itself (barnes & hershon, 1976). in other words, nepotism is a neutral phrase. sarcastically, nepotism reflects the positive or negative charge based on the way one has raised one’s children. generally, the ultimate duty of a parent is to raise responsible adults who have high self-esteem and can function independently in the world (davis & stern, 1981). the values of becoming competent employees include honesty, integrity, dependability, respect for others, being industrious and doing one’s best in every endeavour. these values are crucial to instil in every generation in ensuring that the competent generation is born. on the contrary, the failure of teaching these principles will result in the feeling of entitlement in the children which will lead them to believe that they are privileged and should be given everything. ironically, the stated deficiency eventually turns into a ripe incubator for problems to emerge when the children works in the family business (ford & mclaughlin, 1986). children who come to the business with an attitude of entitlement will think they are exempt from the rules that apply to “ordinary people”. they often do not understand that they must earn their place in the company through hard work and consistently-demonstrated competence. a seemingly small thing like coming to work on time is an example. experience has shown that nepotism works if and only if the values of the family members are congruent and the successor 30 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 is fully qualified (kiechel, 1984). by referring to the literature discussed, it is hypothesized that: h1: there is a significant difference between family ceo firms and non-family ceo firms in terms of firm performance. 2.2 corporate governance and performance corporate governance is a multifaceted concept which has attracted multiple understanding and interpretations; accordingly, there is no general single acceptable definition of corporate governance. different authors have explained corporate governance differently and associated it with quite a range of corporate issues. their explanations surround agency relationship and links between corporate governance and the governance of corporations, paying little attention to shareholders’ interests. gramling and hermanson (2006) state that corporate governance can be mechanisms that direct and control businesses while conducting the business. corporate governance is a set of procedures designed to protect shareholders' interests. due to agency issues, corporate governance has been adopted and has been applied to narrow the gap between shareholders and managers. it also shows that corporate governance studies consists of two categories: using a single component and using multiple components but all focused on the importance of corporate governance (ng’eni, 2015). the malaysian code on corporate governance which was revised in 2007 suggests that directors have certain qualities (skills, knowledge, experience, professionalism, and integrity) in the face of intense responsibility. previous studies found that board chairman with a university degree is positively associated with seven measures of performance which include earnings per share (eps), roa, cumulative returns, cumulative abnormal returns, eps appreciation, roa appreciation, and market-to-book ratio. according to sebora and wakefield (1998), directors with higher education are well adapted to operations and have business acumen compared to their less-educated counterparts. in addition, jalbert, rao, and jalbert (2002) who studied a sample consisting of forbes 800 firms discovered that ceos who graduated from prestigeous schools had a positive relationship with roa of the firm. darmadi (2013) found that board of directors with postgraduate education experiences were positively and significantly correlated with roa. furthermore, a study conducted by ramónllorens, garsía-meca & duréndez (2017) suggested that ceos who had new information and higher knowledge helped family firms to perform well at the international level. a study had been conducted on family and non-family firms in hong kong to see how highly educated ceos’ work affect the performance of firms. the results were surprisingly different for the firms’ performance, as it showed board education, growth and performance of family ceo listed firms in malaysia: 25-46 31 that there were more times that the highly educated ceos became overconfident in making decisions that might have worsened the performance of firms rather than improved them (ying & mie, 2014). furthermore, bhagat et al. (2010) noted that ceos with mba degrees performed no better than those without such qualifications. on a side note, empirical evidence from their study revealed that hiring new ceos with mbas led to short-term improvement in performance. from the results of the study, bhagat et al. (2010) did not consider ceo education as a good proxy for ceo ability. meanwhile, darmadi (2013) revealed that undergraduate and financial certificates were not significant with roa. gottesman and morey (2006a) found no evidence that ceos from more prestigious schools outperformed their counterparts from less prestigious schools. interestingly, mixed results were found in studies that investigated the influence of graduate degree holders toward better firm performance. undeniably, educational background and skills were among the determinants of family firm performance (castillo & wakefield, 2006). however, there is a lack of studies conducted on the relationship between the directors’ educational qualifications with return on equity (roe). based on the given discussion, it is hypothesized that: h2:board degree is significantly and negatively related to the performance of family ceo and non-family ceo firms. h3: board professional qualifications is significantly and negatively related to the performance of family ceo and non-family ceo firms. board size refers to the number of directors who serve on the board. large boards are claimed to be superior to small ones because large groups have more capabilities and resources, and wider networking. haleblian and finkelstein (1993) elaborated that large groups could enhance problem-solving abilities by providing more strategic perspectives and constructive judgement. despite the upside potential, lipton and lorsch (1992) came to the conclusion that too many executive members on board would lead to more problems. furthermore, darkos and bekiris (2010) concluded that there is a negative relationship between board size and firm performance, the larger the board size, the lower the firm performance. other previous studies also found that small board was more effective than a large one in making executive replacement decisions. findings by jensen (1993) revealed that a small board size could increase firm performance. notably, firms with small board sizes have a higher stock market value (yermack 1996). family firms used to have a smaller board size (ibrahim & samad, 2011; hermalin & weisbach, 2003; eisenberg, sundgren, & wells, 1998; lipton & 32 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 lorsh, 1992; wald, 1991). according to bennedsen, kongsted, and nielsen (2008), the optimum number for a family firm board size is less than six members. on the negative side, the source of information, experience, and contact of a small board is limited. from the given evidence, it is hypothesized that: h4:board size is significantly and negatively related to the performance of family ceo and non-family ceo firms. 2.3 firm growth and performance the other variable of the study is firm growth due to family businesses as the major gdp contribution in the malaysian economy (arman & ahmad, 2010). hitt, ireland, and lee (2000) suggested in their study that firm growth depended on knowldge of technology. by understanding and applying technological knowldge, firms are better able to inovate, to lead growth and secure better performance. furthermore, according to the behavioral theory, managers pay more attention to achieve their goals and attempt to apply aspiration levels to each goal. evidence from the insurance industry shows that companies grow more when they are below the aspiration level particularly when performance goals are met (gereve, 2008). other studies conducted on manufacturing companies in turkey which investigated the interaction between firm growth and profitability, showed that there was a positive and significant relationship between them (coban, 2014). by referring to the literature discussed, it is hypothesized that: h5:growth is significantly and positively related to the performance of family ceo and non-family ceo firms. 2.4 ownership structure and performance ownership structure is significant towards implementing the corporate governance system for any firm and the country as well. precisely, the ownership could detect the agency problem whereby the conflict of interest between agents and principals or majority shareholders could be justified. according to samad (2004), the distribution of power between managers and shareholders is determined by the ownership concentration. furthermore, according to hitt, ireland & lee (2000), ownership can be defined as the legal right over the use, disposal and fruits, of means of production indeed in society. there are various studies on ownership structure and firm performance in malaysia and globally. according to demsetz and lehan (1985), their findings of a relationship between stockholdings by 5 to 10 shareholders and roe were found to be insignificant. however, holerness and sheehan indicated positive correlation between shareholdings and firm performance. furthermore, based on board education, growth and performance of family ceo listed firms in malaysia: 25-46 33 empirical evidence regarding the effects of ownership structure such as insider and blockholders’ equity and ownership and firm performance he found that both equity ownership by insiders and blockholders led to better decisions and better firm performance. gorton and schmid (2000) noted a strong positive relationship between concentrated equity ownership and firm performance as measured by roa in germany. the empirical evidence showed that the percentage of family ownership or controlled firms was significant to business organizations in the world as revealed in the study by la porta et al. (1999). family firms have been defined in various ways by theorists in accordance with studies undertaken at the time. first, they defined family firms based on the degree of ownership and/or management by family members and this was supported by previous studies (barry, 1975; barnes & hershon, 1976). besides, according to empirical studies such as by sraer and thesmar (2006) who viewed family controlled firms as firms whose ceos were either the founder or descendent of the founder. family firms with paternalism, trust and altruism could bring commitment and love for the business atmosphere (james, 1999). jensen and randoy (2003) noted that due to incentive alignment, family enterprises gained possible benefits of reducing agency costs. kapopoulos and lazaretou (2007) mentioned ownership structure firstly as an endogenous variable and secondly as they measured ownership structure which reflected ownership as shareholders with conflicting interests. the results of the study showed that more concentrated ownership showed higher performance. the family-owned firms in china where the family of the higher management originated showed that there was a positive relationship in familyowned firms and the performance of the firms (goel & ramanathan, 2011). there was a negative impact of family-owned firms and its performance and this statement was supported by the research conducted on the mena region (ahmed & hadi, 2017). furthermore, if the ownership concentration was low than the firm would perform better but if the concentration of family members was high in the firm’s management than there would be a negative impact on the firm’s performance. according to anderson and reeb (2004), firms with family ceos performed better than firms having non-family ceos. by referring to the literature discussed, it is hypothesized that: h6:the higher the equity ownership of the family firm, the higher the firm performance. 3. methodology this section provides information about the study’s conceptual model. the operationalization and the measurement of each variable will be discussed. a 34 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 total of 90 listed companies from the various sectors: construction, consumer, property, trading and services which had been listed in bursa malaysia from 1999 to 2016 were screened. however, for this study, we took only family firms and grouped them into family ceo and non-family ceo firms. only 37 companies from the list were family firms, out of which only 21 companies were identified as family ceo firms and 16 companies as non-family ceo firms. the data of the time series and cross-sectional studies were collected from the 37 listed companies in bursa malaysia for the period of five years from 2012 to 2016. we defined family firms based on the family ties among the board members and the equity stake held by the family members of at least 20 percent (ibrahim & samad, 2011). 3.1 data source secondary data was the main and only source of data for this study. most of the secondary data were obtained and manually collected from company annual reports and retrieved from the data stream. by referring to those reports, information such as the number of directors with degree qualifications and professional qualifications could be identified and calculated for the proportion. furthermore, board size, growth and the percentage of ownership held by the family members were also extracted from the annual reports. 3.2 measures the response of the dependent variable was predicted, measured and monitored in the event of influence by independent variables (kuo, kao, chang, & chiu, 2012). this study employed two profitability ratios as proxy for the performance which were return on equity (roe) and return on assets (roa). roe and roa were measured based on the net income divided by total shareholders’ equity and the net income over the total assets of firms, respectively (ibrahim & samad, 2011). return on equity (roe) = return on assets (roa) = the board size (bsize) was measured by the total number of directors who served on the board of the company (abor & biekpe, 2007; bokpin & arko, 2009; ibrahim & samad, 2011). according to baldwin (1963), a person holding a degree (bdeg) is considered elite in the region of less developed countries. to proxy that qualification, this study considered any undergraduate degree obtained by the board members and ceos of the listed family firms. graham and net income total shareholders' equity net income total assets board education, growth and performance of family ceo listed firms in malaysia: 25-46 35 harvey (2002) stated that any individual who holds a postgraduate degree must have first obtained an undergraduate degree. on the other hand, professional qualifications (bpro) are titles or awards granted by professional bodies. the proportion of directors holding an undergraduate degree qualification and a professional qualification are calculated using the following formulas: growth opportunities are defined by the annual percentage change of total assets (pandey, 2001; pandey, 2004; eriotis, vasilious, & ventoura neokosmidi, 2007; abor & biekpe, 2009; karadeniz kandir, balcilar, & onal, 2009; chadha & sharma, 2015; ahsan, wang, & qureshi, 2016; ohman & yazdanfar), as shown by: growth = the percentage of equity held by managers is measured using the sum of their direct and indirect share ownership and their stock options outstanding plus share ownership by their immediate families. in this study, family ownership (ownshp) was measured by taking the percentage of equity stake held by family members of at least 20 percent. control variables are also defined as a constant variable. in scientific experimentations, this variable is the experimental element that stays constant and unchanged throughout the investigation in order to test the relative relationship of the dependent and independent variables. many previous studies used firm size (lnfsize) and leverage or debt ratio (dr) as control variables. in this study, the firm size was measured by the natural log of total assets of the company (ibrahim & samad, 2011; sheikh & wang, 2012; vakilifard, gerayli, yanesari, & ma’atoofi, 2011) and debt ratio was measured by total debt divided by total assets (friend and lang, 1998). leverage (dr) = 3.3 panel data of multiple regression analysis this study employed the panel data approach as it eliminated unobservable heterogeneity that different firms in the sample data could present, less collinearity 10 growth opportunities are defined by the annual percentage change of total assets (pandey, 2001; pandey, 2004; deesomask et al., 2004; eriotis, vasilious, and ventoura neokosmidi, 2007; abor & biekpe, 2009; karadeniz et al., 2009; chadha and sharma, 2015; alipour et al., 2015; hussain et al., 2015; hussain et al., 2015 ahsan et al., 2016; ohman and tazdanfar, 2017), as shown by: growth =   100 0 01 x ta tata    the percentage of equity held by managers is measured using the sum of their direct and indirect share ownership and their stock options outstanding plus share ownership by their immediate families. in this study, family ownership (ownshp) was measured by taking the percentage of equity stake held by family members of at least 20%. control variables are also defined as a constant variable. in scientific experimentations, this variable is the experimental element that stays constant and unchanged throughout the investigation in order to test the relative relationship of the dependent and independent variables. many previous studies used firm size (lnfsize) and leverage or debt ratio (dr) as control variables. in this study, the firm size was measured by the natural log of total assets of the company (ibrahim & samad, 2011; sheikh & wang, 2012; vakilifard, gerayli, yanesari, & ma'atoofi, 2011) and debt ratio was measured by total debt divided by total assets (friend and lang, 1998). leverage (dr) = ∑ ����� ���� ∑����� ������ 3.3 data panel of multiple regression analysis this study employed the panel data approach as it eliminated unobservable heterogeneity that different firms in the sample data could present, less collinearity among the variables and a better measurement than pure cross-section or pure timeseries data (baltagi, 2001; gujarati, 2003). panel data analysis enabled the study to consider both time series and cross-sectional effects as it used the collection of observations of cross-section data over several time series. 3.4 model selection model selection is presented as follows: perf = α + β1bdegit + β2bproit + β3bsizeit + β4growthit + β5ownshpit + β6levit + β7lnfsit + εit where: perf = return on assets (roe), return on equity (roe) 10 growth opportunities are defined by the annual percentage change of total assets (pandey, 2001; pandey, 2004; deesomask et al., 2004; eriotis, vasilious, and ventoura neokosmidi, 2007; abor & biekpe, 2009; karadeniz et al., 2009; chadha and sharma, 2015; alipour et al., 2015; hussain et al., 2015; hussain et al., 2015 ahsan et al., 2016; ohman and tazdanfar, 2017), as shown by: growth =   100 0 01 x ta tata    the percentage of equity held by managers is measured using the sum of their direct and indirect share ownership and their stock options outstanding plus share ownership by their immediate families. in this study, family ownership (ownshp) was measured by taking the percentage of equity stake held by family members of at least 20%. control variables are also defined as a constant variable. in scientific experimentations, this variable is the experimental element that stays constant and unchanged throughout the investigation in order to test the relative relationship of the dependent and independent variables. many previous studies used firm size (lnfsize) and leverage or debt ratio (dr) as control variables. in this study, the firm size was measured by the natural log of total assets of the company (ibrahim & samad, 2011; sheikh & wang, 2012; vakilifard, gerayli, yanesari, & ma'atoofi, 2011) and debt ratio was measured by total debt divided by total assets (friend and lang, 1998). leverage (dr) = ∑ ����� ���� ∑����� ������ 3.3 data panel of multiple regression analysis this study employed the panel data approach as it eliminated unobservable heterogeneity that different firms in the sample data could present, less collinearity among the variables and a better measurement than pure cross-section or pure timeseries data (baltagi, 2001; gujarati, 2003). panel data analysis enabled the study to consider both time series and cross-sectional effects as it used the collection of observations of cross-section data over several time series. 3.4 model selection model selection is presented as follows: perf = α + β1bdegit + β2bproit + β3bsizeit + β4growthit + β5ownshpit + β6levit + β7lnfsit + εit where: perf = return on assets (roe), return on equity (roe) board education, growth and performance of family ceo listed firms in malaysia: 25-46 35 proportion of growth opportunities are defined by the annual percentage change of total assets (pandey, 2001; pandey, 2004; eriotis, vasilious, & ventoura neokosmidi, 2007; abor & biekpe, 2009; karadeniz kandir, balcilar, & onal, 2009; chadha & sharma, 2015; ahsan, wang, & qureshi, 2016; ohman & yazdanfar), as shown by: growth = the percentage of equity held by managers is measured using the sum of their direct and indirect share ownership and their stock options outstanding plus share ownership by their immediate families. in this study, family ownership (ownshp) was measured by taking the percentage of equity stake held by family members of at least 20 percent. control variables are also defined as a constant variable. in scientific experimentations, this variable is the experimental element that stays constant and unchanged throughout the investigation in order to test the relative relationship of the dependent and independent variables. many previous studies used firm size (lnfsize) and leverage or debt ratio (dr) as control variables. in this study, the firm size was measured by the natural log of total assets of the company (ibrahim & samad, 2011; sheikh & wang, 2012; vakilifard, gerayli, yanesari, & ma’atoofi, 2011) and debt ratio was measured by total debt divided by total assets (friend and lang, 1998). leverage (dr) = 3.3 data panel of multiple regression analysis this study employed the panel data approach as it eliminated unobservable heterogeneity that different firms in the sample data could present, less collinearity harvey (2002) stated that any individual who holds a postgraduate degree must have first obtained an undergraduate degree. on the other hand, professional qualifications (bpro) are titles or awards granted by professional bodies. the proportion of directors holding an undergraduate degree qualification and a professional qualification are calculated using the following formulas: proportion of = 10 growth opportunities are defined by the annual percentage change of total assets (pandey, 2001; pandey, 2004; deesomask et al., 2004; eriotis, vasilious, and ventoura neokosmidi, 2007; abor & biekpe, 2009; karadeniz et al., 2009; chadha and sharma, 2015; alipour et al., 2015; hussain et al., 2015; hussain et al., 2015 ahsan et al., 2016; ohman and tazdanfar, 2017), as shown by: growth =   100 0 01 x ta tata    the percentage of equity held by managers is measured using the sum of their direct and indirect share ownership and their stock options outstanding plus share ownership by their immediate families. in this study, family ownership (ownshp) was measured by taking the percentage of equity stake held by family members of at least 20%. control variables are also defined as a constant variable. in scientific experimentations, this variable is the experimental element that stays constant and unchanged throughout the investigation in order to test the relative relationship of the dependent and independent variables. many previous studies used firm size (lnfsize) and leverage or debt ratio (dr) as control variables. in this study, the firm size was measured by the natural log of total assets of the company (ibrahim & samad, 2011; sheikh & wang, 2012; vakilifard, gerayli, yanesari, & ma'atoofi, 2011) and debt ratio was measured by total debt divided by total assets (friend and lang, 1998). leverage (dr) = ∑ ����� ���� ∑����� ������ 3.3 data panel of multiple regression analysis this study employed the panel data approach as it eliminated unobservable heterogeneity that different firms in the sample data could present, less collinearity among the variables and a better measurement than pure cross-section or pure timeseries data (baltagi, 2001; gujarati, 2003). panel data analysis enabled the study to consider both time series and cross-sectional effects as it used the collection of observations of cross-section data over several time series. 3.4 model selection model selection is presented as follows: perf = α + β1bdegit + β2bproit + β3bsizeit + β4growthit + β5ownshpit + β6levit + β7lnfsit + εit where: perf = return on assets (roe), return on equity (roe) 10 growth opportunities are defined by the annual percentage change of total assets (pandey, 2001; pandey, 2004; deesomask et al., 2004; eriotis, vasilious, and ventoura neokosmidi, 2007; abor & biekpe, 2009; karadeniz et al., 2009; chadha and sharma, 2015; alipour et al., 2015; hussain et al., 2015; hussain et al., 2015 ahsan et al., 2016; ohman and tazdanfar, 2017), as shown by: growth =   100 0 01 x ta tata    the percentage of equity held by managers is measured using the sum of their direct and indirect share ownership and their stock options outstanding plus share ownership by their immediate families. in this study, family ownership (ownshp) was measured by taking the percentage of equity stake held by family members of at least 20%. control variables are also defined as a constant variable. in scientific experimentations, this variable is the experimental element that stays constant and unchanged throughout the investigation in order to test the relative relationship of the dependent and independent variables. many previous studies used firm size (lnfsize) and leverage or debt ratio (dr) as control variables. in this study, the firm size was measured by the natural log of total assets of the company (ibrahim & samad, 2011; sheikh & wang, 2012; vakilifard, gerayli, yanesari, & ma'atoofi, 2011) and debt ratio was measured by total debt divided by total assets (friend and lang, 1998). leverage (dr) = ∑ ����� ���� ∑����� ������ 3.3 data panel of multiple regression analysis this study employed the panel data approach as it eliminated unobservable heterogeneity that different firms in the sample data could present, less collinearity among the variables and a better measurement than pure cross-section or pure timeseries data (baltagi, 2001; gujarati, 2003). panel data analysis enabled the study to consider both time series and cross-sectional effects as it used the collection of observations of cross-section data over several time series. 3.4 model selection model selection is presented as follows: perf = α + β1bdegit + β2bproit + β3bsizeit + β4growthit + β5ownshpit + β6levit + β7lnfsit + εit where: perf = return on assets (roe), return on equity (roe) 36 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 among the variables and a better measurement than pure cross-section or pure time-series data (baltagi, 2001; gujarati, 2003). panel data analysis enabled the study to consider both time series and cross-sectional effects as it used the collection of observations of cross-section data over several time series. 3.4 model selection model selection is presented as follows: perf = α + β1bdegit + β2bproit + β3bsizeit + β4growthit + β5ownshpit + β6levit + β7lnfsit + εit where: perf = return on assets (roe), return on equity (roe) bdeg = proportion of directors with undergraduate degrees bpro = proportion of directors with professional qualifications bsize = board size growth = growth ownshp = ownership lev= debt ratio lnfs = firm size εit = the disturbance or error term 4. analysis of results table 1 shows the descriptive statistics of the full sample, family ceo and nonfamily ceo firms from 2012 to 2016. the findings demonstrated that firm size was larger for non-family ceo firms compared to family ceo firms. similarly, non-family ceo firms had higher mean performance ratios for both roe and roa compared to family ceo firms. as a summary, non-family ceo firms had a slightly smaller mean in board size (7.725), a higher proportion of board degrees (0.547), and a slightly higher proportion of board professional qualification (0.359), compared to family ceo firms. independent samples t-test results showed that there was only mean proportion of board professional qualifications which was significantly different between family ceo and non-family ceo firms where non-family ceo had a higher mean (0.353) compared to family ceo firms (0.279) at 5 percent level as presented in table 2. this finding was consistent with burkart et al. (2003) where non-family ceo firms with higher board professional qualifications gained significant reliability to perform better than family ceo firms. likewise, it was also consistent with the study carried out by sebora and wakefield (1998), where firm performance were higher as more educated directors were on the board. board education, growth and performance of family ceo listed firms in malaysia: 25-46 37 ta bl e 1. d es cr ip tiv e st at is tic s of o ve ra ll pe ri od (2 01 2– 20 16 ): f am ily c e o f ir m s an d n on -f am ily c e o f ir m s fu ll sa m pl e (n = 37 ) fa m ily c e o f ir m s (n =2 1) n on -f am ily c e o f ir m s (n =1 6) v ar ia bl es m in m ax m ea n sd m in m ax m ea n sd m in m ax m ea n sd g r o w t h -5 5. 91 25 6. 1 11 .5 5 33 .2 2 -5 5. 91 24 1. 8 12 .6 9 33 .4 9 -4 4. 62 25 6. 1 10 .8 3 32 .3 2 b d e g 0. 14 3 1 0. 54 6 0. 16 9 0. 20 0 0. 88 9 0. 54 6 0. 15 7 0. 14 3 1 0. 54 7 0. 18 5 b pr o 0 0. 66 7 0. 31 3 0. 12 8 0 0. 50 0 0. 27 8 0. 09 76 0. 14 3 0. 66 7 0. 35 9 0. 14 8 r o a -0 .6 06 0. 30 6 0. 05 8 0. 08 38 -0 .4 51 0. 30 6 0. 03 64 0. 07 02 -0 .6 06 0. 29 8 0. 04 93 0. 09 90 r o e -3 6. 58 0. 84 4 0. 04 20 2. 70 2 -3 6. 58 0. 31 1 -0 .3 07 3. 57 7 -1 .7 35 0. 84 4 0. 08 16 0. 26 6 o w n sh p (% ) 20 .9 0 77 .8 4 50 .6 1 13 .9 2 20 .9 0 75 .2 1 50 .7 5 12 .3 6 23 .4 7 77 .8 4 50 .4 3 15 .8 2 b si z e 4 15 7. 96 8 2. 26 2 4 13 8. 15 2 2. 36 5 5 15 7. 72 5 2. 11 1 l n fs 10 .9 4 23 .9 8 17 .3 0 3. 70 5 17 .0 2 23 .9 8 20 .1 1 1. 82 3 17 .8 5 24 .3 6 20 .5 2 1. 81 5 l e v 0 8. 18 4 0. 27 5 0. 60 6 0 8. 18 4 0. 32 3 0. 79 4 0 0. 54 3 0. 21 2 0. 13 5 n ot e: b si ze = b oa rd s iz e; b d e g = p ro po rt io n of b oa rd d eg re es ; b p r o = p ro po rt io n of b oa rd p ro fe ss io na l q ua lifi ca tio ns ; r o e = r et ur n on e qu ity ; r o a = r et ur n on a ss et s; g r o w th = g ro w th , l n f s= f ir m s iz e, o w n sh p = o w ne rs hi p, l e v = d eb t r at io . 38 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 table 2. differences in means tests variable family ceo firms non-family ceo firms t-stat mean mean bdeg bpro 0.546 0.279 0.552 0.353 -0.128 -1.943** roe roa 0.039 0.035 0.082 0.050 -0.871 -0.734 lnfs 20.176 20.544 -0.632 *** significant at 0.01 level ** significant at 0.05 level * significant at 0.10 level note:bsize=board size; bdeg= proportion of board degrees; bpro= proportion of board professional qualifications; roe= return on equity; roa=return on assets; growth= growth, lnfs= firm size, ownshp=ownership, lev=debt ratio. table 3. the pearson’s correlation coefficients of the study variables roa roe bsize bdeg bpro lnfs ownship growth lev roa 1 roe 0.705*** 1 bsize 0.0342 0.137 1 bdeg -0.0498 -0.0376 0.0683 1 bpro -0.0451 -0.122 0.109 -0.167 1 lnfs 0.251** 0.127 0.494*** 0.274** 0.165 1 ownshp 0.114 -0.0255 -0.122 -0.0517 -0.0202 0.145 1 growth 0.140 0.0407 -0.0152 0.0300 0.0359 0.0720 0.179 1 lev -0.0619 -0.00284 0.0317 0.125 -0.133 -0.00472 -0.0756 0.00422 1 * p < 0.05, ** p < 0.01, *** p < 0.001 note:bsize=board size; bdeg= proportion of board degrees; bpro= proportion of board professional qualifications; roe= return on equity; roa=return on assets; growth= growth, lnfs= firm size, ownshp=ownership, lev=debt ratio. table 3 shows that roe is positively and highly correlated to roa at 1 percent level indicating that these two variables have strong correlation with each other as it is still in the acceptable level. in conclusion, larger board size, higher roe and higher roa are important complementary factors when firm size is larger. table 4 shows the findings of the study which indicates that the proportion of board professional qualifications (bpro) is negative and significant at 1 percent level for full sample and family ceo firms with roe. this proved that a higher proportion of board professional qualifications would reduce the impact on roe. this was consistent with the findings by bhagat et al. (2010) where they board education, growth and performance of family ceo listed firms in malaysia: 25-46 39 did not consider ceo education as a good proxy for ceo ability in long-term firm performance. in addition, firm size (lnfs) had a positive and significant relationship with roe at 0.01 level this proved that larger firm size as control variable with a significance of 1 percent in this sample influenced roe in the positive direction which was consistent with previous studies (ang & ding, 2005; mcknight & mira, 2003). with regard to growth, there was a positive and significant relationship between growth and firm performance which was consistent with findings by arman and ahmad (2010) who stated that growth would influence firm performance thus contributing to the gdp of the country as exhibited in table 4 and table 5. table 4. regression results for return on equity (roe) variable full sample (n= 37) family ceo firms (n=21) non-family ceo firms (n=16) intercept bsize bdeg bpro lnfs ownshp growth lev -38.20***(10.91) -0.353(0.419) 1.175(2.473) -13.32***(3.805) 2.570***(0.620) -0.000651(0.0356) -0.000575(0.00645) -0.0768(0.345) -61.56***(21.86) -0.330(0.764) 0.896(4.519) -21.54***(8.143) 3.447***(1.076) 0.00294(0.0484) 0.000035(0.0113) -0.0846(0.452) -1.029*(0.583) 0.0007(0.0248) -0.285(0.229) 0.0324(0.293) 0.0638**(0.0312) -0.00182(0.00377) 0.00135*(0.000807) 0.0816(0.342) observation 185 105 80 r-squared 0.160 0.221 0.1443 standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 note:bsize=board size; bdeg= proportion of board degrees; bpro= proportion of board professional qualifications; roe= return on equity; roa=return on assets; growth= growth, lnfsize= firm size, ownshp=ownership, lev=debt ratio. meanwhile, table 5 shows that the relationship between roa and firm size is positive and significant at 0.01 level for all samples which confirms that the larger the size of the firm, the likelihood that it will increase the performance of the firm. furthermore, the relationship between board degrees and roa was negative and significant at 10% level only for full sample. this result was supported by bhagat et al. (2010) who found that ceos who were degree holders performed no better than those without such qualifications. gottesman and morey (2006a) also stated that there was no evidence that ceos from more prestigious schools outperformed their counterparts from less prestigious schools. another study conducted by adnan et al. (2016) on government linked companies and non40 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 government linked companies in malaysia showed similar results, that diversity in education was not associated with better firm performance. the results of study did not show any significant relationship between firm performance and the education level of the board of directors. table 5. regression results for return on assets (roa) variable full sample (n= 37) family ceo firms (n=21) non-family ceo firms (n=16) intercept bsize bdeg bpro lnfsize ownshp growth dr -1.087***(0.271) -0.0112(0.0104) -0.110*(0.0615) -0.0823(0.0946) 0.0757***(0.0154) -0.00012(0.0009) 0.00019(0.00016) -0.0043(0.0086) -0.260**(0.125) -0.0067(0.0053) -0.0586(0.0579) -0.0948(0.0961) 0.0202***(0.0072) 0.000059(0.0007) 0.00016(0.0002) -0.00387(0.0076) -0.416*(0.223) -0.00196(0.0094) -0.0938(0.0848) 0.0503(0.109) 0.0251**(0.0119) -0.0003(0.0014) 0.00049*(0.0003) 0.0398(0.128) observation 185 105 80 r-squared 0.166 0.1115 0.1396 standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 note:bsize=board size; bdeg= proportion of board degrees; bpro= proportion of board professional qualifications; roe= return on equity; roa=return on assets; growth= growth, lnfsize= firm size, ownshp=ownership, lev=debt ratio. 5. conclusion this study shows that board professional qualifications is negative but significantly related to return on equity (roe) of family ceo firms. board qualifications as proxied by only the proportion of board professional qualifications significantly decreased the performance of family ceo firms. this indicated that professional qualifications is less important in improving the performance of a family ceo firm externally (roe) rather than internally (roa). interestingly, there is a significant difference in the proportion of board professional qualifications between family ceo firms and non-family ceo firms. this confirms the presence of altruism and nepotism practices among family members and supports the argument of nepotism characteristics such as granting jobs to friends and relatives regardless of merit. firm size shows a positive and significant relationship with both roe and roa at level 0.01, confirming that a larger size firm contributes to an increase in firm performance. board education, growth and performance of family ceo listed firms in malaysia: 25-46 41 these significant findings can contribute to the advancement of knowledge especially in the importance of board education among directors for better firm performance. as for board qualifications tested in this study, it is divided into board degrees and board professional qualifications by proportion. the results indicated that there is significance but in a negative direction for board professional qualifications among family board members showing that the family members are less concerned with seeking board degrees or professional qualifications to improve their businesses. it was as though the business would be automatically passed down from one generation to the next without consideration for educational background. it is vital that apart from experience, directors are knowledgeable and possess recognized board degrees and/or professional qualifications which collectively contribute toward the expertise or competitive advantage for the firm. flexibility in business acumen can be achieved with in-depth knowledge in the business, especially in accounting and finance. it is encouraged that family ceo firms and non-family ceo firms emphasize the importance of board qualification as previous studies have proven its significance in elevating firm growth and performance . 6. acknowledgements the first author would like to thank universiti sains malaysia for the bridging grant no. 304.pmgt.6316293. references abor, j., & biekpe, n. (2007). corporate governance, ownership structure and performance of smes in ghana: implications for financing opportunities. corporate governance, 7(3), 288–300. adnan, m. f., sabli, n., rashid, m. z. a., hashim, a., paino, h., & abdullah, a. (2016). the impact of educational level of board of directors on firms’ performance. regional conference on science, technology and social sciences (rcstss 2014), 37–48. doi:10.1007/978-981-10-1458-1_4 ahmed, n., & hadi, o. a. (2017). impact of ownership structure on firm performance in the mena region: an empirical study. accounting and finance research, 6(3). ahsan, t., wang, m., & qureshi, m.a. (2016) firm, industry, and country level determinatns of capital structure: evidence from pakistan. south asianjournal of global business research,5(3), 362-384. amran n. a., & ahmad, a. c. (2010). family succession and firm performance among malaysian companies. international journal of business and social science, 1(2), 193 – 203. 42 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 anderson, r. c., & reeb, d. m. (2003). founding-family ownership and firm performance: evidence from the s&p 500. the journal of finance, lviii, (3), 1301–1328. anderson, r. c., & reeb, d. m. (2004). board composition: balancing family influences in s&p 500 firms. administrative science quarterly, 49(2), 209–237. ang, j. s. & ding, d.k. (2006). government ownership and the performance of government linked companies: the case of singapore. journal of multinational financial management, 16(1), 64-88. baldwin, r. (1963). depreciation for non-profit organizations — an opposing view. the new york certified public accountant (august 1963), 549– 557. baltagi, b. (2001). econometric analysis of panel data (2nd ed.). chichester: john wiley and sons. barnes, l. b., & hershon, s. a. (1976). change management: transferring power in the family business. harvard business review, 1976 issue, 1–26. barry, b. (1975). the development of organization structure in the family firm. journal of general managment, 3, 42-60. bennedsen, m., kongsted, h. c., & nielson, k. m. (2008). the causal effect of board size in the performance of small and medium-sized firms. journal of banking and finance, 32, 1098–1109. bhagat, s., bolton, b., & subramanian, a. (2010, 3 august). ceo education, ceo turnover, and firm performance. working paper, university of colorado, boulder. bokpin, g. a., & arko, a.c. (2009). ownership structure, corporate governance and capital structure decisions of firms, empirical evidence from ghana. economics and finance, 26(4), 246–256. burkart, m., panunzi, f., & shleifer, a. (2003). family firm. journal of finance, the journal of the american finance association, 58(5), 2167–2202. castillo, j., & wakefield, m. w. (2006). an exploration of firm performance factors in family businesses: do families value only the “bottom line”. journal of small business strategy, 17(2), 37 –51. chadha, s., & sharma, a. k. (2015). determinants of capital structure : an emperical evaluation from india. journal of advances in managment research, 12(1), 3-14. claessens, s., djankov, s., & lang, l. h. p. (2000). the separation of ownership and control in east asian corporation. journal of financial economics, 58, 81–112. coban, s. (2014). the interaction between firm growth and profitability: evidence from turkish (listed) manufacturing firms. the journal of knowledge economy & knowledge management, 9, 41-50. daily, c. m., & dollinger, m. j. (1992). an empirical examination of ownership structure in family and professionally managed firms. family business review, 5(2), 11–34. board education, growth and performance of family ceo listed firms in malaysia: 25-46 43 darkos, a. a., & bekiris, f. v. (2010). corporate performance, managerial ownership and endogeneity: a simultaneous equations analysis for the athens stock exchange. research in international business and finance, 24(1), 24-38. darmadi, s. (2013). board members' education and firm performance: evidence from a developing economy. international journal of commerce and management, 23(2), pp. 113–135. davis, p., & stern, d. (1981) adaptation, survival, and growth of the family business: an integrated systems perspective. sage journals, family business review. 1(1), 1-99. demsetz, h., & lehan, k. (1985). the structure of corporate ownership: causes and consequences. journal of political economy, 93(6), 1155-1177. eisenberg, t., sundgren, s., & wells, m. (1998). larger board size and decreasing firm value in small firms. journal of financial economics, 48, 35-54. eriotis, n., vasiliou, d., & neokosmidi., z. v. (2007). how firms characteristics affect capital structure: an empirical study. managerial finance, 33(5), 321-331. faccio, m., lang, l., & young, l. (2002). dividends and expropriation. american economic review, 91, 54–78. filatotchev, i., lien, y-c., & piesse, j. (2005). corporate governance and performance in publicly listed, family-controlled firms: evidence from taiwan. asia pacific journal of management, 22(3), 257–283. ford, r., & mclaughlin, f. (1986). nepotism: boon or bane. personnel administrator, 31(11), 78–89. friend, i., & lang, l. h. p. (1988). an empirical test of the impact of managerial self-interest on corporate capital structure. journal of finance, 43, 217281. gersick, k. e., davis, j. a., hampton, m. m., & lansberg, i. (1997). generation to generation: life cycles of the family business. harvard business school press. united states of america. goel, m., & ramanathan, p. e. (2014). business ethics and corporate social responsibilityis there a dividing line? procedia economics and finance, 11, 49-59. gorton, g., & schmid, f. (2000). class struggle inside the firm: a stud of german codetermination. nber working paper no. 7945, october 2000. gottesman, a. a., & morey, m. r. (2006a). does a better education make for better managers? an empirical examination of ceo educational quality and firm performance. working paper, pace university, new york, 21 april (2004). gottesman, a. a., & morey, m. r. (2006b). manager education and mutual fund performance. journal of empirical finance, 13(2), 145–182. graham, j. r., & harvey, c. (2002). how do cfos make capital budgeting and capital structure decisions. journal of applied corporate finance, spring, pp. 8–23. 44 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 gramling, a. a., & hermanson, d. r., (2006), what roles is your internal audit function playing in corporate governance? internal auditing, 21(6), 37 greve, b. (2008). risk and welfare. social policy and administration, 42(6), 710-713. gujarati, n. d. (2003). basic econometrics (4th ed.). singapore:mcgraw. hill. haleblian, j., & finkelstein, s. (1993). ceo succession and stockholder reaction: the influence of organisational context and event content. academy of management journal, 36(3), 544–563. hermalin, b. e., & weisbach, m. s. (2003). boards of directors as an endogenously determined institution: a survey of the economic literature. economic policy review, april, 7–25. hitt, m. a., ireland, r. d., & lee, h. (2000). technological learning, knowledge management, firm growth and performance: an introductory essay. journal of engineering and technology management, 17(3–4), 231– 246. holderness, c. g., & sheehan, d. p. (1988). the role of majority shareholders in publicly held corporations: an explanatory analysis. journal of financial economics, 20(1-2), 317-346. ibrahim h., & samad. f. m. (2011). corporate governance and agency costs. in kose john, anil k. makhija (ed.), international corporate governance advances in financial economics, 14, 09–130. ibrahim, h., & lau, t. c. (2018). capital structure and firms determinants: evidence from surviving family and non-family listed companies in malaysia, asia conference on business and economic studies (acbes 2018), university of economics ho chi minh city, ho chi minh city, vietnam on 8–9 september 2018. jalbert, t., rao, r., & jalbert, m. (2002). does school matter? an empirical analysis of ceo education, compensation and firm performance. international business and economics research journal, 1(1) winter, 83–98. james, h. s. jr. (1991). owner as manager: extended horizons and family firm. international journal of economics of business, 6(1), 41-55. jensen, j. i., & randoy, t.(2003). family firms and good corporate governance: altruism and agency consideration. paper presented at the europen acedmey of managment meeting, september 2003, milan, itlay. jensen, m. c. (1993). the modern industrial revolution, exit, and the failure of internal control systems. journal of finance, 48, 831–880. kapopoulos, p., & lazaretou, s. (2007). corporate ownership structure and firm performance: evidence from greek firms. journal of corporate governance,15(2), 144-158. karadeniz, e., kandir, s. y., balcilar, m., & onal, y. b.(2009). determinanats of capital structure: evidence form turkish lodging companies. international journal of contemporary hospitality managment, 21(5), 594-609. board education, growth and performance of family ceo listed firms in malaysia: 25-46 45 keles, h. n., ozkan, t. k., & beziric, m. (2011) a study on the effect of nepotism, favoritism and cronyism on organizational trust in the auditing process in family buisness in turkey. international buisness & economic research journal, 10(9), 9-16. kiechel, w. (1984) to cpmpute or not to compute: why executives don’t: stop feeling guilty –it may be that the technology wonder has, a yet , little to offer you. sage journal cornell hotle and administration quartely, 24(4) 9-11. kuo, a., kao, m. s., chang, y. c., & chiu, c. f (2012). the influence of international experience on entry mode choice: difference between family and non-family firms. european management journal, 30(3), 248-263. la porta, r. lopez-de-silances, f., & shliefer, a. (1999). corporate ownership around the world. the journal of finance, vol. liv, no. 2, 471–517. lipton, m., & lorsch, j. (1992). a modest proposal for improved corporate governance. business lawyer, 48, 59–77. lubatkin, m. h., schulze, w. s., & dino, r. n. (2005). the effect of parental altruism on the governance of family managed firms. journal of organizational behaviour, 26, 313-330. malaysian code on corporate governance. (2012). security commission malaysia. doi.http://www.sc.com.my/malaysian-code-on-corporategovernance-2012/ mcknight, p. j. & mira, s. (2003). corporate governance mechanisms, agency costs and firm performance in uk firms, http://ssrn.com/ abstract=460300 miller, d., & le breton-miller, i. (2006). family governance and firm performance: agency, stewardship, and capabilities. family business review, 19(1), 73–87. minichilli, a., corbetta, g., & macmillan, i. c. (2010). top management teams in family-controlled companies: ‘familiness’, ‘faultlines’ and their impact on financial performance. journal of management studies, 47(2), 205–222. mishra, c. s., randoy, t., & jenssen, j. i. (2001). the effect of founding family influence on firm value and corporate governance. journal of international financial management and accounting, 12(3), 235–259. ng’eni, f.b. (2015). the corporate governance and firm performance: a review of existing empirical evidence. european journal of business and management, 7(33), 22-35. ohman, p., & yazdanfar, d. (2017). short and long term debts in swedish smes. review of accounting and finance,16(1), 106-124. pandey, i.m (2001). corporate dividend policy and behaviour: the malaysian experince. iima working paper no. 2001-11-01 . 46 the international journal of banking and finance, vol. 15, no 2, 2020 : 25-46 pandey, i.m.(2004). capital structure,profitability and market structure: evidence from malaysia.the pacific journal of economics, 8(2), 7891. ramón-llorens, m. c., garcía-meca, e., & duréndez, a. (2017). influence of ceo characteristics in family firms internationalization. international business review, 26(4), 786–799. samad, f.a. (2004). corporate governance and ownership structure in the malaysian corporate sector. advances in financial economics, 9, 355385. sebora, t. c., & wakefield, m. w. (1998). antecedents of conflict or business issues in family firms. journal of entrepreneurship education, 1, 2–18. sheikh, n. a., & wang, z. (2012). effects of corporate governance on capital structure: empirical evidence from pakistan. corporate governance, 12(5), 629–641. sraer, d.a., & thesmar, d. (2006). performance and behaviour of family firms: evidence from the french stock market. ecgi finance working paper no. 130/2006 stark, o., & falk, i. (1998). transfers, empathy formation, and reverse transfers. the american economic review, 88(2), 271–276. ujunwa, a., nwakoby, i., & ugbam, c.o. (2012). ccorporate board diversity and firm performance: evidence from nigeria. corporate ownership and control, 9(2), 216-223. vakilifard, h. r., gerayli, m. s., yanesari, a. m., & ma’atoofi, a. r. (2011). effect of corporate governance on capital structure: case of the iranian listed firms. european journal of economics, finance and administrative sciences, 35, 165–172. villalonga, b., & amit, r. (2006). how do family ownership, control and management affect firm value? journal of financial economics, 80, 385–417. wald, j. k. (1999). how firm characteristics affect capital structure: an international comparison. journal of financial research, 22, 161-187. yermack, d. (1996). higher market valuation of companies with a small board of directors. journal of financial economics, 40, 185–211. ying, c.h., & mie, y.c. (2014). ceo education and firm performance. evidence from hong kong (unpublished dissertasion degree project proposal). submitted to the school of business in partial fulfilment. zax, j. & ichniowski, c. (1988). the effects of public sector unionism on pay, employment, department budgets, and municipal expenditures. in: freeman, r. b. and ichniowski, c. (eds), when public sector workers unionize. university of chicago press, 323 – 364. isbn: 0-226-26166-2 51 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 how to cite this article: sa’ad, h. n., abubakar, z., & suleiman, s. (2023). accounting conservatism and corporate tax avoidance. international journal of banking and finance, 18(1), 51-66. https://doi. org/10.32890/ ijbf2023.18.1.3 accounting conservatism and corporate tax avoidance 1hamid njiddah sa’ad, 2zaid abubakar & 3salami suleiman 1business school, ahmadu bello university, zaria, nigeria 2chairman, kaduna state internal revenue service, nigeria 3 business school, ahmadu bello university, zaria, nigeria 3corresponding author: suleiman_salami@yahoo.com received: 31/8/2021 revised: 25/12/2021 accepted: 7/1/2022 published: 5/1/2023 abstract this study investigates the effect of accounting conservatism on the corporate tax avoidance of listed non-financial firms in nigeria. this study computes corporate tax avoidance based on the cash effective tax rate (cetr), gaap effective tax rate (getr) and book tax difference (btd). accounting conservatism was measured using negative accruals. the study employed an ex-post factor research design utilizing unbalanced panel data. the study covered 48 listed non-financial firms during the period between 2014 and 2020. three regression models were developed and utilized in the study. the study has revealed that accounting conservatism has a negative and significant effect on both the getr and btd. it is recommended that the financial reporting council of nigeria should encourage promulgation of standards which improve conservatism in financial reporting, as it has been empirically proven to reduce tax avoidance practices by non-financial firms in nigeria. https://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 52 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 keywords: accounting conservatism, tax avoidance, book tax difference, effective tax rates. jel classification: g3, h26, m41. introduction income tax is not desirable from the firm’s perspective (rezaei & dorbehani, 2014). tax consequence, on the other hand is exclusively significant in political space as is evidenced by recent cases of corporate tax avoidance involving giant multinational companies like google and amazon. corporate tax avoidance though, may be legal, but it could tarnish the reputation of a company. tax avoidance will ultimately lower government revenue and subsequently have a negative impact, especially on the fragile nigerian economy which is just on the road to recovery from a recession. the tax on corporate profit yielded nine percent of the revenue for the nigerian government in 2017, a revenue source that has been trending downwards (odhiambo & olushola, 2018). the share of revenue coming from the corporate income tax dropped from one-third of the total in the early 1950s to less than one-tenth in 2017. revenue from the tax has fallen from an average of 3.7 percent of gross domestic product (gdp) in the late 1960s to an average of just 1.7 percent of gdp over the past few years, despite ticking up to 1.9 percent of the gdp in 2014 and 2015. the downward trend in the corporate tax revenue is largely owed to tax avoidance schemes adopted by firms in nigeria. the most recent case of tax avoidance in nigeria was the repatriation by the multinational telephone network (mtn) of over $8.2billion profit out of the country and to avoid paying appropriate taxes to the government (odhiambo & olushola, 2018). accounting conservatism can serve as a tool to discipline management in financial reporting. accounting conservatism is one of the main characteristics of financial reporting, and has been incorporated in accounting theory and practice for a long time (kootanaee et al., 2013). accounting conservatism is the tendency that accountants, when encountering uncertainties in economic transactions, choose to report lower estimates for the values of assets and revenues, but higher estimates for the values of liabilities and expenses. accounting conservatism in financial reporting limits 53 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 management practices on earnings, which gives room for corporate tax avoidance. the asymmetric recognition of gains and losses implies an impairment of the neutrality of financial reports, which is the main argument against conservatism. the argument is essentially that recognition of gains in financial statement has to be delayed until verifiable evidence is obtained. in contrast, losses are incorporated timely into accounts once it arises. despite the significant role of tax avoidance in depressing government revenue, most studies in nigeria, such as those by aminu and hassan (2017), ugwunta and ugwuanyi (2018), suleiman and anifowose (2014) have instead examined its effects on either corporate governance or financial performance. this study attempts to analyze the nature and direction of tax reducing effect of conservatism on nonfinancial institutions in nigeria. to our knowledge, no similar research on accounting conservatism and tax avoidance has been conducted in the nigerian context, therefore our study intends to fill that research gap in the literature. also, this study will not use the basu (1997) conservatism model because it is not firm specific, but rather negative accruals (givoly & hayn, 2000). furthermore, this study will be using unbalanced panel data, as opposed to the widely used balanced panel data, which implies that all elements will be factored in all timeframes and this will upsurge the robustness of the results. the study has focused on the effect of accounting conservatism on tax avoidance. it concentrated on listed non-financial institutions in nigeria which included those producing consumer goods, and conglomerates in the industrial and health sectors. the results of this study will assist future researchers in this area by providing additional empirical explanations on the relationship between conservatism and tax avoidance. it will serve as a blueprint for future researchers in the area and complement the existing body of literature. it will also assist regulators and standard setters in facilitating the development of standards that will discourage tax avoidance activities by companies. literature review and hypothesis development a handful of studies have provided both analytical and empirical evidence that accounting conservatism affects tax avoidance. this section will discuss some tax concepts and also review the empirical evidence and theoretical framework of the subject matter. 54 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 effective tax rate measures effective tax rate (etr), is a measure of tax avoidance which captures the average rate of tax per dollar of income or cash flow. understanding what the numerator captures is essential. there are two types of effective tax rate measures, namely the cash effective tax rate (cetr) and the gaap effective tax rate (gaap etr). the cetr is computed by dividing taxes paid in cash by pre-tax accounting income while the gaap etr is defined as the total income tax expense divided by the pre-tax accounting income. book-tax differences this is a measure of tax avoidance which shows the difference between accounting income and taxable income. the book-tax differences (btd) is usually computed as the difference between the pre-tax income according to the financial statement (also called “book income”) and the taxable income according to the tax return. negative accruals measure givoly and hayn (2000) have proposed a measure of conservatism that focuses on non-operating accruals as a subset of the firm’s book value. non-operating accruals are calculated as total accruals minus operating accruals. total accruals are equal to the firms net income before depreciation minus the cash flow from operating activities. empirical evidence tax avoidance does not have a universally acceptable definition. however, tax avoidance according to aminu and hassan (2017), is as an act by corporate firms to capitalize on those areas in tax laws that are ambiguous to reduce their tax liability. purwantini (2017) analyzed directly and indirectly conservative accounting practices influence towards avoidance on companies listed in the indonesian stock exchange during the period between 2013 and 2015. a sample of 23 companies was taken, making a total of 69 observations. the gaap effective tax rate was used to measure tax avoidance and the negative accrual method was used to measure accounting conservatism. the acquired data was analyzed using path analysis, and the findings of the research pointed to the conclusion that conservatism accounting practices significantly influenced book tax difference, but did not influence tax avoidance. 55 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 yuniarsih (2018) carried out a study to explain the influence of accounting conservatism and corporate governance mechanism on tax avoidance practices of corporations in indonesia. the study sampled 123 companies listed in the indonesia stock exchange (idx), particularly the manufacturing companies which had been listed for a period of three years, between 2014 and 2016. secondary data was collected via the audited financial statements of the companies. to test the hypotheses of the present study, a multiple regression analysis was carried out. the results indicate that conservatism has no significant effect on tax avoidance, a conclusion which is in congruence with the findings of purwantini (2017). bornemann (2018) conducted a study in austria to analyze the relationship between accounting conservatism, future tax rate cuts and the level of book-tax conformity in countries using a panel of firms across 18 countries from 1995 to 2010. the researcher used c-score to measure conditional conservatism and used book tax conformity to measure tax avoidance. the c-score is a measure that reflects the timing of conservatism changes and the variation of conservatism across firms within an industry. the study was able to establish that income statement conservatism was positive and significantly associated with future tax rate cuts when book-tax conformity was high. the effect was particularly manifested in firms that concentrated the majority of their operations in the country in which the tax rate was cut. in contrast, there was no significant relationship between future tax rate cuts and the statement of financial position conservatism. gan (2018) examined the relationship between conditional conservatism and tax avoidance. he took a sample of listed u.s. companies during the period from 2009 to 2016. he computed tax avoidance based on cash effective tax rates (cetr) and employed the c-score method developed by khan and watts (2009) and the skewness method from givloly and hayn (2000) to measure conditional conservatism. the results of the study indicate that the c-score is negatively correlated to the cetr, irrespective of the different models utilized. the negative association of the c-score and the cetr corroborates the hypothesis of the study that ceteris paribus, conditional conservatism is negatively associated with tax burdens. muhsin (2019) was aimed at getting empirical evidence about the effect of accounting conservatism and ownership structure on the 56 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 aggressive tax avoidance actions of listed manufacturing companies in indonesia during the period between 2012 and 2016. secondary data was obtained from financial statements of manufacturing companies listed on the indonesia stock exchange. sampling was done by the purposive sampling method, with a total of 194 samples collected from 49 companies for five years. the result of the multiple linear regression analysis showed that accounting conservatism and foreign ownership had a significant negative impact on aggressive tax avoidance. positive accounting theory has provided a complete theoretical framework for this study. according to watts (2003) “positive accounting theory states that accounting conservatism is an efficient contracting and governance mechanism to mitigate information asymmetries and solving problems associated with agency”. watts (2003) developed four explanations of accounting conservatism based on the postulates of positive accounting theory. these included taxation, litigation, contracting and accounting regulation. the taxation explanation of conservative accounting is that firms engage in conservative accounting practice to lower their taxes. in light of the foregoing discussion, the following hypotheses are proposed: h1 : accounting conservatism has no significant effect on the gaap effective tax rate of listed non-financial firms in nigeria. h2 : accounting conservatism has no significant effect on the cash effective tax rate of listed non-financial firms in nigeria. h3 : accounting conservatism has no significant effect on the book tax difference of listed non-financial firms in nigeria. methodology the section spells out the nature of the research design employed and justifications for the methods and techniques used, and lastly the variable measurement and model specification. the design of this study is an ex-post factor research design utilizing unbalanced panel data. this study has focused on listed non-financial institutions in nigeria, comprising conglomerates producing consumer goods, 57 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 and health and industrial firms. the nature of the data involved in a research normally determines the tool to be adopted for the analysis. the data is secondary in nature, extracted from annual reports and accounts of firms listed on the nigerian stock exchange. accounting conservatism was proxied by negative accruals (givoly & hayn, 2000), while tax avoidance was proxied by the cash effective tax rate (cetr), gaap effective tax rate (getr) and book-tax difference (btd). this study uses firm size, leverage and profitability as the control variables. the present study has used generalized least square regression as the tool of analysis because it involves a dependent variable, three independent variables and three control variables. the reason behind the choice of a multiple regression technique is that it reveals the actual impact the independent variable has on the dependent variable. in this study, multiple regression is the appropriate tool that can clearly explain the effect of accounting conservatism on tax avoidance the population comprised 20 consumer goods firms, 9 health firms, 13 industrial firms and 6 conglomerate firms which came to a total of 48 firms. this represented 96 percent of the total population, the result of the elimination of golden guinea breweries and nigeria-german chemical from the consumer goods and health sectors respectively, as the result of the unobtainability of financial records. however, the population of the study will vary for each year during the period under investigation, depending on the availability of financial records of the firms in each year due the fact that unbalanced panel data was utilized. the population for each year for each of the sectors is as presented in table 1. table 1 shows that the sample collected for the period of study included 287 firms in total. the actual number of firms studied for each year are as follows: 36 in 2014, representing 75 percent (i.e. 36 divided by 48) of the sample; 43 in 2015 and 2019, representing 89.58 percent of the sample; 45 in 2016, representing 93.75 percent of the sample; 42 in 2017, representing 87.50 percent of the sample; 40 in 2018, representing 83.33 percent of the sample, and 38 in 2020, representing 79.17 percent of the sample. this means that there was an average of 41 firms studied across the seven-year period, representing 85.42 percent of the entire sample. 58 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 table 1 population of firms years 2014 2015 2016 2017 2018 2019 2020 total conglomerates sector 5 5 6 6 6 6 5 39 consumer goods sector 17 20 18 16 16 17 12 116 industrial sector 8 10 12 12 12 13 13 80 health sector 6 8 9 8 6 7 8 52 total 36 43 45 42 40 43 38 287 note. source from the nse fact book 2020 this study has assumed a linear relationship between accounting conservatism and the occurrence tax avoidance of listed non-financial firms in nigeria. the study also assumed that conservatism and tax avoidance could also be affected by some firm attributes, and this has led to the adoption of three firm attributes, i.e. firm size, leverage and profitability. in line with these assumptions, the following three regression models were formulated. table 2 measurement of variables variables measurements source na measured by profit before extraordinary items plus depreciation minus operating cash flow divide by total assets. givoly and hayn (2000) getr measured by gaap tax expense divide by profit before tax. guenther et al. (2014) cashetr measured by cash tax paid divide by profit before tax. guenther et al. (2014) btd measured by the difference between accounting profit and taxable profit scaled down by total assets. chyz et al. (2015) size measured by the natural logarithm of total assets. katz et al. (2013) lev measured by total debt divide by total assets. adams & ferreira (2009) roa measured by profit before tax divide by total assets. kubata et al. (2013) 59 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 table 2 shows the variables and their measurement. model 1: getrit = α0 + β1nait+ β2size it + β2levit + β3roait + model 2: cashetrit = α0 + β1nait+ β2size it + β2levit + β3roait + model 3: btdit = α0 + β1nait+ β2size it + β2levit + β3roait + where: getrit = gaap effective tax rate of firm i in year t cashetrit = cash effective tax rate of firm i in year t btdit = book-tax difference of firm i in year t nait = negative accruals of firm i in year t size it = size of firm i in year t levit = leverage of firm i in year t roait = return on assets for firm i in year t = the error term of firm i in year t results and discussion this section encompasses the data presentation, analysis and findings of the study. the descriptive, correlation analysis and the outcome of the ordinary least square regression using robust standard errors, which represents the main findings of the study are presented in this section. descriptive analysis table 3 shows the results of the descriptive analysis, which has looked at the minimum, maximum, mean and standard deviation of the variables under investigation in the present study. the average getr is -19.51, which indicates that on average, the firms have a negative tax expense. this is the result of losses. the lower getr values suggest the prevalence of higher tax avoidance practices. given the negative average value for the getr of -19.51, sampled firms were seen to have engaged in tax avoidance activities during the period of the study. the standard deviation is 307.09, which shows a huge deviation from the mean. the getr in fact has the highest standard deviation amongst all the variables, with the minimum and maximum ranging from -5183.66 to 15.31, respectively. for the cashetr, the average is zero, which is extremely low. the minimum and maximum cetr at -26.78 and 11.79, respectively also means that the government needs to do more in terms of the 6 2018, representing 83.33 percent of the sample, and 38 in 2020, representing 79.17 percent of the sample. this means that there was an average of 41 firms studied across the seven-year period, representing 85.42 percent of the entire sample. this study has assumed a linear relationship between accounting conservatism and the occurrence tax avoidance of listed non-financial firms in nigeria. the study also assumed that conservatism and tax avoidance could also be affected by some firm attributes, and this has led to the adoption of three firm attributes, i.e. firm size, leverage and profitability. in line with these assumptions, the following three regression models were formulated. model 1: getrit = α0 + β1nait+ β2size it + β2levit + β3roait + it model 2: cashetrit = α0 + β1nait+ β2size it + β2levit + β3roait + it model 3: btdit = α0 + β1nait+ β2size it + β2levit + β3roait + it where: getrit = gaap effective tax rate of firm i in year t cashetrit = cash effective tax rate of firm i in year t btdit = book-tax difference of firm i in year t nait = negative accruals of firm i in year t size it = size of firm i in year t levit = leverage of firm i in year t roait = return on assets for firm i in year t it = the error term of firm i in year t table 2 shows the variables and their measurement. table 2 measurement of variables variables measurements source na measured by profit before extraordinary items plus depreciation minus operating cash flow divide by total assets. givoly and hayn (2000) getr measured by gaap tax expense divide by profit before tax. guenther et al. (2014) cashetr measured by cash tax paid divide by profit before tax. guenther et al. (2014) btd measured by the difference between accounting profit and taxable profit scaled down by total assets. chyz et al. (2015) size measured by the natural logarithm of total assets. katz et al. (2013) lev measured by total debt divide by total assets. adams & ferreira (2009) roa measured by profit before tax divide by total kubata et al. (2013) 6 2018, representing 83.33 percent of the sample, and 38 in 2020, representing 79.17 percent of the sample. this means that there was an average of 41 firms studied across the seven-year period, representing 85.42 percent of the entire sample. this study has assumed a linear relationship between accounting conservatism and the occurrence tax avoidance of listed non-financial firms in nigeria. the study also assumed that conservatism and tax avoidance could also be affected by some firm attributes, and this has led to the adoption of three firm attributes, i.e. firm size, leverage and profitability. in line with these assumptions, the following three regression models were formulated. model 1: getrit = α0 + β1nait+ β2size it + β2levit + β3roait + it model 2: cashetrit = α0 + β1nait+ β2size it + β2levit + β3roait + it model 3: btdit = α0 + β1nait+ β2size it + β2levit + β3roait + it where: getrit = gaap effective tax rate of firm i in year t cashetrit = cash effective tax rate of firm i in year t btdit = book-tax difference of firm i in year t nait = negative accruals of firm i in year t size it = size of firm i in year t levit = leverage of firm i in year t roait = return on assets for firm i in year t it = the error term of firm i in year t table 2 shows the variables and their measurement. table 2 measurement of variables variables measurements source na measured by profit before extraordinary items plus depreciation minus operating cash flow divide by total assets. givoly and hayn (2000) getr measured by gaap tax expense divide by profit before tax. guenther et al. (2014) cashetr measured by cash tax paid divide by profit before tax. guenther et al. (2014) btd measured by the difference between accounting profit and taxable profit scaled down by total assets. chyz et al. (2015) size measured by the natural logarithm of total assets. katz et al. (2013) lev measured by total debt divide by total assets. adams & ferreira (2009) roa measured by profit before tax divide by total kubata et al. (2013) 6 2018, representing 83.33 percent of the sample, and 38 in 2020, representing 79.17 percent of the sample. this means that there was an average of 41 firms studied across the seven-year period, representing 85.42 percent of the entire sample. this study has assumed a linear relationship between accounting conservatism and the occurrence tax avoidance of listed non-financial firms in nigeria. the study also assumed that conservatism and tax avoidance could also be affected by some firm attributes, and this has led to the adoption of three firm attributes, i.e. firm size, leverage and profitability. in line with these assumptions, the following three regression models were formulated. model 1: getrit = α0 + β1nait+ β2size it + β2levit + β3roait + it model 2: cashetrit = α0 + β1nait+ β2size it + β2levit + β3roait + it model 3: btdit = α0 + β1nait+ β2size it + β2levit + β3roait + it where: getrit = gaap effective tax rate of firm i in year t cashetrit = cash effective tax rate of firm i in year t btdit = book-tax difference of firm i in year t nait = negative accruals of firm i in year t size it = size of firm i in year t levit = leverage of firm i in year t roait = return on assets for firm i in year t it = the error term of firm i in year t table 2 shows the variables and their measurement. table 2 measurement of variables variables measurements source na measured by profit before extraordinary items plus depreciation minus operating cash flow divide by total assets. givoly and hayn (2000) getr measured by gaap tax expense divide by profit before tax. guenther et al. (2014) cashetr measured by cash tax paid divide by profit before tax. guenther et al. (2014) btd measured by the difference between accounting profit and taxable profit scaled down by total assets. chyz et al. (2015) size measured by the natural logarithm of total assets. katz et al. (2013) lev measured by total debt divide by total assets. adams & ferreira (2009) roa measured by profit before tax divide by total kubata et al. (2013) 6 2018, representing 83.33 percent of the sample, and 38 in 2020, representing 79.17 percent of the sample. this means that there was an average of 41 firms studied across the seven-year period, representing 85.42 percent of the entire sample. this study has assumed a linear relationship between accounting conservatism and the occurrence tax avoidance of listed non-financial firms in nigeria. the study also assumed that conservatism and tax avoidance could also be affected by some firm attributes, and this has led to the adoption of three firm attributes, i.e. firm size, leverage and profitability. in line with these assumptions, the following three regression models were formulated. model 1: getrit = α0 + β1nait+ β2size it + β2levit + β3roait + it model 2: cashetrit = α0 + β1nait+ β2size it + β2levit + β3roait + it model 3: btdit = α0 + β1nait+ β2size it + β2levit + β3roait + it where: getrit = gaap effective tax rate of firm i in year t cashetrit = cash effective tax rate of firm i in year t btdit = book-tax difference of firm i in year t nait = negative accruals of firm i in year t size it = size of firm i in year t levit = leverage of firm i in year t roait = return on assets for firm i in year t it = the error term of firm i in year t table 2 shows the variables and their measurement. table 2 measurement of variables variables measurements source na measured by profit before extraordinary items plus depreciation minus operating cash flow divide by total assets. givoly and hayn (2000) getr measured by gaap tax expense divide by profit before tax. guenther et al. (2014) cashetr measured by cash tax paid divide by profit before tax. guenther et al. (2014) btd measured by the difference between accounting profit and taxable profit scaled down by total assets. chyz et al. (2015) size measured by the natural logarithm of total assets. katz et al. (2013) lev measured by total debt divide by total assets. adams & ferreira (2009) roa measured by profit before tax divide by total kubata et al. (2013) 60 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 assessment and collection of tax. similarly, the lower values of the cashetr have suggested that there were higher tax avoidance practices the average value for the btd is -0.91, while the minimum and maximum is -244.52 and 3.86, respectively. table 3 descriptive analysis variables obs mean std. dev. min max getr 287 -19.51 307.09 -518.66 15.31 cashetr 287 0.00 1.81 -26.78 11.79 btd 287 -0.91 1.40 -244.52 3.86 na 287 -0.23 8.01 -119.61 64.24 firmsize 287 10.08 0.88 7.83 12.23 roa 287 8.25 72.33 -0.31 737.54 leverage 287 0.07 0.30 -1.10 3.32 the na has an average of -0.23 and a minimum and maximum of -119.61 and 64.24, respectively. when accrual values are negative, this suggests the prevalence of conservative accounting practice among firms. the mean for firm size (expressed as the natural log of total assets) is 10.08, which interestingly is having the lowest standard deviation from the mean at 0.88, with the minimum and maximum standing at 7.83 and 12.23, respectively. table 3 also shows that on average, the leverage of the firms (measured as the total term debt scaled by total assets) stands at 825.65 percent, which indicates that the firms depends largely on external financing. this could also be considered as a strategy used by the firms to reduce their taxable profits, as interest on external financing is tax deductible. the minimum and maximum leverage stands at -0.31 and 737.54, respectively. table 3 also shows that the average roa stood at 7.8 percent. the roa measures how effective a firm is in utilizing its assets in generating earnings. the ratio of 7.8 percent indicates that for every n100 invested in assets, the average return is n7.80k, which is a relatively low accounting performance indicator. the highest and lowest roa standing were at 332.82 percent and -110.27 percent, respectively. correlation analysis the correlation matrix for the explained, explanatory and control variables are analyzed and as presented in table 4. 61 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 table 4 correlation matrix variables getr cetr btd na lev f-size roa getr 1 cetr 0.07 1 btd 0.99 0.13 1 na -0.46 -0.21 -0.46 1 lev -0.58 -0.29 -0.60 -0.34 1 f-size 0.14 0.05 0.12 0.03 0.12 1 roa 0.01 0.01 0.03 0.03 -0.04 -0.04 1 the values in table 4 are all the correlation coefficient values of the variables used in the regression models. based on the correlation analysis, the na and getr were negatively correlated. the na was also negatively associated with the cashetr which was congruous with the results in gan (2018). similarly, the na was negatively correlated with the btd while the na was negatively associated with leverage. correlation analysis is a measure of association which does not suggest a cause and effect relationship. however, the roa was positively associated with the na at the 10 percent level of significance. the relationships between most of the explanatory variables were minimal, insignificant and negligible. hence there is no problem of singularity of data. moreover, multicollinearity is not expected to pose a problem to the overall results of the study. table 5 summary of regression results getr. cashetr. btd na -28.49 7.03 -4.05 firm size -3.47 2.13 -1.63 leverage -3.58 1.09 -3.27 roa 14.65 8.11 1.66 no of observation 287 287 287 adj r2 0.83 0.19 0.87 f. value 5.13*** 18.05*** 49.01*** 62 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 regression result this section discusses the regression results of tax avoidance (proxied by the getr, cashetr & btd) on the independent variable (proxied by na) and control variables (the roa, leverage and firm size). the three dependent variables were regressed separately against the explanatory variables. to test whether or not heteroskedasticity exists, the getr model and breusch-pegan/ cook-weisberg tests were performed. the null hypothesis of constant variance amongst the variables was tested, at the 1 percent level of significance. it was safe to reject the null hypothesis and eventually conclude that the regression model contained heteroskedasticity. a robust regression model was run and the result was as depicted in table 5 above giving an adj. r2 = 0.83. this implied that the 83 percent variation in the getr was influenced by the na and the control variables. the f. statistics gives a value of 5.13 which is significant at the 1 percent level of significance. the na had a beta coefficient of -28.49, which indicated that there was a negative relationship between accounting conservatism and the getr of listed non-financial firms in nigeria. the implication of the result is that for every change in the level of accounting conservatism, the getr of the firms will decrease by 28.49. at the 1 percent level of significance, the null hypothesis was rejected, which has stated that accounting conservatism has no significant effect on the gaap effective tax rate of listed non-financial firms in nigeria. this implies that accounting conservatism is associated with reducing the tendency of tax avoidance by firms. the control variable firm size showed a negative impact on the getr with a beta coefficient of -3.47, which is significant at 10 percent. also, the control variable leverage showed a negative impact on the getr, with a value of -3.58 and this is significant at the 1 percent level of significance. the roa however, showed a positive impact, with a value of 14.65 at the 10% level of significance. this is logical because the more returns generated on assets employed by a firm, the higher the tax expense. for every 100 increase in the roa of a firm, the getr will increase by 14.65 and vice versa. the second regression model is in relation to the cashetr model. to test whether or not heteroskedasticity exists, the breusch-pegan/ 63 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 cook-weisberg test was performed and the same results were obtained as the one for the first regression which concluded that their exist heteroskedasticity. a robust regression model was run and the result is as depicted above in table 5, giving an adj. r2 = 0.19. this implies that the 19 percent variation in the cashetr was influenced by the na and the control variables. the remaining 81 percent was explained by other variables, which was represented by the error term. the f. statistics gives a value of 18.05, which is significant at the 1 percent level of significance. the beta coefficient for the na in table 5 shows a positive value of 7.03, which signifies a positive impact. it was however, insignificant with a p value of 0.351. as a result, the null hypothesis which has stated that accounting conservatism has no significant effect on the cash effective tax rate of listed non-financial firms in nigeria is not rejected. this result is consistent with the findings in purwantini (2017), gan (2018) and yuniarsih (2018). for the control variables, firm size has an insignificant effect on the cetr, with a value of 0.021. leverage has a beta coefficient of -0.010, which indicates a negative but insignificant impact on the cetr. the roa has a beta value of 0.060, which implies an insignificant positive impact on the cetr. the third regression is in relation to the btd model. the breuschpegan/ cook-weisberg test was performed to test whether or not heteroskedasticity exists and the test results were consistent with that of the first two regression models, which all confirmed the existence of heteroskedasticity. the result as shown above in table 5 was obtained after running the robust regression to overcome the problem of heteroskedasticity. the coefficient of determination showed a value of 0.8745, which indicates that the independent variable, together with the control variables accounted for 87.45 percent of any variation in the dependent variable. the 12.55 percent of the variation is explained by other variables represented by the error term. this in essence implies that the model is to a great extent healthy. the beta value for the na is -4.05, which indicates a negative impact on the btd at the 1 percent level of significance. hypothesis three which has stated that accounting conservatism has no significant effect on book-tax difference of listed non-financial firms in nigeria is therefore, rejected. leverage has a beta coefficient of -0.175, which 64 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 indicates a negative impact on the btd, similar to the firm size (beta, -0.209). the roa has a beta coefficient of 1.607, which implies a positive impact on the btd. conclusions and recommendations in summary, the findings revealed a negative effect of the na on the three proxies of tax avoidance (the getr, cashetr and btd). the study, therefore has concluded that accounting conservatism reduces the tax avoidance of listed non-financial firms in nigeria. this is not surprising because most firms are on the verge of fully adopting the international financial reporting standards (ifrs). the local statement of accounting standards (sas) is rule based and does not permit flexibility in reporting accounting numbers. rule based standards do not give room to managers to apply discretion in reporting accounting numbers that reduces tax payments. it is recommended that the financial reporting council of nigeria should encourage the promulgation of standards which will improve conservatism in financial reporting, as it is has been empirically shown to reduce tax avoidance practices by non-financial firms in nigeria. the study suffers from the following limitations. the first drawback was the unavailability of annual reports of some firms, which led to the reduction of the population of the study from 50 to 48 firms. furthermore, some firms did not have complete annual reports for the seven-year period under investigation in this study. as such this might affect the strength of the overall results. secondly, the measures of conservatism and tax avoidance selected were from amongst the many other measures available. lastly, the study focused on listed firms, therefore its result could not be generalized to non-listed firms. similarly, the present study only focused on non-financial firms and therefore, its results might not apply to financial firms. for future research, researchers can conduct a study of the same topic, but using listed financial firms. furthermore, other measures of accounting conservatism and tax avoidance could be utilized. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. 65 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 references adams, r. b., & ferreira, d. (2009). women in the boardroom and their impact on governance and performance. journal of financial economics, 94(2), 291-309. https://doi.org/10.1016/j. jfineco.2008.10.007 aminu, l., & hassan, s. u. (2017). accounting conservatism and financial performance of nigerian deposit money banks: an analysis of recent economic recession. scholedge basu, s. (1997). the conservatism principle and the asymmetric timeliness of earnings. journal of accounting and economics, 24(1), 3-37. https://doi.org/10.1016/s0165-4101(97)00014-1 bornemann, t., (2018). tax avoidance and accounting conservatism. wu international taxation research paper series, no. 2018 – 04. chyz, j. a., gaertner, f. b., kausar, a., & watson, a. (2015). overconfidence and corporate tax policy. www.ssrn.com guenther, d. a., matsunaga, s. r., & williams, b. m. (2014). tax avoidance, tax aggressiveness, tax risk and firm risk. lundquist college of business, university of oregon, eugene. www.ssrn.com katz, s., khan, u., & schmidt, a. p. (2013). tax avoidance and future profitability. www.ssrn.com kootanaee, a. j., seyyedi, j., nedaei, m., & kootanaee, m. j. (2013). accounting conservatism and corporate governance: evidence from tehran stock exchange. international journal of economics, business and finance, 1(10), 301-319. http:// ijebf.com/ gan, z. (2018). conditional conservatism and tax avoidance (unpublished master’s thesis). eramus school of economics, university of rotterdam givoly, d., hayn, c. (2000). the changing time-series properties of earnings, cash flows and accruals: has financial reporting become more conservative? journal of accounting and economics, 29(3), 287-320. https://doi.org/10.1016/s01654101(00)00024-0 khan, m., & watts, r. l. (2009). estimation and empirical properties of a firm-year measure of accounting conservatism. journal of accounting and economics, 48(2), 132–150. https://doi. org/10.1016/j.jacceco.2009.08.002 kubata, a., lietz, g. m., & watrin, c. (2013) does corporate tax avoidance impair earnings informativeness? working paper. http://dx.doi.org/10.2139/ssrn.2363873 66 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 51–66 muhsin, m. s. k. (2019). looking for empirical evidence between accounting conservatism and ownership structure towards the aggressive tax avoidance in public listed companies of indonesia. research journal of finance and accounting, 10(10) pp 21-26. 10.7176/rjfa odhiambo, o., & olushola, o. (2018). taxation and economic growth in a resource-rich country: the case of nigeria. taxation and economic growth. 10.5772/intechopen.74381 purwantini, h. (2017). minimizing tax avoidance by using conservatism accounting through book tax differences: case study in indonesia. international journal of research in business and social science, 6(5), 55–67. https://doi. org/10.20525/ijrbs.v6i5.765 suleiman, s., & anifowose, m. (2014). corporate governance and conditional accounting conservatism in nigeria listed food and beverages firms. international journal of accounting and taxation, 2(2), june, pp. 65-84. http://ijatnet.com rezaei, f., & dorbehani, m. (2014). the effect of tax avoidance on the firms’ financial reporting transparency. iranian journal of business economics, 1(3), pp1-12. http://www.scijour.com/ ijbe ugwunta, d. o., & ugwuanyi, b. u. (2018). accounting conservatism and performance of nigerian consumer goods firms’: an examination of the role of accruals. international journal of financial research, 10(1), pp 1-9. 10.5430/ijfr.v10n1p1 watts, r. l., (2003). conservatism in accounting part i: explanations and implications. accounting horizons, 17, 207-221. https:// doi.org/10.2308/acch.2003.17.3.207 yuniarsih, n. (2018). the effect of accounting conservatism and corporate governance mechanism on tax avoidance. academic research international, 9(3), pp 68-78. http://www.savap.org.pk the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 37 market liberalization and trading in korea lloyd p. blenman, dar-hsin chen and chun-da chen university of north carolina, national chiaotung university, taiwan, and dah-yeh university, taiwan ______________________________________________________ abstract this paper reports on the trading behavior of major participants, investment trust companies, banks, and foreigners in south korea in the period after the currency markets were liberalized and the limits on foreign investments were lifted. it was found that trading in the spot currency market was impacted by volatility in the daily won/usd rates. as the daily unexpected range expanded (narrowed), daily spot trading volume and volatility increased (decreased). this is evidence of asymmetric trading behavior on the part of market participants. it was found that only investment trust companies adjusted their spot positions by trading usd futures as a response to unexpected volatility changes of the exchange rate. there is evidence of volatility clustering of the trading volatilities across korean markets and trader types and no signs of market instability was found. keywords: south korea, market liberalization, trading behavior, currency, multivariate garch model jel classi� cation: f21, f31, g14 ________________________________________________________________ 1. introduction the � nancial markets in south korea have grown rapidly over the past decade. research on korea’s economy and its � nancial markets has grown even faster. korea has recently become one of the most active � nancial markets as well. it is categorized as an emerging market. its derivative trading, in terms of trading volume and value is among the highest in the world. despite the growth, very little attention has been given to its currency derivatives market (kim, kim and kim (2004) and an examination of all of its markets in an integrative fashion. a liberalization of the exchange rate system and the capital market of a country impacts the various � nancial markets in many different ways. market liberalization is typically seen as a prelude to � nancial integration so that the full ijbf 1 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 38 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 bene� ts of globalization can be obtained once all capital controls are removed. these bene� ts are presumed to include more stable exchange rates and higher growth rates in the economy, not to mention the lowering of risk. at least this has been the conventional thinking up until recently. bekaert, harvey, lunblad blad and siegal (2007) propose that capital market openness holds the greatest potential for growth opportunities in liberalizing countries. however, there is an emerging viewpoint that holds that such market liberalizations might in fact lead to � nancial instability and impair the ability of the government to deal with negative macroeconomic shocks. for a detailed view of this strand of thinking, one can look at papers by stiglitz (2000, 2004) and others. other papers associate such liberalizations with reduced cost of equity capital, increased returns to equity positions and net capital in� ows: see papers by henry (2000), bekeart and harvey (2000) and kim, landi and yoo (2009). in essence there is controversy over the perceived bene� ts and costs of market liberalization. in february 1980, korea replaced its � xed exchange rate system with a multiple-basket pegged exchange rate system, thus permitting the exchange rate to � uctuate against major currencies relevant for its trade. under this system, the basic exchange rate of the won against the usd was determined as the weighted average of two baskets.1 in march 1990 the multiple-basket pegged exchange rate system was itself replaced by the market average exchange rate system (mars). in december 1997 (during the peak of the asian � nancial crisis period) the daily � uctuation limits for the interbank exchange rate were abolished and thus south korea’s exchange rate system shifted to a totally free� oating mechanism at that time.2 in may 1998, the country’s stock and money markets were also opened to foreign investor ownership. the impact of exchange rate � uctuation on the stock markets can be summarized by at least two major facts. first, the volatility of the exchange rate in� uences � rms’ import and export businesses and other businesses sensitive to foreign exchange rates. this in turn affects their stock prices. second, investors in foreign stock markets are subject to foreign exchange risk. therefore, in order to avoid facing these risks, they may hold currency market positions or trade � nancial derivatives so as to hedge the increased currency risks. such derivative trading may have spillover effects on the stock market. they may also demand higher interest rates to compensate for the increased volatility, if traders choose not to hedge, and are invested in the spot debt markets. 1the two baskets are the sdr basket and a trade-weighted basket composed of major trading currencies, with an adjustment factor which was termed the policy variable: ]1[ ������� ���� ptwbbasketsdrrateexchange . 2 the exchange rate bands are and free before december 1, 1997, between december 1, 1997 and november 20, 1997, and after december 16, 1997, respectively. some papers have examined whether foreigners are at a disadvantage in 25.2� ,� 0.10� ,� 2 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 market liberalization and trading in korea: 37-58 39 the korean market (choe, kho and stulz (rfs), 2005) either because of lack of information available to foreigners or because foreigners face higher trading costs. by examining the cost structure of trades made by foreigners versus domestic traders, they concluded that foreigners face higher costs on their medium and large sized trades in the stock market.3 kim and yoo (2009) investigated the different behavior of equity investors in the korean markets. they concluded that foreign investors are not speculators but long-term value investors. kim, land and yoo (2009) also examined the inter-temporal behavior of foreign investors in the korean equity markets. they showed that increases (decreases) in the net in� ows of foreign funds lead to the appreciation (depreciation) of value of the won. however, the reverse causation link does not hold. the motivation of this study was to explore the trading behavior of investment trust companies (itcs), banks, and foreigners, under the free� oating exchange rate system and open capital and � nancial markets to assess evidence of adverse consequences of liberalization, if any. these groups are the dominant entities in the korean markets and any material change in the markets will be re� ected through their trading activities. hence this will be a good place to search for any evidence of instability in the markets. our paper contributes to the current literature on market liberalization by studying its impact on the spot currency, stock and futures markets in a particular country that has just undergone market liberalization. korea is a good test case as there is reasonable legal protection for foreign investors in her markets.4 the study speci� cally tried to � nd out whether the patterns of trade in these markets changed drastically, which might give clues of some instability in the markets. the primary mechanism for the investigation is the linkages between the volatility of the daily high and low exchange rates (won/usd) and the trading behavior of the different categories of the major market participants.5 3 their data set ends at the year both the capital and foreign exchange markets were liberated. ours starts at the beginning of this period and is therefore a post-liberalization analysis of the markets. 4 there is a substantive literature that points the existence of a substantial “home bias” for investors. hence if the opening of a foreign market will lead to substantial capital in� ows, there must not only be the promise of substantial returns but also enough protection in place to preserve their control rights over their assets. 5 individual investors do not have a major role in the currency, derivatives or stock markets in korea. their share of the trading volume and market capitalization is very small and (choe, kho and stulz (rfs 2005), reported that only 4.1% of individuals held stock market positions. since there is no detailed � rm level data, the track positions of � rms over time. hence the duration of individual positions cannot be tracked in the different markets was not examined. first, the study examined whether the major players have different trading 3 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 40 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 behavior due to expected and unexpected volatilities in the exchange rate, namely to � nd out whether these players exhibit asymmetric trading behaviors.6 second, we analyze the relations between the major players’ trading behavior and the volatility of usd futures volume. this information was used to determine whether these players go long or short by trading usd futures simultaneously for hedging or speculation purposes. this article is organized as follows. section 2 provides previous related research � ndings on exchange rates and stock returns. section 3 discusses the characteristics of the research model. empirical results are presented in section 4. the paper ends with a brief conclusion in section 5. 2. literature review most previous papers focused on the impact of exchange rate volatility on stock price and returns. for example, chen and shen (2004), investigated the inter-linkages between taiwan’s stock and exchange rate markets. their results showed that unrestricted trading volumes reveal more information regarding the market than otherwise because of the distortionary effects of government’s foreign exchange market interventions on the volatility prices in the markets. nevertheless they � nd that a common volatility factor drives stock and exchange markets dynamics. ramasamy and yeung (2005) employ granger causality methodology to consider causality between exchange rates and stock market returns in nine east asian economies. they � nd that the direction of causality can vary according to the period of study. when the entire four years of the asian crisis (19972000) were analyzed, they found that, apart from hong kong, all other countries indicate evidence that stock prices granger cause movements in the exchange rate and imply that capital out� ows trigger the exchange rate declines. aquino (2005) examined whether changes in stock market prices in the philippines were triggered by fx risk during the period 1992-2001, speci� cally before and after the onset of the asian � nancial crisis. their evidence suggested that stock returns did not react signi� cantly to foreign exchange rate � uctuations before the crisis. after the onset of the crisis, however, prices of filipino � rms started to exhibit cross-sectional differences in their reaction to exchange rate movements. during the post-crisis period, market participants began to expect a risk premium on their investments for their perceived added exposure to exchange rate risk. as stock returns did not adequately compensate for the fx risk, riskaverse investors increased their demand for hedging the unpriced fx risk. in the 6 there is already evidence of asymmetric response in stock markets to good and bad news. not only do we want to � nd evidence of such behaviors in the korean markets but we want to � nd out whether all classes of participants operate similarly. ours priors, are, that they would tend to act similarly. larger macroeconomic sense, this implies that market inef� ciencies occurr either 4 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 market liberalization and trading in korea: 37-58 41 in the foreign exchange market or stock market or in both. moreover, local � rms did not hedge adequately for foreign exchange risk. similar research studies are abundant: see for example, valckx (2004), bailey, mao and chang (2003), doukas, lang and hall (1999), malliaropulos (1998), bailey and chung (1995), and solnik (1984). a few tried to analyze the relationship between the volatility of a foreign exchange market and players’ behavior, especially in the south korea market. some studies examined the relation between exchange rate volatility and the trading volume of currency derivatives contracts. in this vein, chatrath, ramchander and fong (1996) examined the relationship between the level of trading in currency futures and the variability in the underlying exchange rates. their results indicate a positive relationship between the level of futures trading activity and the volatility in exchange rate changes. they show that future activity has a positive impact on the conditional volatility in the exchange rate changes, with a weaker feedback from exchange rate volatility to future activity. furthermore, the positive impact of shocks in future trading activity on exchange rate volatility is found to persist over several trading days. another group of studies focused on the impact of volatility of the foreign exchange rate on the trading behaviors of � nancial market participants. wang (2002) investigated the effect of net positions by type of trader on return volatility in the six major foreign currency future markets. the principal � ndings were: (i) volatility is positively associated with unexpected changes (in either direction) in the net positions of speculators and small traders; (ii) volatility is negatively associated with unexpected changes (in either direction) in the net positions of hedgers. chiu, chen, and tang (2005) also studied the effects of south korea’s shift to a free-� oating exchange rate system covering the period november 11, 1997 to june 30, 2004. they found that such an event did not impact foreign players’ trading behavior, or that the move in currency market was negligible, in their view. covrig and melvin (2005) offered that yen/dollar exchange rate quotes adjust to full-information levels three times faster when the informed traders are active versus when they are not. these results are consistent with a view of the foreign exchange market where private information is at times quite important. from the above studies, it was noted that traders’ behavior are apparently unaffected by public or expected information. thus, the linkage between the unexpected components and future trading was tested. 3. data and econometric model 3.1. data the korea futures exchange (kofex), launched on april 23, 1999, has become the largest derivative exchange in the world in 2004, by total annual volume of contracts traded. it registered trading of 2,586,818,602 contracts. it however, does not have the value nor the number of futures instruments in the chicago 5 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 42 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 board of trade. two kinds of currency-related products, won/dollar futures and won/dollar options, are listed on the kofex and they have grown dramatically. the trading volume of usd future contracts now ranks third among listed derivative contracts. however, we do not incorporate usd options contracts into our study due to the fact that were not available at the time of this study. this paper focuses on the south korea stock price index 200 (kospi 200), won exchange rate and usd future contracts. the daily data used in this paper covered the period from april 23, 1999 to february 28, 2005. the data were time-synchronized and drawn from the korea stock exchange (kse), the spot currency market and korea futures exchange (kofex) respectively.7 the data thus covered the period of full market liberalization since the government eliminated the foreign investment ceiling completely on may 25, 1998 and the local bond markets and money markets were completely opened up to foreign traders and investors. usd future trading grew due to an increase in institutions’ longer-term hedging demand and the high volatility of the underlying exchange rate during the fourth quarter of 2004. in korea’s � nancial markets, the three major investor groups with high trading activity in the spot and future foreign exchange rate markets are investment trust companies, banks, and foreigners. investment trust companies and foreigners have traded actively in the won/us dollar futures market, while securities � rms and other institutional participants have reduced their trading volume.8 in order to capture the impact of trading behavior of the three major players for the range volatility of won/usd rate. firstly, derived the base line spot trading volume series data and usd futures volume data together with their rates of change. then process the � rst-order differences of the base line data series to generate stationary )1(i series. finally, de� ned the volatility of the trading volume as a logarithm of the ratio of the daily trading volume: 100)/log( 1,,, �� tititi vvv , (1) for )(),(),(),( ufuturesusdfforeignersbbanksiitcsi � , where tiv , represents the trading volume of the series i at day t , and 1, tiv represents the trading volume of the series i at day 1 t , and tiv , is the rate of change of volume in the data series i . 7 the data reported/are from april 23, 1999. 8 annual reports by korea futures exchange (kofex) showed the trading volume by major investor groups to be investment trust companies, banks, and foreigners excluding individuals. range based volatility estimators are highly ef� cient relative to other returns 6 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 market liberalization and trading in korea: 37-58 43 based estimators of volatility such as daily squared returns using opening and closing prices. furthermore, they are robust to noise generated by market frictions: see brandt and diebold (2006), alizadeh, barandt and diebold (2002) and rogers, christopher and satchell (1991). hence, along the lines of chaboud and lebaron (2001), this paper measures the volatility of the foreign exchange rate (won/usd) with a scaled measure of the daily high-low range. this creates the x variable: � 100)log()log( � � ttt lowhighx , (2) where thigh and tlow represent the day t high and low exchange rate of the won against the usd, respectively. 3.2. econometric model this study explored if the � uctuation of the foreign exchange rate affects the spot and usd futures trading behavior among the three major investor groups and the relationship between spot and usd future volume. employed the autoregressive integrated moving average (arima) model to classify the range of the daily high and low of the foreign exchange rate (won/usd) into expected and unexpected foreign exchange rate range volatilities. some scholars have argued that there are asymmetric effects in � nancial markets due to the expected and unexpected impacts. for example, bessembinder and seguin (1993) found that unexpected volume shocks have a larger effect on volatility than expected volume shocks do. the relation is asymmetric and the impact of positive unexpected volume shocks on volatility is larger than the impact of negative shocks.9 it was assumed that players use their expectations of the range volatility of the exchange rate to adjust their spot or usd future positions in advance. the two expected and unexpected variables are then added into a multivariate garch (1, 1) model. these models are described below. a. arima model the investigation tried to determine whether the major players have asymmetric trading behavior in response to expected and unexpected volatility components. it was de� ned that the daily range of the high-low exchange rate is the volatility variable within the model. this variable, derived from the arima model, is then divided into two variables, the expected and unexpected components of 9 several related studies, such as warther (1995), chen et al. (1999), chang et al. (2000), naranjo and nimalendran (2000), bekeart and wu (2000), and wu (2001) also mention that the expected and unexpected innovation have asymmetric effects. 7 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 44 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 the daily high-low ranges.10 it was then observed that the unexpected high-low range component impacts the trading behavior of these major players more than the expected high-low range component does. in order to use the arima methodology, it is � rst necessary to identify whether each series is stationary. the arima model can only be used on a stationary series. if it is determined that a series is non-stationary, then it could be differentiated repeatedly until a stationary series occurs. the original x variable series is transformed into a stationary series by applying the adf and pp tests, as denoted by )0(i . then employ the general arima (p, q) model to derive the expected and unexpected component variables from the high-low range in the x variable. the arima (p, q) model can be written as follows: �� � � ��� q i iti p i itit xx 01 0 ��� , (3) where, x is composed of a logarithm of the daily high-low range of the won/ usd rate, and (4) (5) where, and tuex represent the expected and the unexpected components of the daily high-low range of the won/usd rate at day t , respectively. the optimal arima model is de� ned. the optimal criteria require that the estimated coef� cients be signi� cant, the model generates a minimum aic value, and the residual terms have no series correlation. it was found that the optimal arima model of the high-low won/usd rate series is arima (5,0). the expected range of the high and low exchange rate can be evaluated by the actual high-low range subtracted by the optimal residual terms. the residuals are estimated from the arima model. the residual values are interpreted as the unexpected range of the high and low exchange rate. by creating these two variables, the model may account for whether asymmetric trading behavior does exist. it was demonstrated that the volatility transmission mechanism is asymmetric in effect. negative innovations (when the high-low range expands) in south korea’s exchange rate market increase the volatility in the spot market and this results in increased trading. positive innovations (when the high-low range narrows), do not result in increased volatility. furthermore, trading is not as robust as is the case for negative innovations. 10 there is another example which employs the general arima (p,0,q) model to divide the expected and unexpected variables from the change in original variables, for example, lee and chen (2005). �� � � ��� q i iti p i itit exeex 01 0 )()( ��� ,� ttt exxuex � ,��� tex � 8 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 market liberalization and trading in korea: 37-58 45 braun et al. (1995), koutmos and booth (1995) and geoffrey, martikainen and tse (1997), all employed the egarch model to investigate that there are volatility asymmetric effects between the futures and spot markets. they assessed whether negative innovations generate a larger impact than positive innovations in these two � nancial markets. the � ndings were that most investors will initiate more hedging strategies in the face of negative impacts than in the face of positive impacts. b. multivariate garch (1,1) model the multivariate garch model is not only is able to test for the time-varying variance or volatility in the spot and usd futures markets, but it can also be used to investigate volatility transmission among the three markets: exchange rate, spot, and usd futures. this paper adopted a multivariate garch (1, 1) model to investigate the dynamic relationships in the three markets to examine the phenomenon of feedback in� uence arising from usd futures. for the three major players, the usd futures can be traded to hedge their spot position risk or to speculate under the free-� oating rate. after computing the expected and unexpected components of the variables from the arima model in the high-low range series, the conditional mean equation was changed so that both variables are included. with the multivariate garch (1,1) model, equations are written as follows: (6) (7) (8) (9) (10) ti i itii i itii i itii i itiiiiti ruexexvv , 2 0 ,4, 2 0 ,3, 2 0 ,2, 2 1 ,,1,, ����� ������ ���� � � � � , tb i itib i itib i itib i itbibbtb ruexexvv , 2 0 ,4, 2 0 ,3, 2 0 ,2, 2 1 ,,1,, ����� ������ ���� � � � � , tf i itif i itif i itif i itfifftf ruexexvv , 2 0 ,4, 2 0 ,3, 2 0 ,2, 2 1 ,,1,, ����� ������ ���� � � � � , tu i itfiu i itbiu i itiiu i ituiuutu vvvvv , 2 0 ,,4, 2 0 ,,3, 2 0 ,,2, 2 1 ,,1,, ����� ������ ���� � � � � , ),0(~| 1, ttti hni � � � � � � � � � � � � � tuutfutbutiu tfutfftbftif tbutbftbbtib tiutiftibtii t hhhh hhhh hhhh hhhh h ,,,, ,,,, ,,,, ,,,, . 9 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 46 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 here ti v , is the logarithm of the ratio of trading volume in the i , b , f , or u series at day t . term tr is the logarithm of the ratio of the kospi 200 index at day t . terms ext and uext are respectively the expected and the unexpectenge components of the daily high-low price of won/usd at day t . term tr is the index return of kospi 200. because shocks of the mean equation are the main drivers in the multivariate model, it is important that the mean equation is not misspeci� ed. the var models up to eight lags were estimated and test the individual and joint signi� cance of the coef� cients were tested. a lag length of two was chosen by the akaike information criteria (aic) (see table 3).11 thus, the time lag length of this model is two days per series. to model the time-varying covariance matrix t h , a multivariate garch model.12 this time-dependent, conditional parameterization is justi� ed by our � nding of heteroscedasticity in the volatility of trading volume. for this methodology, engle and mezrich (1996) provided a concise survey. this case the four-dimensional time-varying covariance matrix contains four variance series and six covariance series. in the diagonalized parameterization without spillovers the positive de� nite and the vech models need 30 parameters. given that we have to include the var and the terms for the transmission of volatility shocks, the number of parameters makes the estimation process intractable. on the contrary, the approach proposed by bollerslev (1990) is less complex, because it restricts the correlation to be constant.13 the parameterization for conditional variances is shown in equation (11) and for covariances in equation (12): 1, 2 1,, ��� tiitiiiti hbach for ufbii ,,,� , (11) tjtiijtij hhh ,,, �� for ji � and ufbiji ,,,, � . (12) in equation (11), tih , are the variances of i , b , f , or u series. here, the spillovers of these four volatilities are included as lagged squared innovations. the coef� cient ib accounts for the identical shock to volatility from the previous day, whereas the coef� cients ia ( ufbii ,,,� ) measure the impact of the sectoral volatility shocks. to preserve a non-negative variance we estimate the coef� cients in the variance equations in absolute value were estimated. in equation (12), the co-variances h ij are driven by the variances p ij and correlation coef� cients. 11 this speci� cation is also employed in karolyi (1995). 12 see, for example, univariate (g)arch introduced by engle (1982) and bollerslev (1986). 13 a similar model is employed by koutmos and booth (1995). 10 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 market liberalization and trading in korea: 37-58 47 the log-likelihood function is de� ned as follows: � � � � � � � � � �� t t tttt hh n 1 15.0||ln5.0)2ln( 2 � . (13) for the numerical optimization of this function, the investigation started with the simplex algorithm to reduce dependence on the starting values. then it was switched to the broyden-fletcher-goldfarb-shanno (bfgs) algorithm. 4. empirical results table 1 reports the descriptive statistics of the volatility of each data series.14 the results obtained with the kurtosis, asymmetry statistics, and the jarquebera normality test showed that their distribution is not like that of a normally distributed series. ljung-box tests have been estimated for 35 lags both for these series in level (q (35)) as well as their squares (q2 (35)), and the results revealed autocorrelation problems in all series. the results of the preceding test on the square of the series, together with the signi� cance of the test based on lagrange multipliers, are a clear indication of the existence of heteroscedasticity problems. the � rst stage of the analysis involves determining the stationarity characteristics of these data. for each of the � ve sample variables, the augmented dickey and fuller (adf) test. the results of the adf tests are shown in appendix 1 (tables a1 and a2) which show that these data contain a unit root and are nonstationary. further analysis requires a stationary variable; hence, the focus was on analyzing the � rst difference of the variables. karpoff (1987) indicated that a � nancial market with relative information re� ects that information in its trading volume. trading volume that is accompanied by high activity (volatility), means that the market is transmitting information. this tends to further increase the volatility in the market. in figure 1 (see appendix 2) it was observed that large volatilities tend to be followed by large volatilities. the attractiveness and empirical success of garch models is that they are able to explain to a large extent the volatility clustering behavior and the excess kurtosis of the empirical distribution of returns. this table gives descriptive statistics for the change rate of spot and futures volumes on investment trust companies (itcs, denoted by ti ), banks (denoted by tb ), and foreigners (denoted by tf ), and usd (denoted by tu ), respectively. tr represents the (return) of kospi 200 index. all volatility measures are in daily percentages for the period april 23, 1999 to february 28, 2005. 14 the data are available from the authors upon request. 11 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 48 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 table 1: descriptive statistics obs. 1438 1438 1438 1438 1438 mean -0.0055 -0.0734 0.0572 0.2141 0.0275 std. error 32.3387 54.0578 35.8637 45.3302 2.2185 skewness 0.0428 0.2405 0.0010 0.0760 -0.3324 kurtosis 1.2142 4.0998 0.8284 1.0513 2.2527 j-b 88.7708 *** 1020.9408 *** 41.1142 *** 67.6111 *** 330.5439 *** 279.9861 *** 343.9147 *** 209.9528 *** 240.3849 *** 50.6750 ** 166.6890 *** 407.4168 *** 196.4368 *** 262.4481 *** 221.8526 *** lm(10) 143.9951 *** 224.5361 *** 60.7274 *** 137.6462 *** 58.6613 *** notes: and are ljung-box tests on each series in levels and squared, respectively, for 35 lags which are distributed as 2 35� � in the null hypothesis of no autocorrelation. lm(10) is engle’s lagrange multiplier test (1982) to contrast the existence of arch effects, which is distributed as 210� � in the null hypothesis of no autocorrelation. j-b is the jarque-bera normality test, which is distributed in the null hypothesis of normality as 2 2� . signi� cance levels of 10%, 5%, and 1% are represented by *, **, and ***, respectively. after all the data transformations, the multivariate garch (1,1) model was used with asymmetric information terms to conduct all the empirical tests. table 2: results of conditional variance tests conditional variance equations: h ii, t i b f u c ii 418.5873 *** c bi 53.6812 172.9110 *** c fi 84.5133 ** 57.5502 10.2198 *** c ui -26.0719 -455.0855 ** -576.2288 *** 246.2909 *** � ii 0.0770 *** � bi 0.0268 * 0.1003 *** � fi 0.0567 *** 0.0467 ** 0.0848 *** � ui 0.0130 ** 0.0273 0.0382 ** 0.0611 *** � ii 0.4121 *** � bi 0.8587 *** 0.8234 *** � fi 0.7375 *** 0.8222 *** 0.8134 *** � ui 0.9494 *** -0.6625 *** -0.7836 *** 0.8266 *** function value -28302.4870 note: *, **, and *** represent 10%, 5%, and 1% signi� cant level, respectively. the critical value refers to dickey and fuller (1981). optimal lags are chosen by the aic criteria. the daily data used in this paper cover the period from april 23, 1999 to february 28, 2005. i t b t f t ut rt )35(q � )35(2q )35(q � )35(2q 12 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 market liberalization and trading in korea: 37-58 49 table 2 presents the estimates for the conditional mean equations of the trading volume volatilities. it shows that the spot trading volatility of the three major trading groups in south korea and the trading volatility of usd futures are strongly in� uenced by their own past innovations. the parameters 1,j� are all negative and signi� cant, which suggests that trading volume is larger (smaller) on this day is followed by trading volume that is smaller (larger) in volatility on the next day. 1,j� and 2,j� are all also signi� cantly negative. together, this implies that these three groups of participants will try to revise their trading positions when the spot trading volatility (in the foreign exchange markets) was unforeseen on the preceding day. on the other hand, they will enter the stock market when the spot trading volatility was stable on the preceding day. it was found that the effect of trading volatility on the spot and usd future volume, has a progressively declining impact (on own trading volume) over time. considering the expected and unexpected terms of conditional mean equations (6)~(8) in table 2, 2.,j� and 3,j� are represented by the expected and unexpected range components of the daily high and low won/dollar rates, respectively. an insigni� cant 2,j� and a signi� cant positive 3,j� together demonstrated that only the daily unexpected range of the high-low won/usd rates impacts the trading activities of these three major groups. they will modify their long or short spot positions only according to the daily unexpected exchange rate range. when the daily unexpected range expands (narrows), their trading volume and volatility would increase (decrease) on such a day.15 it is worth noting that the expected range of the exchange rate has no effect on the trading behavior of any member of the trading classes that were analyze. however, it was conclusively shown that they display asymmetric trading behavior in the spot market after the markets have been liberalized. several coef� cients 4,j� are signi� cantly positive, suggesting that these three major players will hold long or short spot positions when the returns on the kospi 200 index increase (decrease). among the major market participants, trading by the banks has the most impact on the variation of the kospi 200 index return. relative to the banks, the itcs have a smaller but still substantial impact on the index return. it may therefore be reasonably concluded that the kospi 200 index return does not apparently impact nor is it impacted by the trading behavior of foreigners. the index return has the most effect on the banks because the coef� cient 1,4,b� has the largest estimation value in all 1,4,j� . the index return also has a more prolonged effect upon the itcs as the 1,4,i� and 15 it is useful to note that negative innovations coincide with an expansion of the range, so that when the range expands traders are reacting to the arrival of adverse news and hence volatility and trading volumes increase. 13 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 50 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 2,4,i� are all positively signi� cant. the evidence that the trading behavior of foreigners is not signi� cantly affected by the daily movement of the kospi 200 index thus seems to lend weight to the fact that foreign investors are long term investors. it would be expected the trading activities of long-term investors not to be sensitive to daily variation in the index. another explanation could be that of the higher trading costs that they face as suggested by choe, kho and stulz (2005). the estimated results of equation (9) uncover whether these three major categories of investors will hold long or short spot positions by trading usd futures for hedging and speculation purposes under the free-� oating system. according to the results of equation (9), the coef� cients 2,j� are all positive and signi� cant. this indicates that only itcs will buy or sell stocks while trading usd futures. it is presumed that the itcs will rearrange their spot positions against the unexpected range volatility of the exchange rate and then hold long or short usd futures as needed at the same time. this is in order to avoid any loss under the exchange rate � uctuation. the banks and foreigners, however, have no obvious trading behavior like the itcs, relatively speaking. this table reports the maximum-likelihood estimation results of equations (11) to (12), which c is the constant in the variance and covariance equations, a is the coef� cient for the lagged squared residuals, and b is the conditional variance and covariance coef� cient. table 2 shows the results of the conditional variance equations. the coef� cients b are all signi� cant, implying that there are volatility-clustering phenomena among the trading volatilities of these three major groups and usd futures. in addition, it is more like the contemporaneous correlation of volatilities across trader types. the lr1, lr2, and lr3 of table 2 test whether unexpected innovations (the case where the unexpected high-low range of won/usd expands) cause a larger current volatility than expected innovations. the results for itcs and banks are positive and the result of foreigners is negative. however, only lr1, the case for investment trusts, is signi� cantly positive and it � ts the assumption of the information asymmetric theory on � nancial securities (such as braun et al. (1995)). if a market participant ignores the asymmetric characteristics of unexpected innovations, in anticipating the volatility of the exchange rate market, then exchange rate exposure will increase and pro� ts may be impaired. it may be concluded from this evidence that the itcs should hold long or short spot positions in usd futures for purposes of hedging their asset positions under the free-� oating exchange rate system. the banks and foreigners, however, would reduce investment in the stock market under an uncertain and volatile exchange rate. so in summary, if market liberalization results in more stable exchange rates this would be bene� cial if the country wanted to attract foreign capital in� ows. 14 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 market liberalization and trading in korea: 37-58 51 table 3: asymmetric information effect tests likelihood ratio hypothesis test estimated value statistics lr1 13.5526 2.8998* lr2 58.4442 0.0571 lr3 -24.1629 1.4625 note: lr is the likelihood ratio test. 2. lr1~lr3 are all )1(2� distributions. signi� cance levels of 10%, 5%, and 1% are represented by *, **, and ***, respectively 5. conclusion this paper investigated the trading behavior of investment trusts, banks and foreigners under the free-� oating exchange rate system in south korea’s � nancial markets. unlike previous studies, this paper incorporates the trading activities of the participants in the spot currency market, stock market and won/ usd future market. an arima model was � rst employed to divide the daily range of high and low won/usd prices into two types, the daily expected and unexpected ranges. these two variables were incorporated into a multivariate garch model to analyze whether the range volatility of the exchange rate would impact the participants’ spot trading activities. if they did, then they trade usd futures for hedging or speculation accordingly. the overall results indicated that, to some extent, only daily unexpected volatility in won/usd impacts the trading behavior of these three major market participants. they would modify their long or short spot positions only when daily unexpected range innovations occur. that is, when the daily unexpected range expands (narrows), their trading volume and volatility would increase (decrease) on that day. this is quite the type of rational behavior that one would have predicted. traders in the markets only react to unexpected information. expected innovations are already incorporated into their trading plans and hence trigger no revisions in their plans. thus it was shown that the three players have asymmetric and rational trading behavior on the spot market under the free-� oating system in south korea. negative innovations lead to higher volume of trading and higher volatility in all the markets. the traders do not respond to expected volatility, only to unexpected volatility. furthermore, only itcs would trade spot and usd futures simultaneously. it is presumed that the itcs re-arrange their spot positions according to the unexpected range volatility of the exchange rate expected as it occurs, and then go long or short usd futures as needed. 2,2,1,2,2,3,1,3,0 ˆˆˆˆ: iiiih ���� ��� � 22,1,2,2,3,1,3,0 ˆˆˆˆ: bbbbh ���� ��� � 2,2,1,2,2,3,1,3,0 ˆˆˆˆ: ffffh ���� ��� � 15 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 52 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 there is some evidence suggesting that the trading behavior of foreigners in the stock market is not signi� cantly impacted by movements of the index nor does their trading affect the market index returns. this evidence is consistent with two channels of reasoning. either foreigners in korea are long-term value investors who by de� nition will not be sensitive to short-term (daily) movements in the index or they are subject to higher trading costs than are domestic participants. it is hoped that the availability of better data will help sort out which of these explanations is appropriate for the market. however, the was no evidence of irrationality found on the part of the participants or that there is any element of instability in the markets that were analyzed. author information: submitting author is lloyd p. blenman, department of finance and business law, university of north carolina – charlotte, belk college of business, charlotte, nc 28223, u.s.a. email: lblenman@uncc.edu. tel: 704-687-7725. fax: 704-687-6967. dar-hsin chen is a professor at the graduate institute of finance, national chiao-tung university, taiwan and he acknowledges � nancial support from taiwan’s national science council (grant no.: nsc 94-2416-h212-008). chun-da chen is an assistant professor at the graduate institute of international business management, da-yeh university, taiwan. references alizadeh, s., m., brandt, w. and diebold, f.x. (2002). range-based estimation of stochastic volatility models. journal of finance, 57(3), 1047-1092. aquino, r. q. (2005). exchange rate risk and philippine stock returns: before and after the asian � nancial crisis. applied financial economics, 15(11), 765-771. bailey, w., and chung, p.y. (1995). exchange rate � uctuations, political risk, and stock returns: some evidence from an emerging market. journal of financial and quantitative analysis, 30(4), 541-561. bailey, w., c., mao, x. and zhong, r. (2003). exchange rate regimes and stock return volatility: some evidence from asia’s silver era. journal of economics and business, 55(5&6), 557-584. bekeart, g., and wu, g. (2000). asymmetric volatility and risk in equity markets. review of financial studies, 12(1), 1-42. bekeart, g., harvey, c., lunblad, c., and s. siegel (2007). global growth opportunities and market integration. journal of finance, 62, 1081-1137. bessembinder, h., and seguin, p. j. (1993). price volatility, trading volume, and market depth: evidence from futures markets. journal of financial and quantitative analysis, 28(1), 21-39. bollerslev, t. (1986). generalized autoregressive conditional heteroscedasticity. journal of econometrics, 31, 307-327. 16 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 market liberalization and trading in korea: 37-58 53 bollerslev, t. (1990). modeling the coherence in short-run nominal exchange rates generalized arch. review of economics and statistics, 72(3), 498-505. brandt, m. w. and f. x. diebold. (2006). a no-arbitrage approach to rangebased estimators of return covariances and correlations. journal of business, 79(1),61-74. braun, p. a., nelson, d. b., and sunier, m.a. (1995). good news, bad news, volatility, and beta. journal of finance, 50(5), 1575-1603. chaboud, a., and lebaron, b. (2001). foreign exchange trading and government intervention. journal of futures markets, 21(9), 851-860. chang, e., chou, r. y., and nelling, e. f. (2000). market volatility and the demand for hedging in stock index futures. journal of futures markets, 20(2), 105-125. chatrath, a., ramchander, s., and song, f. (1996). the role of futures trading activity in exchange rate volatility. journal of futures markets, 16(5), 561-584. chen, c. r., mohan, n. j., and steiner, t. l. (1999). discount rate changes, stock market returns, volatility, and trading volume: evidence from intraday data and implications for market ef� ciency. journal of banking and finance, 23(6), 897-924. chen, s.-w., and shen, c.-h. (2004). price common volatility or volume common volatility? evidence from taiwan’s exchange rate and stock markets. asian economic journal, 18(2), 185-211. chiu, c.-l., chen, c.-d., and tang, w.-w. (2005). political elections and foreign investor trading in south korea’s � nancial market. applied economics letters, 12(11), 673-677. choe, h., kho, b., and stulz, r. m. rfs (2005). do domestic investors have an edge?: the trading experience of foreigners in korea. review of financial studies, 18(3), 795-829. covrig, v., and melvin, m. (2005). tokyo insiders and the informational ef� ciency of the yen/dollar exchange rate. international journal of finance and economics, 10(2), 185-193. desantis, g., and gerard, g. (1998). how big is the premium for currency risk? journal of financial economics, 49(3), 375-412 dickey, d. a., and fuller, w.a. (1981). the likelihood ratio statistics for autoregressive time series with a unit root. econometrica, 49(4), 10571072. doukas, j., lang, l., and hall, p. (1999). the pricing of currency risk in japan? journal of banking and finance, 23(1), 1-20. engle, r., and mezrich, j. (1996). garch for groups. risk, 9(8), 36-40. engle, r. f. (1982). autoregressive conditional heteroscedasticity with estimates of the variance of united kingdom in� ation. econometrica, 50, 9871008. 17 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 54 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 geoffrey, b., martikainen, t., and tse, y. (1997). price and volatility spillovers in scandinavian stock markets. journal of banking and finance, 21, 811-823. henry, p. b. (2000). stock market liberalization, economic reform and emerging market equity prices. journal of finance, 55, 529-564. karolyi, g. a. (1995). a multivariate garch model of international transmissions of stock returns and volatility: the case of the united states and canada. journal of business and economic statistics, 13(1), 11-25. karpoff, j. m. (1987). the relation between price changes and trading volume: a survey. journal of financial and quantitative analysis, 22(1), 109126. kim, j., landi, j., and s. s. yoo. (2009). market liberalization and foreign portfolio selection in korea. journal of multinational financial management, 19, 206-220. kim, j., and yoo, s.s. (2009). inter-temporal examination of the trading activities of foreign investors in korea. paci� c basin journal, 17, 254-256. kim, m., kim, g. r., and kim, m. (2004). stock market volatility and trading activities in the kospi 200 derivatives markets. applied economics letters, 1(1), 49-53. koutmos, g., and booth, g. g. (1995). asymmetric volatility transmission in international stock markets. journal of international money and finance, 14, 747-762. lee, m., and chen, c.-d. (2005). the intraday behaviors and relationships with its underlying assets: evidence on option market in taiwan. international review of financial analysis, 14(5), 587-603. malliaropulos, d. (1998). international stock return differentials and real exchange rate changes. journal of international money and finance, 17(3), 493-511. naranjo, a., and nimalendran, m. (2000). government intervention and adverse selection costs in foreign exchange markets. review of financial studies, 13(2), 452-477. pan, m.-s., r. t. hocking, and h. k. rim. (1996). the intertemporal relationship between implied and observed exchange rates. journal of business finance and accounting, 23(9&10), 1307-1317. ramasamy, b., and yeung, m.c.h. (2005). the causality between stock returns and exchange rates: revisited. australian economic papers, 44(2), 162169. rogers, l., christopher, g., and satchell, s. e. (1991). estimating variance from high, low, and closing prices, annals of applied probability, 1(2), 504512. solnik, b. (1984). stock prices and monetary variables: the international evidence. financial analysts journal, 40(2), 69-73. stiglitz , j. e. (2004). capital market liberalization, globalization and the imf. oxford economic policy, 20(1) 57-74. 18 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 market liberalization and trading in korea: 37-58 55 stiglitz. (2000). capital market liberalization, economic growth and instability. world development, 25, 1975-1086. valckx, n. (2004). the decomposition of us and euro area stock and bond returns and their sensitivity economic state variables. european journal of finance, 10(2), 149-173. wang, c. (2002). the effect of net positions by type of trader on volatility in foreign currency futures markets. journal of futures markets, 22(5), 427-450. warther, v. a. (1995). aggregate mutual fund � ows and security returns. journal of financial economics, 39(2&3), 209-235. wu, g. (2001). the determinants of asymmetric volatility. review of financial studies, 14(3), 837-859. 19 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 56 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 appendix 1: table a1. augmented dickey fuller (adf) tests for spot trading volume of itcs, banks, and foreigners and futures volume of usd variables without drift and trend with drift and trend constant only lag(s) statistics lag(s) statistics lag(s) statistics log levels spot volume of itcs 18 -0.9421 18 -4.3771 *** 18 -4.0601 *** spot volume of banks 15 -4.4038 *** 15 -7.3298 *** 15 -7.0726 spot volume of foreigners 19 -0.4401 20 -5.1326 *** 20 -2.9916 ** usd futures volume 19 -0.4184 *** 18 -4.3506 *** 18 -3.3156 ** first differences spot volume of itcs 17 -14.2327 *** 17 -14.2239 *** 17 -14.2275 *** spot volume of banks 20 -12.1301 *** 20 -12.1193 *** 20 -12.1253 *** spot volume of foreigners 18 -13.5656 *** 18 -13.5681 *** 18 -13.5718 *** usd futures volume 18 -14.8211 *** 18 -14.9810 *** 18 -14.8879 *** note: *, **, and *** represent 10%, 5%, and 1% signi� cant level, respectively. the critical value refers to dickey and fuller (1981). optimal lags are chosen by the aic criteria. the daily data used in this paper cover the period from april 23, 1999 to february 28, 2005. table a2, akaike and schwarz information criteria appropriate model selection criteria are the akaike information criterion (aic) and the schwarz information criterion (sic). we choose the value of p that minimizes the aic and sic. the aic selects 2�p , whereas sic selects 3�p . it is well known that the sic penalizes additional parameters more heavily than the aic, as the sic prefers a more parsimonious model. based on the selection criteria and the results of the statistical tests, we choose the var(2) speci� cation. 20 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 market liberalization and trading in korea: 37-58 57 p aic sic 1 27.71567 27.92105 2 26.09832 * 27.68363 3 26.10565 27.67109 * 4 26.22056 27.76634 5 26.94054 27.86686 6 26.88849 27.99555 7 26.87544 28.16345 8 26.84778 28.31694 note: p denotes the lag in var( p ). * denotes the minimum value of the information criteria. appendix 2: figure 1. the trend of daily usd futures volume and kospi 200 index � 0 20 40 60 80 100 120 140 160 0 5000 10000 15000 20000 25000 30000 35000 40000 4/23/1999 2/9/2000 12/4/2000 9/28/2001 7/29/2002 5/20/2003 3/16/2004 12/30/2004 indexvolume usd futures volume kospi 200 index 21 blenman et al.: market liberalization and trading in korea produced by the berkeley electronic press, 2010 58 the international journal of banking and finance, vol. 7. number 2: 2010: 37-58 t ab le 4 . r es ul ts e st im at io n of t he m ul ti va ri at e g a r c h ( 1, 1) m od el c on di ti on al m ea n e qu at io ns : v i, t � j � j, 1, 1 � j, 1, 2 � j, 2, 1 � j, 2, 2 � j, 2, 3 � j, 3, 1 � j, 3, 2 � j, 3, 3 � j, 4, 1 � j, 4, 2 � j, 4, 3 p an el a : f or e q u at io n ( 6) v i 4. 08 55 ** -0 .4 84 4* ** -0 .2 26 3* ** -1 3. 47 31 25 .9 34 3 -2 0. 23 88 22 .9 86 9* ** 3. 02 68 -1 3. 07 53 * 1. 95 80 ** * 1. 13 78 ** * -0 .3 77 7 p an el b : f or e q u at io n ( 7) v b 5. 65 24 -0 .5 29 9* ** -0 .2 49 6* ** -3 1. 16 28 -4 .7 24 6 26 .2 13 6 15 .9 95 9* ** 6. 56 09 -2 .8 71 2 2. 81 57 ** * 0. 67 96 -0 .7 69 0 p an el c : f or e q u at io n ( 8) v f 1. 76 20 -0 .3 60 3* ** -0 .2 94 2* ** 15 .6 86 8 21 .5 27 0 -3 9. 93 13 16 .1 42 1* ** -3 .0 91 2 -1 1. 36 98 0. 62 83 0. 65 20 -0 .2 06 2 p an el d : f or e q u at io n ( 9) v u -0 .0 39 3 -0 .4 56 2* ** -0 .2 28 5* ** 0. 87 15 ** * 0. 46 38 ** * 0. 19 29 ** * -0 .0 72 5 -0 .0 43 3 -0 .0 05 7 0. 09 78 0. 00 67 0. 05 85 n o te : s ig ni � ca nc e le ve ls o f 10 % , 5 % , a nd 1 % a re r ep re se nt ed b y *, * *, a nd * ** , r es pe ct iv el y. t hi s ta bl e re po rt s th e m ax im um -l ik el ih oo d es ti m at io n re su lt s of e qu at io ns ( 6) t o (9 ). t he s am pl e da ta s pa n th e pe ri od a pr il 2 3, 19 99 t o f eb ru ar y 28 , 20 05 . t h ey c on ta in 1 ,4 39 o b se rv at io n s. d ai ly c on ti nu ou s ch an ge r at es a re c on st ru ct ed u si ng t he f or m ul a 10 0 ) / lo g( 1 , , , � � t i t i t i v v v , w he re tv i s th e vo lu m e of e ac h in ve st or g ro up a t ti m e t, j � i s th e co ns ta nt i n th e m ea n eq ua ti on , an d 1,j � is t he o w n la gg ed v ar ia bl es f or e ac h se ri es . in e qu at io ns ( 6) t o (8 ), 2,j � a nd 3,j � a re t he c oe f� c ie nt s fo r th e ex pe ct ed a nd u ne xp ec te d w on /u s d r an ge s, r es pe ct iv el y, a nd 4,j � i s th e re tu rn o f k o s p i 20 0 co ef � ci en t. i n eq ua ti on ( 9) , 2,j � , 3,j � , a nd 4,j � a re t he s po t tr ad in g be ha vi or s fo r th e th re e in ve st or g ro up s, r es pe ct iv el y. 22 international journal of banking and finance, vol. 7, iss. 2 [2010], art. 3 http://epublications.bond.edu.au/ijbf/vol7/iss2/3 recommended citation international journal of banking and finance 8-1-2010 market liberalization and trading in korea lloyd p. blenman dar-hsin chen chun-da chen corporate spin-offs, their price reactions and determinants in malaysia 83the international journal of banking and finance, vol. 5.(number 1): 2007: 83-112 ijbf corporate spin-offs, their price reactions and determinants in malaysia chung-sin yoon and mohamed ariff alan yoon associates and bond university, australia abstract spin-off as a form of financial restructuring has been examined in the us and the uk but not in other markets. this is a first study outside those markets. the evidence from a sample of 85 spin-off cases in malaysia reveals that both the parent and the spin-off company stocks gain significant positive abnormal returns: parent firms earn smaller value while the spin-off firm gains substantially, much greater than is documented in other markets. examining the factors correlated with the size of the spin-off effect, we find the abnormal returns are positively correlated with market capitalization and negatively correlated with age. the larger is the company or the newer is the company the greater is the magnitude of the positive abnormal returns. jel classification: g14 key words: spin-offs, restructuring, emerging markets, tax motivation, size, and age. 1. introduction to spin-off this paper reports new findings from a study on corporate spin-off phenomenon, an increasingly popular financial restructuring method, and their determinants in an emerging economy. while few studies exist on the spin-off price effect in the us and the uk markets, this is the first study of its kind outside those countries, which also tests if certain firm characteristics are correlated with price behavior at the announcement period. hence, this paper is motivated to fill a gap in the literature to contribute to studies of financial restructuring. restructuring is becoming popular also for the other reason that firms in recent years are moving away from the conglomeration model to focused business model * editorial note: this paper was one of three best papers selected by the then editor of ijbf at the fma-afa 2005 conference. the co-author of this paper commenced duties as co-editor of ijbf in 2007, and he did not have any part in the selection/review of this paper the international journal of banking and finance, vol. 5.(number 1): 2007: 83-11384 to create value in a more competitive business environment of the last two decades. understanding the factors driving this phenomenon in a developing economy context will be useful for corporate strategy and financing matters. the growth of the emerging markets in such economies has received much attention in the past few years: see barry, peavy and rodriguez (1997). investors have been attracted to the potential for high returns along with diversification benefits of such markets.1 managers and trustees of australian, us and uk pension funds have begun for the first time to commit portions of their pension assets to emerging market debt and equity securities. the unique characteristics of emerging markets are helping academics to better understand the development and application of appropriate financial management techniques in companies operating in such environment. this is another aspect of this paper to provide an understanding of the forces affecting spin-offs. the earliest recorded spin-off case in this paper was in 1970 relating to a popular conglomerate namely the magnum corporation. since then, there have been 125 spin-offs. spin-off is occurring at increasing frequency in several asian countries as public-listed firms have come of age over the last four decades, and are creating more focused businesses by spinning-off from their existing firms. the capital market in malaysia is classified by the world bank as an emerging market for the simple reason that per capita income in this country is less that the threshold2 for warranting a classification as developed. however, market specialists such as dow jones and others have included this capital market as among the 20 plus markets in the world for international diversification of investments. there are about 900 plus firms on the bursa malaysia, the single malaysian exchange for the capital market.3 of these, most firms are listed on the main board while 300 plus firms are listed on the second board. in total, these companies have market capitalization of over us$ 160 billion. there is also a bond market with a capitalization equal to about 95 percent of the gdp, which would place this bond market as the fourth largest in asia. the economy is an industrialized economy with industrial output accounting for 35 percent of the output. for many years, malaysia has become the most open economy among the developing countries to foreign investment, international trade and also the ex1 based on the ifc’s emerging markets data base for a recent period, a commonly held view is emerging stock markets are characterized by high returns and high volatility. malaysian market is such a market and is ranked among the top-20 or thereabout, and is among the 65-odd emerging markets. it is ranked in the financial press as being among the top five in terms of its desirability for cross-border investments. 2 the world bank defines a developing country as one having a per capita gnp that would place it in the lower or middle-income category; a developing country had an annual per capita gnp less than about us$10,000. capital markets in such a country would be named an emerging market. the capital market is ranked by oecd in 1999 report as among the top-20 investible market. 3 two stock exchanges, the then kuala lumpur stock exchange (klse) and malaysian exchange of securities dealing and automated quotation bhd (mesdaq) had merged to form bursa malaysia. klse was incorporated in 1976 and mesdaq was incorporated in 1997. 85the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 tent of multinational presence. 4 it was quick to recognize the powerful role that foreign investors could play in fuelling export-led growth, and it is also wellplaced to attract such investment during the years of regional structural adjustment taking place in the late 1980s. partly as a result of fdi inflows, this economy was among the world’s fastest growing economies prior the 1997 financial crisis. in the wake of the financial crisis which swept through the asian region,5 it is useful to note the new focus of the various asian countries in their economic development approach. in all the four countries malaysia, indonesia, philippines and thailand development strategies now includes a selective approach to investment promotion with a clearly circumscribed role for foreign direct investors. such partial openness allows foreign firms to contribute to rapid economic growth driven by exports, but these days countries have tended to lessen the welcome to short term capital flows, which has been the underlying cause of the economic flap in the late 1990s. the legal concept of tax-free spin-off can be found in the inland revenue service (irs) code section 355 of the securities and exchange commission tax code and the treasury regulations section 1, 355-2 in the u.s.6 there was an amendment to the legal framework after 1969 to restrict tax avoidance. in fact, after the tax reform act of 1986, a spin-off or other divisive reorganization is the only way a company can distribute appreciated property to shareholders without incurring a corporate-level tax. an added appeal is that corporations have considerable latitude in reporting spin-offs in their financial statements. the regulation on spin-off became more pronounced in europe recently as spin-offs became more popular during the 1990s. in malaysia, legal representation initially rested on the capital issues committee and the foreign issues committee. since 1996, spin-off is regulated under the arrangements and reconstruction section of the companies act 1965. following the procedure of companies going for listing, a spin-off application is reviewed by the securities commission, companies commission and the bursa malaysia as the regulators.7 spin-offs are also regulated under section 176 of the 4 new foreign and domestic investment played a significant role in the transformation of this economy. manufacturing share increased from 13.9% (1970) to 34% (2005) as agriculture declined from 42.7% to under 7% over the same period: see 2005 bank negara malaysia annual report. 5 after nearly two decades of strong economic growth averaging 8.7 percent annually, the financial crisis had long-lasting impact by halving the growth prospect, when the regional financial crisis of 1997-8 swept the region. the gdp suffered a sharp 7.5% contraction in 1998: it rebounded in 1999 to grow by 5.6%. 6 irs section 355 includes four key requirements for tax-free treatment: controlled corporation; securities distribution; active businesses; and not a distribution device. treasury regulations section 1, 355-2 adds two additional requirements: business purpose; and continuity of interest. see appendix a for details. 7 prior to 1995, capital issues committee (cic) and the foreign issues committee were reviewing. cic was set up in 1968 by the minister of finance to ensure an orderly development of the capital market by regulating the issue of securities by public companies and the listing of such securities on bursa malaysia including spin-offs. see the requirements for spin-offs in appendix a. the international journal of banking and finance, vol. 5.(number 1): 2007: 83-11386 companies act. in any scheme of arrangement or reconstruction which results in a transfer of assets or undertakings of a foreign incorporated company to malaysia, the securities commission requires that the entries in the books of accounts of “mirror company” comply with the provision of section 60 of the companies act. the listing of spin-off subsidiary and/or associated companies of a listed holding company could be considered if the pretax profits and net tangible assets of the subsidiary and/or an associated company to be listed should not account for more than 35 percent respectively of the consolidated pretax profits and net tangible assets of the group in the previous five years (main board listing) or three years (second board listing). the threshold level of after-tax profits and/or net tangible asset contribution for the listing of subsidiary and/or associated companies has been raised from 35.5 percent to 50 percent from 1997. another consideration is the parent company excluding the subsidiary and/or associated company to be listed and existing listed subsidiary and/or associated companies should on its own meet securities commission’s requirements for listing as if it were a new company seeking listing. the applicant subsidiary and or associated company should also be involved in a particular business of its own with profits able to meet securities commission’s criteria for listing with respect to profit track record and future prospects. the relationship between the subsidiary and/or associated company seeking listing and the other companies within the group, including the holding company, should not give rise to intra-group competition or conflict of interest situation. the subsidiary and/or associated company to be listed should demonstrate that it is not overly dependent on the other companies with the group including the holding company, in term of its operations, including purchases and sales of goods, management policies and finance. there has been no capital gains tax on share disposal in this economy. however, an acquisition of shares in a real property company shall be deemed to be an acquisition of chargeable assets, and where such shares are disposed of shall attract real property gains tax (rpgt) – schedule 2, section 34a rpgt act (1976). it is difficult to rely on section 17.1 (c) to claim exemption for this rpgt. after 1997, it is not possible to utilize the “bonus shares”8 approaches to exempt rpgt on share disposal of real property companies. 2. literature on spin-off cases it is pertinent to review a set of literature in this section in order to undertake this study. earlier theories on capital market research are relevant to this study as are also the specific theories, methods applied and the variables used in studies of 8 it was possible to use distribution of bonus shares as consideration to save rpgt on selling of land together with the real property companies before 1977. 87the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 spin-off case in this literature: table 1 is a summary of existing literature on spinoff. hence we review both sets of ideas in this section. in all there are 11 published studies on spin-off. most of them, in fact ten, are in the us markets. there is one study in the european market. these authors applied the very common method of estimating the abnormal price behavior measured as abnormal returns around the time of announcements of the spin-off. they all reported significant abnormal returns to this event both in the us and europe. those studies also examined a number of variables listed in the last column in the table to verify if these factors are correlated with the observed price behavior. broadly, in these matured markets, these variables have been found to be significantly correlate with the price change behavior. these studies are limited to spinoff in different legal environments within matured market places. it is unknown if the factors associated with the event are also likely to be correlated with price change behavior in emerging market chosen for this study. the relevant theories to be reviewed from broader literature are itemized here. 1) valuation as in hearth and zaima (1986) and leshchinski (2002). 2) agency theory as in jensen and meckling (1976). 3) wealth transfer as in galai and masulis (1976) and maxwell and rao (2003). 4) clientele as in vijh (1994) and schnabel (1992). 5) divergence of opinion hypothesis as in miller (1977). 6) information asymmetry as in habib, johnson and naik (1997), table 1: a summary of relevant literature and variables on spin-off the international journal of banking and finance, vol. 5.(number 1): 2007: 83-11388 9 despite the widespread use of the named variables in this section, there is still no agreement among the researchers on the factors influencing the spin-off effect. we place on record this list as the possible starting point for this study. krishnaswami and subramaniam (1999) and veld and merkoulova (2003). empirical tests on the spin-offs events have confirmed information content of the news of spin-off had effects on share prices in terms of cumulative abnormal returns (car). the seven theoretical ideas prevailing about the value of the firm for the study of the management decision effect from spin-off are relevant.9 the market model is the most common research model used in the previous studies as shown in table 1 while variables such as tax, size and focus are the most common variables analyzed in the context of these theoretical ideas. 3. data, variables and methodology the data for this study are from the scans data base of security prices for the kuala lumpur market. the period covered in this study is over a 23-year period, where we found 125 spin-offs. of these we selected 85 spin-off cases using the criteria: both firms must have price history of at least five years; the thin-trading problem is minimal in that the daily price observations are complete; that the firms are local companies. the stock index used is the value-weighted composite index for this market, and the stock returns are computed after adjustments for capitalization and dividends. we selected 85 non-spin-off firms matched by industry and firm size to form a portfolio of firms not affected by spin-off to test if the spin-off sample is different from the non-spin-off as a robustness method to the usual risk-adjusted methods applied in this type of studies. no other studies used the matched sample procedure. the event impact is estimated over a 100 and +50 days. this length of the test window is longer on account of the longer period over which the spin-off-related regulatory decisions are made in this market. sharpe (1963) developed the standard form of the general equilibrium relationship of asset returns now known as the market model (see equation 1). it is a statistical model, which relates the return of any given security to the return of the market portfolio. considering the information content of spin-off studies with daily data, the event of interest will be the spin-off announcement and the analysis period consists of a defined period of price information around the announcement day. this would permit an examination of the price change, if any, to be statistically tested, for the selected periods surrounding the events. the analysis period in this event study is defined as –100 to +50 days around the announcement day. notation and time line are defined to facilitate the measurement and analysis 89the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 of abnormal returns. consistent with event study methods (see brown and warner, 1985 for example). defining t = 0 as the event date, then t = -100 days to t = +50 days represent the event window for analysis. the rate of returns over each day is computed for each of the parent and the spin-off firms. from each of the daily returns, a factor is subtracted as systematic return (to be explained later as riskadjusted; market adjusted; matched sample adjusted) to adjust the gross returns of the days for market-wide impacts as well as risk-adjustment effects. for example, we apply the market model of sharpe to derive the risk-adjusted returns in two steps. in the first step, the market model parameters α j and β j for each parent firm are estimated using monthly stock returns (r jt ) and monthly market returns (r mt ) from market index using the model: r jt = α j and β j r mt + e jt (1) the abnormal returns to each of the event for the parent company are estimated as in equation 2: ar jt = r jt α j β j r mt (2) where, ar is the abnormal return over -100 to +50 days with day 0 being the event date when announcement of spin-off is made. the abnormal returns are then cumulated over a given window of time to measure the total impact of the event over a period prior and after-the-event date. this is the standard methodology followed in all event studies; hence the individual formulas are not reproduced here. one thing is need to be explained in the case of matched sample. we estimate the returns for a matched sample of companies as the gross returns over the test window of -100 and +50 days. most spun-off companies took 2-3 months to list their companies after announcing to the public: there was a case of 4 years. to adjust the market-wide and risk-difference-effect, we subtract these returns from the spin-off case returns so that the abnormal returns of the spin-off could be tested. we cumulate the abnormal returns over different time periods as the cumulative abnormal returns or cars. again this method is well-entrenched in event study process. the cars are then used in the regression to test if the firm-specific variables suggested in the literature in fact are correlated with the price change measured as cars. the study adopts three different models in the estimation of the abnormal return as follows: 1)risk-adjusted market model, 2)market-adjusted market model, and 3)control sample adjusted market model. in the controlled sample model, we match for each spin-off case, another the international journal of banking and finance, vol. 5.(number 1): 2007: 83-11390 firm that is not spin-off case in the same industry with approximately the same asset size. since the accuracy of our estimates are subject to a number of unknown variables, a matched sample is perhaps a better method to see the underlying difference, in case the traditional methods shown as (1) and (2) are prone to error in the emerging market. the return observations from t = -63 months to t = -3 months are used to estimate the market model parameters α and β for the parent firm: parent firm’s beta is needed to compute the risk-adjusted returns along with the alpha on a daily basis.10 the alpha, α, measures the unexplainable performance relative to the market model. the, β, beta is the slope of the characteristic line estimated using the monthly data. as α is estimated from monthly returns data α m needs to be recomputed as daily alpha α d as in: α d = (1 + α m )1/22 – 1 (3) the α d is the alpha specified in equation 2 in an earlier paragraph. the main hypotheses are developed here are now explained. there are two main hypotheses tested using three different methods of computing the abnormal and cars. the hypotheses 1is as follows: null; the market value of the postspin-off companies in the period following the announcement of spin-offs is the same as the pre-spin-off entities. this result would suggest that there is no wealth effect from the announcement of spin-offs after the announcement is made but prior to the court decision about asset transfers. if the null hypothesis is rejected if the observed abnormal return is greater for the post-spin-off period than in the pre-spin-off entity during the analysis period, this means that there is wealth effect created by spin-offs. this will be measured using the three methods of computations. hypotheses 2: null; the selected firm-related variables and the cars from the event are not significantly correlated: the variables selected are tax, focus, age and market capitalization. if the result supports the null that would suggest that the selected variables are not at all likely to determine the size of the abnormal price effect. rejection of the null hypothesis would reveal, as has been the case in the u.s. and european studies, the firm characteristics are likely to determine the size of the price effect measured as the cars. there are four subsidiary hypotheses to be tested as well. these relate to the individual variable’s correlation with the cars. the magnitude of the capital market reaction to the spin-off announcement depends on the size of the spin-off company in relation to the size of the parent company before spin-off. the longer is the age of the parent company at spin-off time, the larger is the wealth effect. 10 despite the widespread use of the named variables in this section, there is still no agreement among the researchers on the factors influencing the spin-off effect. we place on record this list as the possible starting point for this study. 91the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 tax factor is a significant determinant of the corporate spin-off price effect. finally, improved focus is significantly correlated with the corporate spin-off price effect. a priori, these are the expected results, based on the priors from the 11 studies referred to in an earlier section. given the institutional differences in the tested market, it may be that some of these results may be otherwise than hypothesized. for the regression model, the dependent variables are the various cars over selected intervals. the car is then regressed against the independent variables: car j = α 0 + α 1 focus j + α 2 tax j + α 3 age j + α 4 size j +∈ j ……. (4) the independent variables are suggested by the literature. focus is the variable that suggests that firms separating a spin-off to create a more focused business. the tax staus is indicated by the amount of tax benefit from the spinoff case. age is the length of time the parent has been in business to the date of the spin-off event while the size is the market capitalization of the firm. 4. findings and discussion a: descriptive statistics in table 2, the descriptive statistics are presented on the abnormal returns of spin-off parent companies. there are three sets of numbers, for the 85 observations, making a total of 255 data points on the spin-off companies. there is a corresponding set of 85 non-spin-off firms matched with the spin-off sample. the statistics for all the three data sets indicate that 75 percent of the abnormal returns were positive while the matched sample of non-spin-offs had just 39 percent positive price changes. this indicates that there is a high probability of stock price increases with spin-off decision compared to the portfolio of firms that were not undergoing spin-off event. the mean of the spin-off abnormal returns is 21 percent compared to 4 percent for the non-spin-off sample. the standard deviation is quite close across the firms, which seems to suggest that these companies have similar risk patterns probably because spin-off parents are well established firms with long experience, with larger capitalization, which would lead to lower price volatility. the range between the minimum and the maximum for the spin-off samples is more than five times than that for the non-spin-off sample. this disparity is explainable as being caused by the spin-off effect on the measured variable around the time of announcement. there is no spin-off impact on the non-spin-off companies: implicitly, spin-off firms must have greater variability. the high jarque-bera test value in both samples indicates normality of the distribution of both the samples selected. the international journal of banking and finance, vol. 5.(number 1): 2007: 83-11392 these statistics indicate significant positive abnormal returns of 8 percent higher than that of the parent companies. if median is a more reliable statistics given the distribution is not normal, the difference is 17 percent. by presenting these average abnormal returns values, we can see that there is a better performance in terms of wealth generation for the spun-off companies as compared to the parent companies. the skewness value shows more of the spin-off companies have abnormal returns more than 27 percent. this validates there are actual returns from the spin-off phenomenon for the spin-off companies. the range between the minimum and the maximum abnormal returns is more than seven times, indicative of again a non-normal distribution. from the statistics, it can be deduced that, in bursa malaysia which is an emerging market, there has been wealth gains from spin-offs. in addition, wealth garnered from the spin-off companies are more certain and greater in absolute terms than that of the parent companies in general. b: spin-off announcement and ex-date effects in table 3 the reader will find the test statistics on the cars for selected windows. the first window of -50 to -2 is the pre-event period; the on-event period is -1 to 0; and the post-event period is for the post-event period. the cars in the post-event period are expected to be not significant in an efficient capital market as all the information effect is absorbed either on the date of announcement (t=0) or in the days prior to the announcements. the abnormal returns, which are usually given for each day, are reported as means over the test window: a table going back -50 to +50 days will be too lengthy. as is evident, the cars for the parent firms amount to 22.7 percent (0.227) on the on-event period. the size of the price effect is very large compared to means of under 7 percent reported in other studies in developed markets. the cars for the on-event and pre-event windows are significant as well. the absence table 2: descriptive statistics of parent and spin-off companies 93the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 of significant post-event car is consistent with the tested market’s known famaefficiency, so it is as expected. overall, these results appear to suggest that the shareholders of parent firms spinning new firms had significant increases in their firm values. this result is consistent with the literature and the theories about focused businesses as well as removal of hurdle rates constraints on subsidiaries increasing value in a focused business environment. besides, there is less agency problem of the parent firm with a substantial portion of the parent being taken off from the current management: the agency theory predicts a positive price effect. the two figures 1 and 2 are plots of the cars from the risk-adjusted and market adjusted returns. both show significant spikes in the returns in the period prior to the announcements. in figure 1, the post-event trend, though it appears as large, is not statistically significant. in figure 2, the trend is more marked. however, it is to be noted that the market adjustment method is likely to be biased table 3: risk-adjusted abnormal returns surrounding spin-off announcements significant at 0.1 (*), 0.05 (**) and 0.01 (***) levels. figure 1: plot of risk-adjusted cars over –100 through +50 for the parent companies the international journal of banking and finance, vol. 5.(number 1): 2007: 83-11394 since the spin-off parent firms have betas lot different from 1, hence, the lack of risk-adjustment would make this measure introduce some bias in the plotted graph. the plots appear to suggest that the on-event and pre-event trends are more pronounced, and are likely to be significant. the plot in figure 3 is by taking the difference between the gross returns fo the spin-off parent firm returns from and the control sample of matched nonspin-off firms. in a sense, the use of this method distinguishes this study from others. as mentioned in the methodology section, we deem this method to be more reliable for studying the event effect of spin-off in emerging markets where there are many factors that could not be controlled, and empirical evidence shows that the price behavior have peculiar impacts from regulatory regime in place as figure 2: plot of market adjusted cars –100 to +50 days for the parent companies figure 3: plot of cars using control sample-adjustment for the parent companies 95the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 well as the transient effects from portfolio flows into this market. the shape of the plot suggests that the on-event impact between day -10 and day +1 is very pronounced. it is more pronounced than the impact in the pre-event period from -95 days to -11 days. the post-event impact is much more muted, very unlikely to be significant. summarizing the behavior of the event impact on share prices of parent firms, it is evident that there is significant shareholder value increase in the parent firm in the period up to the event date. now, the results on the spin-off firms are presented in the ensuing discussion. this plot refers to the increase in value of the separated firm from the date of listing and trading of the spin-off firms. a priori, it can be argued that the shareholders of the spin-off companies are the same as the shareholders of the parent company at the time of listing of the spin-off. spin-off shares are given in proportion to the existing share owners of the parent firm: unlike in the developed countries, spin-off in malaysia is a closed shop affair as no members of the public are given any shares in the new firm. hence, any observed increase in the share prices of spin-off firm after the listing day arises from the market’s re-valuation of the spin-off firm by the subsequent purchasers of the shares after listing, as is consistent with their valuation they place on this new firm coming to the public for the first time. thus, as in hypothesis 1, any increase in the value of the spinoff is a transfer of wealth from the event to the shareholders. the trend clearly evident in this plot suggests a 50 percent increase in value to the original shareholders’ value in the spin-off firm. the market reacts to the spin-off announcement for both entities. however, the reaction is at different days during the analysis period. this phenomenon is figure 4: plot of market adjusted car for the spin-off companies the international journal of banking and finance, vol. 5.(number 1): 2007: 83-11396 consistent with observed pattern in a study of stock price reaction to share repurchases: lim and obiyathulla (2002). however, in a study of the impact of accounting earnings disclosure on stock prices in malaysia these findings can be explained by the familiarity of the market participants towards spin-offs. in any case, this can be interpreted as the revaluation by investors buying into the firm that was closed to them before as being valued higher in the hands of the public. this is not evidence of market inefficiency as some would suggest. we estimated this new finding as being relevant to the study of spin-offs as previous studies only examined the announcement day impact. part of the reason for this price increase is also the leakage of information that comes from post-listing news of the worth of the spin-off companies from the analysts as well as the news of the court hearings that would be completing just about the time of the listing. following the reason often advanced for a spin-off that the parent company wants to reduce the diversity of its business operations, the parent spins off the subsidiary because its core business is very different from that of the rest of the parent’s operations. focus may be a determinant of spin-off. in most efficient economies, resources flow to where they are best utilized for the benefit of society. therefore, since material resources are generally scarce and inequalities exist in entrepreneurial skills, some members of society will own means of production while others will not, even though they may lose their focus to do so. owners of corporate organizations and their agents should understand that they are using these resources to produce goods and services to improve the quality of life of society. corporate organizations are to use these resources efficiently for the benefit of the society. to the consumer, efficiency means the product is of good quality, affordable and has the ability to provide the services it was designed for. to move into a responsible position to carry out such business philosophy, a company must first be established. hence, the age of a company is offered here in this study as a predictor to such a motive. generally, the older a company, the more establish she becomes. the companies have been announcing spin-off for the purposes of consolidating efforts on core operations. the genting group’s spin-off of resort world focuses on the casino operations. the tan chong group’s spinning-off its spareparts division is another example. such spin-offs may benefit the parent company in three primary ways. the parent company may be better able to concentrate on its managerial, personnel, distributional and production resources in one or a small number of operational areas. spinning off areas of operations unrelated to the parent company’s core business will decrease diversification. this decrease will raise the firm’s volatility of earnings and subsequently, increase the value of the leveraged firm. spinning off the company into multiple operating units will enable investors to purchase just those operations which will provide a proper fit into their own portfolios. this “clientele” effect may serve to increase the total value of the company. 97the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 during the 1990s, some of the largest public companies spun-off major chunks of their assets in an effort to increase their share values. reversing their efforts towards conglomerate formation begun in the 1980s and continued into the early 1990s, berjaya group bhd spun-off three distinct companies; berjaya capital; cosway corporation; and nam fatt corporation. the conglomerate sunway holding incorporation separated sunway building technology when it spun-off this division. ytl corporation, in a more conventional manner, spunoff its ytl cement. telekom malaysia spun-off its subsidiary celcom after acquiring the unit two years earlier. from the above mentioned cases involving large companies, it can be seen that market capitalization can be a factor to spinoff. under certain circumstances, spin-off can reduce the overall tax burden of a corporation. that is to say that the combined taxes of the parent and spin-off company being reduced. this is because the spin-off can derive tax benefit from unused capital allowances and investment tax credits. spin-off activity can prevent the permanent loss of tax credits that would otherwise accrue to companies with negative incomes. companies losing money frequently cannot benefit from tax credits. spinning-off such firms to join as a profit making firms can create value by preventing the permanent loss of such credits. there are situations that exist where after the spin-off, it can be more highly geared, thus reducing overall tax burden. however, tax gains cannot be realized more than in the case of mergers and acquisitions as there is no group relief for tax i.e. each entity is taxed separately. there can still be a number of other factors for the spin-off activity. spin-off increases shareholders wealth, presumably at the expense of the bondholders wealth though no evidence can be forwarded as of yet. the argument is that by taking a portion of the assets fo the parent, the bondholders to the parent has now lost collateral, and hence the risk of the debt increases, thus the bond value goes down. consequence of that is the increase in shareholder value. these extemporaneous factors are not included in the regression model in view of their ambiguity as outlined in previous studies: see fro example hite and owers (1983) and schipper and smith (1983). c: findings from regression analysis the summary statistics on the variables are given in table 4. the median for the variables of focus and tax were 0 because these are dummy figures of 1 and 0. following the same note, minimum and maximum figures of 0 and 1 are used for focus and tax respectively. as for the market capitalization and age, the median is not far from the mean indicating that the related values are quite close in absolute amounts. here, the large absolute values for market capitalization were in common logarithm for accuracy and simplicity reasons. except for the age variable, where the kurtosis value is very large and the jarque-bera test the international journal of banking and finance, vol. 5.(number 1): 2007: 83-11398 table 4: descriptive statistics of focus, tax, market capitalization and age table 5: statistics on variables regressed against the car of parent companies 99the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 value is small, the statistics for all other variables show normal distribution of the data set. the kurtosis value is closed to 3 and the jarque-bera test records a high value, showing normality as the result. the statistics in table 5 reveals that for the parent firms, the regression model is significant with an f-ratio of 5.549 and above. the r-squared values are very high suggesting a good amount of explanatory power of these variables in the model. the first variable, focus of business, is not significant and the sign is as predicted. the second variable behaves the same way with the correct sign, except for the test window of one day, t = 0. the two other variables size and age of company are significant. the bigger the size of the parent the bigger is the price effect; the age is negatively correlated as older firms being larger would have less risk, and hence the returns would be less. prior studies in the matured markets in europe and the u.s. had similar results. thus, we have only two of the variables affecting significantly the pricing behavior in spin-off parent firms in this emerging market place. the spin-off firms may behave differently, so the model was applied to the spin-off firms: see the statistics in table 6. from the tabulation of the regression results in tables 6, it can be deduced that the regression methodology is valid. the coefficient of determination is pretty high as the adjusted r-squared of average 59.6-82.7 percent is much higher than in the cases of parent firms. the f-statistic of 11.362 and higher are also found to be significant as the probability of this value is 0.00 and lower thus, the regression model is valid and there is no risk in rejecting the null hypothesis of no relationship with the independent determinants. on the significance of the independent variables, tax and focus are not statistically significant although, as with the parent firms, the signs are in the right direction. the coefficients of market capitalization and age (except in one window) are also not significant: the signs are also perverse for the age factor. unlike prior studies by hite and owers (1983), veld and merkoulova (2003), focus factor is not proven to be significantly relevant. the same goes to tax factor, it does not appear as a predictor even though it is well taken as a determinant in prior studies in the us by hite and owers (1983), copeland, lemgruber and mayers (1987). this is in line with the argument that emerging has no specific tax regulations on spin-off and does not practice group tax loss relief, thus the impact ought to be not significant. the regulations that were used for guidance for spin-off are confined to listed companies listing their subsidiaries and rpgt concerns for transferor transferring shares in land-based companies. thus, in the absence of a generous tax relief as in the matured markets, the tax factor in the tested market is perhaps unimportant as suggested by the test result (except for the sign). the higher the market capitalization, the higher is the car. this result is similar to that in the us: schipper and smith (1983); hite and owers (1983); the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113100 copeland, lemgruber and mayer (1987); kudla and mcinish (1988); and krishnaswami and subramaniam (1999). generally, market capitalization grows with the age of a company in that the longer is the firm in existence, hence, frequently more established is the company with market capitalization increasing. if this case is prevailing in this study, then it will give rise to high correlation between the variables, capitalization and age. in order to ensure that the regression analysis is not affected by multicollinearity problem caused by these two variables, a different run of the regression was done with the age variable, and then the capitalization variable dropped. however, the results of the second and third regressions show no differences with the initial regression output in table 5 and 6. therefore, the initial regression results where the four independent variables were reported. the market capitalization factor can be supported by freeman’s (1987) argument that larger firms provide a greater variety of information than smaller firms and larger firms have greater exposure in the media. large firms are more table 6: statistics on variables regressed against the car of parent companies 101the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 likely to have additional information reported in the form of interim financial reports, analyst forecasts, industry forecasts, management forecasts and even litigation. another explanation is that trading by informed investors reveals private information. atiase (1985) argued that the partial revelation of private information by informed investors reduces the potential for profits in small firms to a greater extent because private information is more noticeable in thinly traded stocks. this factor limits the potential to exploit the knowledge of a misinformation in a small firm, and is an added incentive to undertake research on large firms rather than small firms. institutional investors are likely to concentrate on large firms due to liquidity constraints. for example, institutions cannot hold a large percentage of stock of a small firm and expect to be able to sell the stock immediately without price discounts. further, because institutions are a major source of demand for information, financial analysts may concentrate their search activities on larger firms. 5. conclusions this study was motivated to contribute by way of newer method of analysis restricted to the study of spin-off in emerging markets. the market chosen is ranked as among the top-20, although it is in an emerging economy. we analyzed a sample of 85 parent companies and their spin-off firms using the announcement information over a recent 23 years. the average car over the -100-day event window was more than 20 percent for the parent companies and more than 25 percent for the spun-off companies. the trend of the abnormal returns charted in the study shows that the wealth increase due to spin-off is some 5-6 times higher than is the case in studies in developed economies. the first two hypotheses about the wealth effect accruing to the parent shareholders and the information effect at the time of announcement on the share prices of parent companies have been shown to hold significantly. tax, age, market capitalization and focus factors were considered as possible determinants of spin-off performance, as is suggested in the literature. the study reveals that market capitalization and age factors are significant to the market price variation particularly for the parent firms, not so for the spin-off firms. these findings are, in our opinion, robust as we applied a matched-sample procedure to control for unknown factors in the emerging market. in fact the results from the matched sample procedure appear to provide a robust set of results. more studies of this kind are needed to generalize the findings to other emerging markets as being much more pronounced than in the developed capital markets studied so far. acknowledgment: this is one of three papers at the fma-asian finance association 2005 17th conference that won best prize award as selected by the the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113102 editors of the journal. the authors thank the editorial office for the revision of the paper as per the review comments. for errors, the authors take full responsibility. author statement: yoon chung-yin is a partner of the accounting firm and the co-author holds a chair in finance at the named university. yoon presented the paper, and won the best paper prize for this paper at a conference. ariff holds the renong endowed chair as a visiting professor at the university putra malaysia. the authors wish to thank taufiq hassan chowdury at the upm for the invaluable assistance in the research process. references ariff m., and johnson, l. w. (1990). “securities markets and stock pricing evidence in a developing equity market in asia” longman professional books, london, singapore and sydney. annuar, m. n., ariff m and shamsher m., (1994) “is kuala lumpur’s emerging share market efficient?” journal of international financial markets, institutions and money (4) 2 89-100 barry c b, peavy j w, and rodriguez m (1997) “performance characteristic of emerging capital markets” emerging markets quarterly (spring) 41-48. brown, s. j., & warner, j. b., (1985). “using stock returns the case of event studies” the journal of financial economics 14: 3-31. copeland t, lemgruber, e. f., and mayers, d., (1987) “corporate spin-offs: multiple announcement and ex-date abnormal performance” modern finance and industrial economics, chapter 7 t e copeland, ed. new york: basil blackwell press. cusatis p.j., miles. j. a., and woolridge, j. r., (1993). “restructuring through spin-offs” the journal of finance economics 33: 293-311. fama, eugene (1980). “agency problems and the theory of the firm,” journal of political economy 88: 288-307 fama, e., fisher, j., & roll, r., (1969). “the adjustment of stock prices to new information” international economic review 1-21. fama (1970) “efficient capital markets: a review of theory and empirical work” the journal of finance 383-417 government of malaysia “the companies act of malaysia 1965” government printers government of malaysia “the real property gains tax act 1976” government printers galai, d., and masulis, r., (1976). “the option pricing model and the risk factor of stock” journal of financial economics 3: 53-82. 103the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 habib, m. a., johnsen, d. b., and naik, n. y., (1997) “spin-offs and information” working paper, london business school. hite, g. l., and owers, j. e., (1983). “security price reactions around corporate spin-off announcement” journal of financial economics 12: 409-436. jensen m., and meckling, w., (1976). “theory of the firm: managerial behavior, agency costs and ownership structure” journal of financial economics 3: 305-360. johnson and klein, j., (1996). “the effects of spin-offs on corporate investment and performance” the journal of financial research 19(2): 293-307. john, t. a., (1993). “optimality of spin-off and allocation of debt” journal of financial quantitative analysis 28: 139-160. schipper, k., and abbie smith (1983). “effects of recontracting on shareholder wealth the case of voluntary spin-offs” journal of financial economics 12: 437-467. klein a (1986). “the timing and substance of divestiture announcements: individual, simultaneous and cumulative effects” the journal of finance 41 (july): 685-697 krishnaswami, s., and subramaniam, m., (1999). “ information asymmetry, valuation, and the corporate spin-offs decision” journal of financial economics 53: 73-112. kuala lumpur stock exchange “companies handbook” volume 1-3. kudla, r. j., and mcinish, t. h., (1988). “divergence of opinion and corporate spin-offs” quarterly review of economics and business 28(2): 20-29. lim, e. h., and obiyathulla, i. b., (2002). “price reaction to stock repurchase; evidence from the klse” working paper, international islamic university malaysia. miller, m., (1976). “debt and taxes” the journal of finance 31: 279-308. miles, m., and rosenfeld, j., (1983). “the effect of voluntary spin-off announcements on shareholder wealth” the journal of finance 38(5): 1597-1606. schipper, k., and smith, a., (1983). “effects of re-contracting on shareholder wealth: the case of voluntary spin-offs” journal of financial economics 12: 437-467. seward, j., and walsh, j., (1996). “the governance and control of voluntary corporate spin-offs” strategic management journal 17: 25-39. sharpe, w. f., (1963). “capital asset prices: a theory of market equilibrium under condition of risks” the journal of finance 3: 309-327. vijh, a. m., (1994). “the spin-off and merger ex-date effects” the journal of finance 19(2): 581-609. the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113104 appendix a key requirements of spin-off of a listed company and real property company, malaysia 1. profitability of parent the pretax profits, profits after tax and net tangible assets of the subsidiary or associated company to be listed should not account for more than 35 percent respectively of the consolidated pretax profits, profits after tax and net tangible assets of the group in respect of the past five years for a main board listing or three years for a second board listing. in addition, if the consolidated pretax profit of the holding company is reduced by more than 15 percent in the financial year following that in which the subsidiary or associated company is listed, the parent company, consequently upon listing of its subsidiary or associated company, will have to make good the shortfall within six months in the following year, failing which the parent company will be suspended; 2. capital issuing committee (cic) the parent company excluding the subsidiary or associated company to be listed and existing listed subsidiary or associated companies should on its own meet cic’s requirements for listing as if it were a new company seeking listing. 3. separate business the applicant subsidiary or associated company should be involved in a particular business of its own with profits able to meet cic’s criteria for listing with respect to track record and future prospects; 4. no intra-group competition the relationship. between the subsidiary or associated company seeking listing and the other companies within the group, including the parent company, should not give rise to intra-group competition or conflict of interest situation. 5. independence the subsidiary or associated company to be listed should demonstrate that it is not overly dependent on the other companies within the group, including the parent company, in terms of its operations, including purchases and sales of goods, management policies and finance; 6. real property companies prior to 1997, real property company (rpc) shares are deemed to be acquired on the date the relevant company becomes a real property company international journal of banking and finance, vol. 5, iss. 1 [2007], art. 4 105the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 or on the date of acquisition of the rpc shares, whichever is later. if the relevant company acquires additional real property or shares or both equivalent to 50 percent or more than those which it already owns, the date of acquisition of the rpc shares will shift forward to the date of acquisition of the additional real property or shares. after 1997, the deemed date of acquisition of rpc shares will no longer shift forward on purchase of additional real property or shares. if at that date of acquisition of shares, the relevant company was already a rpc, the deemed acquisition price will be either the consideration paid together with the incidental costs for the rpc shares or the deemed market value of the rpc shares in the event that the shares were not acquired at arms length. the acquisition price of bonus rpc shares will be zero since no consideration has been paid for the shares. however, the acquisition price of shares acquired through right issues will be the consideration paid for the rights issue. the new rule after 1997 may be unfair to disposers of rpc shares if the appreciation in value of the shares is not due solely to appreciation in value of the underlying property but to the retention of profits in the rpc and appreciation of other properties purchased subsequent to the date of acquisition of the rpc shares. yoon and ariff: corporate spin-offs, their price reactions and determinants in ma the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113106 appendix b key requirements of tax-free spin-off, united states 1. controlled corporation to qualify as a tax-free spin-off, the distributing corporation must distribute the stock of a controlled corporation preexisting or newly created to its shareholders. provided the companies meet the active business requirement (see requirement 3 below), the distributing corporation has significant latitude in the assets it transfers to the controlled corporation. for purposes of this requirement, control is defined as owning at least 80 percent of the voting power and at least 80 percent of the shares of each class of nonvoting stock. 2. securities distribution the distributing corporation generally must distribute all its controlled corporation stock and securities immediately before the transaction. revenue procedure 96-30, however, allows the distributing corporation to retain a limited amount of the stock or securities if the stock is widely held, the retention satisfies a significant business purpose, the controlled corporation officers and directors are officers and directors of the distribution corporation, the retained stock and securities are disposed of as soon as possible but no later than five years after the separation and the distributing corporation votes the retained stock in the same proportion as the stock distributed. in all cases, stock meeting the control definition must be distributed. 3. active businesses following the distribution, both the controlled and distributing corporations must be actively engaged in a trade or business with a five-year history. regulations section 1.355-3(b) defines an active trade or business as one in which all the necessary steps or activities take place to earn a profit. an active business does not include ownership of investments such as stock or land or the leasing of real or personal property unless the corporation provides significant services related to the property. 4. not a distributing device neither the distributing nor the controlled corporation can use the spin-off as a device for distributing earnings and profits. because of its vagueness, this requirement usually is the most troublesome. the key issue is whether the spin-off is indistinguishable from an ordinary dividend. regulations section 1.355 –2 (d) tries to help by listing factors that indicate when the spin-off (or other corporate division) is or is not a device for distributing 107the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 earnings. a spin-off starts with one strike against it: the first device factor is “pro rata distribution,” the very essence of a spin-off. regulations section 1.355-2 (d)(5), however, discusses distributions that have no tax avoidance potential and thus may satisfy the device requirement even if one or more device factors are present. for example, a distribution ordinarily would not be a device if the distributing and controlled corporations have no accumulated earnings and profits. 5. business purpose regulations section 1.355-2(b) requires a spin-off to be “motivated, in whole or substantial part, by one or more corporate business purposes.” the purposes must be “real and substantial” and “germane” to the distributing or controlled corporation’s business or to the affiliated group of which the distributing corporation is a member. neither reducing federal income taxes nor satisfying a shareholder purpose is a corporate business purpose. however, a distribution made primarily to achieve a corporate business that also achieves a shareholder purpose is not disqualified. a corporate business purpose also fails the test if it could be accomplished in a nontaxable transaction that is not “impractical or unduly expensive.” the subjectivity inherent in the above rules, as well as intense irs scrutiny, has made the business purpose requirement one of the most difficult hurdles a company must overcome to ensure tax-free treatment. the revenue procedure 96-30 removes some of the subjectivity by discussing at length the qualifying criteria and the information companies must submit for advance rulings for each of nine business purposes that may qualify a spin-off for tax-free treatment. 6. continuity of interest regulations section 1.355-2 (c) says that following the distribution of the controlled corporation’s stock, the distributing corporation shareholders must maintain continuity of interest in both companies. revenue procedure 9630 further says this requirement generally is met if one or more persons who directly own the distributing corporation before the distribution also own 50 percent or more of the stock in each of the modified companies after the separation. the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113108 109the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113110 111the international journal of banking and finance, vol. 5.(number 1): 2007: 83-113 1 based on the ifc’s emerging markets data base for a recent period, a commonly held view is emerging stock markets are characterized by high returns and high volatility. malaysian market is such a market and is ranked among the top-20 or thereabout, and is among the 65-odd emerging markets. it is ranked in the financial press as being among the top five in terms of its desirability for cross-border investments. 2 the world bank defines a developing country as one having a per capita gnp that would place it in the lower or middle-income category; a developing country had an annual per capita gnp less than about us$10,000. capital markets in such a country would be named an emerging market. the capital market is ranked by oecd in 1999 report as among the top-20 investible market. 3 two stock exchanges, the then kuala lumpur stock exchange (klse) and malaysian exchange of securities dealing and automated quotation bhd (mesdaq) had merged to form bursa malaysia. klse was incorporated in 1976 and mesdaq was incorporated in 1997. 4 new foreign and domestic investment played a significant role in the transformation of this economy. manufacturing share increased from 13.9% (1970) to 34% (2005) as agriculture declined from 42.7% to under 7% over the same period: see 2005 bank negara malaysia annual report. 5 after nearly two decades of strong economic growth averaging 8.7 percent annually, the financial the international journal of banking and finance, vol. 5.(number 1): 2007: 83-112112 crisis had long-lasting impact by halving the growth prospect, when the regional financial crisis of 1997-8 swept the region. the gdp suffered a sharp 7.5% contraction in 1998: it rebounded in 1999 to grow by 5.6%. 6 irs section 355 includes four key requirements for tax-free treatment: controlled corporation; securities distribution; active businesses; and not a distribution device. treasury regulations section 1, 355-2 adds two additional requirements: business purpose; and continuity of interest. see appendix a for details. 7 prior to 1995, capital issues committee (cic) and the foreign issues committee were reviewing. cic was set up in 1968 by the minister of finance to ensure an orderly development of the capital market by regulating the issue of securities by public companies and the listing of such securities on bursa malaysia including spin-offs. see the requirements for spin-offs in appendix a. 8 it was possible to use distribution of bonus shares as consideration to save rpgt on selling of land together with the real property companies before 1977. 9 despite the widespread use of the named variables in this section, there is still no agreement among the researchers on the factors influencing the spin-off effect. we place on record this list as the possible starting point for this study. 10 in this study we chose day –100 to day +50 as the window for analysis for the parent company because preand post spin-off days are needed for testing a hypothesis. also, in a less matured market with approval procedures that take longer time than in the matured market, we needed to go back to about 4.5 months to observe the price effect. 11 the argument offered here is that there is an absence of tax motivation for spin-off in the malaysia compared to the united states. see appendix a and b for the legal framework in these countries. 12 the results of these second and third regressions are produced in appendices c, d, e and f. international journal of banking and finance 1-1-2007 corporate spin-offs, their price reactions and determinants in malaysia chung-sin yoon mohamed ariff recommended citation credit risk of islamic banks in gcc countries credit risk of islamic banks in gcc countries hamid a. h. al-wesabi and nor hayati ahmad university utara malaysia ________________________________________________________________ abstract this paper is about factors affecting credit risk of islamic banks in the gulf cooperation council countries using website data covering 25 islamic banks over 2006 to 2010. this study uses non-performing loans as a proxy for credit risk, which is the dependent variable with three macro-economic, and six firmspecific independent variables. we find income is significantly negatively related to credit risk, which is consistent with findings in other countries about credit risk. some firm-specific variables such as leverage, liquidity are also relevant variables for credit risk, which results are also consistent with bank behavior reported in other studies. credit risk is also broadly affected by both macro and firm-specific factors as found in other regions. inflation and interest rates do not appear to be relevant. these results would suggest non-performing loan is broadly correlated with factors identified in other studies of banks. keywords: credit risk; risk management; islamic banks; gulf countries; nonperforming loans jel classification: g32 ________________________________________________________________ 1. introduction islamic banking has developed as a niche banking serving customers with special financial needs, and islamic banking products are compliant with a number of binding restrictions in the pricing, contracting, investments, and return characteristics compared with the mainstream banks, which dominate banking transactions. over some 50 years, these islamic niche banks in 76 countries have developed systems of operations that rely on the principles of islamic sharīʿah (legal system) pertaining to financial transactions. in those banks, sharīʿahcompliant banking products are securitized, offered to meet the special needs of customers. predominantly, these products are offered by the same banks that also sell products without the need for such compliance. in order to achieve the perceived benefits to society as per the compliance with the broader social needs of the community, islamic banks have spread around the world during the past five decades. it has a significant presence in the middle eastern countries, with saudi arabia, dubai and bahrain having ijbf al-wesabi and ahmad: credit risk of islamic banks in gcc countries 96 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 the biggest islamic banking assets compared with the other slightly smaller centers in kuwait, malaysia, indonesia, sudan, and pakistan, etc. bahrain with 27 banks has about half the number of islamic banks in the gulf cooperation council or gcc countries. though it has the largest number of islamic banks in gcc, in terms of assets, it is not the biggest center. this paper is about finding the factors associated with credit risk of islamic banks in the gcc countries. the paper is divided into five sections. in the next section, we describe the banks in the gcc countries and identify the research problem in the islamic banking. the concept development and empirical results are found in section 3. the findings are reported in section 4 and conclusions are presented in section 5. 2. islamic banks in the gcc countries islamic banks have been welcomed in many legal jurisdictions especially after the bank of england decided about 12 years ago to license such banks as providing a niche banking services to special needs of customers. prior to that acceptance of this niche banking, islamic banks had slowly established in the persian gulf region, southeast asia, and few other places. the gcc countries as the recipient of oil-and-gas money found outlet as growth of the conventional and this niche banking. for example, dubai bank and the national bank of sharjah have become islamic banks in the uae. national bank of saudi arabia, and kuwait real estate bank (kuwait international bank) recently converted their entire operations to be islamic.1 in the light of competition in conventional banking and the increasing customer demand for islamic banking services, banks such as dubai islamic bank, qatar islamic bank, and alrajhi bank2 in the gcc region expanded globally. at the end of 2009,3 the islamic banks’ total assets were estimated to be over us$350 billion, which by estimate is 43 per cent of total islamic banking assets in the world compared with less than 10 per cent in 2003. that indicates a very high growth rate, and is expected to trend even higher on the back of the wider acceptance and increasing demand for banking products and services. the most important factors explaining the fast growth of islamic banking assets in the gcc region are strong government involvement4 supporting their economies with ever increasing oil-and-gas wealth. economists classify gcc countries as developing countries. oil and gas are the most important sources of income which contributing to upwards of 60 per cent of the economic activities. furthermore, the energy sector is characterized by a high degree of energy-financing and higher average per capita income. the richest muslim majority countries are located in this region: the average population of these countries is about 8 million per country, and 1 booz & company www.booz.com and www.booz.com/me. 2 previously known as al-rajhi banking and investment corporation. 3 a report released by kuwait finance house (2010). 4 standard & poor’s ratings services. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 8 credit risk of islamic banks in gcc countries: 95-112 97 the gdp per capita is upwards of us$ 24,000 (laabas & limam, 2002). as a result, gcc countries face many urgent challenges to decrease their energysector dependency and develop non-energy sectors (saif, 2009). it is obvious that macroeconomic performance would be affected by the changes in the oil/ gas prices or the fast developing alternatives to oil and gas. instability in oil revenues would have an impact on the rates of growth in the economy through the effect on government expenditures. these countries are diversifying their exports in order to decrease the dependency on oil, while they are also trying to decrease their imports by growing basic manufacturing capacities (despite the influence of cheaper products from china that works against this effort). economic diversification is also important for these countries to reduce risk or at least spread it, to limit the effect of a wide volatility in oil-and-gas prices, to create job opportunities and to promote economic development (fasano & iqbal, 2003). 2.1 credit risk in islamic banks islamic banks in these countries still face several challenges in introducing workable reforms and processes to improve the development of the islamic finance on a larger scale. one obstacle to improving performance of these banks is the lack of knowledge about the dynamics of their credit risk. hence, a study of their credit risk may provide policy-relevant findings for their intended reforms to be considered. credit risk especially in mushārakah, murārabah, salam and istisnāh contracts is studied. the expected risk in these contracts arises as a result of inability of debtors to meet their financial commitments on the dates of maturity. consequently losses are incurred by islamic banks. to minimize risk, islamic sharī’ah allows collateral which is considered as secured against the debt risk while third-party guarantees have also been introduced in addition to the loan loss reserve (llr) to forestall credit risk. islamic banks are still no exception to the four main types of risks that lead to systemic failure. of the four, the most critical one is the credit risk (fleifel, 2009). according to chapra and khan (2000) and khan and ahmed (2001), credit risk is an important financial risk that is included within default risk, which covers more than three-quarters of the risks in an average islamic bank’s asset portfolios in the banking book. credit risk causes more than three-quarters of islamic bank failures. credit risk appears greater than other risks, especially with pls (profit-and-loss-sharing) modes of financing. furthermore, default risk also introduces market risk and liquidity risk. on the other hand, in the modes of non-pls contracts such as murābahah and ijārah, market risk is higher than the credit risk. this is, given the fixed income mode adopted in these contracts, the rate of bank profit is affected by the market’s volatility (khan & ahmed, 2001). according to al-jarhi and iqbal (2001), most gcc islamic banks prefer fixed return modes, especially murābahah modes, which represent the dominant form of financing. they found that the share of mushārakah-mudārabah in the total financing is smaller than the share of murābahah (that is to say, murābahah al-wesabi and ahmad: credit risk of islamic banks in gcc countries 98 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 and ijārah form 75 per cent of the total of financing with only 25 per cent mushārakah-mudārabah, istisnāh and salam).5 the dominant contracts are not based on pls principle. these are akin to purchasing good/services or leasing of capital goods/consumer durables. the returns to investors are fixed, while the pls-based instruments are less dominant. therefore, returns to investors would depend on the volatility in the market. hence, islamic banks try to aim for a balance between the two modes of non-pls and pls financing modes. there is expectation that the two modes have significantly different impacts on banking credit risk formation (al-jarhi & iqbal, 2001). therefore, islamic banks should be operationally well-prepared to manage credit risks to ensure the maintenance of their performance as well as gain growth. there are many factors affecting credit risks. this topic has been studied widely in the case of conventional banking (ahmed & nizam, 2004). this study examines the factors affecting credit risk of banks in the gcc countries. another major significant and unique risk that islamic banks face is the sharī’ahcompliance risk. however, that is not included in the present study. the research questions that guide the research are about what factors affect credit risk of the sample of islamic banks in the gcc countries. the secondary question is to examine how the credit risk has been affected by the 2008-9 global financial crisis. 3. literature review 3.1 delegated monitoring theory delegated monitoring theory explains the process by which banks are justified as providing financial intermediation services to transfer deposit funds of surplus units (mainly households) making deposits enabling banks to act as their delegated monitors by making loans to deficit units (producers): diamond (1984; 1996). monitoring is possible through the collection of required information about the firm (the investor) on the grounds of safeguarding the loans from defaulting. further, the banks impose restrictive conditions on the borrower to forestall firms from putting other parties ahead of paying the debt and the returns to the banks. that process incurs costs of monitoring on banks as a result of monitoring and executing the debt contracts. therefore, delegated monitoring theory is one of the most significant theories to explain why banks have to rely more on capital and less on labor (matthews & thompson, 2005; 2008). delegated monitoring theory as applied to islamic banking has to be reinterpreted using the profit-and-loss-sharing mode to take into consideration the asymmetric information problem that would occur between islamic bank (the agent) and the counterparty (firm) when the bank has insufficient information about the actual profits of the counterparty. also the asymmetric information 5 statistics of the general council for islamic banks, bahrain, (2001). international journal of banking and finance, vol. 10, iss. 2 [2013], art. 8 credit risk of islamic banks in gcc countries: 95-112 99 problem could also occur in the non-pls mode such as in murābahah, in the event of external events such as systemic risk from other banks that may cause non-performance of the contracts of a borrower (khan & ahmed, 2001). 3.2 review of findings many studies identify financial risk as one of the most important challenges for islamic banking. islamic banks are affected by many issues related to this risk. according to makiyan (2008), risk management in islamic banking is still not effective enough to face the major challenges of financial risk. one needs to understand many issues better to be able to apply innovative and appropriate solutions to prevent financial risk attributes of financial products. islamic permissible modes of financing (pls such as murārabah or non-pls such as murābahah) need many measurements of risk such as income recognition, adequacy of collateral to be available, etc. (makiyan, 2008). according to hasan and dridi (2010) whose research is covering both types of banks in many countries, islamic banks were affected by the crisis quite differently than conventional banks. the decline in profitability of some islamic banks in 2008 was actually a result of weakness in the practices of risk management. but generally, credit and asset growth in islamic banks were better than in conventional banks in 2008-9. iqbal, ahmad and khan (1998) present evidence in their study to the effect that most of the islamic banks in gcc countries use both fixed return modes and variable return modes. even though pls modes are preferable but the share of these funding modes in the total financing is very small, about a quarter or less than the share of non-pls modes. murābahah mode dominates all other modes. this occurs perhaps due to bankers wanting to avoid the greater credit risk in the pls modes, apart from the acute asymmetric information problem in such contracts. khan and ahmed (2001) define credit risk as a failure of the counterparty to meet bank obligations according to the agreed terms. this risk would arise in islamic banks as a form of settlement/payment risk when the banks pay money before they receive the assets (e.g., salam or istisnāh) or deliver assets before receiving cash (as in murābahah contract). khan and ahmed (2001) also show the diversification effect of granting loans on the portfolio. they mention that the effective means of reducing credit risk is to actually reduce the risk by forming a minimum risk portfolio in the loan book. in this manner, risk management can be minimized to handling the systematic risk and avoid (better still lessen) the need for managing non-systematic risk in the bank’s portfolios. in this sense, the conventional wisdom of portfolio risk minimization works in all banks. thiagarajan, ayyappan and ramachandran (2011), who studied indian economy, find that the lagged non-performing asset (npa) has a positive influence on the current npa and that there is a significant inverse relationship between credit risk and gdp. thus, the gdp growth helps the banking system while having non-performing loans at an acceptable level. in fact, al-wesabi and ahmad: credit risk of islamic banks in gcc countries 100 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 both macroeconomic factors and bank-specific factors play important roles in determining the credit risk of the banking system. ahmed, akhtar and usman (2011) state that the credit risk is highly affected by size of the bank, capital adequacy and debt equity ratios. they also find a positive and statistically significant relationship with credit risk at 0.05 per cent and 0.01 per cent levels respectively. the asset management quality has a significant negatively relationship on credit risk. therefore all explanatory variables appear to affect credit risk. they find a statistically insignificant relationship between npl and credit risk in the indian banks. ahmad and ariff (2007) search across two types of conventional banking systems, banks in emerging economies and their counterparts in several developed countries by comparing them. they find that several factors are significant and relevant to credit risk, whether the banks are from emerging economies or developed economies. regulatory capital would be increased to adequately absorb any losses so the capital is significantly positively relevant to credit risk. however, their results suggest that some countries with undercapitalized banks still face more risks, meaning that the credit risk is affected significantly negatively. management quality, which is a ratio of earning assets to total assets affect credit risk significantly positively. loan loss provision as a ratio of total loans affects credit risk significantly positively. these researchers use npl, which is bad loans, as a measure for credit risk, as in this paper. they highlighted that emerging economy banks have a higher credit risk than their counterparts in developed economies. since, gcc countries are classified as emerging economies so the banking system is likely to be facing higher credit risk as well. das and ghosh (2007) examine both macroeconomic and microeconomic factors that affect credit risk of indian banks, a case of banks from emerging economy. generally, the results show that real loan growth, gdp growth and (at the bank level) bank size, real interest rate and operating expenses have significant influences on credit risk. the effects of these factors are respectively mixed. they employ npl instead of credit risk as a proxy for dependent variable. khemraj and pasha’s (2010) study ascertains what determines the npl in the guyanese banking system. they find gdp growth is significantly related inversely to npl and the effective exchange rate has a positive impact on npl, given currency depreciation. they also find that real interest rate has a significantly positive effect on npl, as also is the case for inflation and loan growth. however, the size of bank has mixed positive and negative relationship. the ratio of loans to total asset was negatively related to npl. al-smadi (2010) applies risk index to measure exposure to risk of several jordanian banks, using data over 1995 to 2008. his findings indicate that three major macroeconomic variables are statistically significant. they are gdp, inflation rate and market interest rate. he provides evidence that internal variables have effects on credit risk more than external variables. he finds that the relationship between gdp and credit risk is significantly negative, while it is positive on inflation and also positive on interest rate. there are five bankinternational journal of banking and finance, vol. 10, iss. 2 [2013], art. 8 credit risk of islamic banks in gcc countries: 95-112 101 specific variables: npl, loan concentration in risky sectors, loan growth, bank size and net interest margin in their study. these five variables have significant relationship with credit risk. loan growth and loan concentration in risky sectors have positive effects as well. bank size has a negative effect on credit risk. a study on the determinants of credit risk (ahmad & nizam, 2004) in malaysia for the period 1996 to 2002 compares islamic banks and conventional banks. they use non-performing loans for credit risk exposure. their results show that three variables are significantly related to credit risk of islamic banking: management efficiency, risk weighted asset, and natural logarithm of total assets. all these have positive relationships with credit risk. additionally, four variables significantly affect credit risk in the cases of conventional banking: these are risky sector loan exposure, regulatory capital, loan loss provision, and risk-weighted assets. they find positive relationship with credit risk except in the case of regulatory factor. in addition, the authors find a higher r-square value in conventional banking compared to islamic banking: 75.6 per cent and 34 per cent respectively. this denotes that the combined impact of these variables is stronger on conventional bank credit risk than islamic bank credit risk. 4. data and methodology this section explains how this study is conducted in terms of research design, sample, data collection, and analysis. the main questions that guide the study are restated as follows: what are the variables which influence credit risk? what is the impact of global financial crisis on the credit risk? all questions apply to islamic banks in the gcc countries. once these variables are identified, the management of credit risk will be manageable using the variables to mitigate risk and improve credit risk management. these variables are selected as suggested in several studies as being important. these are explained in this section, together with all the measurements of variables, the model, framework and the hypotheses development. 4.1 theoretical framework and model this section explains the modeling of the variables that relates npl as credit risk to a number of assumed independent variables widely used in such studies: lnta (natural log of total assets), mgteff (management efficiency), regcap (regulatory capital), l/d (proportion of loan to deposit), rskast (risky asset which is real estate asset in gcc countries according to the published reports) and llp (loan loss provision). these variables represent internal independent variables (ahmad and nizam, 2004). gdp (gross domestic product), inf (inflation rate) and libor (london inter-banks offered rate) because islamic banks in gcc countries usually link profit margin with that ratio) are external independent variables. regression model is used to test the variables affecting credit risk. according to past studies of das and ghosh, (2007); ahmed and ariff, (2007), al-wesabi and ahmad: credit risk of islamic banks in gcc countries 102 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 there are nine variables in studies on credit risk. some descriptive statistics are given in appendix a. the test model is in equation (1): crit = b 0 + b 1 gdp it + b 2 inf it + b 3 libor t + b 4 lnta it + b 5 mgteff it + b 6 regcap it + b 7 , l/d it + b 8 rskast it + b 9 llp it + e it (1) where: cr it : non-performing loan (plns) for bank (i) in year (t), b 0 : constant coefficient, gdp it : gross domestic product for the country (i) in year (t), inf it : inflation rate for the country (i) in year (t), libor t : london inter-bank offered rate in year (t). lnta it : natural log of total assets for bank (i) in year (t), mgteff it : management efficiency for bank (i) in year (t), regcap it : regulatory capital for bank (i) in year (t), l/d it : proportion of loan to deposit for bank (i) in year (t). rskast it : risky asset for bank (i) in year (t), llp it : loan loss provision for bank (i) in year (t), e it term of the random error; and b i : the coefficient respectively of variable i. 4.2 hypotheses development the main aim of this investigation is to identify selected variables that affect credit risk. these variables are divided into two categories with one representing specific variables of banking and the other representing macroeconomic factors. both have significant impact on credit risk according to literature. h 1 : there is a negative relationship between gdp and credit risk in islamic banks of gcc countries. h 2 : there is a positive relationship between inflation rate and credit risk in islamic banks of gcc countries. h 3 : there is a positive relationship between libor and credit risk in islamic banks of gcc countries. h 4 : there is a negative relationship between size and credit risk in islamic banks of gcc countries. h 5 : there is a negative relationship between management efficiency and credit risk in islamic banks of gcc countries. h 6 : there is a positive relationship between regulatory capital and credit risk in islamic banks of gcc countries. h 7 : there is a positive relationship between proportion of loan to deposit and credit risk in islamic banks of gcc countries. h 8 : there is a positive relationship between a risky asset and credit risk in islamic banks of gcc countries. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 8 credit risk of islamic banks in gcc countries: 95-112 103 h 9 : there is a negative relationship between the loan loss provision and credit risk in islamic banks of gcc countries. 4.3 measurement of variables according to ahmed and nizam (2004), das and ghosh (2007), and al-smadi (2010), credit risk is a dependent variable and is measured by npl as a proxy. definition of npl in this study follows assumption in the form sent to previous studies and is measured as in the definitions that follow. this study has two types of independent variables (external and bankspecific variables) the external variables are the macroeconomic factors and they are found as a figure or ratio form in the macroeconomic annual reports, thus easing the burden of measuring these variables by researchers. however, sometimes researcher can measure some variables in few cases. for example for the gdp, they use this equation: gdp = c + i + (x m) + g. where c: consumption, i: investments, x: exports, m: imports and g: government’s expenditure. regarding inflation rate, they use consumer price index (cpi) and inf rate equal the rate of change of cpi and taken as the annual basis. as for the specific banking independent variables, they are measured in this study following measurements as in past studies. size of bank (lnta) is measured as follows: (2) (3) management efficiency (mgteff) variable: (4) regulatory capital (regcap) variable: (5) the proportion of loan to deposit (l/d) variable: (6) risky asset (rskast) variable: (7) loan loss provision (llp) variable: (8) financing, besides loan, means it is based on pls modes, which is different from the loan based on interest rate in conventional banking. figure 4.1 shows the extent of the rise of credit risk especially in kuwait in 2009. it is npls = lnta = log (total assets) mgteff= regcap= l/d= rskast= llp = npls = lnta = log (total assets) mgteff= regcap= l/d= rskast= llp = npls = lnta = log (total assets) mgteff= regcap= l/d= rskast= llp = npls = lnta = log (total assets) mgteff= regcap= l/d= rskast= llp = npls = lnta = log (total assets) mgteff= regcap= l/d= rskast= llp = npls = lnta = log (total assets) mgteff= regcap= l/d= rskast= llp = npls = lnta = log (total assets) mgteff= regcap= l/d= rskast= llp = al-wesabi and ahmad: credit risk of islamic banks in gcc countries 104 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 noticed that credit risk generally increases over the period from 2006 to 2010, meaning that there is an imperfection in the credit risk management which is required to be examined and identified. figure 1: npl in the islamic banks in gcc countries from table 1, it is clear that the mean credit risk estimated is 0.043 for all of the banks in the test period, 2006-10. this is also shown in figure 4.5 where the npl increases from 3 percent in 2006 to 6 percent in 2010. this increase in the npl is because of the rising in the impaired loans (financing in islamic banks) compared to the total loans. the variance between banks in term to its credit risk exposure (standard deviation) estimated is 0.040 as shown in table 1. also the means of gdp, inf and libor are 6.46, 4.94, and 0.029 respectively. the standard deviations are 5.53, 3.34 and 0.018 respectively. table 2 provides a summary of the regression results obtained from running a regression using npl as dependent variable on the 9 independent variables as in the model developed in this paper. as can be observed, four factors are statistically significant: management quality; l/d; risk assets; gdp. the other factors are not significantly correlated with the credit risk. overall, the regression provided significant model fit as judged by the f-ratio of 4.31 (not shown in the table). the coefficient of determination was 0.6578 meaning that the variables explained 65.78 per cent of the variation in the credit risk as measured by the npl. as is clear from the results, we have identified four factors out of the nine factors as being significantly correlated with the credit risk. these factors appear to make sense. liquidity is of course a major issue in all banking systems, and this is of importance in managing credit risk. management quality has consistently come up as a major factor in other studies. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 8 credit risk of islamic banks in gcc countries: 95-112 105 table 1: descriptive statistics of variables used in gcc countries variables mean std. deviation n npls 0.0433 0.0402 125 lnta 22.281 1.324 125 mgteff 0.771 0.1979 125 regcap 0.237 0.195 125 l/d 2.496 5.455 125 rskast 0.211 0.142 125 llp 0.086 0.432 125 gdp 6.464 5.531 125 infl 4.939 3.337 125 libor 0.028 0.0181 125 table 2: coefficients from regression analysis model unstandardized coefficients standardized coefficients t-values significance b std. error beta 1 constant -0.018 0.097 -0.186 .853 lnta 0.005 0.004 0.157 1.114 .268 mangeff -0.065 0.020 -0.318 -3.189** .002 regcap 0.010 0.031 0.049 0.326 .745 l/d 0.002 0.001 0.300 2.852*** .005 rskast 0.053 0.023 0.188 2.323** .022 llp -0.007 0.007 -0.077 -0.980 .329 gdp -0.002 0.001 -0.208 -2.179** .031 inf 0.000 0.001 -0.002 -0.023 .982 libor -0.119 0.209 -0.053 -0.569 .571 this study confirms that for islamic banks, the higher the risky assets in the total asset portfolio the higher is islamic bank’s credit risk. finally, gdp or income is a significant factor. if gdp growth is declining, the chances are high that credit risk will go up. we did not find the other five factors to be al-wesabi and ahmad: credit risk of islamic banks in gcc countries 106 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 significant, although the signs are sometimes revealing of the directions of the effect. nevertheless, not finding some of the factors to be relevant is consistent with other studies such as ahmad and ariff op cit. 5. conclusion the aim of this paper is to make a preliminary study of how credit risk is associated with key macroeconomic and bank-specific factors in the fast-growing islamic banking region, namely the gulf region. findings from this study are likely to be the first on this policy-relevant topic. unlike in the conventional banking, no one knows how credit risk is driven by what factors in islamic banks. we developed a simple cross-sectional model using annual data of 25 established commercial banks over a reasonable length of time to relate credit risk to the already verified and widely used nine potential factors used in studies of credit risk of conventional banks. the results of this study reveal that the islamic banks in the gulf region have significant credit risk as measured by the size of the non-performing loans. also, the credit risk is largely determined by such common factors (that is common with conventional banking) as liquidity, management quality, risky assets in portfolios and gdp. these results may be treated as preliminary from six countries in one region of this fast growing banking industry that has spread to 76 countries. more research of other banks in other regions may reveal how credit risk is formed in this new niche banking across the world. when such further studies are made, it would be possible to frame a strategy on how islamic banks would manage credit risk. author information: this paper is one of three papers that were selected as the best papers considered for award at the fifth foundation of islamic finance conference at langkawi, malaysia in july, 2012. the authors wish to record their appreciation for this award and also thank the discussant and participants at the conference for their generous comments. nor hayati ahmad is a professor in uum college of business, and may be contacted at ayati@uum.edu.my. hamid abdulkhaleq hasan al-wesabi may be contacted at hameedalwesabi@ yahoo.com. references ahmad, n. h. & ariff, m. (2007). multi-country study of bank credit risk determinants. the international journal of banking and finance, 5, 135152. ahmad, n. h. & nizam. a, s. (2004). key factors influencing credit risk of islamic bank: a malaysian case. review of financial economic 2. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 8 credit risk of islamic banks in gcc countries: 95-112 107 jeddah: islamic development bank-islamic research and training institute. ahmed, n., akhtar, m. f., & usman, m. (2011). risk management practices and islamic banks: an empirical investigation from pakistan. interdisciplinary journal of research in business, 1(6), 50-57. al-jarhi, m. & iqbal m. (2001). islamic banking: answers to some frequently asked questions. jeddah: islamic development bank-islamic research and training institute. al-smadi, m. o. m. (2010). credit risk, macroeconomic and bank specific factors in jordanian banks. phd thesis, universiti utara malaysia. chapra, m. u. & khan, t. (2000). regulation and supervision of islamic banks. jeddah: islamic development bank-islamic research and training institute. das, a. & ghosh s. (2007). determinants of credit risk in indian state-owned banks: an empirical investigation. economic issues, 12(2), pp 48-66. diamond, d. w. (1984). financial intermediation and delegated monitoring. the review of economic studies, 51(3), 393–414. diamond, d. w. (1996). financial intermediation as delegated monitoring: a simple example. federal reserve bank of richmond economic quarterly, 82(3), pp 51-66. fasano, u. & iqbal, z. (2003). gcc countries: from oil dependence to diversification. washington: international monetary fund. fleifel, b. a. (2009). risk management in islamic banking and finance: the arab finance house example. an unpublished thesis submitted to the university of north carolina wilmington, usa. greuning, h.v., & z. iqbal. (2008). risk analysis for islamic banks. washington: world bank. hasan, m. & dridi, j. (2010). the effects of the global crisis on islamic and conventional banks: a comparative study. imf working paper, wp 10/201. iqbal, m., ahmad, a. & khan, t. (1998). challenges facing islamic banks. jeddah: islamic development bank-islamic research and training institute. iqbal, z. (2009). impact of global financial crisis on idb member countries: the case of gulf cooperation council and sub-saharan africa. the pakistan development review, 47(4), pp. 583-601. khan, t. & ahmed, h. (2001). risk management: an analysis of issues in islamic financial industry. jeddah: islamic development bank-islamic research and training institute. khemraj, t. & pasha, s. (2009). the determinants of non-performing loans: an econometric case study of guyana. paper presented at the caribbean centre for banking and finance bi-annual conference on banking and finance, st. augustine, trinidad. laabas, b. & limam i. (2002). are gcc countries ready for currency union? kuwait: arab planning institute. al-wesabi and ahmad: credit risk of islamic banks in gcc countries 108 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 makiyan, s. n. (2008). risk management and challenges in islamic banks. journal of islamic economics, banking and finance, 4(3), 45-54. matthews, k. & thompson, j. (2008). the economics of banking. chichester: john wiley and sons. saif, i. (2009). the oil boom in the gcc countries, 2002–2008: old challenges, changing dynamics. washington: carnegie endowment for international peace. thiagarajan, s., ayyappan, s., & ramachandran, a. (2011). credit risk determinants of public and private sector banks in india. european journal of economics, finance and administrative sciences, 34, 147154. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 8 credit risk of islamic banks in gcc countries: 95-112 109 appendix a table a1: anova model sum of squares df mean square f sig. 1 regression .064 9 .007 6.035 .000 residual .136 115 .001 total .201 124 al-wesabi and ahmad: credit risk of islamic banks in gcc countries international journal of banking and finance, vol. 10, iss. 2 [2013], art. 8 credit risk of islamic banks in gcc countries: 95-112 111 t ab le a 2: c ro ss c or re la ti on m at ri x n p l s l n t a m g t e f f r e g c a p l /d r s k a s t l l p g d p in f l l ib o r p ea rs on c or re la ti on n p l s 1. 00 0 -. 10 5 -. 39 2 .2 33 .2 71 .3 16 -. 05 1 -. 20 7 -. 06 7 -. 24 6 l n t a -. 10 5 1. 00 0 .4 83 -. 78 5 -. 30 6 -. 01 4 .1 26 -. 11 3 .0 30 -. 20 8 m g t e f f -. 39 2 .4 83 1. 00 0 -. 42 6 -. 09 2 -. 19 8 .0 32 .2 41 .0 45 .2 00 r e g c a p .2 33 -. 78 5 -. 42 6 1. 00 0 .5 51 .1 44 -. 10 6 .1 17 -. 04 2 .0 97 l /d .2 71 -. 30 6 -. 09 2 .5 51 1. 00 0 .1 65 -. 05 3 .3 58 .0 45 -. 02 7 r s k a s t .3 16 -. 01 4 -. 19 8 .1 44 .1 65 1. 00 0 .0 45 -. 02 6 .0 06 -. 15 7 l l p -. 05 1 .1 26 .0 32 -. 10 6 -. 05 3 .0 45 1. 00 0 -. 09 6 -. 10 9 -. 17 2 g d p -. 20 7 -. 11 3 .2 41 .1 17 .3 58 -. 02 6 -. 09 6 1. 00 0 .2 87 .3 65 in f -. 06 7 .0 30 .0 45 -. 04 2 .0 45 .0 06 -. 10 9 .2 87 1. 00 0 .3 08 l ib o r -. 24 6 -. 20 8 .2 00 .0 97 -. 02 7 -. 15 7 -. 17 2 .3 65 .3 08 1. 00 0 s ig . (1 -t ai le d) n p l s . .1 21 .0 00 .0 04 .0 01 .0 00 .2 87 .0 10 .2 30 .0 03 l n t a .1 21 . .0 00 .0 00 .0 00 .4 38 .0 81 .1 05 .3 69 .0 10 m g t e f f .0 00 .0 00 . .0 00 .1 54 .0 13 .3 63 .0 03 .3 09 .0 13 r e g c a p .0 04 .0 00 .0 00 . .0 00 .0 55 .1 21 .0 97 .3 19 .1 42 l /d .0 01 .0 00 .1 54 .0 00 . .0 33 .2 79 .0 00 .3 09 .3 84 r s k a s t .0 00 .4 38 .0 13 .0 55 .0 33 . .3 10 .3 85 .4 73 .0 40 (c on ti nu ed ) al-wesabi and ahmad: credit risk of islamic banks in gcc countries 112 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 n p l s l n t a m g t e f f r e g c a p l /d r s k a s t l l p g d p in f l l ib o r l l p .2 87 .0 81 .3 63 .1 21 .2 79 .3 10 . .1 43 .1 14 .0 28 g d p .0 10 .1 05 .0 03 .0 97 .0 00 .3 85 .1 43 . .0 01 .0 00 in f .2 30 .3 69 .3 09 .3 19 .3 09 .4 73 .1 14 .0 01 . .0 00 l ib o r .0 03 .0 10 .0 13 .1 42 .3 84 .0 40 .0 28 .0 00 .0 00 . n n p l s 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 l n t a 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 m g t e f f 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 r e g c a p 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 l /d 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 r s k a s t 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 l l p 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 g d p 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 in f 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 l ib o r 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 12 5 international journal of banking and finance, vol. 10, iss. 2 [2013], art. 8 ijbf7-marina.indd the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 51 the input requirements of conventional and shariahcompliant banking1 mariani abdul majid universiti kebangsaan malaysia _____________________________________________________ abstract islamic banking activities are limited within the scope of shari’ah which is within the scope of socially responsible and ethical banking activities, different from that based on interest-based banking. this paper attempts to measure the input data required by shari’ah-compliant banking in comparison with conventional banking to estimate their relative effi ciencies and economies of and returns to scale. cost and output distance functions were estimated for a sample of banks in 10 countries which operate both types of banking. the results showed that shari’ah-compliant banking has higher input requirements relative to interestbased banking, but exhibit superior average effi ciency only in malaysia but inferior average effi ciency in cross-country analysis. there is little evidence of differences in economies/returns to scale between shari’ah and conventional banks. keywords: islamic banking, stochastic frontier analysis, effi ciency, frontier analysis, conventional banks. jel classifi cation: g21, g28. _____________________________________________________ 1. introduction the development of modern islamic banking arose from adherents to islam becoming conscious about rejection of the interest element in conventional banking. islamic banks which started to operate in the early 1960s were initially concentrated in the middle east before spreading to other regions such as asia and europe, due to demand mainly from the muslim communities. banks ijbf 1note from editors: this paper is one of three best papers selected by a review panel of three professors at a symposium held in november, 26-28, 2008 in melbourne, australia. the symposium was funded by the australian research council grant, 2007-2009/10, for research on islamic banking and finance. ht tp :// ijb f.u um .e du .m y 52 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 started to offer this as a choice to bank customers. although islamic banking is sometimes perceived as a limiting choice, it is actually broadening the banking choice. compared to conventional banking, islamic banking activities are limited within the scope of shari’ah hence, the mechanism involved in islamic banking is different from that on interest-based banking. this gives bank customers an alternative to interest-based banking. in addition, islamic banking is not viewed as threatening the existing business. instead, it opens opportunity for new business as its operation is within the scope of ethical banking activities (wilson, 2007). islamic banking services have now been offered by both full-fl edged islamic banks, as well as conventional banks that choose to operate islamic banking windows, and they can either be foreignor domestic-owned fi rms. as islamic banking has been in operation for over 47 years and is viewed as an alternative to interest-based banking, the performance of islamic banking needs to be assessed. moreover, as islamic banking is part of a country’s banking system, the performance of islamic banks may affect the soundness and stability of the banking system. furthermore, islamic banking infl uences the performance of conventional banks, if they choose to operate islamic banking windows in addition to conventional windows. hence, determination of the relative performance of islamic and conventional banks will help policy makers to devise policies in order to improve the performance of a country’s banking system as well as to provide some guidelines for managers of conventional banks with islamic banking windows to improve bank performance. in addition, the rising number of islamic banks has increased the competition between fullfl edged islamic banks and conventional banks. therefore, the determination of their relative performances will encourage both full-fl edged islamic and conventional bank managers to improve their performance in order to compete with each other. given the above issues, the aim of this paper is to measure the effi ciency of islamic banking and compare it with conventional banks, concentrating on the impact of operational characteristics. specifi cally, the fi rst objective is to compare the effi ciency of islamic banking relative to conventional banks in malaysia, focussing on the impact of operating characteristics. the second objective of the paper is to compare the effi ciency of islamic banks relative to conventional banks in countries operating islamic banking. 2. methodology a. the econometric specifi cation in achieving the fi rst objective, translog cost and output distance functions were applied to study the commercial banks to measure their effi ciency. the measured effi ciency of a fi rm is calculated as the difference between its observed ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 53 input and output levels and the corresponding optimal values (given a country’s fi tted frontier). given that islamic banks cannot charge or pay interest and are hence, likely to face higher capital costs2 and meet objectives other than profi t maximization, a cost function has been employed which allows the potential higher costs of capital faced by islamic banks to be controlled. furthermore, if the non-profi t oriented activities of islamic banks are carefully controlled for, it is reasonable to assume that islamic banks will try to minimize their costs of operation. in specifying the model, the intermediation approach, which has been widely employed in most bank studies (brown and skully, 2003; hassan, 2003; saaid, saiful, mansor and naziruddin., 2003, yudistira, 2004), and islamic and conventional bank studies (e.g., alshammari, 2003; el-gamal and inanoglu, 2005) was employed in this paper: see ariff, badar, shamser and taufi q. (2008). given this discussion, stochastic frontier analysis (sfa) will be employed in order to estimate a total cost function for malaysian commercial banks. a single-equation stochastic cost function model can be described as: inc n,t = (y n,t ,w n,t ,z n,t )+ n,t (1) where c n,t, ,y n,t, , w n,t , and z n,t are the observed total cost of production for the n-th fi rm at time t, a vector of outputs is y n,t , an input price vector is w n,t , and an exogenous factor vector is z n,t . the assumption of the composed error term is as below (aigneret, knox and peter 1977),  n,t = v n,t + u n,t (2) where v n,t represents random uncontrollable error and is assumed to be normally distributed with zero mean and variance, and is drawn from a one-sided distribution that is assumed to capture ineffi ciency. similar to many previous studies, u n,t is assumed to be drawn from a half-normal distribution with mean zero and variance (berger and mester, 1997, mester, 1996). v n,t and u n,t are independently distributed. given this assumption, the log likelihood for ineffi ciency is expressed in terms of the two variance parameters, which captures the variance of composed error and, which is a measure of the amount of variation originating from ineffi ciency relative to statistical noise (jondrow, knox lovell, ivan and peter, 1982). 2 v . 0 ,u tn 222 uv 2 as they cannot issue or hold interest-bearing loans or securities but use alternative contract arrangements such as musharaka (karim, 2001). however, as the available investment avenues using contracts are very limited, and most of them concentrate on short term investments, they may yield lower returns. in contrast, interest-based banks have wide choice of both shortand long-term investments thus potentially yield higher returns. ht tp :// ijb f.u um .e du .m y 54 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 maximum-likelihood estimates (mle) are obtained by estimating a translog cost function as below, after including environmental variables, imposing the standard assumption of homogeneity in input prices, and allowing for the composed error terms: (3) where, p k,n,t = w k,n,t w k,n,t and k=1,…,k, and s=1,…,k are indices for input prices; m=1,…m and j=1,…,m are indices for output prices; h=1,…,h is an index for environmental variables; while the greek letters (except v and u) represent unknown parameters to be estimated. output and input variables follows the existing literature (allen and rai, 1996; casu and girardone, 2002; mester, 1996), are normalized around their means and the values are in real 2000 myr. total costs (c) are defi ned as operating and fi nancial costs and are calculated as the sum of labour expenses, physical capital expenses, and either income paid to depositors for islamic banks or interest expense for conventional banks. input prices w1, w2, and w3 are the price of labour, price of fi nancial capital and price of physical capital, respectively. w1 is labour expenses divided by the number of full time workers, w2 is the amount of income paid to depositors divided by total deposits, and w3 is the physical capital expenses divided by the fi xed assets. bank outputs, are defi ned as the sum of total loans (y1), and total other earning assets (y2). the latter comprise deposits with other banks, securities and equity investments. standard symmetry is imposed to the second order parameters: and  ks =  sk and  mj =  jm . given the above model specifi cation and assumptions, a measure of cost effi ciency can be derived as the ratio of observed costs to predicted effi cient costs, which is theoretically equivalent to: ce n,t = exp(u n,t ) (4) these relative effi ciency measures range from one to infi nity with a score of one indicating full effi ciency. however, ce n,t relies on the unobservable ineffi ciency, u n,t . following jondrow, et al. (1982), the conditional expectation of u n,t given tnktntn wcc ,,,, ~ 1 1 sk, 1 1 1 1k k plnplnplnln tn,s,tn,k,.50tn,k,+, ~ k s k k k tnc m 1 ,,,,jm, m 1m m 1m ,,m lnlnlny 0.5 j tnjtnmtnm yy 1 11 , 1 1 ,,,,,, lnlnln k k k m m mk k k tpyp tnktnmtnk ttty m m m tnm 2 111 1 5.0,,ln tntn h h tn,h,h uvz ,, 1 ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 55 the observed value of the overall composed error term,  n,t , can be expressed as: (5) where,  is the standard normal cumulative distribution function and ø is the standard normal density function. failure to account for differences between bank groups may yield inappropriate conclusions about bank performance (bos and kool, 2006). therefore, differences in operating characteristics that may affect the effi cient level of costs or output in this paper have been controlled for, by including environmental factors directly in the function, hence the resulting effi ciency scores are net of the impact of environmental infl uences on effi cient input requirements. as a result, these effi ciency measures permit one to predict the ranking of fi rms under the assumption that fi rms operate in an equivalent environment. however, these exogenous factors are possibly an indicator of differences in effi ciency rather than differences in effi cient costs or outputs. the estimated economies of scale enable banks to identify potential costs savings if they change the operation scale which can be obtained by fi rst calculating the m output elasticities: (6) from which a scale elasticity can be calculated as: (7)  scale,n,t >1 ,  scale,n,t = 1,  scale,n,t <1, indicates economies of scale, constant returns to scale and diseconomies of scale respectively. banks that produce at constant returns to scale realise the lowest average costs in which any increase (decrease) in output will increase (decrease) costs proportionately. cost effi ciency, as mentioned above, measures how effi cient banks minimise inputs, given outputs. as effi ciency estimates obtained from different approaches should generate consistent estimates of effi ciency and effi ciency rankings, as well as give consistent results over time, an alternative method employing an output-oriented distance function, which estimates how effi cient banks transform inputs into outputs, is also applied. by using this function, the paper has also the benefi t of employing a quantity measure to identify bank inputs and outputs, thus avoiding possible problems leading to distorted and inaccurate price estimates that might occur given divergences in asset classifi cation among islamic and conventional banks, hence, potentially resulting in unreliable estimates of cost effi ciency. moreover, this function does not call for the strong behavioural assumptions of a profi t maximisation tn tntn tn tn ue ,,, , , 2 11 tmmy c tnm k k tnkmktnj m j jmtnm tn py 1 1 ,,,,, 1 ,,, , ~ lnlnln ln ,, m i tnm tnscale 1 ,, 1 ,, . ht tp :// ijb f.u um .e du .m y 56 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 or cost minimisation approach and is therefore appropriate for islamic banks as they have dual objectives of fulfi lling non-profi t obligations for the society and profi t or revenue maximisation for the depositors and shareholders. moreover, if behavioural objectives between islamic and conventional banks differ, the weaker behavioural assumptions of the output distance function approach may allow more consistent estimates of relative effi ciency. a production technology that transforms inputs into outputs can be represented by the technology set, which is the technically feasible combination of inputs and outputs (coelli, prasada and george, 1998; cuesta and orea, 2002; fare and primont, 1957). if the vector of k inputs, indexed by k is denoted by x=(x 1 ,x 2 ,…,x k ) and the vector of m outputs, indexed by m , is denoted by y=(y 1 ,y 2 ,…,y m ), the technology set can be defi ned as: (8) where and are the sets of non-negative, real k and m-tuples respectively. for each input vector, x, let p(x) be the set of producible output vectors, y, that are obtainable from the input vector x: p(x) = {y: (x,y) t)} (9) the output distance function can then be defi ned in terms of the output set, p(x) as: (10) the output distance function is defi ned as the maximum feasible expansion of the output vector given the input vector which is non-decreasing, positively linearly homogeneous and increasing in y, and decreasing in x (cuesta and orea, 2002). given an output distance function with two outputs and a given input vector, x, the production possibility set is the area bounded by the production possibility frontier (ppf), which indicates the maximum feasible output given x, and the y 1 and y 2 axes. if the output vector, y, is an element of the feasible production set, p(x), do(x,y)≤1, fi rms which produce on the ppf, d 0 (x,y) = =, thereby indicating technical effi ciency. in contrast, for a fi rm operating inside the ppf, d 0 (x,y) = <1, thereby indicating the proportion by which output is below potential output. this also illustrates that farrell is (1957) output-oriented measure of technical effi ciency, defi ned as the maximum producible radial expansion of the output vector, and can be represented as: oe 0 = 1/ d 0 (x,y) (11) yproducecanxryrxyxt mk ,,:), .)(:0.min, xp y yxdo ob oa kr mr ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 57 oe 0 increases with ineffi ciency and lies between one and infi nity. if y is located on the outer boundary of the production possibility set, oe 0 = 1, indicating effi ciency. on the other hand, if y is in the interior of the production possibility set, oe 0 >1 indicating ineffi ciency. following fare and primont (1957) and cuesta and orea (2002), and allowing for exogenous factors, the general form of a stochastic output distance function can be shown as follows: 1 = d 0 (y n,t ,x n,t , z n,t ,) h( n,t ) (12) where h( n,t ) = exp (u n,t + v n,t ), y n,t is a vector of outputs, x n,t is an input vector, z n,t is an exogenous factor vector and  is a vector of parameters. ineffi ciency is accommodated in the specifi cation of h (.) as n,t is a composed error term comprised v n,t which represents random uncontrollable error that affects the n-th fi rm at time t, and u n,t , which is assumed to be attributable to technical ineffi ciency. in order to facilitate estimation, the author followed the standard practice of imposing homogeneity of degree one in outputs on the distance function, which implies that d 0 (z,x, y) = d 0 (z,x,y),  > 0. by arbitrarily choosing the m-th output, the author can then defi ne and write: (13) from equation 13 and after assumingy* n,t = (y 1,n,t /y ,m,n,t ,y 2,n,t /y m,n,t , ...,y m-1,n,t / y m,n,t ). and rearranging terms yields the general form: (14) finally after assuming the standard translog functional form3 to represent the technology, the output distance can be represented as: (15) my 1 m o m o y yxzd y y xzd ,, ,, tntntntno tnm hzxyd y ,,, * , ,, ,,, 1 3 in the literature, the translog function is preferred in estimating a parametric distance function because it is fl exible, easy to calculate and permits the imposition of homogeneity (fuentes, emili and sergio, 2001). tnstnk k s sk k k tnm m m mtnk k k kotnm xxyxy ,,,, 1 , 1 * ,, 1 1 ,, 1 ,, lnln5.0lnlnln * ,,,, 1 1 , 1 * ,, * ,, 1 1 1 1 , lnlnlnln5.0 tnmtnk m m mk k k tnjtnm m m m j jm yxyy ht tp :// ijb f.u um .e du .m y 58 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 where, y* m,n,t = y m,n,t / y m,n,t , k=1,2,..k and s=1,2,..k are indices for inputs; m=1,2,…m and j=1,2,..m are indices for output. the selection of the input and output variables follows the existing literature (cuesta and orea, 2002; cuesta and zofío, 2005; iqbal, kizhanathan, and aigbe, 1999). the outputs y 1 , y 2 are loans and total other earning assets, and the inputs x 1 , x 2 , x 3 are labour deposits and capital (fi xed assets), respectively. x 1 is the number of full time workers, x 2 is total deposits including customer funding and short term funding, and x 3 is the total expenses on fi xed assets. it is noted that linear homogeneity in outputs is imposed using y 2 as a numeraire and these variables have been mean-corrected prior to estimation. the approach of jondrow, lovell, materov, and schmidt (1982) is followed to derive the log likelihood which is expressed in terms of the two variance parameters, following from equation 12, and given current model assumptions, an estimate of output distance can be derived as d 0 (y n,t ,x n,t ,z n,t, ) = exp(–  . equivalently an estimate of farrell output oriented effi ciency is obtainable as: (16) however, oe n,t relies on the unobservable ineffi ciency, u n,t ,. following jondrow et al. (1982), the conditional expectation of u n,t given the observed value of overall composed error term,  n,t can be expressed as: (17) where, ,  (.) is the standard normal cumulative distribution function and  (.) is the standard normal density function. given the estimated model, estimated scale elasticity can be calculated as the negative of the sum of the input elasticities (cuesta and orea, 2002): (18) scale n,t > 1, scale n,t < 1, and scale n,t = 1, are when a bank is operating with increasing returns to scale (irs), decreasing returns to scale (drs) and constant returns to scale (crs) respectively. the second objective of how islamic banks perform relative to conventional banks internationally is examined by employing a translog output distance function. the relative effi ciency and returns to scale of islamic and conventional banks have been investigated in countries that operate islamic banking namely malaysia, sudan, bangladesh, tunisia, jordan, lebanon, yemen, indonesia, bahrain and iran. except for sudan and iran which only 222 uv and 222 / uvu . )exp( ,,, ,,, 1 0 , tntntn zxyd oe tn a tn atntn atn atn ue , ,, , , 1 21a k k tnk tnktnmo tn x xyd scale 1 ,, ,,,, , ln ,ln ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 59 operate islamic banking, banks from other countries operate both islamic and conventional banking. a common frontier with country-specifi c environmental variables is estimated after allowing for country specifi c differences in estimated ineffi ciency and the analysis puts emphasis on the impact of operating characteristics, including islamic banking and country-specifi c conditions on the relative outputs of banks. the frontier is controlled for variations in economic and regulatory environments between countries that may justify differences in effi ciency, by including country-specifi c variables directly in the distance function, and also allowed country dummies to directly infl uence output ineffi ciency. these country dummy variables simultaneously capture other country-specifi c environmental conditions and determine relative effi ciency between countries. this implies that the resulting effi ciency scores are net of the impact of controlled for environmental infl uences on effi cient input requirement, and the differences in these scores are directly infl uenced by country-specifi c ineffi ciency distributions. as a result, these effi ciency measures enable one to determine how fi rms are ranked under the assumption that fi rms operate in an equivalent environment, while at the same time measuring how bank effi ciency in one country differs from another. employing the output distance function approach, the cross-country analysis specifi es battese and coelli's (1995) truncated normal sfa model with the mean of the truncated normal distribution made an explicit function of country dummy variables. the illustration is in equation 18 and the formulation of the model detailed in (coelli, 1996) is followed. previous studies that employ the intermediation approach found that equity is signifi cant in defi ning bank output but many (girardone, philip and edward, 2004; kasman and yildirim, 2006) include it either as an environmental variable or a netput (fully interactive with input and output). nevertheless, in fi nancing the operation of banks, equity capital is an alternative to deposits and inter-bank borrowings (bonaccorsi di patti and hardy, 2005). furthermore, islamic banks that apply an equity participation principle rely heavily on their equity to fi nance loans (metwally, 1997). therefore, it is appropriate that equity is considered as part of bank inputs for studies employing the intermediation approach. the author therefore, included equity as an input, because of both its role in islamic banking and because all banks can potentially raise funds to fi nance their loans through equity, rather than deposits. the specifi cation therefore extends the standard intermediation model by including two outputs, y 1 , y 2 representing loans and total other earning assets, and three inputs, x 1 , x 2 , x 3 representing total operating expense, deposits, measured by total deposits including customer funding and short term funding, and equity, measured by total equity. thus, v n,t is assumed to be normally distributed with zero mean and variance and independently distributed of the u n,t , where u n,t ≥ 0 is assumed to be drawn from a truncation (at zero) of the normal distribution with mean, em n,t and variance ,   is a parameter to be estimated, f=1,2...,f is an index for countries, and c is a country dummy. hence, given the absence of a constant 2 v 2 u ht tp :// ijb f.u um .e du .m y 60 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 in equation 18, each country  is estimated to have ineffi ciency drawn from a distribution with mean   , that is truncated at zero. the parameters in the translog function as defi ned in equation 14, the composed error parameters , and the estimated means of the country specifi c ineffi ciency distributions (  ) specifi ed in equation 18 are estimated simultaneously using maximum likelihood estimation (mle) techniques. (18) given the model assumptions and following equation 18, an estimate of output distance can be derived as d 0 (y n,t ,x n,t ,z n,t ,  ) = exp() . equivalently, an estimate of farrell (1957)’s output oriented effi ciency is obtainable as: (19) oe n,t relies on the unobservable ineffi ciency, u n,t , hence following the approach of battese and coelli (1995) and frame and coelli (2001) to estimate the unobservable ineffi ciency, u n,t , the conditional expectation of u n,t given the observed value of the overall composed error term,  n,t , can be expressed as: (20) following cuesta and orea (2002), returns of scale for the banks in the sample can be estimated using the estimated scale elasticity which is calculated as the negative of the sum of the input elasticities: (21) where if scale n,t > 1, and scale n,t < 1, scale n,t = 1, when a bank is operating at irs, drs and crs respectively. b. data and input variables for malaysian banks, data on 33 banks were drawn from bureau van dijk’s (bvd’s) bankscope database for the period 1996-2002 and were verifi ed against the banks’ annual reports. the data are expressed in malaysian ringgit (myr) and are adjusted for infl ation using the malaysion gdp defl ator, which was extracted from imf (2004). the number of full-time workers and ownership information is taken from the central bank of malaysia (2002) and association of banks in malaysia (various years). as some banks have incomplete information, this has resulted in an unbalanced panel of 168 observations. tnff f f tn cem ,, 1 , )exp( ,,, , ,,, 1 0 , tn tntntn zxyd oe tn *,, **,, 2 *,, ,, 1 15.01exp exp tntn tntntntn tntn m mm ue k k tnk tnktnmo tn x xyd scale 1 ,, ,,,, , ln ,ln 222 uv and 222 / uvu ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 61 mergers during the sample period have caused a marked reduction in the number of malaysian commercial banks. over this period, ten mergers and acquisitions took place: two in 1999, one in 2000, six in 2001 (involving 14 banks) and one in 2002. given these trends, each pre-merger commercial bank is included as a separate bank and these banks are assumed to have merged into one of the premerger banks. for cross-country analysis, data on 23 islamic and 88 conventional banks from 10 countries that operate islamic banking were drawn from the bankscope database for the period 1996-2002 resulting in an unbalanced panel of 558 observations expressed in constant 2000 us dollars.4,5 c. environmental environments focussing on malaysian banks while applying the cost function, the fi rst operating environment variable is an indicator of loan quality, and is proxied by the ratio of the non-performing loans (npl)-to-total loans (berger and mester, 1997; clark, 1996, girardone et al., 2004; mester, 1996; williams and nguyen, 2005). the second operating environment variable is measured by the equity-tototal assets ratio (berger and mester, 1997; clark, 1996; girardone, et al., 2004; mester, 1996; williams and nguyen, 2005). the remaining environmental variables are dummy variables that are designed to capture potential differences in bank characteristics, and operating environment that may infl uence costs. the dummy variable indicating fullfl edged islamic banks, is to control for the potential impact of full-fl edged islamic banking on bank costs. as changes in bank scale should be captured through the impact of output growth on estimated costs, the impact of mergers will be net of the impact of changes in bank scale attributable to the merger. a dummy for observations in 1998 is included to control for the east asian fi nancial crisis. the author considered including a foreign-owned dummy, for banks with more than 50% foreign ownership. however, while almost all domestic banks operate an ibs window relatively few foreign banks do. the author therefore, chose to interact a foreign dummy variable with a dummy variable for conventional banks that operate ibs windows and include the resulting set of dummy variables. therefore, the model includes dummy variables for foreign banks without ibs, foreign banks with ibs, domestic banks with ibs, and leaves domestic banks 4in the estimation, all input and output variables were normalized around their means and the linear homogeneity in outputs was imposed using the output y2 as a numeraire. 5all data employed in this analysis were converted into constant international dollars according to the purchasing power parity hypothesis (lozano-vivas, jesus and jose, 2002). ht tp :// ijb f.u um .e du .m y 62 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 without ibs as the base case measured in the constant.6 finally, z 9 provides a dummy variable indicating public ownership, and is expected to have a positive sign indicating higher costs.7 turning to the operating environments for malaysian banks employing output distance function, the fi rst operating environment variable is loan quality, the dummy variable indicating full-fl edged islamic banks is to control for the potential impact of full-fl edged islamic banking on bank output. the model also includes a dummy variable for foreign banks, foreign banks with ibs and all banks with ibs, leaving conventional domestic banks without ibs as the base case measured in the constant, where banks with ibs are conventional banks offering islamic banking products through a separate islamic banking window. a dummy variable for observations in 1998 is included to control for the east asian fi nancial crisis. finally, given that some banks have gone through mergers, one can control for this effect by using a merger dummy variable. the author also tests for the potential effects of individual mergers, fi nding that the dummy is signifi cant for three individual mergers, namely merger 1, merger 2 and merger 3.8 for cross-country analysis, in order to identify a common frontier, variables describing distinctive features of the economy, the banking industry as well as the geography of each country were identifi ed. these variables are grouped into three categories. the fi rst category includes macroeconomic conditions, and consists of a measure of population density, per capita income, density of demand (deposits per kilometer squared) and real gdp growth. these indicators explain the macro conditions under which banks operate. population density is measured by the ratio of inhabitants per square kilometre, and it is expected that with high population density, the retail distribution of banking services becomes less costly. high per capita income, measured by gross national income (gni) per inhabitant, is usually associated with countries having a mature banking environment, and thus, competitive interest rates and profi t margins which lower banking costs and increase bank outputs. density of demand is measured as total deposits per square kilometre. a less concentrated demand for banking services is costly because demand is more dispersed. as a result, bank customers are less 6as all islamic banks in the sample are domestically owned, and by defi nition are not conventional banks, the impact of islamic banking on costs is also relative to the base case of a domestic bank that does not operate ibs. 7publicly-owned banks are defi ned as banks with more than 50 percent government ownership through its agencies such as the employees provident fund (epf) and permodalan nasional berhad (pnb). by defi nition, no foreign banks are included in the publicly owned category. 8merger 1, 2, 3 refer to mergers between oriental bank and eon bank, between chung khiaw bank and uob bank, and between international bank malaysia, sabah bank and multi-purpose bank respectively. ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 63 informed and banks tend to achieve lower output.9 finally, real gdp growth is expected to increase bank outputs due to increasing economic activities. the second group of environmental variables identifi es differences in banking structure and therefore provides measures of both banking concentration and the intermediation ratio. the concentration ratio is defi ned as the ratio of the total assets of the fi rst three largest banks in a country to total banking assets. higher concentration may be associated with higher or lower output. if higher concentration of banks is a result of market power, then the banks may become ineffi cient in producing outputs (leibenstein, 1966). on the other hand, if higher concentration is a result of effi ciency, then bank costs are reduced and bank outputs increase (demsetz, 1973). in order to control for differences in regulation or allow factors that may affect the ability to convert deposits to loans among banking industries, the intermediation ratio, as measured by the loan-todeposits ratio is employed. it is expected that the higher the intermediation ratio, the higher bank outputs will be. thus, the fi rst two groups of variables follow closely those of dietsch and lozano-vivas (2000), and carvallo and kasman (2005). the fi nal group of environmental variables includes proxies for accessibility of banking services. the proxy variables are roads paved and telephone lines per 100 inhabitants. roads paved is the percentage of road being paved in total roads, and is expected to positively impact bank outputs. finally, the author expects that easier access to telephone lines will also increase potential bank outputs. one fi nal control variable is a dummy variable indicating whether a bank is an islamic bank. inclusion of this variable allows the author to test whether full-fl edged islamic banks have a different operating environment from conventional banks. therefore, a dummy variable is included in the model to capture for this difference, but no a priori assumption is made due to mixed results in the literature on the direction of the infl uences of islamic banking on ineffi ciency (al-jarrah and molyneux, 2005; el-gamal and inanoglu, 2005; mokhtar, naziruddin and syed m. al-habshi, 2006) and none has assumed islamic banking to infl uence potential bank output. the author also noted that while this modelling assumption maintains the assumption that adherence to shariah causes a shift in potential output obtainable from given inputs, it could also be argued that any difference in output between conventional and islamic banks is evidence of differences in effi ciency. however, the author adopts this approach because it is believed that the restrictions imposed by shariah require islamic banks to operate a modifi ed banking technology that is not equivalent to that of conventional banks. 9countries with population concentrated in small habitable area(s) warrant careful judgement with regard to these results. ht tp :// ijb f.u um .e du .m y 64 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 3. results and interpretation a. efi ciency table 1 maximum likelihood estimates for parameters of the environmental factors:1996-2002 coeffi cient parameters estimated value a std. error a. costs function for malaysian banks ζ 1 loan quality 0.309*** 0.103 ζ 2 equity/asset ratio -0.736*** 0.229 ζ 3 islamic bank 0.150*** 0.041 ζ 4 foreign without ibs -0.218*** 0.028 ζ 5 financial crisis -0.048** 0.023 ζ 6 merged bank 0.108*** 0.026  lambda 1.501*** 0.439 σ sigma 0.096*** 0.014 b. output distance function for malaysian banks ζ 1 loan quality 0.380*** 0.048 ζ 2 islamic bank 0.066*** 0.021 ζ 3 foreign owned bank -0.140*** 0.027 ζ 4 foreign with ibs 0.118*** 0.031 ζ 5 financial crisis -0.027*** 0.012 ζ 6 merged bank 1 0.083*** 0.035 ζ 7 merged bank 2 0.097*** 0.034 ζ 8 merged bank 3 0.063* 0.038 σ 2 sigma-squared 0.005 0.001 γ gamma 0.826*** 0.143 (continued) ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 65 for malaysian banks employing cost function, recalling that = u /  v , table 1 shows the highly signifi cant estimate of 1.501 implies that estimated deviation from the frontier is due mainly to ineffi ciency rather than statistical noise. loan quality (z 1 ) is positive as predicted and indicates that the lower output quality (higher the npl-to-loan ratio), the higher the cost incurred by banks, which may refl ect higher monitoring costs. the equity-to-asset ratio (z 2 ) has a negative relationship with costs, indicating that as the equity-to-asset ratio increases, costs are lower relative to those banks that depend more on deposits. the positive coeffi cient for the islamic bank dummy (z 3 ) indicates that full-fl edged islamic banks are found to have costs that ceteris paribus are 15.0% higher than for other banks. this may result from constrained opportunities in terms of investments and limited expertise in islamic banking. merged banks coeffi cient parameters estimated value a std. error c. output distance function for banks internationally ζ 1 islamic bank 0.141*** 0.022 ζ 2 density of population 2.82x10-4*** 7.83x10-5 ζ 3 density of demand -0.035*** 0.008 ζ 4 telephone lines 0.015*** 0.003 δ 1 malaysia -0.541*** 0.096 δ 2 sudan 0.537*** 0.082 δ 3 bangladesh -0.366*** 0.097 δ 4 tunisia 0.210*** 0.047 δ 5 jordan -0.047 0.095 δ 6 lebanon 0.112*** 0.041 δ 7 yemen 0.412*** 0.083 δ 8 indonesia 0.212*** 0.053 δ 9 bahrain -0.353*** 0.121 δ 10 iran -0.987* 0.555 σ2 sigma-squared 0.029*** 0.002 γ gamma 0.491*** 0.076 a *,**,*** signifi cance at 90, 95 and 99% confi dence levels, respectively. source: extracted from (abdul-majid, 2008) table 1 maximum likelihood estimates for parameters of the environmental factors:1996-2002 ht tp :// ijb f.u um .e du .m y 66 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 (z 4 ) are found to have costs that are 10.8 percent higher, after controlling for other variables.10 the dummy variable for the fi nancial crisis (z 5 ) is positive, indicating that costs fell by 4.8% in 1998 after controlling for other variables. finally, foreign banks without ibs windows (z 6 ) are found to have costs that are 21.8% lower than the combined group of all domestic banks, publicly owned banks, and foreign owned banks with ibs windows. focussing on table 2, the cost effi ciency of malaysian commercial banks is on average 1.066, and ranges from 1.019 to 1.217. the yearly average as well as the range of the effi ciency scores has increased. the trend in effi ciency suggests a decline in average effi ciency over the sample period, but also the presence of a group of fi rms that were steadily slipping further away from the cost frontier. thus, average effi ciency deteriorated from 1.064 in 1996 to 1.075 in 2002 and the maximum effi ciency score increased from 1.142 in 1996 to 1.206 in 2002. this may indicate that there are high gains achieved by bestpractice banks but declines in effi ciency as other banks struggle to keep up with best practice. the effi ciency scores is judged against an effi cient frontier, which for example allows full-fl edged islamic banks to have 15% higher costs and requires foreign banks without ibs windows to have 21.8% lower costs. these results can be compared to the previous literature: islamic banks are found to have no difference with conventional banks in malaysia (abdul-majid, mariani and fathin, 2005; mokhtar, et al., 2006), but are equall if not more effi cient in turkey (el-gamal and inanoglu, 2005), are more effi cient in arabian countries (al-jarrah and molyneux, 2005) and in gcc countries (alshammari, 2003), when compared to conventional banks. these differences may potentially be due to the absence of environmental variables in some previous studies employing the intermediation approach, different input and output specifi cations, and cross-country differences in islamic banking that may infl uence relative cost effi ciency.11 concentrating on the estimated output distance function parameters for malaysian banks as reported in table 1b, recalling that , the highly signifi cant estimate of 0.826 for this parameter suggests that the portion of technical ineffi ciency in total variance is high. thus, the estimated deviation from the frontier is mainly due to ineffi ciency rather than statistical noise. the estimated coeffi cients of all variables have the expected signs. loan quality (1) is positive as predicted, and indicates that lower output quality (higher nplto-loan ratio) reduces output, thereby refl ecting the higher input requirement needed to monitor default loans. 222 / uvu 10berger and humphrey (1997) noted that some mergers improve cost effi ciency whereas others worsen their performance. orea (2002) found that merged banks have negative effi ciency change in contrast to the unmerged banks in the initial period of merger activities. 11for example, islamic banks in other countries may employ more equity-based fi nancing rather than debt-like fi nancing which is more common in malaysia. ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 67 b. malaysian banks using output distance function descriptive statistics: all banks average 1.042 1.061 1.054 1.052 1.060 1.050 1.060 1.055 standard deviation 0.023 0.027 0.034 0.026 0.052 0.037 0.044 0.036 minimum 1.016 1.015 1.014 1.015 1.011 1.015 1.015 1.011 maximum 1.104 1.109 1.161 1.123 1.220 1.144 1.211 1.220 average effi ciency of conventional, conventional with ibs and islamic banks all banks 1.042 1.061 1.054 1.052 1.060 1.050 1.060 1.055 without ibs 1.043 1.062 1.056 1.055 1.069 1.063 1.069 1.060 with ibs 1.041 1.060 1.055 1.052 1.054 1.041 1.052 1.052 islamic 1.037 1.066 1.017 1.028 1.061 1.062 1.086 1.057 c. banks in 10 countries using output distance function descriptive statistics: all banks average 1.087 1.106 1.102 1.106 1.102 1.120 1.112 1.105 standard deviation 0.121 0.158 0.173 0.159 0.151 0.173 0.167 0.158 minimum 1.014 1.012 1.011 1.010 1.014 1.014 1.019 1.010 maximum 1.756 1.949 2.352 1.918 1.743 1.882 2.114 2.352 average effi ciency of conventional and islamic banks conventional banks 1.076 1.076 1.081 1.081 1.076 1.096 1.094 1.082 islamic banks 1.187 1.289 1.195 1.204 1.214 1.215 1.200 1.215 1996 1997 1998 1999 2000 2001 2002 all years a. malaysian banks using cost function descriptive statistics: all banks average 1.064 1.057 1.064 1.071 1.075 1.056 1.075 1.066 standard deviation 0.029 0.026 0.033 0.039 0.048 0.036 0.041 0.037 minimum 1.033 1.022 1.025 1.026 1.02 1.019 1.024 1.019 maximum 1.142 1.124 1.155 1.181 1.217 1.157 1.206 1.217 average effi ciency of conventional, conventional with ibs and islamic banks all banks 1.064 1.057 1.064 1.071 1.075 1.056 1.075 1.066 without ibs 1.071 1.057 1.066 1.082 1.078 1.057 1.083 1.071 with ibs 1.061 1.057 1.062 1.068 1.076 1.057 1.072 1.065 islamic 1.058 1.056 1.072 1.061 1.062 1.042 1.059 1.057 table 2 effi ciency estimates for banks and by types, 1996-2002 source: extracted from (abdul-majid, 2008). ht tp :// ijb f.u um .e du .m y 68 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 the positive estimate for 2 implies that full-fl edged islamic banks are found to have outputs that ceteris paribus are 6.6% lower than other banks and this may be due to constrained opportunities in terms of investments and limited expertise in islamic banking. the coeffi cient for foreign-owned banks is negative, indicating that output increases by 14.0% relative to domestic banks. however, foreign-owned banks with ibs (z 4 ) are found to have potential output that is 11.8% lower than foreign banks without ibs. the coeffi cient for the fi nancial crisis dummy variable (z 5 ) is negative, indicating that output increased by 2.7% in 1998 after controlling for other variables. this fi nding is consistent with the reactions of banks towards the fi nancial crisis, which was to lay off substantial number of workers and to cut other operating expenses. the individual mergers (z 6 , z 7 , z 8 ) are found to be associated with output that is 8.3%, 9.7% and 6.3% lower respectively, after controlling for other variables. the effi ciency estimates of malaysian commercial banks using output distance function as shown in table 2b is on average 1.055, and ranges from 1.011 to 1.220, hence on average, banks only produce 94.8%12 of the output they could produce if they operated on the effi cient frontier. the effi ciency scores demonstrate that while there is little variation in the estimated effi ciency once differences in the environmental variables are controlled for. in other words, if effi ciency is judged against an effi cient frontier, which for example, allows full-fl edged islamic banks to have 6.6 percent lower output, it should be expected that the resulting effi ciency scores exhibit small difference across bank types. the yearly average and the range of the effi ciency scores have risen and it implies a deteriorating in average effi ciency over the sample period, but also the existence of a group of banks that were steadily deviating from the output frontier. hence, average effi ciency worsened from 1.042 in 1996 to 1.060 in 2002 and the maximum effi ciency score deteriorated from 1.104 in 1996 to 1.211 in 2002. the author fi nally compares the output distance function with the results the cost function in order to check the consistency of results. with the cost function approach, slightly higher average ineffi ciency estimates of 1.066 percent are found as compared to 1.055 when using an output distance function. on balance however, the author believes that an output distance function approach is a better method because the behavioural assumptions being made with the output distance function are less likely to create biases when jointly evaluating islamic and conventional banks, and this approach also allows the author to avoid the further potential pitfall associated with price endogeneity. concentrating on the estimated output distance function parameters for cross-country analysis as reported in table 1c,13 recalling that, , the highly signifi cant estimate of 0.491 for this parameter, suggests that the 12oe=(1/ 1.055)100. 13the author notes that a log likelihood ratio test for the joint signifi cance of the 6 parameters related to equity is 17.98, thus the author can reject the null hypothesis that these parameters are jointly insignifi cant at the 99 percent confi dence level. 222 / uvu ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 69 14bank specifi c loan quality and merger dummy variables were also found to be statistically insignifi cant when they were included in the distance function. 15the fi nding is consistent with cost function studies in which higher population density contributes to an increase in banking costs in france and spain (dietsch and lozanovivas, 2000), and latin american and caribbean countries (carvallo and kasman, 2005). 16all countries in the sample are developing economies except for bahrain (world bank, 2007). estimated deviation from the frontier is equally due to both ineffi ciency and statistical noise. besides the statistically signifi cant islamic bank dummy variable, the only signifi cant country-specifi c environmental variables are density of population, density of demand, and telephone lines per 100 inhabitants. many country-specifi c variables become insignifi cant when country dummy variables are included in the model, thereby suggesting that these factors serve as proxies for cross country differences in bank effi ciency, rather than legitimate determinants of potential output.14 the islamic bank dummy (z 1 ) has a positive coeffi cient, indicating that full-fl edged islamic banks are found to have potential effi cient outputs that ceteris paribus are 14.1 % lower than other banks. therefore the results suggest a systematic reduction in potential output that can be attributed to islamic banking, which may result from constrained opportunities in terms of investments and limited expertise in islamic banking. however, because the estimated model effectively assumes that the reduced outputs associated with islamic banking result from legitimate differences in operating environment that reduce potential output, the effi ciency scores reported below for islamic banks must be carefully interpreted as they net out this impact. in contrast to expectations, the sign of the coeffi cient of the population density variable (z 2 ) is positive indicating that, ceteris paribus countries with high population density have lower bank output.15 a possible explanation for this fi nding is that in non-price bank competition, banks may open branches in large cities, in which real estate and labour costs are high, for strategic reasons, and thereby reduce their potential outputs (dietsch and lozano-vivas, 2000). as expected, lower density of demand (z 3 ), tends to increase expenses thereby, limiting potential output. the fi nding of reducing potential output is consistent with (dietsch and lozano-vivas, 2000) and (carvallo and kasman, 2005), which found that lower density of demand raises bank costs, and hence reduces effi ciency. finally, in contrast to the a priori assumption, the positive sign of telephone lines per 100 inhabitants’ variable (z 4 ) indicates that greater availability of telephone lines decreases bank outputs. this is possibly because most countries in the sample are developing economies16 in which electronic communications including phoneand internet-banking are not fully developed. hence, telephone usage may raise relative bank costs within the sample of countries. ht tp :// ijb f.u um .e du .m y 70 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 table 1c demonstrates that the country dummy variables illustrate systematic and signifi cantly differences in the relative ineffi ciency of banks across countries. thus, for example,  jordan is found to be insignifi cantly different from zero, thereby suggesting that ineffi ciency for jordanian banks is drawn from a standard half-normal distribution. however, banks in malaysia, bangladesh, bahrain17 and iran are found to have   <0 and hence, ineffi ciency in these countries is estimated as being drawn from truncated normal distributions with lower expected ineffi ciency than in a half normal distribution. in contrast, sudan, tunisia, lebanon, yemen, and indonesia all have   >0, and hence are estimated to have higher expected ineffi ciency than that drawn from a halfnormal distribution, with given variance . furthermore, table 1c suggests that while iranian banks have on average the best output performance, sudanese banks experience the worst output performance. this is consistent with two previous dea studies, which fi nd that iranian banks are among the most effi cient banks (brown, 2003; brown and skully, 2003) and sudanese banks are among the least effi cient banks (brown, 2003).18 the parameters suggest a clear hierarchy of estimated effi ciency across countries, with higher indicating greater ineffi ciency. table 2c reports the estimated effi ciency of all, conventional and islamic banks on average for cross country analysis, respectively. the effi ciency of all banks is on average 1.105, and ranges from 1.010 to 2.352. the yearly average as well as the range of the average effi ciency scores, has only slightly increased over time. thus, average effi ciency deteriorated from 1.087 in 1996 to 1.112 in 2002. the trend in both conventional and islamic banks suggests only a slight decline in average effi ciency over the sample period. hence, the conventional bank average effi ciency score increased from 1.076 in 1996 to 1.094 in 2002 and the islamic bank average effi ciency score increased from 1.187 in 1996 to 1.200 in 2002. across all countries, the average conventional and islamic bank effi ciency measures are 1.082 and 1.215, respectively. this suggests that on average, even after having netting out the 14.1% lower output associated with islamic banking, potential output of conventional banks is only 8.2 % higher than actual output, while for islamic banks this difference is 21.5%. sudan and yemen, which have only islamic banks in the sample, have extremely low average estimated effi ciency, even after netting out the impact of the statistically signifi cant environmental characteristics and islamic banking. put differently, while the results do clearly demonstrate a signifi cant 14.1% decrease in potential output attributable to islamic banking, the further particularly poor performance 17al-jarrah and molyneux (2005) also found that bahrain is relatively effi cient when compared to jordanian banks. 18 even within sudanese banks, wide ineffi ciency difference exists (hussein, 2004). 2 u ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 71 of islamic banks in sudan and yemen must be attributed to country specifi c banking ineffi ciency.19 the author fi nally emphasized that because the methodology assumes that differences in operating environment infl uence potential output rather than effi ciency, the resulting effi ciency estimates should in principle be interpreted as allowing for legitimate difference in potential output associated with compliance with shari'ah. therefore, as argued by (coelli, sergio and elliot, 1999), as this approach nets out the impact of operating environments, it provides a measure of managerial effi ciency. thus, based on this argument, islamic banks are substantially more effi cient in tunisia and marginally more effi cient in malaysia, but less effi cient in all other countries where both islamic and conventional banks operate. however, this interpretation is dependent on the assumption that all of the reduced output of islamic banks is attributable to differences in technology rather than systematically greater ineffi ciency amongst islamic banks. these results can be compared to the previous literature that does not allow for exogenous variables in either the frontier or as an infl uence on ineffi ciency: islamic banks are found to be no different with conventional banks in malaysia (abdul-majid, mariani, nor ghani and fathin, 2005; mokhtar, et al., 2006), and equally if not more cost effi cient in turkey (el-gamal and inanoglu, 2005). modelling for bank types of the islamic bank, commercial, investment banks, country dummy, assets, liquidity, concentration ratio, and market share to directly infl uence ineffi ciency effects in arabian countries, islamic banks are found to be more cost effi cient (al-jarrah and molyneux, 2005). controlling for loan quality and capital in the function and modelling for bank type, country dummy, assets, liquidity, concentration ratio, and market share to directly infl uence ineffi ciency effects in arabian countries using profi t function, islamic banks are also more effi cient (al-jarrah and molyneux, 2005). alshammari (2003) also found relatively effi cient islamic banks in gcc countries when loan quality and capital are included in the function, and bank type and country dummies are assumed to directly infl uence ineffi ciency. the differences in results may potentially be due to different environmental variables in the function, different input and output specifi cations, and cross-country differences in islamic banking operation that may infl uence relative effi ciency.20 b. returns to scale table 3 provides fi rm specifi c scale economy estimates for all banks and by bank types using the three methodology. using cost function, table 3a demonstrates the range of the estimated scale economies is between 0.911 and 1.218 and this is consistent with the previous literature (carvallo and kasman, 2005; orea, 2002). on average, these estimated scale economies have declined from 1.066 19effi ciency estimates by country is available from the author by request. 20for example, islamic banks in countries other than malaysia may have a higher percentage of equity-based fi nancing which has been controlled for in this study. ht tp :// ijb f.u um .e du .m y 72 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 1996 1997 1998 1999 2000 2001 2002 all years a. malaysian banks using cost function descriptive statistics: all banks average 1.066 1.061 1.059 1.042 1.026 1.026 1.025 1.043 standard deviation 0.036 0.042 0.041 0.040 0.053 0.039 0.049 0.048 minimum 0.990 0.973 0.965 0.944 0.925 0.936 0.911 0.911 maximum 1.115 1.140 1.150 1.166 1.218 1.084 1.104 1.218 average return to scale of conventional, conventional with ibs and islamic banks all banks 1.066 1.061 1.059 1.042 1.026 1.026 1.025 1.043 without ibs 1.070 1.080 1.073 1.054 1.032 1.013 1.015 1.045 with ibs 1.064 1.056 1.054 1.038 1.027 1.038 1.038 1.045 islamic 1.051 1.045 1.056 1.023 0.992 0.992 0.980 1.010 b. malaysian banks using output distance function descriptive statistics: all banks average 1.018 1.017 1.004 0.989 0.974 0.969 0.967 0.990 standard deviation 0.035 0.032 0.038 0.035 0.047 0.040 0.046 0.044 minimum 0.943 0.945 0.912 0.894 0.869 0.880 0.856 0.856 maximum 1.062 1.061 1.081 1.067 1.092 1.034 1.051 1.092 average return to scale of conventional, conventional with ibs and islamic banks all banks 1.018 1.017 1.004 0.989 0.974 0.969 0.967 0.990 without ibs 1.015 1.030 1.006 0.988 0.968 0.957 0.957 0.985 with ibs 1.020 1.012 1.003 0.990 0.978 0.977 0.975 0.993 islamic 1.016 1.013 1.004 0.989 0.972 0.963 0.945 0.978 c. banks in 10 countries using output distance function descriptive statistics: all banks average 1.045 1.044 1.040 1.032 1.025 1.023 1.022 1.034 standard deviation 0.021 0.022 0.023 0.025 0.020 0.023 0.023 0.024 minimum 0.989 0.995 0.996 0.945 0.983 0.984 0.981 0.945 maximum 1.093 1.117 1.128 1.106 1.097 1.103 1.096 1.128 average return to scale of conventional and islamic banks conventional banks 1.044 1.040 1.035 1.027 1.021 1.018 1.019 1.030 islamic banks 1.061 1.066 1.065 1.054 1.040 1.040 1.036 1.052 table 3 return to scale for all, islamic and conventional banks note: a: no mergers between islamic banks have occurred during the sample period. if return to scale >,< or =1, there are increasing return to scale; decreasing return to scale or constant returns to scale respectively. source: extracted from (abdul-majid, 2008) ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 73 in 1996 to 1.025 in 2002. similarly, within all of the bank types summarized, very moderate economies of scale and a slight downward trend in estimated scale economies is evident. thus, there is little evidence for a difference in scale economies across the groups identifi ed in table 3a. moreover, even though fullfl edged islamic banks are the only type with average economies of scale less than one in any year, this result is also consistent with the broader fi nding that most banks in the sample appear to operate at or near crs.21 table 3b illustrates the average estimated return to scale is 0.990 for malaysian banks using output distance function, thereby indicating the presence of mild decreasing return to scale. the range of estimated returns to scale is between 0.856 and 1.092, and is consistent with the previous output-oriented literature (cuesta and orea, 2002). on average, this estimated scale elasticity has decreased from 1.018 in 1996 to 0.967 in 2002. likewise, within all bank categories summarised in table 3b, very mild decreasing returns to scale and a slight downward trend in estimates is observed. thus, there is little evidence for a difference in returns to scale across the bank types identifi ed in table 3b.22 using both methods however, banks experience almost constant returns to scale. table 3c provides fi rm specifi c returns to scale estimates cross –country analysis for all, conventional and islamic banks on average. estimated returns to scale averages 1.034 for all banks, ranges between 0.945 and 1.128, and is consistent with the previous literature (abd karim, 2001; carvallo and kasman, 2005; cavallo and rossi, 2001). on average, these estimated returns to scale have declined from 1.045 in 1996 to 1.022 in 2002. the average estimated returns to scale for conventional banks is lower (1.030) than for islamic banks (1.052) and this applies to all countries except for malaysia and jordan. this suggests that generally a larger scale of operation will be useful if islamic banks wish to eliminate disadvantages attributable to their relatively small size. however, there is little evidence of substantial returns to scale to be gained, nor is there substantial difference in potential returns to scale between conventional and islamic banks.23 the trend for both conventional and islamic banks also suggests a decline in average returns to scale over the sample period. hence, conventional bank average returns to scale declined from 1.044 in 1996 to 1.019 in 2002 and islamic bank average returns to scale declined from 1.061 in 1996 to 1.036 in 2002. compared to other countries, sudanese banks exhibit relatively strong returns to scale, which is consistent with the very small bank size in this country. this is consistent with kasman (2005) who found economies of scale in small-sized banks in poland and the czech republic. 21yudistira (2004) found that small and medium-sized islamic banks in most countries have diseconomies of scale but alshammari (2003) found that bank type has no effect of economies of scale in gcc countries. 22yudistira (2004) found that small and medium-sized islamic banks in most countries have diseconomies of scale but alshammari (2003) found that bank type has no effect of economies of scale in gcc countries. 23alshammari (2003) found almost constant returns to scale in banks (including islamic banks) in gcc countries and no difference across bank types. however, yudistira (2004) found that small and medium-sized islamic banks in most countries have diseconomies of scale. ht tp :// ijb f.u um .e du .m y 74 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 4. conclusions the aim of this paper is to examine the effi ciency and economies of scale of islamic banks relative to conventional banks using sfa. operating characteristics such as shari’ah compliant banking could capture validated differences in costs or systematic differences in effi ciency. similar to cost function approach, using output distance function on malaysian banks, higher input requirements for full-fl edged islamic banks relative to average banks have been found. in crosscountry analysis, having netted out the 14.1 % lower output, the potential output of conventional banks is only 8.2 % higher than actual output, while for islamic banks this difference is 21.5 %. however, as these effi ciency estimates are net of the measured effect of islamic banking, the inferior average performance of islamic banks must be in part attributed to the low country-specifi c effi ciency scores for certain countries. furthermore, it has demonstrated that country effects play a signifi cant role in explaining effi ciency distributions between countries, even after controlling for country-specifi c environment conditions, including islamic banking. the paper has however concluded that bank compliance with shari’ah which operates ethical banking activities has higher input requirements and it is possible that the reduced potential output is proof of systematic ineffi ciency. although studies on malaysian banks found banks to operate at almost constant return to scale, the cross-country analysis demonstrates that the average estimated returns to scale for conventional banks are lower than those for islamic banks, with the exception of malaysia and jordan. therefore, moderate benefi ts will be realized even if islamic banks attempt to increase their scale size. finally, the main conclusion derived from the paper, in which islamic banks have relatively higher input requirements compared to conventional banks should however, motivate policy makers involved in islamic banking and islamic bank managers to identify and overcome factors leading to these higher input requirements. in addition, they should aggressively work to create a more encouraging banking environment for islamic banking, if they plan to further expand islamic banking. in order to expand our knowledge and understanding concerning the investigated issues, more research especially applying advanced techniques needs to be carried out in different time and country settings and the investigation should also extend to foreign-owned islamic banks. author information: mariani abdul majid is a teaching staff, school of economics, faculty of economics and management, universiti kebangsaan malaysia, 43600 ukm bangi, selangor, malaysia. tel: +(60-3)8921-5780; fax: +(60-3)-8921-5789. email: mariani@ukm.my. ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 75 references aigner, d. j., c. a. knox lovell, and peter schmidt (1977). formulation and estimation of stochastic frontier production function models. journal of econometrics, 6, 21-37. al-jarrah, idries, and philip molyneux, (2005). effi ciency in arabian banking. in munawar iqbal and rodney wilson (eds.), islamic perspectives on wealth creation ( pp.97-117). edinburgh: edinburgh university press. allen, linda, and anoop rai (1996). operational effi ciency in banking: an international comparison. journal of banking & finance, 20(4), 655-72. alshammari, s.h., (2003). structure-conduct-performance and effi ciency in gulf co-operation council. university of wales, uk. ariff, m., m bader, shamsher, m. and taufi q hassan (2008). cost, revenue, and profi t effi ciency of islamic versus conventional banks: international evidence using data envelopment analysis. islamic economic studies, 15(2), 23-76. association of banks in malaysia. abm bankers directory. kuala lumpur, various years. battese, g.e., and t.j. coelli, (1995). a model for technical ineffi ciency effects in a stochastic frontier production function for panel data. empirical economics, 20( 2), 325. berger, allen n., and david b. humphrey (1997). effi ciency of fi nancial institutions: international survey and directions for future research. european journal of operational research, 98, 175-212. berger, allen n., and loretta j. mester (1997). inside the black box: what explains differences in the effi ciencies of financial institutions? journal of banking and finance, 21, 895-947. bonaccorsi di patti, emilia, and daniel c. hardy (2005). financial sector liberalization, bank privatization, and effi ciency: evidence from pakistan. journal of banking & finance, 29(8-9),2381-406. bos, j.w.b., and c.j.m. kool, (2006). bank effi ciency: the role of bank strategy and local market conditions. journal of banking & finance, 30(7), 1953-74. brown, kym (2003). islamic banking comparative analysis. the arab bank review , 5( 2), 43-50. brown, kym, and michael skully (2003, september). a comparative analysis of islamic bank effi ciency. paper presented at the international banking conference, prato, italy. carvallo, oscar, and adnan kasman (2005). cost effi ciency in the latin american and caribbean banking systems. journal of international financial markets, institutions and money, 15(1), 55-72. casu, barbara, and claudia girardone (2002). a comparative study of the cost effi ciency of italian bank conglomerates. managerial finance, 28(9), 3-24. cavallo, laura, and stefania p s. rossi (2001). scale and scope economies in the european banking systems. journal of multinational financial management, 11( 4-5),515-31. ht tp :// ijb f.u um .e du .m y 76 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 central bank of malaysia (2002). annual report. malaysia. clark, jeffrey a. (1996). economic cost, scale effi ciency, and competitive viability in banking. journal of money, credit and banking, 28(3), 34264. coelli, tim, (1996). a guide to frontier 4.1: a computer program for stochastic frontier production and cost function estimation. retrieved from http://www. uq.edu.au/economics/cepa/frontie and htm.www.uq.edu.au/economics/ cepa. coelli, tim, d. s. prasada rao, and george e. battese (1998). an introduction to effi ciency and productivity analysis. london: kluwer academic publishers. coelli, tim, sergio perelman, and elliot romano (1999). accounting for environmental infl uences in stochastic frontier models: with application to international airlines. journal of productivity analysis, 11 (3), 251-73. cuesta, rafael a., and josé l. zofío (2005). hyperbolic effi ciency and parametric distance functions: with application to spanish savings banks. journal of productivity analysis, 24(1), 31. cuesta, rafael a., and luis orea (2002). mergers and technical effi ciency in spanish savings banks: a stochastic distance function approach. journal of banking and finance, 26(12), 2231. demsetz, h., (1973). industry structure, market rivalry, and public policy. journal of law and economics, 16, 1-9. dietsch, michel, and ana lozano-vivas (2000). how the environment determines banking effi ciency: a comparison between french and spanish industries. journal of banking and finance, 24, 985-1004. el-gamal, mahmoud a., and hulusi inanoglu (2005). ineffi ciency and heterogeneity in turkish banking: 1990-2000. journal of applied econometrics, 20(5), 641-65. fare, rolf, and daniel primont (1957). multi-output production and duality: theory and applications. london: kluwer academic publishers. farrell, m. j., (1957). the measurement of productive effi ciency. journal of the royal statistical society, 120(series a part iii), 253-81. frame, w. scott, and tim j. coelli (2001). u.s. financial services consolidation: the case of corporate credit unions. review of industrial organization, 18(2), 229. fuentes, hugo j., emili grifell-tatjé, and sergio perelman (2001). a parametric distance function approach for malmquist productivity index estimation. journal of productivity analysis, 15(2), 79. girardone, claudia, philip molyneux, and edward p. m. gardener (2004). analysing the determinants of bank effi ciency: the case of italian banks. applied economics, 36(3), 215-27. hassan, m. kabir (2003). cost, profi t and x-effi ciency of islamic banks in pakistan, iran and sudan. paper presented at international conference on islamic banking: risk management, regulation and supervision. jakarta, indonesia. ht tp :// ijb f.u um .e du .m y the input requirements of conventional and shariah-compliant banking: 51-78 77 hussein, khaled a., (2004). operational effi ciency in islamic banking: the sudanese experience. islamic research and training institute, islamic development bank. retrieved june, 30, from http:// www.irti.org/ sudan%20-banks.pdf. imf (2004). international fi nancial statistics. retrieved june, 9, from http:// imfstatistics.org/im/logon.aspx iqbal, zahid, kizhanathan v ramaswamy, and aigbe akhigbe (1999). the output effi ciency of minority-owned banks in the united states. international review of economics & finance, 8(1), 105. jondrow, james, c.a. knox lovell, ivan s. materov, and peter schmidt (1982). on the estimation of technical ineffi ciency in the stochastic frontier production function model. journal of econometrics, 19, 233-38. karim, rifaat ahmed abdel (2001). international accounting harmonization, banking regulation, and islamic banks. the international journal of accounting, 36(2), 169-93. kasman, adnan (2005). effi ciency and scale economies in transition economies. emerging markets finance & trade, 41(2),60-81. kasman, adnan, and canan yildirim (2006). cost and profi t effi ciencies in transition banking: the case of new eu members. applied economics, 38(9), 1079. leibenstein, h. (1966). allocative effi ciency vs. x-effi ciency. american economic review, 56, 392-415. lozano-vivas, ana, jesus t pastor, and jose m pastor (2002). an effi ciency comparison of european banking systems operating under different environmental conditions. journal of productivity analysis, 18(1),59. mariani abdul-majid, , david, s, and giuliana , b. (2008). effi ciency of islamic and conventional banks. aston business school working paper series rp0811. aston university. mariani abdul-majid, nor ghani, m. n. and fathin, f. s. (2005). effi ciency of islamic banks in malaysia. in munawar iqbal and ausaf ahmad (eds.), islamic finance and economic development (pp.94-115). new york: palgrave macmillan. mariani abdul-majid. (2008). effi ciency of islamic and conventional banks (unpublished doctoral dissertation). aston business school. aston university. mester, loretta j., (1996). a study of bank effi ciency taking into account riskpreferences. journal of banking and finance, 20, 1025-45. metwally, m. m., (1997). differences between the fi nancial characteristics of interest-free banks and conventional banks. european business review, 97(2). mohd zaini abd karim. (2001). comparative bank effi ciency across selected asean countries. asean economic bulletin, 18(3), 289-304. mokhtar, hamim s. ahmad, naziruddin abdullah, and syed m. al-habshi (2006). effi ciency of islamic banking in malaysia: a stochastic frontier approach. journal of economic cooperation, 27(2),37-70. ht tp :// ijb f.u um .e du .m y 78 the international journal of banking and finance, vol. 7. number 1: 2010: 51-78 orea, luis, (2002). parametric decomposition of a generalized malmquist productivity index. journal of productivity analysis, 18(1), 5-22. saaid, abd elrhman elzahi, saiful azhar rosly, mansor h. ibrahim, and naziruddin abdullah (2003). the x-effi ciency of the sudanese islamic banks. iium journal of economics and management, 11(2), 123-41. williams, jonathan, and nghia nguyen (2005). financial liberalisation, crisis, and restructuring: a comparative study of bank performance and bank governance in south east asia. journal of banking and finance, 29(89), 2119-54. wilson, rodney (2007-8). islamic fi nance in europe. european university institute. world bank (2007). country groups. retrieved june 13, from http:// www. worldbanl.org/data/countryclass/classgroups.htm. yudistira, donsah (2004). effi ciency in islamic banking: an empirical analysis of eighteen banks. islamic economic studies, 12(1). ht tp :// ijb f.u um .e du .m y determining the role of debt in the economy and a new approach for solving sovereign debt crises 94 the international journal of banking and finance, volume 10 (2), 2013: pages 94-133 determining the role of debt in the economy and a new approach for solving sovereign debt crises munawar iqbal king abdul aziz university, saudi arabia _________________________________________________________ abstract this paper aims to begin a dialogue on how to seek a longer term solution to the sovereign debt problems in general and those of eu in particular. although the history of debt crises is quite old, none of the several solutions proposed and tried in the past have been successful to curb recurring debt crisis. this issue has assumed critical importance as the eurozone debt crisis, which followed after the 2007-09 global financial crisis. several governments have been outvoted in europe due to this crisis and the cohesion of eurozone is at stake. a rethinking on debt creation and its macroeconomic effects are being seriously studied. it seems that traditional options available to policy makers have lost much of their luster. it is high time that unconventional measures may have to be offered for consideration to provide longer term solution. this paper is a brief on the islamic approach to the role of debt, and has potential to limit debt creation in the long term. we present some basic tenets of that approach referring in particular to the current dev eloped nation sovereign debt crisis. keywords: sovereign debt, islamic finance, debt overhang, asset backing, leveraging and growth. jel classification: h63 ___________________________________________________________ ―when national debts have once been accumulated to a certain degree, there is scarce, i believe, a single instance of their having been fairly and completely paid.‖ adam smith (1776). 1. introduction our understanding of islamic rulings about debt inspires us to suggest that those rulings could be of immense help in determining the proper role of debt in the economy. this paper makes an effort to explain that approach about the role of debt in wealth creation 95 and the appropriate limits to debt usage. analyzing instances taken from the ongoing eurozone economic crisis (2011-todate), we endeavour to derive some principles from islamic teachings on financial management that can help to solve the problems created from excessive debt. the decision to take debt has historically been viewed differently in different cultures. perceptions about the extent of benefits and harms from debt have also varied in various economic situations. over the last fifty years, debt-taking has been on the rise. there is credible literature in the web sites of several scholars with substantive data on how bad the debt burden had become under the no-risk-shared interest-based debtgiving. a recent book (still, 2011) provides extensive data on sovereign debt, so we do not provide the easily available statistics from this and other sources freely available (imf, world bank, and bis web sites). the financial crisis that started in 2008 is the hottest political economy topic being avidly followed by human societies around the world even after five years since the onset of that crisis. various policy measures have been suggested, of which some have been implemented. politician and economists alike are trying their best to suggest ways to control or minimize its adverse effects on economic and social well-being of peoples in several economies. several reasons for the crisis have been presented by analysts as well as policy makers. these reasons have been discussed in the economics literature (the financial times, 2013). the debate has given rise to many controversies both in economic theory and policy circles. while conservative economists hailed a return to stringency, interventionists are calling for less belt-tightening extolling keynesian solutions, which would make debt crisis even worse. all of these controversies hover around the role of debt in the economy, especially in the public sector. taking debt has always been a controversial issue. in this paper, we intend to explore a novel approach, no doubt, a third approach as a longer term solution to debt overhang suggesting some unconventional measures in the public policy arena. since debt is an issue that concerns everyone, shakespeare‘s famous advice is: ―… neither a borrower nor a lender be.‖, a wise statement that modern societies cannot live by unless economic entities are weaned off from their dependence on debt. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 96 many economists took a tolerant view, commending the advantages of debt and dismissing its well known macroeconomic disadvantages. simple minded individuals, sharp corporate managers, and vote-hungry, often corrupt politicians, went on a borrowing binge. due to the easy-money approach of suppliers and demanders of debt, international debt market crossed $100 trillion mark in 2011 of which more than 50 per cent is sovereign debt. that means the sovereign debt of all countries is about 75 per cent of worlds‘ gross domestic product (gdp), whereas the governments‘ revenue is a mere 31 per cent of gdp (ariff, 2012). undoubtedly such high levels of debt are unsustainable. the situation in the private sector is also quite serious as we will explain in subsequent sections. the interaction between various sectors of the economy has made the picture much gloomier. there is an influential school of thought that considers that leveraging as an useful way of increasing wealth. accounting theory demonstrates that the return on equity is increased (leveraged) by the addition of debt. if a firm‘s roe (return on shareholders‘ fund) is say 10 per cent with 100 per cent equity, the roe with 50 per cent debt and 50 per cent equity would be about 20 per cent. furthermore, because interest cost is exempted from tax, debt is the cheapest source of capital. it is claimed that debt can open up many opportunities for individuals while corporations and governments could fulfill funding needs or capture opportunities that their capital cannot afford. national income can be increased by putting more idle resources into use by leveraging the productive force of equity capital with some amount of borrowing. another school has long since advocated that debt (except in very exceptional cases) leads to many long term problems such as heavy debt servicing burden, debt overhang, debt-addiction, higher interest rates as risk of more debt makes further debt more risky. all these ills could likely crowd out good investments and create false expectations leading to speculation. borrowing could also make firms and governments become financially weak if there is debt beyond an optimal level (miller and modigliani, 1961; warner, 1977). some even claim that economic bubbles in various sectors build up with over-liberal availability of debt (as credits supplied by fractional banking). ultimately bubbles burst, recession occurs and cause huge loss of wealth and 97 socio-political unrest as happened from the easy-credit policy of us banking during 1994-05. perhaps both of these are extreme views. the public debate about the eurozone debt crisis is part of a worldwide debate on debt overhang. global financial crisis (20079), the worst since the great depression of the 1929-33, is having a curtailing impact on the economic and social thinking about the role of debt and the risk of debt. the 201113 eurozone debt crisis raised a number of questions about the tolerant view on debt of the interventionist. many of those who took liberal stands with respect to debt are being forced to reconsider their opinions. it is generally agreed that the root cause of the eu debt crisis is ‗excessive‘ debt, that is, debt overhang. many elected governments have fallen in europe partly due to crisis-driven election platforms on the issue of public debt problem and the adverse consequences from continuing with proposed policies of governments in power. the very existence of eu as a single currency region since 1992 is increasingly at stake. we believe that the world is ready, per force, to consider alternative solution under the islamic view on debt. in section 2 we discuss the role of debt and its limits for households, corporate and public sectors. in section 3 we explain the anatomy of a financial and economic crisis based on historical experiences. section 4 gives a brief account of the rescue operations being undertaken by governments and central banking authorities. we present components of a non-conventional approach for solving the problem in section 5. six principles for handling debt based on islamic teachings are explained. in the last section we present conclusions of this paper in the form of eight axioms on debt and economic stability that emerge from the analysis in this paper, based on islamic economics and finance literature. 2. role of debt and its limits in this section, a brief survey of debt as a problem is explained. living within one‘s means is a golden principle that applies to individuals, corporations and governments alike, the three actors in the economy. this does not mean that there is absolutely no role for debt in economic matters. nevertheless, it must be realized that debt is a double edged sword. there are circumstances when debts may become unavoidable. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 98 furthermore, if used in moderation and ‗wisely‘, it may have some benefits. however, except in exceptional cases of disability, the general rule is that if one expects a future stream of income (or assets), the receipt of which is fairly assured, that will enable one to pay off debt, hence incurring debts (leveraging) can be a useful short run technique to improve one‘s economic conditions. optimal leverage hypothesis advocated in financial economics signifies the same rule. this rule applies to individuals, corporations and governments alike. while the main focus of this paper is public debt, some comments on consumer and corporate debts may be helpful because of the inter-linkages between these and public debt. 2.1 consumer debt one of the reasons that policy makers failed to predict the seriousness of the present debt crisis (and for that matter all previous ones) is under-realization of the interactions between private and sovereign debt. adair turner, former chairman of the united kingdom financial services authority spoke about this incredible lack of foresight. experts failed to understand that high — and growing — debt burdens, especially in the private sector, endangered economic stability. policymakers thought private debt had no impact on macroeconomics. ―that assumption was dangerous, because debt contracts have important implications for economic stability. they are often created in excess, because in the upswing of economic cycles, risky loans look risk-free. and, once created, they introduce the rigidities of default and bankruptcy processes, with their potential for fire sales and business disruptions… private leverage levels, as much as the public-debt burden, must therefore be treated as crucial economic variables.‖ 1 for individuals, economic justification for the rule that leveraging can improve one‘s economic conditions was provided by consumer behaviour theories propounded by milton friedman, in his permanent income hypothesis (1957) and modigliani and blumberg‘s life cycle hypothesis (1954). according to these theories of consumer behaviour, consumers can borrow money to finance expenditures particularly housing 1 quoted in kling michael (2013). 99 and schooling earlier in their lives and pay back that debt in later years which are usually higher-earning periods. assets built through higher earnings and leveraging in early working age allow a smoother consumption pattern, including in the post retirement periods. this can be useful only if the rule stated above is strictly adhered to. unfortunately, due to liberal consumer loans policies and practices, especially the emergence of ‗plastic money‘, the amount of debt outstanding versus the consumer‘s disposable income has risen to unsustainable levels in most advanced countries. household debt in the united kingdom rose from less than 15 percent of gdp in 1960 to more than 90 percent in 2008. in the united states, total private credit increased from about 70 percent of gdp to well over 200 percent in 2008. 2 table 1: ratio of household debt to disposable income in eu 17 year debt/gdp ratio 2000 0.749 2001 0.751 2002 0.771 2003 0.796 2004 0.830 2005 0.876 2006 0.915 2007 0.940 2008 0.950 2009 0.976 2010 0.998 2011 1.000 source: eurostat; http://ec.europa.eu/eurostat. the number of households who carry consumption debt is quite high in almost all advanced countries including those in the eu, and hence the amount of debt. when would consumer debt be considered too high? a number of parameters are used to determine that. the most commonly used measure is the debt/disposable income ratio, 2 kling, ibid. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 http://click.infospace.com/clickhandler.ashx?du=http%3a%2f%2fec.europa.eu%2feurostat&ru=http%3a%2f%2fec.europa.eu%2feurostat&ld=20130507&ap=1&app=1&c=srchresrow3&s=srchresrow3&coi=239138&cop=main-title&euip=212.26.103.207&npp=1&p=0&pp=0&pvaid=3088c8da2b6d4755a1939519888ababe&ep=1&mid=9&hash=66981032517a05f9257f3ced1567da3b 100 the so-called consumer leverage ratio. it is generally believed that a ratio of more than 0.6 indicates household debt to be high. statistics presented in table 1 show that euro zone countries have crossed that limit long since and the ratio reached 1.0 in 2011. in the uk also, it is close to 1.0. 3 in the us, statistics released by the bureau of economic analysis and the federal reserve show that the consumer leverage ratio has been between 1.2 to 1.3 throughout the 2005-2009 period. another indicator used to determine whether or not consumer loans are within safe limit is the debt servicing as a percentage of gross household income. in this respect, a study 4 on new zealand household debt defines ‗vulnerable households‘ as those having a debt servicing obligations exceeding 30 per cent of their gross income. the same study found that the percentage of vulnerable household in 2008 was more than 16 per cent. that means one in every six households was in the vulnerable household category. in the uk, a report by bis 5 published in june 2011 showed that 12 per cent of households may be defined as being in financial difficulties and another 11 per cent (total of 33 per cent) of households are considered to be at risk of financial difficulties. another report 6 on the uk household debt states that 6.2 million households are financially vulnerable. in the us, the household financial obligations ratio (for) peaked at 17.7 per cent of disposable income in 2007. thereafter, it has had a downward trend but as at 2012, it was around 14.0 per cent. 7 easy access to credit may have some benefits, but once one lifts the guard, debt sneaks in like a snake and quickly grows into a dangerous serpent. the worst part of it is that most, if not all, debt is unnecessary and could be avoided. hardly anyone will question the undesirability of what is known as gambling debt. yet the volume of such debt runs into billions of dollars. the average (gambling) debt in us is between $55,000 and $90,000 and is increasing. 8 we know gambling is an addiction but easy access to debt plays an important role in facilitating and continuing in that addiction. it would be no exaggeration to state that debt itself is addictive. 3 creditaction.com, debt statistics released in april 2013. 4 henderson katherine and grant m. scobie (2009). 5 department of business information and skills, uk (2011) 6 financial inclusion centre (2011). 7 data released by the federal reserve board, march 2013. 8 debt.org 101 let us consider a supposedly more benign type of debt, the credit card debt, where one can charge one‘s dinners and holidays to credit cards, not to speak of furnitures, television and almost every consumer durable. according to nelson report, which reports credit card data for us consumers on a regular basis, the volume of credit card purchases in 2011 amounted to $2,050 trillion. 9 benjamin franklin, who earned the title of the first american for his foundational role in defining the american ethos, once said: ―rather go to bed without dinner than to rise in debt.‖ now as one famous american columnist earl wilson said: ―modern man drives a mortgaged car over a bond-financed highway on credit-card gas‖. among the 185 million card-holding u.s. consumers, the average person carries three bankissued credit cards, four retail credit cards and one debit card, according to cardweb.com. according to a survey by the american bankers association and dove consulting conducted in 2005-06, plastic payments accounted for 53 per cent of consumer purchases, compared to 43 per cent in 1999. the average credit card debt per card holder is estimated to be more than $6,500. use of plastic money is not bad per se. the problem is that access to easy money, does not allow people to make their calculations right. some circles claim that with credit card, one can track one‘s expenses more easily. that may be true. but what it makes easy to track is one‘s past expenses. at the moment of making the expense, one hardly thinks whether she can afford that $500 purchase or not. it is like putting one‘s car on ‗cruise control‘, with the brakes not working. would that not put one to a big risk of a crash, a debt overhang crisis? 2.2 corporate debt in the area of corporate debt, the natural point to begin discussion is the famous modigliani-miller theorem (mmt). this theorem states that a firm‘s value, under no interest deduction for tax, is independent of how it is financed, i.e., leveraging does not affect its value. it only determines the distribution of expected cash flow streams from operations among creditors and owners. the mmt paradigm has been extensively studied. even though, the results are not conclusive, the debt proliferators 9 nelson report (2012). international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 102 have conveniently ignored its main message. without any conclusive evidence, they have relied on some partial and conflicting results to justify borrowing as heavily as their credit rating would allow. the school of thought that advocates leveraging as a mean of increasing wealth often advocates using other peoples‘ money to prosper. this school believes that it is the easiest way to get rich. that is attractive in theory, but when it comes to practice, it is quite different. when someone takes a loan to start a new business instead of equity, he alone faces the high risk if the business fails. that risk is quite substantial. as shown in table 2, in the us twenty five per cent of new businesses (average of business and industrial sector) fail within the first year and fifty per cent in the first four years. table 2: failure rates of new businesses in the united states source: statisticbrain.com: verified by university of tennessee research. it is pertinent to note here, that an islamic debt-like instrument, called sukuk, can overcome this problem. it enables risk-sharing and gives new entrepreneurs better terms for investing. by barring interest-on-interest practice, it prevents borrowing beyond capacity to pay back as the partners become more vigilant due to their stakes involved and are also obliged to participate in loss (ariff et. al. 2012). another factor is that leveraging creates a feeling of ‗deceptive richness‘ among investors which creates false booms. on the business cycle up-turn, rising leverage gives a false impression of increasing wealth. this is what subprime mortgage lending did to americans when actually they were suffering from stagnant or falling real wages. in the downturn that followed, overleveraged businesses and consumers reduce investments and consumption in order to pay off debts. japan's lost decades after 1990 are a stark example of such behaviour. year percent failed 1 25 2 36 3 44 4 50 5 55 103 after the financial crisis of 2008-09, there have been influential voices questioning the usefulness, or at least the level of, corporate leveraging that one observes. it would be useful to quote here some statements from an ongoing dialogue that started from the discussions at the brookings panel on economic activity 2010. at that forum, alan greenspan, the famous long-serving former chairman of us federal reserve (1987-06), presented a paper "the crisis". he made a number of reform proposals in that paper. one of those was raising capital requirements and hence reducing leverage. however, he suggested that there are limits to how much we can do so. he was of the opinion that if businesses reduce leverage too much, financial intermediaries will not be sufficiently profitable to remain viable. in other words, he (a) believes that in principle reducing leverage ratio is good economics but (b) considers the scope of such reduction in practice to be limited. 10 his proposals have generated a lot of debate. in that debate, a large majority agrees on reducing leverage. furthermore, there are influential voices which questioned greenspan‘s reservation on the limits to such reduction. one such voice is that of harvard economist greg mankiw whose position is worth quoting. he says, ―indeed, i think it is possible to imagine a bank with almost no leverage at all. suppose we were to require banks to hold 100 per cent reserves against demand deposits. and suppose that all bank loans had to be financed 100 per cent with bank capital. a bank would, in essence, be a marriage of a super-safe money market mutual fund with an unleveraged finance company. (this system is, i believe, similar to what is sometimes called ―narrow banking‖). it seems to me that a banking system operating under such strict regulations could well perform the crucial economic function of financial intermediation. no leverage would be required.‖ another recent study 11 states: ―the fact that banks choose high leverage does not imply that this is socially optimal, and, except for government subsidies and viewed from an ex ante perspective, high leverage may not even be privately optimal for 10 that is practicing bad economics for so long has made getting out of it very difficult. 11 anat r. admati et al., ―fallacies, irrelevant facts, and myths in the discussion of capital regulation: why bank equity is not expensive‖, graduate school of business, stanford university, manuscript, march 2011. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 104 banks‖. 12 these views are, and will continue to spur a lot of debate. however, abstracting from the value of a such a firm idea and considering the interest of the economy in general, we believe that if we factor in the above-mentioned risk factors (and not all risk in the contemporary world have been mentioned), they will far outweigh any benefit that could be gained by leveraging not securely linked to the real sector in an economy. if and when leveraging is to be considered, it must be considered keeping in view ‗sustainable‘ growth in the real economy. 2.3 public debt economic justification for governments to borrow comes from the purported growth-boosting potential of debt. the keynes‘s theory of aggregate demand is behind that supposition. lord maynard keynes, perhaps the most influential economist in the history of economic thoughts, had suggested in the theory of aggregate demand that governments can use fiscal policies (budget deficits) as well as monetary policies (by lowering the interest rates) during depressions to boost aggregate consumer and investment demands to grow, which in turn increase employment and national income. other economists drew attention to many loopholes in keynes‘ theory but in the depth of great depression, his theory became popular among policy makers: government debt was not too high at that time. in table 3, we present data released by the bank of international settlement on gross government debt as a percentage of gdp for the period 1990-2013. 13 in recent years, witnessing the worst depression after the great depression, his followers named by some economists as vulgar keynesians have been recommending similar measures. once again, under the burden of recession, policy makers followed many of those policies. as a result government debt increased sharply. policy makers hoped that the conventional (but unreliable) pump priming via debt technique will help them getting out of the crisis. however, the relationships between low rates of interests high budgetary deficits and growth have all been questioned. according to leigh skene of lombard street 12 http://gregmankiw.blogspot.com/ 13 statistics for 2013 are estimations. http://click.infospace.com/clickhandler.ashx?du=http%3a%2f%2fgregmankiw.blogspot.com%2f&ru=http%3a%2f%2fgregmankiw.blogspot.com%2f&ld=20130507&ap=1&app=1&c=srchresrow3&s=srchresrow3&coi=239138&cop=main-title&euip=212.26.103.207&npp=1&p=0&pp=0&pvaid=4c9b60f7bc5242e8849647e667375dca&ep=1&mid=9&hash=bff7888df87a8fe554ed9ca4e09b9bf5 105 research, 14 each additional dollar of debt is associated with less and less growth. a working paper issued in september 2011 by the bank of international settlements 15 summarizes the findings of an extensive research based on a huge dataset containing long time series/cross-sectional data as follows: ―at moderate levels, debt improves welfare and enhances growth. but high levels can be damaging. when does debt go from good to bad? we address this question using a new dataset that includes the level of government, non-financial corporate and household debt in 18 oecd countries from 1980 to 2010. our results support the view that, beyond a certain level, debt is a drag on growth. for government debt, the threshold is around 85% of gdp. the immediate implication is that countries with high debt must act quickly and decisively to address their fiscal problems. the longer-term lesson is that, to build the fiscal buffer required to address extraordinary events, governments should keep debt well below the estimated thresholds. our examination of other types of debt yields similar conclusions. when corporate debt goes beyond 90% of gdp, it becomes a drag on growth. and for household debt, we report a threshold around 85% of gdp.‖ there were signs from early 2007 onwards that economies were reaching the limit of their ability to absorb more borrowing that could not yield positive effects on growth. debt-addition prevented them to return to economic fundamentals. in addition, as reported later in the paper, there is evidence of positive correlation between corruption and higher public debts. in brief, it was more for political rather than economic reasons that the debt pile up went for so long. it seems that many rich countries exhausted the growth-boosting potential of debt. with ageing populations and shrinking workforces, their economies may only grow more slowly than in the past. they may have borrowed from the future, using debt to enjoy a standard of living that is unsustainable presently. greece provides a stark example. standard & poor‘s estimates that her gdp will not regain its 2008 level until 2017. 16 14 leigh skene (2012). 15 cecchetti, g. stephen et.al. (2011). 16 ibid. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 106 table 3: gross government debt as percentage of gdp source: bank of international settlements, annual report, 2012. the prevailing uncertainty in the markets created an opportunity for the gamblers (speculators) to fish in troubled waters. the derivatives market ballooned and crossed in 2012 the quadrillion dollars mark. that is 20 times the world economy. according to many analysts the global financial crisis of 2007-09 was caused in part by the proliferation of non-performing derivative products tied to us home loans that triggered hundreds of billions of dollars in write downs (ariff et al., 2012). bankruptcy of lehman brothers holdings inc., in september of 2008, the largest ever year advanced economies us uk germany france italy spain greece portugal ireland japan 1990 58.90 63.14 32.31 37.02 38.57 97.17 47.68 74.71 60.52 108.78 63.89 1991 61.67 67.88 32.80 37.69 39.46 99.94 49.55 77.15 63.10 110.38 63.21 1992 65.49 70.27 38.97 40.81 43.85 106.24 52.08 82.42 56.56 106.89 67.58 1993 70.45 71.88 48.73 46.11 50.80 115.55 65.54 103.37 61.40 109.89 73.86 1994 71.52 71.08 46.84 46.48 60.08 120.35 64.31 101.31 64.51 103.52 78.97 1995 73.99 70.67 51.58 55.68 62.60 122.03 69.28 102.03 66.77 95.89 86.24 1996 75.54 69.90 51.20 58.82 66.37 128.13 75.98 104.07 66.55 85.84 93.81 1997 75.14 67.43 52.02 60.36 68.93 129.62 75.00 100.92 65.32 75.18 100.45 1998 75.48 64.21 52.53 62.33 70.40 131.67 75.33 98.57 63.27 62.60 113.17 1999 73.93 60.50 47.39 61.83 66.86 125.50 69.39 102.42 60.51 51.69 127.04 2000 71.19 54.52 45.16 60.85 65.73 120.97 66.51 116.35 60.23 39.96 135.38 2001 71.36 54.45 40.40 60.10 64.30 120.12 61.89 119.16 61.68 37.44 143.67 2002 73.67 56.82 40.84 62.50 67.54 118.74 60.32 118.64 65.04 35.72 152.28 2003 75.81 60.16 41.50 65.87 71.74 116.34 55.34 113.34 66.76 34.51 157.98 2004 77.53 61.25 43.79 69.26 74.11 116.66 53.34 115.80 69.29 33.14 165.54 2005 79.22 61.50 46.39 71.78 76.04 119.44 50.74 113.35 72.84 32.94 175.27 2006 77.48 60.93 46.04 69.83 71.21 116.87 46.20 116.91 77.61 29.19 172.15 2007 76.27 62.12 47.17 65.57 72.96 112.12 42.35 115.04 75.42 28.65 167.05 2008 83.42 71.40 57.43 69.74 79.27 114.69 47.70 118.07 80.69 49.59 174.10 2009 96.37 85.03 72.45 77.44 90.82 127.10 62.86 133.47 93.27 71.11 194.05 2010 103.33 94.17 82.19 87.06 95.18 126.12 67.06 149.07 103.58 98.46 199.97 2011 107.85 97.60 89.95 86.88 98.62 127.74 74.13 165.10 111.94 112.57 211.73 2012 112.67 103.64 97.20 87.34 102.35 128.11 77.23 181.24 121.93 118.82 219.05 2013 116.19 108.45 102.28 86.44 104.05 126.59 78.97 183.89 123.68 122.42 226.82 107 in history, was only the tip of the iceberg. derivatives caused havoc all around the world. the worst part of the story is that the bulk of the otc derivatives market is largely un-regulated. 3. making of financial crises why are financial crises often associated with debt overhang? many researchers believe that interest-based borrowing, especially speculative borrowing (apparently hedged by derivatives) and budgetary deficits are among the root causes of crises, so is the case of the eu debt crisis. the history of financial and economic crises is centuries old (kindleberger, 2005). however, the intensity of these crises has been increasing over time primarily due to derivatives, debt overhang and the breakdown of bretton woods rules. in the economic literature, causes of past crises have been studied in great detail. reinhart and rogoff (2009) 17 provide an excellent survey going back eight centuries. though there are some unique features of every financial crisis and within each special aspects for each country affected, however, there are some characteristics that are common. in the opinion of the present author, three factors are most prominent in all crises occurring in the last 50 years. these are: (i) reliance rather over-reliance on interest-based borrowing; (ii) excessive public sector involvement in the economy; and (iii) speculation in stock, foreign exchange and commodity markets. the problems are compounded because these three elements have strong inter-linkages among them magnified by the use of derivatives. let us review financial crisis as a case study over the 2007-todate. immediately before the crisis the interest rates were low. individuals increased their borrowing in the form of mortgage loans, many of those being sub-prime mortgages. real estate prices started rising so much so that a real estate bubble built over 12 years had to burst by a 48 per cent price falls over 2007-09. since real estate is one of the most commonly used collateral for bank loans, banks were ready to offer other types of consumer loans. moreover, with higher real estate prices, property owners‘ net worth 17 reinhart, carmen m. and kenneth s. rogoff (2009). international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 108 also increased. that encouraged them to take higher credit card loans. the consumer leverage ratios in all advanced countries increased. this ratio in the eu increased from below 75 per cent in 2001 to 95 per cent in 2007. 18 by that time it had become obvious that many countries around the globe including eu nations, the uk, japan and the us were experiencing property bubble. in the corporate sector, enterprises encouraged by low interest rates increased their borrowing with the result that their capital asset ratios sharply fell. many mega banks and other financial enterprises relied on the theory of ‗too big to fail‘ went on a borrowing binge to lend. banks having the ability to create credit by the stroke of a pen on the famous money-multiplier formula and profiting by interest rate margins and multiple credit creation were obliging the borrowers. low interest rates, reducing the cost of borrowing, also provided a golden opportunity to speculators. easy money encouraged speculation in the commodity and stock markets. they were earning handsome returns, not through dividends, but mostly through short selling. that led to a phenomenal growth of the derivatives market. making of a bubble was quite obvious but given a rising stock market generally considered being a sign of a booming economy, the governments turned their eyes the other way. as early as june 2005, the economist magazine warned, ―the worldwide rise in house prices is the biggest bubble in history. prepare for the economic pain when it pops.‖ 19 as in the case of earlier bubbles, the property prices started decreasing. by 2007, the property bubble had burst like an economic bomb. while the value of the properties decreased, the level of debt did not. by 2009 the ratio of household debt to disposable income in eu countries had crossed 97 per cent. the burden of repaying/defaulting on the loan depressed aggregate demand, which in turn reduced rates of growth in gdp around the world making things more difficult for governments. imf statistics would have us believe that the 3.6 per cent on-long-trend growth rate of the 2005 was shaved to 1.6 per cent since then. if one takes the world gdp as 18 eurostat. 19 the economist. june 16, 2005. http://en.wikipedia.org/wiki/the_economist http://en.wikipedia.org/wiki/aggregate_demand 109 us$65 trillion, the shrinkage due to the financial bubble in the world gdp has been us$1.3 trillion times 5, that is us$6.5 trillion over five years to 2013. just think of the opportunity cost of debt overhang to the wealth of nations! the number of bankruptcies and bail outs increased, putting a heavy burden on tax payers, and the skilled workers whose incomes stopped through no fault of theirs. the holders of household debt, mortgage companies and commercial banks had increasing bad debts. 4. rescue operations by governments fearing a collapse of the financial sector, many governments stepped in to rescue the ailing financial institutions. on october 3, 2008, the us passed the emergency economic stabilization act of 2008 authorizing the treasury to buy risky and nonperforming debt from various lending institutions. in europe, a mix of policy responses was undertaken at about the same time. in the uk, the government announced a £500 billion bailout package on 8 october, 2008. the initial british plan had three pillars: (1) recapitalization through a bank recapitalization fund, for £50 billion; (2) a credit guarantee scheme, a government loan guarantee for new debt issued between british banks for up to £250 billion; (3) liquidity provision through short term loans made available through a special liquidity scheme operated by the bank of england, for £200 billion. in france, a rescue plan was enshrined into law on october 16, 2008 (loi de finances rectificative pour le financement de l‟économie, no. 2008-1061). it created two ad hoc institutions: the société de financement de l‟economie française (sfef), set up to raise capital on financial markets and provide liquidity to ailing financial institutions, and the société de prise de pariticipation de l‟etat (sppe), through which the government would buy equities from the french banks and thus help to recapitalize them. the government agreed to guarantee bank bonds issued by the sfef up to €360 billion. at the same time, the sppe would invest €10.5 billion to recapitalize french banks by january 2009. the combination of rescue measures for the consumer and the corporate sectors, spilled into public sector. government had to borrow from the market issuing interestbased bonds or from the central banks and ran excessive budget deficit in an attempt international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 http://www.investopedia.com/terms/t/troubled-asset-relief-program-tarp.asp http://www.investopedia.com/terms/t/troubled-asset-relief-program-tarp.asp http://www.investopedia.com/terms/u/ustreasury.asp 110 to pump prime the economy. they hoped that the supposed positive link between budget deficit and growth in gdp will increase rates of growth, which will in turn increase tax revenues and help balancing the budget over the business cycle. however, several empirical studies have now shown that the positive relation between high budget deficit and high growth rates is not as simple as previously believed. 20 based on empirical evidence, the best that can be said if the initial public debt levels had been low with low budget deficits, governments would have been successful in pump priming the economy. however, that is not the case in the present circumstances. imf fiscal monitor, 2013 states: ―most studies find that high debt levels (above 80–90 percent of gdp) have a negative effect on growth (some 0.15–0.20 per cent per 10 percentage points of gdp). high debt also makes public finances more vulnerable to future shocks, because it constrains governments‘ ability to engage in countercyclical policies and because the larger the initial debt ratio, the bigger the increase in the primary surplus required to stabilize that ratio after an adverse shock to growth or interest rates. indeed, when debt is high, there is a risk of falling into a bad equilibrium caused by self-fulfilling expectations. high debt is unsustainable because markets believe it is so‖. table 4 presents average data for the 17 eu countries. it may be seen that the averages given are more than the reasonable threshold. as a matter of fact, in some individual countries the ratios are much higher. as stated above, from an economic point of view budget deficits, except in very special cases and for short periods of time, are bad economics. however, the shackles of debt once on, are not easy to break out from. strong lobbies are created as is evident in most western countries. political parties; right, left and center, all agree on the need to reduce budget deficits in the shortest period of time. the signing of ―the fiscal compact‖ on march 2, 2012 by all member states of the european union (eu), except the czech republic and the united kingdom was hailed as a landmark towards european fiscal integration. the treaty entered into force on 20 for example see, ursua, j and d. wilson (2012) and baum, a, c. et.al. (2012). http://en.wikipedia.org/wiki/member_states_of_the_european_union http://en.wikipedia.org/wiki/member_states_of_the_european_union http://en.wikipedia.org/wiki/czech_republic http://en.wikipedia.org/wiki/united_kingdom 111 january 1, 2013 for the 16 states which completed ratification prior to this date. for subsequent ratifiers, it will enter into force on the first day of the month following the deposit of ratification instruments. german chancellor, ms merkel, main proponent of the fiscal discipline school was buoyed. she said at that time that the pact, which binds euro zone countries to keeping their deficits below 3pc, was ―non-negotiable‖ and would ―last forever‖. later developments showed that her statement was a wishful thinking. table 4: debt/gdp ratios in eurozone (17) source: eurostat: http://ec.europa.eu/eurostat. in 2012, french president nicolas sarkozy, seeking re-election went to the extent to promise that he intended to insert the "balanced budget" rule into the french constitution, subject to approval by french electorate in a referendum on the issue. electorate did not buy that and he lost the election. françois hollande won the french presidency on an anti-austerity platform. he pledged to pursue efforts to trim the country‘s budget deficit to avoid fueling the euro-zone sovereign-debt crisis. however, once his honeymoon in power ended, he had to face the tricky balance year debt/gdp 2000 74.91 2001 75.14 2002 77.12 2003 79.60 2004 83.01 2005 87.60 2006 91.54 2007 93.99 2008 94.96 2009 97.56 2010 99.76 2011 99.36 international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 http://click.infospace.com/clickhandler.ashx?du=http%3a%2f%2fec.europa.eu%2feurostat&ru=http%3a%2f%2fec.europa.eu%2feurostat&ld=20130507&ap=1&app=1&c=srchresrow3&s=srchresrow3&coi=239138&cop=main-title&euip=212.26.103.207&npp=1&p=0&pp=0&pvaid=3088c8da2b6d4755a1939519888ababe&ep=1&mid=9&hash=66981032517a05f9257f3ced1567da3b 112 between appeasing the tax payers and the benefit seekers. with france‘s sluggish economy, the euro zone‘s second-largest after germany, running short of growthstimulating power of its own, he was forced to seek help from the european central bank. he faced fierce opposition from the most influential and the richest member of the euro zone, germany. however, the european central bank (ecb) is trying to find ways to keep france afloat. across the atlantic, in the american presidential elections of 2012, both front runners, barack obama and mitt romney, worried about the reaction of the electorate who considered budget deficit to be their top worry, tried to lure voters promising to cut the federal deficit, though through different plans. they were forced into this by a public outcry after standard & poor‘s downgraded the us credit rating and raised the fears that the united states may default on its obligations. if that could happen to a ―currency issuer‖, who pockets billions of dollars annually in ‗seigniorage‘, simply by running the printing press, the other countries are much more vulnerable. 5. a nonconventional approach using islamic debt principles in this section, the reader will find a third approach to managing debt based on sound debt-taking principles balanced in extreme situations with debt forgiveness. debt forgiveness has been practiced repeatedly in historical times in greece, rome, china and in the us. we then derive the lessons as axioms for debt taking under this approach as a long term solution to the current debate on how to eliminate debt overhang to restore economies to sustainable growth path. balancing the budget in present situations of most advanced countries is easier said than done, under wither the classical or interventionist policies. it requires either austerity measures or tax increases both of which are politically unpopular: a third method is stealing wealth in place, which has been shown as unpalatable from the cyprus experiment in 2013. dissenting voices about austerity plans as a solution of the crisis are already being heard, and newly elected governments are distancing that choice and street protests against austerity plans began in many countries. because of political expediencies rather than economic fundamentals, most governments are now thinking of other alternatives, in place of austerity. 113 several solutions to the sovereign debt crises have been recommended and tried in the past. some of these have merits. however, by and large, these are only ad hoc adjustments; temporary pain killers rather than permanent cure. as in the case of palliative medicine, most often these temporary pain killers aggravate the disease. dosage of pain killers is increased and in many cases the end result is terminal illness of the patient. in case of the eu debt crisis, per force, economists and politicians alike are ready to consider a non-conventional measure to handle the crisis? this author believes that islamic teachings and principles for financial transactions can be helpful. 5.1 six principles on debt creation and discharge the best place to start any discussion on islamic principles relating to finance is a reference to verses 278-280 of chapter 2 of the qur‘an, islam‘s guidepost. these verses state: “o you who believe! be afraid of god and give up what remains (due to you) from riba (usury) (from now onward), if you are (really) believers. and if you do not do it, then take a notice of war from god and his messenger but if you repent, you shall have your capital sums. deal not unjustly (by asking more than your capital sums), and you shall not be dealt with unjustly (by receiving less than your capital sums). and if the debtor is in a hard time (has no money), then grant him time till it is easy for him to repay, but if you remit it by way of charity, that is better for you if you did but know”. these verses establish three important principles with respect to borrowing and lending which we discuss briefly. 1. the principle of ‗principals‘ the first principle specifies that the creditor must get his principal back (except in exceptional circumstances discussed below). justice demands that is to be done. if the principal emerged from a loan contract, the lender did him a favour (expecting a reward only from god) and if it emerged from a sales contract he earned a profit once but thereafter his money has the same status as that of a lender. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 114 there are a number of islamic texts that enshrine this principal. the most important of these is the following qur‘anic verse: ―if you repent, you are entitled to your principal. deal not unjustly nor shall you be dealt with unjustly‖. [2: 279] the principle of principals is also established from the following hadith (tradition): ―anything given for benefitting from its usufruct is to be returned, a she-camel given temporarily for milking must be given back, a debt must be discharged, and one who stands surety is responsible to honor it‖. [sunan abu dawood: 296/3.] there are deposit insurance schemes in many advanced countries. however, as pointed out by askari et al. (2012), ―while deposit insurance can deter runs on solvent banks temporarily facing a liquidity crisis from an asset-liability mismatch, banking crisis come about also because of insolvency (loan losses exceeding bank capital) resulting from bad (or mispriced) loans, speculation, and even fraud on the part of banks. insolvencies can either be allowed to run their course, leading to bankruptcies and loss of shareholder value and creditor loans, or the government can bail out its banks. bailouts could be ominous and shift bank losses to the taxpayers, workers, pensioners and the poor.‖ 21 2. the principle of ‗non-principals‘ this principle requires that any addition to the principal amount that is not exposed to risk-taking must not be protected. once a capital owner has earned a return having exposed his capital to risk, that should be the end of the deal. if an investor wants to earn continuing returns, he must also share in risk. the two principles: (a) as a matter of principle, capital must earn a return in order for the economic growth to be sustainable and (b) as a matter of principle also, the capital owner must participate in risk to which the capital is applied as part and parcel of everyday business. insisting on the first principle and disregarding the second is unjust and are at the core of 21 askari, hossein, et. al.(2012). 115 generating economic instability. note here that current world practice is to exact a reward (agreed interest charge) without the capital provider taking part in the risk of the venture. if the purpose is to establish justice, principals and non-principals must be treated differently. it is not only the careless or carefree debtor alone who is to be blamed, the insatiable reckless creditor must also share part of the blame and the results: without risk-shared contracting, creditor gets scot-free from his erroneous lending. nonprincipals need not be guaranteed. economics tells us that and history gives us that lesson. one of the prominent writers in islamic finance, umer chapra, comments on the verses quoted above as follows: ―the principal reason why the qura‘n has delivered such a harsh verdict against interest is that islam wishes to establish an economic system where all forms of exploitation are eliminated, and particularly, the injustice perpetuated in form of the financier being assured of a positive return without doing any work nor sharing in the risk, while the entrepreneur, in spite of management and hard work, is not assured of such a positive return. islam wishes to establish justice between the financier and the entrepreneur‖. 22 again note that exacting interest (reward) without partaking in the risk makes lending a risk-less action, which is made worse by government‘s largesse to take over lender‘s liability using taxpayers‘ money. in this respect, we may draw attention to several research studies that have shown that risk-sharing contracts are superior to interest-based contracts. one such study by hossein askari, et al. 23 makes a convincing case for risk sharing contracts as opposed to no-risk-shared but interest-based contracts. analysing several studies on the causes of financial crises, they opine that ―debt is at the core of all these banking and financial crises‖. they also point out the reasons why in practice interest-based contracts have dominated the global financial markets. referring to the mmt that states that in the absence of frictions (e.g., taxation, information asymmetry, transaction costs), a firm‘s financial structure would be indifferent between debt and equity, they point out that ―in the real world there are a number of frictions that bias 22 chapra, m. umer (1985), p.64. 23 askari, hossein, et.al. op.cit. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 116 financial structures in favour of debt and debt-based contracts, with the two most important factor being tax and information. the tax treatment of equity returns and interest in industrial countries is heavily biased against equities. informational issues (information asymmetry and the subsidies and policies that encourage moral hazard and adverse selection) are conceived in favour of debt and debt-based contracts. broadly speaking, legal-financial systems in advanced countries are structured to favour of debt and debt-based transactions‖. 24 another recent study 25 concludes that: ―policies that subsidize debt and indirectly penalize equity through taxes and implicit guarantees are distortive. any desirable public subsidies to banks‘ activities should be given directly and not in ways that encourage leverage. and while debt‘s informational insensitivity may provide valuable liquidity, increased capital (and reduced leverage) can enhance this benefit. finally, suggestions that high leverage serves a necessary disciplining role are based on inadequate theory lacking empirical support. we conclude that bank equity is not socially expensive, and that high leverage is not necessary for banks to perform all their socially valuable functions, including lending, deposit taking and issuing moneylike securities. to the contrary, better capitalized banks suffer fewer distortions in lending decisions and would perform better. the fact that banks choose high leverage does not imply that this is socially optimal, and, except for government subsidies and viewed from an ex ante perspective, high leverage may not even be privately optimal for banks.‖ even if we assume that debts were incurred with valid justifications, yet there are circumstances when despite genuine efforts on the part of debtor, he is unable to fulfill his obligations. at that point, a third islamic principal becomes relevant. 3. the principle of relief as stated above, debt once incurred must be discharged, except in very exceptional circumstances. this third principle explains the need, rationale and the rules for providing relief to debtors who are in temporary and genuine difficulties. in 24 ibid. page 57. 25 anat r. admati et al. (2011). 117 general, creditors‘ right to get their principals back must be protected, but they cannot go beyond certain limits. an important distinction must also be made between a delinquent debtor and a distressed debtor due to unexpected external factors. the former is a debtor who has enough assets or fairly certain flows of future income or wealth to discharge his debts. such debtor must be dealt with strictly. later in the paper, we will mention legal measures to do that. the latter debtor is a debtor who despite his willingness to pay and sincere efforts made to discharge his debt, is not in a position to honour his payment liabilities because of market failures. such a debtor deserves to be provided relief. this principle is also well-established in islamic law. two islamic texts are quoted below, one of which has already been quoted but is repeated to draw the attention to this principle: (a) ―and if the debtor is going through a hard time (has no money), then grant him time till it is easy for him to repay, but if you remit it by way of charity, that is better for you if you did but know.‖ [2:280] (b) ―during the time of prophet (pbuh), the fruits of one companion‟s garden got destroyed due to which his debts piled up. prophet (pbuh) asked the companions to collect funds to help him pay his debts. they did, but the collection was not enough to pay the debts. upon that he told the creditors “take this and you do not have any more claim against him.‖ [sahih muslim] some important points that emerge from the hadith quoted above need to be noted. these are:  the right of the creditor to his principal was recognized [principle of principals].  all possible efforts were made to discharge the debt in full. for this purpose, even charity was collected.  having done that, when ‗available‘ money was not enough, debt write-off was granted. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 118 however, it has to be ensured that the principal of relief is not misused. in that respect the distinction between a delinquent debtor and a distressed debtor becomes important. the principal applies only to the latter. as for delinquent debtors, islamic law provides very strict penalties. these include:  delinquent debtor can be jailed (given physical punishment by a judge).  delinquent debtors can be stopped from using their assets for buying or selling, gifting, or transferring through any other manner until they pay their debts.  delinquent debtors can be declared ―persona non grata‖.  delinquent debtors can be humiliated and dishonored, e.g., publishing their names in public media.  a delinquent debtor can be prohibited from moving out of his city. in modern times, his name can be put in the ―exit control list‖.  creditors will have the first claim over the inheritance of the debtor. for distressed debtors, islamic system has many positive action provisions also. these include: (i) prohibition of interest that the amount of debt cannot be increased, whether it originated from a loan or an exchange transaction, gives the debtor time to put his house in order. tolerance shown to the distressed debtor by not increasing the amount of debt and extending the period of repayment is good not only for the debtor. it is also good for the creditors and the economy in general. it has both ethical and economic justification. when markets turn down, decline of collateral value triggers downward spiral as creditors try to protect their positions. however, this would make the market self-destructive, transforming a downturn into a crash. forbearance relieves debtors, giving them better chances for repayment which will go to the advantage of the creditors. preserving the value of collaterals and preventing market crash will benefit the entire economy and the society. the practice of debt-restructuring with unpaid interest as principal is disallowed under this rule. 119 (ii) reassignment of debt (hawalah): reassignment is voluntary and is a charitable act where someone undertakes to pay the debts of another person. no fee, compensation or benefit can accrue to the one taking up this responsibility. 26 in this regard it is stated in sahih bukhari that prophet (pbuh) said, ―procrastination (delay) in paying debts by a wealthy person is injustice. so, if your debt is transferred from your debtor to a rich person, you should agree.‖ (ii) share of gharimeen (those in debt) in zakah as is well-known to all, charity (zakah) is a compulsory levy imposed by god on every muslim who has a net-worth above a minimum level. of the five pillars of islam, charity is the third one. it is an earmarked levy voluntarily given across the world and managed outside the government revenue. its beneficiaries are fixed who are grouped into eight categories. one of these eight relates to those who have debts that they are unable to pay. in this way, the rich people in the whole society contribute to payment of debts, in case of distress as a charitable act, a voluntary act. (iii) given enough available resources, the state is responsible to discharge the debt. prophet stated (as a head of the state) that: ―i am more rightful than other believers to be the guardian of the believers, so if a muslim dies while in debt, i am responsible for the repayment of his debt, and whoever leaves wealth (after his death) it will belong to his heirs.‖ (sahih bukhari). it is important to note that state‘s responsibility to pay citizens‘ debts depends on availability of enough resources in the public exchequer. the above saying relates to the period when the islamic state became prosperous. earlier, it was the individual or the society who were obliged to bear that burden. as mentioned in an earlier hadith, charity was collected to pay off debts of farmers who met a calamity. when a massive flood wiped off $4 billion infrastructure in the queensland australia, a tax was 26 this may not be confused with ‗money transfer‘, whereby an institution transfers money from one place to another against a fee. charging such a fee is permissible. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 120 imposed at about 1 per cent of incomes of citizens as a forced charity for one year in 2013! while selective and well-deserved debt write-off is perfectly in line with islamic principles, the temptation that it provides to unscrupulous debtors is very strong. a debtor can choose to default intentionally if he so desires. despite legal provisions in most countries that try to prevent that, bankruptcy fraud is rampant and calls for reform are loud and clear. such fraud may take the form of concealment of assets by the debtors, multiple bankruptcy filings in different legal jurisdictions, intentional inaccuracies in bankruptcy petitions in order to prolong the proceedings to the disadvantage of creditors and similar other acts of camouflaging. bailouts such as fannie mae and freddie mac in 2008 raise the expectations of future bailouts. that would lend support to the too big to fail theory. in this and other similar cases, rescue operations by central banks by reducing interest rates encourage corporations to ‗make hey while the sun shines‘. they are tempted to delay putting their house in order in time to avoid eventual collapse. the creditors will get deeper in the abyss by advancing further loans to them considering them safe. the financial institutions will take a happy ride on their low-cost debt to leverage even more. that vicious cycle has no bright side to it. it is an established fact that enterprises like fannie mae and freddie mac expanded substantially their holdings of sub-prime mortgages concealing facts on their balance sheets. such experiences could be avoided by another principal derived from islamic teachings. (iv) the principle of transparency this principle provides safeguards to ensure that the principle of relief is not misused. if the greed of creditors is insatiable, that of debtors is obsessive. ‗debt handlers‘ make billions out of this cheating game. it must be ensured that relief is not misused. it requires strict laws to punish unscrupulous debtors, especially those who manipulate evidence to win relief from the courts. islamic stance in this regard is very strict. in an authentic hadith, prophet (pbuh) stated: ―one who cheats, does not belong to the community of muslims.” (sunan abu dawood). 121 the cases of corporate cheating to win relief are very frequent. for example, in the case of the lehman bankruptcy in september of 2008 were stuffed with fictitious valuations through hudson castle, an entity set up to get the bad stuff off lehman‘s balance sheet. in 2006, merrill lynch used a technique known as a ‗total return swap‘ to hide its riskier than declared cdos 27 through firms like pyxis. citigroup used similar arrangements that the security exchange commission later said should have been disclosed to shareholders in the summer of 2007. in all these cases, the ceos admitted the facts in court proceedings but pleaded making mistakes and being wrong is no crime. the court appointed examiners have been alleged to have been bribed to help these concealments. the ―buy now, pay later (or never)‖ culture plagues consumer and corporate borrowers alike. the above four principles deal with post-debt situations. islamic teachings also provide guidance for pre-debt circumstances. taking debt is not prohibited in islam in recognition of the fact that there may be circumstances under which taking debt becomes inevitable. however, three conditions are prescribed:  debts have to be non-interest bearing.  debtor has a firm intention to repay the debts and strives for that.  debts are taken only to the extent that is inevitable. the last condition gives us the fifth principle with respect to debt taking. (v) the principle of restraint this principle prescribes that, while taking debt is permissible, may even be desirable in some cases, the reasons of incurring debt and its aftermath must be carefully considered. exercising restraint is the rule. this is perhaps the most important of the principles with respect to the role of debt derived from islamic teachings that are presented in this paper. huge debt has several disadvantages; it 27 cdo stands for ‗collateralized debt obligation‘. it is a security; backed by a pool of bonds, loans and other assets. it includes different types of debt, often referred to as 'tranches' or 'slices'. each slice has a different maturity and risk associated with it. the higher the risk, the more the cdo pays. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 122 dishonours; it disables and kills people and worst, it enslaves people. in this respect, it is pertinent to draw attention to books/stories and terminology that has gained currency: debt is slavery; 28 debt is virus; 29 debt bondage, debtor‘s prison; wage garnishment, etc. being in debt limits one‘s decision-making power. consider a person william, who has no debts to pay. he is laid off. he has to find a job. his friend peter is also laid off. unfortunately, he has a lot of debt to pay. he also has to find a job, but his need for a new job is much more urgent than that of william. he will most likely accept a job that may not fit his career ambitions. urgency of his debt instalments will force him to accept something that he does not like. william also has to find a job, sure, but he can wait for something to come up that fits his priorities. if living beyond one‘s means is bad for an individual, it is equally bad for corporations and for governments. in resorting to debt finance, restraint is the rule for all, individuals, corporations and governments alike, the three economic agents. easy money leads people to ignore the precautions that one must take even when taking debt is considered to serve some useful purpose or becomes inevitable. one must never forget that once incurred, debt must be discharged, except in very exceptional circumstances mentioned above. when debt reaches crisis proportions, which it invariably does if restraint is not exercised, attempts are made to sell-off debt which creates bubbles. sooner or later these bubbles burst and the resulting ‗burns‘ wreck not only the debtors but the economy in general. in the wealth of nations, adam smith speaking of public debts stated, ―when national debts have once been accumulated to a certain degree, there is scarce, i believe, a single instance of their having been fairly and completely paid.‖. 30 though the actions given in the legal measures against delinquent debtor stated above generally do not apply to governments, the six principles given in this paper equally apply to governments as they do to individuals and corporations. 28 mihalik, michael (2007). 29 jaikaran, jacques s. (1992). 30 smith, adam, (1904). 123 reducing budget deficits is the most important issue in elections in the us and european countries including france. election promises aside, balancing the budget in present situations of these countries is easier said than done. it requires austerity measures that the electorate cannot digest after they have got used to public supports. dissenting voices are already being heard, because of political expediencies, rather than economic fundamentals. from an economic point of view budget deficits, except in very special cases and for short periods of time, are bad economics. however, as we hinted above, the shackles of debt once on, are not easy to break out from. strong lobbies are created as is evident in most western countries for perpetuating the status quo. american president, and the author of ‗united states bill of rights‘, james madison, once said, ―i go on the principle that a public debt is a public curse, and in a republican government a greater curse than any other‖. today the us national debt stands at $16.9 trillion. the democrats hold republicans responsible for the massive increase and blame them to be big spenders, but the fact remains that total federal debt grew by about us $4.6-trillion under obama‘s first term. during the us 2012 elections, the issue of fiscal cliff became a major topic of debate. it was a program of simultaneous increases in tax revenues and across-theboard government spending cuts scheduled to become effective dec. 31, 2012. the debate went on during the election campaign with democrats resisting expenditure cuts and republicans resisting tax increases. the congress could not reach an agreement before the november 2012 elections. in january 2013 a compromise was reached and on january 2, when president barack obama signed the compromise bill into law. however, the most important matter of debt ceilings to be imposed on the government could not be resolved. noble laureate maurice allais pointed out that one of the bad features of debt is that it permits postponement of the decision to adjust. 31 by the time one realizes the severity of the situation, it is already too late: in few years, the debt-service would amount to 40 per cent of revenue, which is a diversion of funds from economically useful public investments. 31 maurice allais (1993). international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 124 it is commonly known that debt has not grown for economic reasons. a large majority of american citizens, each one of whom shares around us$53,000 in the national debt, believes that the reason for the huge debt built up was the so-called war on terror, not the economy. in the light of the above discussion, it is safe to conclude that reasons for high public debt are largely political not economic. therefore, it will be useful at this point to discuss the role of state and the islamic view on that. 5.2 on the role of state the role of state in any society depends on the prevailing philosophy that it follows. the subject has been one of the most controversial issues in economics. until 1930s, the dominant philosophy was laissez faire, whereby governments were advised not to intervene in the operation of market. it was believed that rational individuals pursuing their self-interests in competitive markets will achieve both private and public interest simultaneously. the invisible hand of market forces would ensure symmetry between public and private interests and the best allocation of resources. hence there was no need for the government to assume any economic role accept to set the rules of the game according to the requirements of perfect competition. government intervention could be accepted only to remove market distortions or to offset market failures. from „market failure‟ to „government failure‟: in the aftermath of great depression this paradigm was challenged by keynesian economists on the grounds that markets are generally characterized by imperfections and rigidities and hence may not always ensure full employment and the best allocation of resources. moreover, the market forces left to themselves may not produce the distribution of income desired by the society. partly due to rising levels of unemployment and income inequalities and also partly as a reaction to communist philosophy, governments started taking upon themselves the responsibility of supporting the unemployed, the disabled, the old and the very young. the period starting from early 1940s and stretching well into 1970s was in general dominated by big government philanthrophy all over the world. 125 a more serious challenge to the laissez faire paradigm came from marxist economists who argued that the interaction of the institutions of market and the private property perpetuates injustice and exploitation leading to class conflicts. hence markets were replaced with central planning and private property with public control of the means of production. the collapse of socialism in the soviet union and eastern europe around 1990s shifted the balance once again in favor of economic liberalism. this, however, proved to be short lived. the imperfections of the market, the unrest created by wide disparities in the distribution of income and wealth, the pressing concerns for social justice, unethical use of the enormous powers that the market bestows upon the transnational corporations, and non-sustainability of long run economic growth without active monetary and fiscal policies have once again focused the attention of economists and policy makers on the need for some critical role for the state in economic matters. as a matter of principle, governments should only ‗govern‘. 32 they should not do business. except for exceptional cases, business should be left to individuals and corporations. because governments are non-profit organizations and use tax payer‘s money to provide services, they do not pay as much attention as required to costbenefit analysis. there are also issues relating to good governance. there are severe problems in that area also. three of these are of critical importance.  governments do not use money wisely (economically).  governments use money to gain political advantages.  government officials and politicians may be corrupt. it is interesting to note that in the corruption perceptions index (cpi), prepared annually by the transparency international, many euro zone countries, particularly those affected by the financial crisis, including greece and italy, are doing worse and worse. many of the lowest-scoring european countries are those 32 that of course includes the area of regulation and supervision. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 126 hardest hit by the financial and debt crises. a recent quantitative study on determinants of public debt documented what is common street knowledge, a positive correlation between corruption and high public borrowings. the study found that, even when the other fundamentals of the economy suggest that the optimal public debt level should be zero, the presence of corruption can cause significant government borrowing. the corruption-public debt interaction generates endogenous periodic equilibria, where debt cycles between high and low values. these debt cycles offer a possible explanation for why it is common for countries to accumulate debt for extended periods of time, only to abruptly carry out reforms designed to reduce government borrowing. 33 it seems that we have moved from „market failure‟ to „government failure‟. does that imply moving from bad to worse? the modern concept of a welfare state, whereby the state guarantees certain welfare benefits to individuals has its own drawbacks. without going into that debate, let us point to the fact that government benefits become addictive and work as a drag on productive involvement of able-bodied work force. highly progressive taxes needed to fund those benefits kill incentives to productive investments. political reasons prevent the governments touching white elephants in the area of health care, unemployment benefits etc. any cut in ‗welfare benefits‘, even if parasitic, is very hard to carry through. welfare benefits cannot fall, the government can fall instead. i call it ―downward rigidity of welfare benefits‖. a classic and most recent example is the on-going protests by workers, police, doctors, judge and even military against the salary cuts agreed as part of the ‗austerity measures‘ against €130 billion rescue package granted to greece. the so-called modern states are facing so many problems that a new workfare model as an alternative to the social welfare systems currently in vogue in many advanced countries is being proposed. other studies for reforming the welfare benefit system are also being discussed in the literature. 34 33 maksym ivanyna, et.al. (2010) 34 for example, see bisaillon, chris, ― (1993). 127 under the islamic system, a hierarchal schema for fulfillment of basic needs is designed. its implementation is overseen by the state. in such a schema, first and foremost it is the individual who is responsible to meet his and his dependents‘ expenses. the role of state here is to provide individuals with education, skills and employment opportunities. then the neighborhood is charged with the responsibility of fulfillment of basic needs of households living in a particular area. after that responsibility falls on the society at large. figure 1: public private partnership for welfare here islamic system provides a blueprint for public private partnership for achieving economic welfare goals. islamic institution of waqf 35 which mobilizes resources on non-governmental basis has historically played an important role in 35 waqf refers to appropriation or tying up a property in perpetuity for specific purposes. no property rights can be exercised over the corpus. only the usufruct is applied towards the objectives (usually charitable) of the waqf. state society neighborhood individual to meets his dependents‘ expenses international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 128 achieving social and economic goals of a society. rich individuals and not-for-profit organizations (npos) have been active throughout history from 624 ad to 1924 ad in islamic countries. going from below to the top, ultimately, it is the state which guarantees the fulfillment of basic needs. islamic schema is presented in figure 1. in the area of corporate finance, banks only tell us of their bad debts. what we are trying to argue in this paper is that almost all debt is bad. 36 if living beyond one‘s means is bad for an individual, it is equally bad for corporations and governments. in resorting to debt finance, restraint is the rule for all, individuals, corporations and governments alike. easy money leads people to ignore the precautions that one must take even when taking debt is considered to serve some useful purpose or becomes inevitable. one must never forget that once incurred, debt must be discharged, except in very exceptional circumstances that we will mention under the next principle. when debt reaches crisis proportions, which it invariably does if restraint is not exercised, attempts are made to sell-off debt which creates bubbles. sooner or later these bubbles burst and the resulting ‗burns‘ wreck the debtors down to their souls. another problem is created by evaluation process used by creditor institutions which, being keen to improve their deployment ratios, 37 rely more on collateral than on careful project evaluation. chapra (2000) 38 correctly argues that: ―interest-based lending makes the banks rely on the crutches of the collateral to extend financing for practically any purpose, including speculation. the collateral cannot, however, be a substitute for a more careful evaluation of the project financed. this is because the value of the collateral can itself be impaired by the same factors that diminish the ability of the borrower to repay the loan. the ability of the market to impose the required discipline thus gets impaired and leads to an unhealthy expansion in the overall volume of credit, to excessive leverage, and to living beyond means‖. thus the principal of restraint is important for the creditors also. 36 that is not to say that debt has no role whatsoever. the issue has been discussed in detail in section 2. 37 deployment ratio is a performance evaluation measure to see how best a bank is using its resources. its defined as total investment/total equity + total deposits. for details please see iqbal (2005), p. 76. 38 chapra, m. umer, (2000). 129 6. the principle of monetary-real linkage in this section, we expand our discussion on how to link funding to the real economic activities. a principle making debt creation contingent upon linking the monetary to the real sectors of the economy is explained. if there is a grain of truth in the argument that leveraging can be a means to increasing wealth, it must necessarily be qualified. money creation, whether through borrowing from the central bank by a government (running the printing machine) or through credit creation by commercial banks must be linked to the real sector. the position statement of the international association for islamic economics issued at the occasion of eighth international conference on islamic economics and finance, held in doha, qatar, during december 19-21, 2011, states: ―to build a sustainable system, debt must grow in tandem with real (as opposed to financial) wealth; leverage must be capped by productivity potentials. this allows for real wealth to grow without being hurdled with too heavy debt‖. one of the most prominent islamic economists, khurshid ahmad explains the issue more clearly: ―the delink between money and production activity has led to the emergence of two parallel economies, a money economy and other the real physical economy. money that was an instrument of exchange and a measure of value became the sole embodiment of value. an instrument that was designed to facilitate exchange and production became an object in itself. real economy slumbered but financial economy expanded beyond all limits, enriching those alone who were engaged in the business of finance. this created a fiduciary world of its own: money generating more money without producing goods and services in the economy, without increasing the stock and flow of assets in the society. in this new fangled economic order, money became the main player as well as the prize. the process of real value-addition in the economy has been slowed if not disrupted. money and creation of fiat money became the real game, resulting in the creation of wealth that only produces billionaires without adding much to the well being of the people. the result is that the delinked system has succeeded in creating an economy wherein presently while the world gdp in current prices is around $65 trillion, we have an ever expanding realm of fiduciary money, wherein annual trade in derivatives now exceeds $900 trillion. in this game the real players and also beneficiaries are only international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 130 a few thousand pseudo-investors while the whole global economy is at the suffering end. the same is true of the foreign exchange markets where daily trade in foreign currencies is fifty times more than the volume of daily movement of trade and tourists. the present day vulnerability and instability of the capitalistic economy owes much to this fatal delink to the real economy.‖ 39 6. conclusions the discussion in the paper pointed out a number of factors that interact to build an economic/financial crisis like the one going on in the eu as well as in several developing and low income countries. six principles derived from islamic teachings for debt-taking and debt management have been mentioned as possible long-term solution to recurring debt/economic crises. now we present our conclusions in the form of eight axioms on debt and economic stability inviting other researchers to participate in this dialogue. axiom 1. borrowing/lending on the basis of interest is the root cause of most economic and financial crisis. axiom 2. reward sharing contracts are superior to debt contracts based on interest. [the later dominate in practice largely due to unfair tax treatment of equity]. axiom 3. economic instability is a consequence of the delink of money from the real economy, while money is used to fuel speculative transactions including short-selling disconnected to production of credit and production of physical wealth. axiom 4. welfare benefits are downward rigid, a reason for debt addiction. axiom 5. budgetary deficits are largely politically motivated. axiom 6. participation of ‗society at large‘ is necessary for achieving welfare goals. [we call it private public partnership for welfare (pppw).] 39 ahmad, khurshid (2012). 131 axiom 7. government failure is as bad as market failure and has been a major reason of debt crisis. axiom 8. ―too big to fail‖ theory in the corporate sector has a parallel in the public sector, namely ―too big to manage‖ and the two are interlinked. [present day governments (public sectors) are too big. size of the public sector should be closely linked to public revenue. balanced budgets as a constitutional requirement with some monitorable flexibility is a good policy option for long term economic stability.] author information: munawar iqbal phd is a professor at the institute of islamic economics, king abdul aziz university, saudi arabia. he is the founder of the international association of islamic economics in 1974, and has taught and researched on this topic for decades. he lectures and conducts training in islamic finance throughout the world. the author wishes to record his gratitude to the two editors of the journal for providing review comments on an earlier version of this paper. munawar may be contacted at: e-mail: munawariqbal@gmail.com. references ahmad, khurshid, (2012). ―global economic crisis and the role of islamic economics‖, idb prize laureate lecture, april 21, 2012. anat r. admati, peter m. demarzo, paul pfleiderer (2011). ―fallacies, irrelevant facts, and myths in the discussion of capital regulation: why bank equity is not expensive‖, graduate school of business, stanford university, manuscript, march 2011. ariff (2012). elgarblog.wordpress.com, october 24. ariff, m, munawar iqbal and shamsher mohamed, eds. (2012). islamic debt markets for sukuk securities: theory, practice and issues. skaekwr new york us and cheltenham, u.k: edward elgar publishing limited ariff, m., john farrar and ahmed khalid eds. (2012). regulatory failure and the global financial crisis. skaekwr new york us and cheltenham, u.k.: edward elgar publishing limited. askari, hossein, zamir iqbal, noureddine krichene and abbas mirakhor, (2012). risk sharing in finance: the islamic finance alternative, singapore: john wiley and sons (asia) pte. ltd. baum, a, c. checherita-westphal and p. rother (2012). ―debt and growth: new evidence for the euro area‖, working paper c34 no.1450, frankfurt: european central bank. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 132 bisaillon chris (1993). ―causes and effects of welfare dependency‖, illinois wesleyan university, honours project. cecchetti,stephen g, madhusudan mohanty and fabrizio zampolli (2011). the real effects of debt, bank of international settlement, working papers no 352, september 2011. chapra, m. umer, (1985). towards a just monetary system, leicester, uk: the islamic foundation. chapra, m. umer, (2000). ―international financial stability: the role of islamic finance‖, policy perspectives, vol. 4, no.2, islamabad: institute of policy studies. creditaction.com, debt statistics released in april 2013. department of business information and skills (2011). ―credit, debt and financial difficulty in britain (2009/10)‖, june 2011. financial inclusion centre (2011) ―report on debt and household incomes‖. friedman, milton (1957). ―the permanent income hypothesis‖, chapter 3 in a theory of the consumption function, princeton: princeton university press. also available at, http://www.nber.org/books/frie57-1. henderson katherine and grant m. scobie (2009). household debt in new zealand, new zealand treasury working paper 09/03. iqbal, munawar and philip molyneux (2005) thirty years of islamic banking: history, performance and prospects houndsmill, uk and new york usa: palgrave mcmillan. ivanyna maksym, alex moumouras and peter rangazas 2010. jaikaran, jacques s. (1992). the independent, friday, april 13, 2012. ―now euro zone debt virus starts to overwhelm italy‖. kindleberger, charles, p., and robert z aliber (2005: fifth edition). manias, panics and crises: a history of financial crisis. john wiley & sons, new york, ny, usa. kling michael, http://www.moneynews.com, september, 6, 2013 leigh skene (2012). ―repent at leisure‖, the economist, may 24, 2012. maurice allais (1993). the monetary conditions of an economy of markets: from the teachings of the past to the reforms of tomorrow, jeddah: islamic research and training institute. miller, m. h. and modigliani, f. (1961). dividend policy, growth and the valuation of shares. journal of business, 34, 411-33. modigliani, f. & brumberg, r. (1954). utility analysis and the consumption function: an interpretation of cross-section data. in: kurihara, k.k (ed.): post-keynesian economics. new brunswick, nj: rutgers university press. nelson report (2012), issue no. 989. available at http://www.nilsonreport.com/ reinhart, carmen m. & kenneth s. rogoff (2009), this time is different: eight centuries of financial folly, princeton: princeton university press. smith, adam, (1904) an inquiry into the nature and causes of the wealth of nations, london: methuen and co., ltd., ed. edwin cannan, fifth edition. still, bill, (2011). no more national debt. bill-still interactive book. accessed 15 september, 2013: www.billstill.com. the economist. june 16, 2005. the financial times. september, 12, 2013. london, united kingdom. this article suggested banking reforms as a solution. http://www.bis.org/author/stephen_g_cecchetti.htm http://www.bis.org/author/madhusudan_mohanty.htm http://www.bis.org/author/fabrizio_zampolli.htm http://www.nber.org/books/frie57-1 http://www.moneynews.com/ http://click.infospace.com/clickhandler.ashx?du=http%3a%2f%2fwww.nilsonreport.com%2f&ru=http%3a%2f%2fwww.nilsonreport.com%2f&ld=20130507&ap=1&app=1&c=srchresrow3&s=srchresrow3&coi=239138&cop=main-title&euip=212.26.103.207&npp=1&p=0&pp=0&pvaid=ed1bf804171f47139fd02008142bd610&ep=1&mid=9&hash=a5316db7e178c331edb6311de5b9a58b 133 the independent, friday, april 13, 2012. ―now euro zone debt virus starts to overwhelm italy‖. ursua, j and d. wilson (2012). ―risks to growth from build ups in public debt‖, global economics weekly, no 12/10, new york: goldman sachs warner, j. b. 1977. ‗bankruptcy costs: some evidence‘ the journal of finance, 32(2): 337-348. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 6 international prudential regulation, regulatory risk and cost of bank capital 27the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 ijbf international prudential regulation, regulatory risk and cost of bank capital phong t. h. ngo australian national university abstract we define regulatory risk as regulation that leads to an increase in the cost of capital for a regulated firm. in a general equilibrium setting, scholars have shown that uniform increases in capital requirements lead to an increase in the cost of capital. we extend their model to show that when regulatory standards differ across countries, financial integration leads to positive spillovers that reduce the cost of capital mark up for a given increase in bank capital. accordingly, regulatory risk may be greater under a regulatory agreement such as the basel accord, which imposes international uniformity in capital ratios. jel classication: g15, g21 & g28 keywords: basel accord, bank capital regulation, cost of capital, regulatory risk, international harmonization. 1. introduction international finance has witnessed remarkable changes in the past two decades. the removal of restrictions on cross-border capital flows and the opening up of domestic markets to foreign competitors have seen an ever-growing internationalization of financial and banking activities. this is all good given the substantial literature documenting that well-functioning financial markets accelerate economic growth, which in turn, alleviates poverty.1 nevertheless, looking at international financial markets over the past two decades, some have suggested that a potential cost of financial globalization may be the staggering scope of financial crises that have occurred in argentina, indonesia, mexico, russia, thailand, and some south and east asian economies, as the more prominent examples that come to mind.2 1 see levine (1997) for a discussion on finance and growth while dollar and kraay (2000) provide discussion on poverty and growth. the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5828 the economic and social costs of such crises have been studied and are well documented. this has led to calls for reforming financial regulation.3 as such, there has been a wave of new regulations that have hit financial sectors across the globe in recent years. in some cases, regulatory demands have come not just from domestic regulators, but also from international initiatives such as the basel capital accords and european union (eu) directives. the 1988 basel accord was a landmark regulatory agreement. for the first time, regulations affecting banks in many different countries were jointly established (wagster, 1996).4 the ostensible goals of the basel accord were to minimize the systematic risk of the international banking system by limiting the competitive advantage that banks which had belonged to regulatory regimes with less stringent capital standards. this had the effect of leveling the playing field.5 ultimately, the goal of the accord was to remove the funding-cost advantage that japanese banks had, which saw them capture over a third of international lending during the 1980’s. in january 2001, the basel committee issued a proposal for a basel ii capital accord that, once finalized, would replace the 1988 basel i accord. it is now slated for implementation in 2007 by a number of countries. the proposal is based on three pillars. the first is improved minimum capital standards. the second focuses on better, supervisory practices. the third argues for greater market discipline through increased information disclosures by banks. with these, there has been an ever-growing perception of regulation as an additional risk, henceforth known as regulatory risk, which corporations must manage. a recent article published by the banker reporting the results of the 2003 survey conducted by the centre for the study of financial innovation claimed that there is a long term trend of regulation increasingly being viewed as a risk, not only by bankers, but also by non-bankers and analysts. one banker was quoted as saying that “banks will end up managing basel ii rather than managing risk,” while one bank chairman told the magazine the following: “…political dogma, enforced by socialist governments in europe and the u.k. is increasingly damaging the commercial viability of the banking industry. on the other hand, it is common place for the regulators, such as bernie egan of the australian prudential regulation authority, to remark the following: ‘the net cost to shareholders would be quite modest. i would have thought a small price for shareholders to pay....’”. though this may sound like a contradiction in terms, regulation is supposed to minimize risk, not exacerbate it. the concerns of the regulated firms are real and should be examined more closely. in the taxation literature, there is a sound 2 it should be noted that these crises were not born out of financial liberation per se, rather, they have more to do with these countries pursuing .xed exchange rate regimes and other interventionist policies such as the rescue of insolvent banks and poor prudential standards 3 on the costs of recent crises see boyd, kwak, and smith (2000) and hoggarth, resi, and saporta (2001). 29 4 the basel committee on banking supervision oversees the basel accord. the committee was formed in 1974 by the central bank governors of the g-10 countries, along with switzerland and luxembourg, in the aftermath of the failures of the franklin national bank in new york and bankhaus hersatt in germany. the committee is under the auspices of the bank for international settlements and has no formal authority, thus its agreements are carried out on a voluntary basis by the member countries (wagster, 1996). 5 the term systematic risk belongs to the standard rhetoric of economic policy discussions relating to the finance industry. despite the fact that systematic risk is one of the most popular buzzwords in the debate about banking regulation, it is fair to say that there is no precise definition (summer, 2002). prima facie, the literature shows that systematic risk is used as a description of many different phenomena. it is used to describe crises related to financial markets, to bank runs and bank panics, to contagion effects between financial markets. body of evidence that has since the mid-1990s led to parliaments requiring estimates of compliance costs of all tax legislation, before a tax law is enacted. nothing like that has been done for the basel accord cost. at the very least, one should ask whether all the costs of regulation have been taken into account when devising and implementing new policy. this paper adds to the debate by investigating a potential cost arising from regulatory risk of the international harmonization of financial regulation. this paper defines regulatory risk as being any regulatory action that leads to an increase in the ex-ante cost of capital for the regulated. thus, the paper examines the impact that the international capital adequacy requirements introduced by the basel accord may have on the cost of bank capital. the results indicate that uniform increases in capital requirements lead to an increase in the cost of capital. however, when regulatory standards differ across countries, financial integration leads to positive spillovers that reduce the cost of capital mark up for a given increase in bank capital. accordingly, regulatory risk may be greater under a regulatory agreement such as basel. we also find that independent regulators will react to lax regulations of opposing countries by strengthening their own regulatory standards. this is an interesting result since the literature addressing the issue of regulatory competition concludes on the lines that when responding to lax financial regulations in another country, regulators will always choose to reduce regulatory requirements, leading to the so called “race to the bottom”. the next section provides a definition of regulatory risk and discusses the intuition behind and postulates the effects that capital regulation has on the cost of bank capital. a model is then proposed in section 3, while section 4 concludes this paper. 2. regulatory risk and the cost of capital there is much ambiguity surrounding the definition and sources of regulatory risk. to date, there does not appear to be a precise definition. the term regulatory the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5830 risk has been bantered around in policy discussions from just about every regulated industry one would care to think of. from the traditionally regulated, natural, industrial monopolies like telecommunications, water, electricity, gas, and railways, to regulations of environment, banking, finance, and pharmaceuticals, to more obscure regulations such as the alcohol content in the wine industry and the content and disclosure of information on websites. the definitions of regulatory risk span from the risk of adverse government interventions to the risk of a regulatory breach, none of which provide a very useful framework for making predictions or providing us with testable implications. one would expect that, if regulation did indeed lead to greater risks, it would be reflected in the regulated entities’ stock price or the cost of capital. thus, the simplest and most useful definition is: regulatory risk arises whenever regulation affects (increases) the ex-ante cost of capital for a regulated firm, in this case a regulated bank. the cost of capital is the rate of return that the suppliers of capital require from businesses in return for the provision of those funds. any company that cannot achieve these returns will not be able to attract new funding. this required return is determined within the supply-demand equilibrium for funds in an economy. it will also vary between companies and projects according to their risk class, meaning that the higher the level of risk, the higher is the required rate of return. in other words, for a given risk of capital deployment in a firm or a project, the cost of capital can be thought of as the opportunity cost of funds invested in that business. thus, an increase in the cost of capital is analogous to an increase in the risk of a firm or project. when regulators impose higher capital requirements that are binding, capital deficient banks are forced to hold more capital.6 consequently, their value will decline if their capital structure moves away from its optimal level and/or their ability to expropriate deposit-insurance subsidies from the insuring agency is reduced, leading to an increase in their cost of capital. additionally, gorton and pennacchi (1990) argue that bank equity is uniquely costly, and that this cost comes from the role of demand deposits as an efficient means of exchange. the crucial point is, unlike deposits, bank equity is very sensitive to private information, 6 banks with capital holdings above the minimum requirement should not be affected as much as banks with capital holdings below the new requirement. although banks currently hold capital well in excess of what is officially required by regulatory authorities, at the time of inception of the basel accord, almost all banks were capital deficient and were required to raise significant amounts of capital. 7 general equilibrium in the gorton and winton (2000) sense where increases in capital requirements necessitates some agents in the economy to rebalance their portfolios by substituting one kind of bank liability for another. it should be noted that in this paper, the focus is on the cost of bank equity capital leaving aside considerations of the cost of debt capital. it is generally agreed upon that the determination of the cost of debt is relatively straightforward and is generally not a controversial issue (grout, 1995). furthermore, it is frequently established that the cost of equity exceeds that cost of debt and all of the complications related to the estimation of the cost of capital crop up with the measurement of the cost of equity capital owing to tax considerations, asymmetric information, agency costs, and moral hazard problems arising from deposit insurance. 31the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 8 in this framework, gorton and winton (2000) analyzed the impact of a system wide increase in capital requirements on the cost of bank equity. they used a general equilibrium setting where banks produce demand deposits by creating and holding loans. the model is general equilibrium in the sense that market clearing conditions require that bank capital increases are matched by decreases in total deposits as some agents in the economy rebalance their portfolios. their work is unique in their emphasis on the role of bank deposits, which lead to a unique cost of bank equity capital in general equilibrium. much of the earlier, theoretical work in bank capital was partial equilibrium and took the cost of bank equity capital as exogenous. 9 this earlier work provided conflicting answers and tended to look at whether banks increased the risk of their asset portfolios when faced with an increase in the required capital ratio. the regulation of bank capital is often justified to attain a balance between the conflicting objectives of preventing costly bank runs with the moral hazard induced by the deposit insurance (berger, herring and szego, 1995; kaufman, 1991; furlong keeley, 1989; keeley and furlong, 1990). nevertheless, whether the implementation of capital requirements actually reduces bank risk-taking incentives is still a moot point amongst researchers. as an example, authors such as kahane (1977), koehn and santomero (1980), lam and chen (1985), kim and santomero (1988), flannery (1989), genotte and pyle (1991), rochet (1992), besanko and katanas (1996), and blum (1999) argue that actual capital requirements can lead to an increase in bank risk taking behavior. making it a bad medium for exchange. thus, forcing banks to increase capital means that in general equilibrium, consumers must hold more bank equity (the less preferred medium of exchange) in aggregate, and therefore demand a higher expected return on equity in compensation, or put simply, a higher cost of capital.7 3. the model the model proposed in this paper draws from the wisdom of several authors. in their seminal paper, diamond and dybvig (1983) advocated that banks provide liquidity by acting as risk-sharing arrangers to insure against depositors’ random consumption needs. this model, the so-called “diamond-dybvig model”, can be likened with the idea that banks provide liquidity for consumers with uncertain consumption preferences. however, unlike diamond and dybvig (1983), where early liquidation of real assets causes an exogenous loss, the model presented here follows gorton and pennacchi (1990) where information asymmetries between investors lead to a wealth transfer from liquidity traders to strategic traders.1 gorton and pennacchi (1990) point out with reference to jacklin (1987) and cone (1983) that a pivotal assumption made by diamond and dybvig (1983) was that agents could not trade equity claims on physical assets. contrary to this, it is the presence of informed traders that motivates the formation of an intermediary in their model. they suggest that investors with uncertain consumption needs may be required to finance unexpected, high consumption by selling any securities they hold before their cash flows have been realized. to the extent that strategic trading can occur, the investor with such liquidity needs will be exposed to trading losses. however, if investors hold risk-free securities whose payoffs are fully known to all investors, they will not suffer trading losses to strategic traders. consequently, there is a demand for low risk trading securities and banks meet this demand by issuing demand deposits. the important point to note from their paper is that the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5832 relative to bank deposits, bank equity is very sensitive to private information, thus making it relatively costly, bad medium of exchange. the discipline of general equilibrium is essential for their results as it is for the results of this paper. in general equilibrium, two aspects of banking create a tension in setting bank capital levels. on one hand, bank capital is (presumed) needed to prevent the social and private costs of bank failure. on the other hand, if capital requirements are binding, then some agents in the economy must be induced to hold bank equity, which means that they must be compensated for the additional costs associated with having to sell the equity when liquidity needs arise. if regulators increase system wide capital requirements (or as is the case with the basel accord, global increases in capital requirements in the aggregate), investors must hold more equity in their portfolios, increasing the chance that they must sell it to meet consumption needs, thus increasing the cost of capital. in what follows, we model the relationship between private and social incentives to increase bank capital with the extent of regulatory forbearance, looking at the impact of a global increase in capital requirements on the cost of bank capital. the model is largely inspired by gorton and winton (2000). however, we extend their model by (i) looking more closely at how capture and competition in regulation influences regulatory forbearance, (ii) endogenizing regulatory forbearance into the choice of capital requirements, and (iii) incorporating multiple economies. 3.1 single economy model there are four dates in the model economy, t = 0; 1; 2; 3, and a single consumption good. there is a single banking sector with a continuum of homogeneous banks, a continuum of risk neutral consumers, a bank regulator, and many competitive, risk neutral market makers. the following assumptions detail the model. consumer preferences: all consumers are identical as of t = 0. as in diamond and dybvig (1983), each faces a privately observed, uninsurable risk of being an ‘early’ consumer. that is, consumers either consume at date 2 or 3, varying on how likely they are to consume early (i.e. suffer a liquidity shock). at t = 1 consumers realize their probability of being an early consumer (i.e. consume at t = 2), with say probability t. however, the shock itself does not occur until the beginning of t = 2. there is continuum of risk neutral consumers with total mass of 1 who each have a unique type (probability) t, and the total distribution of types is uniform over [0; 1]. it follows that the total mass of consumers who suffer the liquidity shock and must consume early is precisely t. given this framework and defining c t (t) as type t consumption at time t, a consumer of type t has an expected utility at t = 1 of: 33the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 endowments and initial portfolios: at t = 0, all agents receive endowments of one unit of a capital good which is invested to earn a return in the form of a consumption good at t = 3. additionally, at t = 2, agents receive an extra endowment of g units of the consumption good. the focus is on established banking systems that have outstanding loans and equity capital. consequently, the representative bank has issued d 0 deposits promising to pay an exogenous interest factor r d at either t = 2 or t = 3 per unit of capital good invested and n 0 shares, with each shareholder holding n 0 shares. since consumers are risk neutral, they will optimally choose portfolios that are either all deposits or all equity. we will see that selling equity at t = 2 involves a discount, thus, consumers with the lowest probability of consuming early (lowest t) will hold portfolios consisting entirely of equity. as a result, there must exist a marginal shareholder of type t* who is indifferent between holding shares or deposits, such that all consumers with t < t* will hold equity so that in general equilibrium, n 0 = n 0 t*. market clearing implies that the remaining consumers hold deposits, that is d 0 = 1 t*. bank technology: capital is homogeneous, and each unit produces the same random return, . it is assumed that is non-negative and belongs to a symmetric distribution with mean r, distribution h (.), and density h (.), which implies that . the assumption of symmetry is not essential. however, it does simplify later analysis. let us specify an exogenous private bank charter value cp. if the bank is solvent at t = 3 then is cp preserved, and thus divided amongst the shareholders. otherwise, the bank fails and cp is completely lost. gorton and winton (2000) point out that this private charter value is aimed at capturing the informational quasi-rents that banks are assumed to accrue in the lending process by having private information about their lending opportunities. this private information about borrowers is an intangible asset that has value contingent on the bank’s continued operation. moreover, the social cost of bank failure is generally regarded as being greater than the private costs. generally speaking, the reason for this widely held belief stems from the so called ‘contagion effect’, or simply systematic risk, the notion that a bank failure could have knock on effects that cause difficulties for other financial institutions. accordingly, we define the bank’s social charter value as cs ≥ cp. as with private charter value, this social charter value dissipates if the bank fails. regulator: there is a regulator in the economy, such as the central bank who designs regulatory mechanisms with the goal of maximizing aggregate social welfare. deposits are insured, regulators levy lump sum taxes, l, to support deposit insurance and enforce capital standards to mitigate the moral hazard costs of insurance. deposit insurance and capital requirements are both ex-ante mechanisms aimed at reducing the likelihood of bank failure. later, we relax the the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5834 assumption that the regulator’s objective is to maximize social welfare and to include the possibility of capture or competition in regulation. market makers: for simplicity it is assumed that there exist many competitive, risk neutral, market makers who compete for trades by offering bid and ask prices, pb and pa, respectively. 9 bertrand competition ensures that competitive market makers earn zero economic rents so that each price fully reflects the expected value of the share, conditional on it being bought or sold.10 timing: at t = 0, banks exist with d0 deposits offering a rate of rd and n0 shares of equity. at t = 1, regulators announce new capital standards and banks raise more capital. the representative bank raises an amount of capital k1=n1p1, where n1 is the total number of new shares issued and p1 is the issue price per share. since existing shareholders hold portfolios made up entirely of equity, this new capital must come from existing depositors. once again, consumers with the lowest probability of suffering a liquidity shock will find it optimal to hold this new equity. so as capital is raised, the marginal shareholder changes, precisely, there must be a new marginal shareholder who is indifferent between holding deposits or equity so that all consumers with strictly prefer holding only equity. accordingly, depositors strictly prefer equity. if each depositor buys n1 shares such that (i.e. use up all their deposits) then it follows that market clearing implies that k1=d0 d1 (i.e. the increase in capital must equal change in deposits), but we also know that the new deposit level is so it logically follows that . at t = 2, all consumers receive an extra endowment of g and regulators levy lump sum taxes of l to cover expected shortfalls in deposit insurance. new shareholders decide whether to expend resources to acquire private information about their bank’s asset returns at t = 3. the utility cost of gathering information is exogenously set at c units of consumption. we assume for simplicity that old shareholders as ‘insiders’ can acquire this private information with virtually little or no cost whatsoever. shareholders with private information receive a signal (at t = 2) as to whether the bank’s asset returns will be above or below the mean return r at t = 3. immediately thereafter, consumers find out whether they suffer a liquidity shock (consume early) or not (consume late). following this, early consumers exchange their financial claims for units of consumption. shareholders sell their equity to late consumers in exchange for part of the late consumers t = 2 endowment and 9 it is assumed that there is a single bid and ask price regardless of trade size. 10 banks may raise capital either on their own accord or at the demand of the regulator. although, banks can choose either to raise capital or exit the industry (or be forced to exit if they do not meet capital requirements). for now we assume that capital requirements bind and regulators can credibly enforce them. we can then concentrate on the symmetric outcome where banks raise the same amount of capital. 35the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 depositors withdraw their deposits from their banks. note however, that bank assets do not yield consumption until t = 3, thus banks must buy consumption from late consumers to pay for withdraws from early consumers issuing demand deposits to the late consumers in exchange. in aggregate banks simply act as a means for early consumers to trade their deposits to late consumers in exchange for consumption goods at t = 2. at the climax of t = 2, early consumers consume while it is assumed that late consumers can store (hoard) any residual t = 2 endowments with no cost. this ensures that insured deposits will trade at a one for one ratio with consumption at t = 3.11 at t = 3, bank asset returns are realized and divided among investors. if a bank is unable to honor its deposits at face value, then it fails and the private charter value is lost; otherwise, the shareholders implicitly consume the private charter value. finally, late consumers consume. the model is solved recursively. the behavior at t = 3 is elementary, so we begin the analysis at t = 2. 3.1.1 strategic and liquidity trading at t = 2 old shareholders and new shareholders who gather information at a cost of c units of consumption receive a signal informing them whether the bank asset returns at t = 3 are above or below the mean return r. we call this group of investors informed since they can use this information advantageously to trade shares. shareholders who suffer liquidity shocks must sell their shares to meet early consumption needs (regardless of whether they are informed or not). as a result, the informed traders might exploit them. we call this group of shareholders liquidity traders (sellers). note however, that gathering information is a necessary but insufficient condition for a shareholder to be able to act strategically on this information. to be a ‘strategic’ trader, a shareholder needs to have gathered information as well as be a late consumer.12,19 there is perfect information on the ask side of the market since only informed shareholders will buy shares. the model has no liquidity buyers. since informed traders only buy shares if the shares are priced fairly, market makers will be able to infer from the demand for shares (or lack there of) whether returns will be above or below average and set pa accordingly. however, the model does entertain the notion of liquidity sellers, thus, strategic traders can profit from selling shares at bid price when they know that bank returns will be below average.13 let us define as the expected value of the bank’s equity conditional on asset returns being above and below the mean return respectively. assuming there is equal probability that asset returns will be above or below 11 equivalently, we can assume that late consumers are able to invest their residual t = 2 endowments in a risk free asset which yields zero net return. 12 incidentally, old shareholders are the most likely to be able to trade strategically since they can gather information at no cost, and have the lowest probability of early consumption. the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5836 average, the expected value of equity at the end of t = 1 (before any information is received) is and the standard deviation across realizations is . the values of e1and σ depend on whether or not the banks’ promised payments on deposits, rdd1, are above the mean return r. the focus here is on the case where rdd1 < r, which implies that the bank’s ex-ante chance of failure is less than one half, which seems a reasonable assumption. given this assumption, we have the following:14 thus, if a good signal is obtained, the bank survives with certainty, honors deposits at par, retains private charter value, and the returns exceed the mean by . if a bad signal is received, then the bank survives with probability , in which case it honors its debt at par, retains its charter value, and returns are below the mean by . it is straightforward to see that: we now consider the equilibrium information gathering decision and market clearing price at t = 2. suppose a new shareholder (those who buy shares at t = 1) of type t becomes informed, with probability t they will face a liquidity shock and are forced to sell their stock at the bid price pb. thus, with probability 1—t they will be able to act strategically. half of the time, banks asset returns will be above average, so it will pay to hold on to the shares and receive . the rest of the time, bank returns will be below average, in which case they will rationally sell the shares and receive pb. accordingly, their total return is:1 5 13 short sales are ruled out for simplicity. 14 gorton and winton (2000) consider the case where rdd1 > r, however we abstract from this complication in this paper. 37the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 a new shareholder who does not become informed has probability t of selling their shares at bid price to finance early consumption and has probability 1-t that they will be able to hold onto their shares till maturity and earn e1. thus, their total return is: a rational shareholder will gather information if tr i exceeds tr u , which is the same as: the bid price pb is the expected value of a share, conditional on it being sold. now, defining s and λ to be the volume of strategic and liquidity trades respectively, it follows that: consequently, at t = 2, instead of receiving a price of e 1 , because of their relative lack of information, liquidity sellers receive a discounted price, where ceteris paribus, the discount is increasing both in the relative risk of bank equity , and the relative volume of strategic trades . substituting (9) into (8), we have that a new shareholder of type t will gather information if and only if: 15 derivations of equations can be obtained upon request. the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5838 that is, informed sellers receive a profit over and above the share’s true value, which is proportional to the level of risk of bank equity and the relative volume of liquidity trades it is possible that some, none, or all new shareholders satisfy (10). define the new shareholder with the highest t for whom (10) is satisfied (or t* otherwise) to be . we can now calculate s and λ. recalling that informed shareholders sell strategically with probability , and that all old shareholders are informed, then: in sum, given the marginal informed shareholder ti, the equilibrium stock price pb at t = 2 is determined by (9) where strategic trading s is given by (11), and liquidity trading λ is given by (12). the information gathering decision is summarized by the marginal informed shareholder t i . if the left hand side of (10) is negative at t = t* , then no new shareholders become informed and .if the left hand side of (10) is positive at then all new shareholders become informed and . otherwise, ti is a unique value between t* and for which (10) holds with equality, and all new shareholders in become informed. 3.1.2 equilibrium when banks raise capital at t = 1 we now move from the secondary stock market at t = 2 to the primary market at t = 1 in which banks issue additional equity. this section determines the equilibrium issue price for equity p 1 given that all banks raise the same amount of capital. this in turn determines the banks’ cost of capital. as one would expect, p 1 takes into account the equilibrium at t = 2 where early consumers who must sell their shares face possible losses to informed shareholders’ strategic behavior. suppose that all banks raise an amount of capital k 1 by issuing n1 shares at an issue price of p 1 per share. recall that market clearing requires that total deposits fall by the same magnitude as the increase in capital so we have k 1 = p 1 n 1 = d0 = d1. the new marginal shareholder must be indifferent between buying 39the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 a share at price p1 or holding onto p1 units of deposits and eventually earning r d p 1 . since the marginal shareholder depends in part on whether or not they choose to gather information, gorton and winton (2000) assume that (i) the cost of gathering information is sufficiently low so that at least some new shareholders choose to acquire information, and (ii) the marginal new shareholder does not become informed. that is, they assume that . given these assumptions, the typical consumer maximization problem at t = 1 is tooptimally choose their amount of new shares n1 and new level of deposits d 1 to maximize expected utility taking as given the issue price of equity p 1 and initial level of deposits d 0 . that is, the consumer problem is: subject to the first constraint is an early consumers’ budget constraint at t = 2. they consume by withdrawing deposits at rd per unit, selling their share in equity n1 at bid price p b = (1− δ)e 1 , consuming their t = 2 endowment g, less lump sum taxes l. the second constraint is a late consumer’s budget constraint at t = 3, save they do not face potential losses to strategic traders and get the full value of their total equity. three says that the consumer purchases their new shares with existing deposits. four and five are non-negativity constraints. the first-order condition yields the following result. proposition 1: (equilibrium and the cost of bank capital at t = 1) suppose at t = 1 all banks raise capital by k1, then all consumers with t < ^t use all their deposits to buy shares such that each buys shares. then the equilibrium price of bank equity p 1 satisfies the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5840 proof. see gorton and winton (2000). as noted, the new marginal shareholder must be indifferent between buying one share out of n 0 +n 1 shares or holding onto p 1 units of deposits, eventually earning r d p 1 (the value of the former is given by the right hand side). this marginal shareholder’s indifference condition is what yields equation (14). also, because consumers are risk neutral, all inframarginal shareholders strictly prefer holding portfolios consisting only of deposits. proposition 1 states that, due to the presence of strategic traders when some consumers face a liquidity shock at t = 2, knowing this at t = 1, depositors demand mark up in return of per dollar of deposit to induce them to buy new shares (they demand a lower price per share or equivalently a higher expected return on equity in compensation). since banks need to raise k 1 dollars of equity in total, the aggregate, expected value of new shareholders’ share of the total value of equity is 3.1.3 social and private welfare at t = 1 we now turn the analysis to the incentives of regulators and banks to raise capital at t = 1. for now, it is assumed that regulators have the objective of maximizing social welfare, that is, maximize aggregate welfare of all agents assigning equal importance to the welfare of shareholders and depositors. define w s (k 1 ) to be the aggregate, expected, social welfare function as a function of the new amount of capital raised at t = 1. in this model, raising capital is beneficial to social welfare because it reduces the probability of bankruptcy. thus, net of information gathering costs we assume that raising capital improves social welfare.2 more precisely, we are assuming that the social welfare function that regulators are attempting to maximize is increasing in capital. that is, the decision as to whether an individual bank raises capital or not is made by the banks old shareholders (those who already held shares at t = 0). assume that shareholders act to maximize their private welfare w p (k 1 ), which is a function of the amount of capital raised at t = 1. it has been shown that raising capital is costly for banks and we assume that these costs outweigh the potential benefits. that is, we assume the combination of liquidity effects and strategic trading means that issuing equity is always bad for old shareholders. intuitively, due to the existence of negative externalities from bank failure, debt overhang effects created 41the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 16 gorton and winton (2000) show that information-gathering costs are a dead weight loss. 17 gordon and winton (2000) derive specific functional forms for ws (k1) and wp (k1). by deposit insurance (myers, 1977) and the combined costs of liquidity and strategic trading effects of equity issuance makes issuing additional equity to support higher capital standards always less attractive for shareholders than it is for the regulator. that is 16 the preceding discussion has presented the baseline of gorton and winton (2000) model where we were able to show that increasing capital requirements leads to a discount on the newly issued share price, or equivalently, an increase in the cost of bank equity capital. 17 it was also assumed that the social incentive to increase capital is greater than that of private incentives, which naturally create tensions between regulators and banks when setting capital requirement standards. in this setting, we next analyze the impact of capture and competition in regulation, linking it to regulatory forbearance. 3.2 capture, competition and forbearance in regulation in the preceding sections, we presented a model in which the cost of bank equity capital is endogenously determined in general equilibrium. an important assumption made was that regulatory objectives were to maximize aggregate social welfare ws (k1), and that regulators were able to enforce banks to raise an exogenous amount of capital k1. in this section, i allow for differences in regulatory objectives that endogenizes the exercised levels of forbearance, and therefore, the levels of capital that banks must raise. i define forbearance to be regulators reneging of previously tough policies, or simply, forbearance is the lax enforcement of capital requirements. we begin the analysis in this section by appealing to two facts. first, the political economy of regulation implies that some regulators are more closely aligned with an interest group, such as bank owners. this ‘public choice’ or ‘capture’ view emphasizes the important role of interest groups in determining regulatory outcomes. although most authors attribute this intellectual wisdom to stigler (1971) and later peltzman (1976) and becker (1983), its origin has deep roots and can be traced back to marx who argued that big businesses sought and paid for control of important economic institutions (laffont and tirole, 1991). empirical evidence supporting this hypothesis can be found in kane (1990) who documented proof of regulatory capture in the united states, the savings and loans debacle, and white’s (1982) commentary on the evolution of banking regulation in the united states. more recently, slinko et al. (2004) provide evidence of capture across a number of industries in russia. the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5842 second, to the extent that regulators are concerned about the old shareholders of their domestic banks, they may lower regulatory requirements in order to provide them with an advantage over foreign banks (acharya, 2003; dell, ariccia, and marquez, 2001; holthausen and rønde, 2003). niskanen (1971) amongst others provide another rationale for competition in regulation. given the possibility of international banks engaging in regulatory arbitrage by switching to lower-cost regulators to reduce the costs of compliance, for relatively stringent (high-cost) regulators, this regulatory competition will see their regulatory market shares eroded. accordingly, in view of protecting their market share, these high-cost regulators will either reduce their regulatory burdens or increase regulatory subsidies. this type of regulatory race to the bottom or competition in laxity as regulators attempt to attract business into their domain with ever more large regulatory standards has been the catalyst for arguments in favor of international harmonization of regulatory standards before reaching the bottom of the barrel. although this issue is not formally dealt with until the next section when multiple economies are incorporated into the model, it should be noted here that this competition in laxity can also enter the regulator’s objective function. accordingly, we generalize the regulator’s objective function to one that maximizes the weighted average of aggregate welfare for old shareholders and aggregate social welfare, with weights respectively, thus: the parameter can be thought of as the degree of regulatory capture (or regulatory competition). if , then , which corresponds to a situation where regulators are completely captured and care only about existing shareholders. while implies that , which represents the normative case of aggregate social welfare maximization described thus far in the paper. because of government ownership of banks, government influences over regulatory agencies as well as the time inconsistency in enforcing ex-ante optimal regulatory policies, which typically benefits bank owners, many regulators would be classified as having a weight of . we recognize the impact of regulatory capture or competition (or some combination of both) on the exercised level of forbearance and consequently the amount of capital that regulators require banks to raise at t = 1 in the following result: lemma 1 (capture and forbearance) as the degree of regulatory capture increases, the level of exercised forbearance increases, and thus the amount of capital that banks are required to raise by regulators decreases. that is, the amounts of capital banks are required to raise at t = 1 is decreasing in regulatory capture or dk1d . 43the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 proof. see appendix. the intuition for this result is straightforward. we assumed that the increase in aggregate social welfare for a given increase in the level of capital is always greater than the change in aggregate private welfare. therefore, as regulatory capture increases, the regulators objective function becomes increasingly aligned with that of the old shareholders who have a preference for lower capital requirements. whether it is because regulators are more captured or are competing more intensely (or some combination of both), this setup allows us to analyze situations where regulators from different countries have differing ‘tastes.’ for regulation and place greater or less emphasis on existing bank shareholders versus aggregate social welfare. this is the issue i take up in the rest of the paper. 3.3 two-country model financial integration generates the potential for ‘spillovers’ from one country to another arising from differing regulatory practices. to study these potential spillovers from one country’s regulations to other countries and their regulations, i extend the model to two countries. consider two countries, a and b. the banking sector in each country consists of a continuum of homogeneous banks, a continuum of risk neutral consumers uniformly distributed on [0; 1], many competitive, risk neutral, market makers, and a regulator, as in the single economy case already presented. banks operate across countries, offer the same exogenous deposit rate r d , and have equal access to deposit and lending opportunities. there are no restrictions on whether consumers invest (buy shares or deposits) in their domestic or foreign bank. the consumers in the two countries suffer independent liquidity shocks, however, these shocks occur with the same probability t in each country. regulators may adopt regulatory policies with differing levels of forbearance toward the banks chartered in their respective countries. the differing levels of forbearances reflect the degrees of capture in each country and are denoted and , respectively.. under this setting, i ask myself what kind of effect the level of capture in country a will have on the cost of equity capital of the banks in country b. first, i assume that the regulator in country a is relatively more captured, that is . in this situation, i show that the cost of equity capital for country b in banks is decreasing in the level of capture (forbearance) of country a’s regulator. this effect is called the spillover from country a to country b. the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5844 lemma 2 (the new marginal shareholder) ceteris paribus, if then and the new marginal shareholder for banks chartered in country b is proof: see appendix. allowing for a difference in forbearance between the two countries captures the institutional reality that although the basel accord has taken large steps in attempting to harmonize capital standards globally, individual countries have maintained large discretion on the determination of which assets can be used to meet capital requirements.18 lemma 1 states that the exercised level of forbearance increases as the level of regulatory capture increases resulting in a lower required increase in bank capital. it follows that if the regulator in country a is relatively more captured then they must also exercise a higher level of forbearance, thus demanding a lower increase in bank capital at t = 1. in other words, assuming implies that . it is straightforward to see from proposition 1 that if these two countries were segmented then the cost of capital in country a will be lower than the cost of capital in country b. this result follows from the fact that because regulator a is relatively more captured and therefore requires less capital to be raised at t = 1, the new marginal shareholder in country a, ^ta, has a lower probability of suffering a liquidity shock relative to country b, and thus, values bank equity more highly than the marginal shareholder in country b, ^tb. however, when the two countries are financially integrated, then banks from country b can (in a cost of capital sense) benefit from country a’s lax enforcement of capital requirements. recall that when the countries are segmented, then . it follows that, relative to country a, there must be more consumers in country b who become new shareholders at t = 1. given that consumers in both countries are identical, conditional on their probability of sobering a liquidity shock, they will have the same valuation of the banks in country b. consequently, when the countries become integrated, it is rational for banks in country b to issue new stock to consumers who value it most highly (have a lower probability of consuming at t = 2) regardless of whether they are from country a or b. accordingly, instead of issuing all new equity only to consumers in country b, banks in country b will raise capital domestically up to a level that is equal to the amount that country a’s banks raise (domestically), and then raise the additional required amount of capital demanded by the regulator in country 18 gorton and winton (2000) argue that actions by national countries to broaden the definition of capital has considerably weakened capital adequacy standards and led to significant heterogeneity in the definition of capital across countries 45the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 b. that is, the amount of capital over and above the amount demanded by the regulator in country a by issuing new equity to consumers in both countries. in fact, banks in country b will raise exactly half of the remaining required capital from consumers in country b and the other half from consumers in country a. in other words, the new marginal shareholder for country b’s banks will lie exactly half way between the new marginal shareholder for country a, , and the new marginal shareholder for country b when the countries are segmented, ^tb. explicitly, the new marginal shareholder for country b’s banks when the countries are integrated will be recall however that this mark up in the cost of capital demanded by new shareholders at t = 1 stems from the presence of strategic traders when some consumers face a liquidity shock at t = 2. so, aside from the marginal consumer’s probability of suffering a liquidity shock (which determines the marginal valuation of new bank equity), we also need to consider the possible affect that this spillover of depositors from country a to country b may have on the lemons discount on the share price at t = 2 in the secondary market. remember that the liquidity sellers receive a discounted bid price pb determined by equation (9) where the discount was given by . with this in mind, we have the following result. lemma 3 (equity discount at t=2 when countries are integrated) relative to the scenario where the countries are financially segmented, with an unchanged relative risk of the bank in country b, , financial integration decreases the volume of liquidity trades, _b, proportionately more than the decrease in volume of strategic trades, sb, resulting in the relative volume of strategic trades, sb sb+_b, increasing. consequently, the t = 2 discount for country b.s bank equity increases. that is economies discount. proof: see appendix. it is straightforward to see that the relative risk of the bank in country b does not change. the expected value of equity at the end of t = 1, , and the standard deviation across realisations, , are determined by equations (4) and (5) respectively. we can see from equations (4) and (5) that the only endogenous factor that will change the expected return or risk of the .rm is the new level of deposits that they hold at t = 1, d 1 , all other variables are assumed exogenous. although integration allows banks in country b to issue their stock to consumers who value it most highly in both countries, the absolute number of consumers who give up their deposits in exchange for equity remains unchanged, and thus, the new deposit level is unchanged post integration. integration means that banks in country b will sell some of its t = 1 equity to consumers in country a. this reduces the probability of the marginal (and the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5846 inframarginal) shareholder suffering a liquidity shock and collectively decreases the aggregate probability that new shareholders will have to consume early leading to a reduction in the volume of liquidity traders. this reduction in the volume of liquidity traders will, ceteris paribus, reduce the incentive for new shareholders to gather information as given by equation (10). of course, this leads to a reduction in the number of strategic traders. however, one would logically expect the fall in volume of strategic traders to be smaller than the fall in the volume of liquidity traders. this is because we originally assumed that there was a unique shareholder for which the information gathering condition – equation (10) held with equality. that is, the new marginal informed shareholder lay somewhere between the old marginal shareholder and the new marginal shareholder. thus, it would be logical to expect that post integration the marginal shareholder would once again locate somewhere between the new marginal shareholder and the old marginal shareholder which determines the number of strategic traders. accordingly, the t = 2 discount on country b’s bank equity will be higher. the preceding discussion established that post integration the marginal shareholder ~tb has a lower probability of suffering a liquidity shock than the segmented economies marginal shareholder ^tb, and therefore, values country b’s bank equity more highly, thus demands a smaller discount on the issue price or equivalently a lower mark up in the expected return (cost of capital) in compensation. on the other hand, the increase in the t = 2 discount will lead to an increase in the cost of capital. these two effects offset each other, so before any result can be presented we need to make the following qualifying assumption. it is assumed that the effect on the cost of capital from the reduction in the marginal shareholder’s probability of consuming early is greater than the effect of the increase in the t = 2 discount. this assumption seems plausible since the reduction in the new marginal shareholder’s probability of suffering a liquidity shock enters directly into the cost of capital calculation as given by equation (14), while the increase in the t = 2 discount is less direct. the increase in the t = 2 discount comes about because of the change in the new marginal shareholder’s probability of consuming early, which then reduces the number of liquidity traders, then impacts on the number of new shareholders who choose to gather information, which finally reduces the volume of strategic trades before the increase in the t = 2 discount enters the cost of capital calculation. with this in mind, we have the following result. proposition 2 (international spillover) ceteris paribus, in financially integrated economies, the cost of capital for banks chartered in country b is decreasing in the level of capture of country a.s regulator. 47the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 proof. see appendix. this result tells us that, other things being equal, as the regulator in country a becomes more captured or is competing more intensely with other regulators to either maintain their regulatory market share or favor banks chartered in their country or some combination of all three. whatever the reason, the level of exercised forbearance increases. with this, the amount of capital that the regulator in country a requires to be raised declines. by lemma 2 this leads to a reduction in the probability of the new marginal shareholder of country b’s banks experiencing a liquidity shock while lemma 3 suggests that the discount on country b’s bank equity at t = 2 increases. in aggregate, it is argued that the former effect outweighs the latter leading to an overall decrease in the cost of bank equity for banks chartered in country b. specifically, we argued earlier that in a single economy model proposition 1 tells us that due to presence of strategic traders when some consumers face a liquidity shock at t = 2, knowing this at t = 1, depositors demand a marked up expected return (cost of capital) of per dollar of deposit to induce them to buy shares. proposition 2 tells us that when countries are financially integrated and capital regulations are not credibly enforceable, then differences in regulatory capture and forbearance lead to depositors demanding a smaller mark up on the cost of capital of per dollar of deposits, where . this discussion also gives us the following welfare result. corollary 1 (international spillover and welfare) relative to the case when the two countries are segmented, the spillover of depositors from country a to country b which results in the reduction in the mark up in the cost of capital demanded by new shareholders of b’s banks leads to an increase in both private and social welfare in country b. however, there is no change in private welfare in country a and the impact on social welfare in country a is ambiguous. since consumers are risk neutral, summing across all shareholders the expected trading profits of strategic traders cancels out with the expected losses of liquidity traders, that is, trading is a zero sum game. nonetheless, since old shareholders have the lowest probability of suffering a liquidity shock they are the best placed to trade strategically on their private information. consequently, on average, old shareholders have a net trading gain while new shareholders have a net trading loss. recall our assumption . consequently, although old shareholders face a net trading gain at t = 2, we implicitly assume that the discounted share price (higher cost of capital) on the issue of new equity at t = 1 more than offsets the t = 2 trading gains.19 19 gorton and winton (2000) show that this is indeed the case. the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5848 with this in mind, lemma 3 tells us that the spillover of depositors from a to b reduces the relative volume of liquidity trades and thus reduces expected trading profits to old shareholders. this effect is however, countered by the reduction in the aggregate discount on new equity (proposition 2) and by assumption, the reduction in the aggregate discount on new equity will be greater than the fall in the expected trading profits (including trading costs), leaving the old shareholders in country b better off, post integration. social welfare in country b also increases due to a reduction in the aggregate cost of gathering information since there is a decline in the number of consumers in country b who firstly, buy new equity, and secondly satisfy equation (10), choosing to become informed. there is no change in the level of capital held by a’s bank’s post integration, nor is there a change in a’s new marginal shareholder’s probability of suffering a liquidity shock. thus, the valuation of a’s banks remains constant. on the other hand, some depositors in country a now hold new equity in country b’s banks, leading to a fall in social welfare. this occurs for two reasons. first, the new shareholders of b’s banks who belong to country a have the highest probability of suffering a liquidity shock and are thus more likely going to be exploited by b’s old shareholders. they have an expected net trading loss while b’s old shareholders have a net expected trading profit. second, some of b’s new shareholders in country a will choose to gather information, which as previously mentioned is a dead weight loss, imposing a further welfare cost on country a. this negative impact on social welfare in country a must be weighed up with a positive spillover that arises from the fact that some depositors in country a now hold shares in b’s banks which are less likely to fail than banks in country a. these contrasting influences leave the impact on the social welfare of country a ambiguous. two qualifying points are in order before concluding this section. first, in a world where countries are symmetric and a bank’s investment decision (risk taking) is exogenous, capital adequacy requirements have no effect on bank risk taking incentives. this lack of flexibility is the primary source of spillover of depositors from a to b.30 of course, if we were modeling the investment side of the story, then lower capital standards will potentially induce greater risk taking by a’s banks leading to negative spillover from a to b. though whether capital regulations actually reduce bank risk taking is far from being agreed upon. (see for example kahane (1977), koehn and santomero (1980), lam and chen (1985), kim and santomero (1988), flannery (1989), genotte and pyle (1991), rochet (1992), besanko and katanas (1996), and blum (1999). second, we have established that regulation adopted by country a has an externality on the banks in country b. if each regulator is concerned only about maximizing their own welfare function, w_, this externality will in general, not be internalized in the absence of coordination. the question we need to ask (similar in vain to the conclusions reached by acharya, 2003; dell, ariccia and marquez, 2001; holthausen and rønde, 2003) is whether or not both countries increase their 49the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 forbearance, thereby, producing welfare externalities for the other country, potentially a new equilibrium outcome. that is, will regulators race to the bottom in an attempt to either promote their domestically chartered banks or protect their own regulatory market share? this question is addressed and formalized below. 3.4 a race to the bottom? the previous discussion was made on ceteris paribus grounds. however, when regulators have discretionary powers, they can respond to regulatory choices of other regulators. the literature addressing this issue has to the best of my knowledge has almost unanimously concluded on the lines that, when responding to lax financial regulations in another country, regulators will always choose to reduce regulatory requirements, leading to the so called race to the bottom or regression to the worst regulation. in this section, i address whether the spillover caused by financial integration presented in the preceding discussion will lead to increased forbearance on a global scale. the assumptions in the previous section remain here, so the two countries are identical save that country a is more captured than country b, or . given the potential for spillovers to occur, each independent regulator’s welfare function now depends not only on their chosen level of capture or forbearance (and hence level of capital required to be raised at t = 1), but also on the level of capture in the opposing country. hence, the maximization problem for regulator b is to choose optimally their level of capture given that regulator a has optimally chosen its level of capture. we see that: with this in mind, what is b’s optimal response to a’s lax enforcement of capital requirements? we have the following result: proposition 3 (no race to the bottom) in equilibrium, the regulator in country b decreases capture and thus forbearance upon the increase in the capture of the regulator in country a. that is @_b @_a < 0. proof: see appendix. this is a striking result and is in stark contrast to any conclusion previously reached addressing this issue. this result states that, from a cost of capital point of view, instead of reacting to a’s laxity with an increase in its own forbearance vis-à-vis capital requirements, regulators in country b will in fact become less captured, and thus be less forbearing towards banks chartered in country b, the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5850 resulting in a demand of a higher level of capital to be raised at t = 1. intuitively, this proposition holds for two reasons. relative to the segmented country scenario, integration increases private and social welfare effects of a chosen level of capture. it follows that for country b to reap as much benefit from integration as possible it makes sense for regulator b to align itself less with bank shareholders and demand stricter capital requirements. second, the spillover of depositors from a to b increases the foreign ownership of b’s banks. one would logically expect that domestic bank regulators to be less concerned about shareholders the higher the fraction of foreign ownership, consequently putting a greater weighting on social welfare and demand higher capital levels. this result casts doubt (at least in the context of this model) on the argument that independent regulators will react to lax regulations of opposing countries by weakening their own regulatory standards. i have demonstrated here that it is at least theoretically possible that a country will react to lax regulations of other countries by strengthening their own regulations. 4. conclusion in recent years, the progressive integration of international financial markets has brought with it an increasing focus on the coordination of regulations across national borders. this trend has spurred a debate on the costs and benefits of the international harmonization of banking regulations. the basel capital accord of 1988 provides a case in point in regard to the voracity of the debate between the opposing sides. this paper adds to this debate by investigating the now common claim by banks that the regulations they face induce an increase in risk .a regulatory risk. the paper defines regulatory risk to be regulations that lead to an increase in the cost of equity capital. accordingly, the cost of capital is modeled in a general equilibrium setting where demand deposits serve a unique role in providing an efficient medium of exchange. although bank capital reduces the probability of failure, bank equity is relatively sensitive to information, thus making it a poor hedge against liquidity risk. consumers initially hold deposits to the extent they need coverage against liquidity shocks. those who have the highest probability of being early consumers optimally choose to hold deposits. a system wide increase in capital will, in the context of the model, force some depositors to exchange their deposits for their less preferred bank equity. this increases the probability that the marginal shareholder will suffer a liquidity shock and thus leads to a mark up on the cost of capital. however, when regulatory standards differ across countries, a spillover of depositors from the lax to more stringent country leads to a reduction in the cost of capital mark up demanded by new shareholders for the stricter country. the bank belonging to the country with stricter enforcement of regulatory standards 51the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 is, in effect, able to diversify their new equity issuance across countries by selling their shares to consumers with the lowest risk of suffering a liquidity shock. this result suggests that an international regulatory agreement such as the basel capital accord that imposes uniformity on capital requirements may indeed lead to a higher cost of capital for banks and thus induce greater regulatory risks relative to a decentralized solution. it suggests that the claims made by banks and other financial institutions are at least a theoretical possibility. another interesting result was the finding that countries will respond to lax enforcement of regulations by neighboring countries by strengthening their own regulatory standards. this result is in stark contrast to the almost unanimous conclusion reached by other authors tackling this issue that national regulators will react to lax regulations of opposing countries by weakening their own regulatory standards leading to the commonly coined phrase .a race to the bottom vis à vis regulatory standards. the findings of this paper support the arguments of the regulated entities. however, it does not go so far as to make a value judgment as to whether a decentralized approach to capital adequacy regulation is optimal relative to the harmonizing of capital standards or vice versa. it may be the case that a unified framework such as basel is optimal, but it is also possible that banks in some countries are more efficient and stable than others, thus necessitating differing regulatory standards. there is no reason a priori why we should choose the centralized solution over the decentralized one and vice versa. the paper simply presents testable implications that warrant empirical investigation, only in conjunction with comprehensive empirical studies will the tensions between regulators and the regulated firms are resolved. acknowledgement: this article is an edited version of one of the three best papers selected after blind review process at the 17th fma-asian finance association meeting, auckland, new zealand, 2006. the author is grateful for the reviewer’s editorial comments, which assisted the editors to revise my original version of this paper. all remaining errors are my responsibility. author statement: phong t. h. ngo, school of economics, hw ardnt building 25a, australian national university, act 0200, australia. phone: +612-61254487. email: phong.ngo@anu.edu.au. the paper benefited from the comments given by tom smith, chris jones, shane evans, colleen cassidy, pornpen sodsrichai and seminar participants at the australian prudential regulation authority, school of finance and applied statistics (anu) and the cbe/apseg seminar series at the anu. this research was funded by the generous support of an anu research scholarship as well as the brian gray scholarship which is jointly funded by the australian prudential regulation authority and the reserve bank of australia. the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5852 references acharya, v. v., (2003). is the international convergence of capital adequacy regulation desirable? the journal of finance 58: 2745-2781. berger, a. n., r. j. herring and g. p. szego (1995). the role of capital in financial institutions, journal of banking and finance 19: 257-276. becker, g. s., (1983). a theory of competition among pressure groups for political influnce, quarterly journal of economics 98: 371-400. besanko, d., and g. katanas (1996). do capital adequacy requirements reduce risk in banking? journal of financial intermediation 5: 160-183. blum, j., (1999). do capital adequacy requirements reduce risks in banking? journal of banking and finance 23: 755-771. boyd, j. h., s. kwak and b. d. smith (2000). banking crises: what comes after?. mimeo, university of minnesota, department of finance (carlson school of management). cone, k., (1983). regulation of depository institutions. ph.d. thesis, standford university. dell.ariccia, g., and r. marquez (2001). competition among regulators and credit market integration, working paper, university of maryland. diamond, d. w., and p. h. dybvig (1983). bank runs, deposit insurance and liquidity, journal of political economy 91: 401-419. dollar, d., and a. kraay (2000). growth is good for the poor. mimeo, the world bank. flannery, m., (1989). capital regulation and insured banks. journal of monetary economics 24: 235-258. furlong, f. t., and m. c. keeley (1989). capital regulation and bank risk taking: a note, journal of banking and finance 13: 883-891. genotte, g., and d. pyle (1991). capital controls and bank risk, journal of banking and finance 15: 805-924. gorton, g., and g. pennacchi (1990). financial intermediaries and liquidity creation, the journal of finance 45: 49-71. gorton, g., and a. winton (2000). liquidity provision, bank capital and the macroeconomy, working paper 9806, university of minnesota, institute of financial studies (carlson school of management). grinblatt, m. s., and s. a. ross (1985). market power in a securities market with endogenous information, quarterly journal of economics 100: 11431167. grout, p., (1995). the cost of capital in regulated industries in regulatory challenge, edited by m. bishop, j. kay and c. mayer, oxford university press, chapter 16 pp.386-407. 53the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 hoggarth, g., r. reis and v. saporta (2001). costs of banking system instability: some empirical evidence, working paper, bank of england. holthausen, c., and t. rønde (2003). cooperation in international banking supervision, working paper 2004-02, centre for industrial economics, institute of economics, university of copenhagen. jacklin, c., (1987). demand deposits, trading restrictions, and risk sharing, in e. d. prescott and n. wallace, (eds) contractual arrangements for intertemporal trade, university of minnesota press, minneapolis. kahane, y., (1977). capital adequacy and the regulation of financial intermedi aries. journal of banking and finance 1: 207-218. kane, e. j., (1990). principle-agent problems in the s&l salvage, the journal of finance 45: 755-764. kaufman, g. g., (1991). capital in banking: past present and future, journal of financial services research 5: 385-402. keeley m. c., and f. t. furlong (1990). a re-examination of mean variance analysis of bank capital regulations, journal of banking and finance 14: 69-84. kim, d., and a. m. santomero (1988). risk in banking and capital regulation, the journal of finance 35: 1219-1233. koehn, m., and a. m. santomero (1980). regulation of bank capital and portfolio risk, the journal of finance 35: 1235-1250. kyle, a. s., (1985). continuous auctions and insider trading, econometrica 53: 1315-1336. lafont, j., and j. tirole (1991). the politics of government decision making: a theory of regulatory capture, the quarterly journal of economics 106, 1089-1127. lam, c. h., and a. h. chen (1985). joint effects of interest rate deregulation and capital requirements on optimal bank portfolio adjustments, the journal of finance 45: 563-575. levine, r., (1997). financial development and economic growth: views and agenda, journal of economic literature 35: 688-726. myers, s. c., (1977). determinants of corporate borrowing, journal of financial economics 5: 147-175. niskanen, w. a., (1971). bureaucracy and representative government, aldineatherton: chicago. peltzman, s., (1976). towards a more general theory of regulation, journal of law and economics 20: 229-263. rochet, j., (1992). capital requirements and behaviour of commercial banks, european economic review 36: 1137-1178. the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5854 slinko i., e., yakovlev and e. zhuravskaya (2004). laws for sale: evidence from russia, mimeo, stockholm school of economics; center for economic and financial research in moscow. summer, m., (2002). banking regulation and systematic risk.. working paper 57, oesterreichische national bank. stigler, g. j., (1971), .the theory of economic regulation, bell journal of economics and management science 2: 3-21. wagster, j. d., (1996) impact of the 1988 basel accord on international banks, the journal of finance 51: 1321-1346. white, e. n., (1982), .the political economy of banking regulation, 1864-1933, journal of economic history 42: 33-40. 55the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 appendix: proofs proof. (lemma 1) consider the regulator’s unconstrained design problem given by (15). a regulator chooses a level of capital k 1 in order to maximize their welfare function wp (k1) given the level of capture θ . the first order condition yields taking the derivative of the first order condition w.r.t. θ gives the following strategic interaction condition proof. (lemma 2) suppose that θ a > θ b then by lemma 1 . consider first the case where the two countries are segmented. let ^ta and ^tb be the new marginal shareholder for country a and b respectively then we must have . thus relative to a, there must be more new shareholders in b. however,, when the two countries become integrated then b’s banks will sell its new stock to the consumer that values it most highly irrespective of which country they belong to. since equity valuation is inversely proportional to the probability of suffering a liquidity shock it pays for b’s banks to issue some of its new stock to the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5856 consumers in a who have and will therefore value b’s shares more highly. consequently, b’s banks should sell exactly half of the excess capital required over an above the amount raised by a’s banks to a’s consumers and raise the rest domestically. that is, b’s banks should sell shares to a’s consumers and raise the rest domestically. it follows that the new marginal shareholder for b’s banks post integration will be = . proof. (lemma 3) the expected value of equity an the end of t = 1, eb1 , and the standard deviation across realizations, _b, are determined by equations (4) and (5) respectively. from equations (4) and (5), the only endogenous factor that will change the expected return or risk of the firm is the new level of deposits that they hold at t = 1, d 1 , all other variables are assumed exogenous. from previous discussion we have that , so the only factor determining d b 1 is . prior to integration , post integration , but so substituting this in gives us which is the same as the pre-integration level. to see the reduction in the volume of strategic trades we restate equation (12) for a single economy here however, when the two countries are integrated and there is spillover of depositors from a to b then 57the international journal of banking and finance, vol. 5.(number 1): 2007: 27-58 however, since n1t is a linear monotonic increasing function and with this cannot be true, thus by contradiction the reverse statement must be true, that is this reduction in the volume of liquidity traders will, ceteris paribus, reduce the incentive for new shareholders to gather information as given by equation (10). of course this leads to a reduction in the number of strategic traders. however, one would logically expect that the fall in volume of strategic traders to be smaller than the fall in the volume of liquidity traders. this is because we originally assumed that there was a unique shareholder for which the information gathering condition equation (10) – held with equality. that is, the new marginal informed shareholder lay somewhere between the old marginal shareholder and the new marginal shareholder. thus, it would be logical to expect that post integration the marginal shareholder would once again locate somewhere between the new marginal shareholder and the old marginal shareholder which determines the number of strategic traders. accordingly, the t = 2 discount on country b.s bank equity will be higher. proof. (proposition 2) if then by lemma 1 . consider once again the case where the two countries are segmented. let and be the new marginal shareholder for country a and b respectively then proposition 1tells us that the mark up on the cost of capital for new equity issuance at t = 1 is rd 1__^t_ per dollar of deposit. therefore, the cost of capital mark up per dollar of deposit for a is less than that of b.s, that is, . now if the countries become integrated then lemma 2 tells us that post integration , that is, the probability of the marginal shareholder in country b experiencing a liquidity shock decreases. while lemma 3 suggests that post integration , that is, the t = 2 discount (due to the existence of strategic traders) on b.s equity traded in the secondary market increases. assuming that it follows that, post integration, . in other words, the cost of capital mark up per dollar of deposit is lower when the two countries are integrated with . moreover, the larger _a is, ceteris paribus, the greater the spillover of depositors and thus the greater the decline in b.s new marginal shareholders liquidity risk (lemma 2) and consequently an even larger reduction in the cost of capital mark up. proof. (proposition 3) consider the regulator’s unconstrained design problem given by (16). regulator b chooses a level of capture _b which in turn determines the level of capital raised kb 1 in order to maximize their welfare function given that regulator a has chosen its level of capture optimally . the first order condition yields the international journal of banking and finance, vol. 5.(number 1): 2007: 27-5858 first, most of the welfare gains post integration accrue to the old shareholders of the banks in country b, they receive a reduction in the aggregate discount on new equity while the only benefit to social welfare is a reduction in aggregate information gathering costs thus second, we have assumed concavity of welfare functions so that and note also that leading to the desired result. to see this consider once again the first order condition given by (a13). differentiating this w.r.t. again gives now likewise recall from our assumptions that and lemma 1 tells us which implies that and leading to the desired result. taking the derivative of the first order condition w.r.t. θ a gives the following strategic interaction condition international journal of banking and finance 1-1-2007 international prudential regulation, regulatory risk and cost of bank capital phong t. h. ngo recommended citation east asian financial contagion under dcc-garch the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 17 east asian financial contagion under dcc-garch j. h. cho and a. m. parhizgari florida international university abstract we reconsider the definition and measurement of contagion by analyzing the 1997 east asian financial crisis in the equity markets of eight countries using dynamic conditional correlation (dcc). taking thailand and hong kong as alternative sources of contagion, a total of fourteen source-target pairs is analyzed. we define contagion as the statistical break in the computed dccs as measured by the shifts in their means and medians. in the dcc process, the parameters of each pair of source-target country contagion are allowed to vary and be dictated by the data. contagion is tested using dcc means and medians difference tests. our findings indicate the presence of contagion in the equity markets across all the fourteen pairs of source-target countries that are considered. key words: contagion, east asian financial markets, dynamic conditional correlation jel classification codes: g15, f36, c51 1. introduction defining an evolutionary process in the context of cross-country analyses, the term contagion has gone through a gradual refinement and measurement process only in the last two decades or so. in the early days, a simple (static) measure of correlation, for instance, between the stock return series of two countries, was deemed informative enough to establish the relation between their respective equity markets, and thereby aid in decisions on cross-country portfolio diversification. the construction of such portfolios has indeed been geared, for the most part, to a static measure of correlation. further developments in correlation analyses have progressively led into new measures and techniques including co-movements, causality, errorcorrection models, and co-integration among cross-country return series (see, among others, pascual, 2003; darbar and deb, 1997; karolyi and stulz, 1996; and parhizgari et al., 1994). by now, it is well recognized that estimates of correlations may require further statistical refinements (forbes and rigobon, ijbf 18 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 2002) and that such estimates should consider the dynamic, i.e., the time-varying, aspect of correlations (engle, 2002). this latter feature may be exploited in identifying and measuring contagion among cross-country markets. the anatomy of a cross-country financial crisis and thereby contagion is not expected to be the same in all instances and for all time periods. yet, given a set of countries with some common interaction variables, financial distress or bad news in a segment of a county’s market, for instance in its equity market, generally leads into gradual increases in the volatility in the returns of that country. during the early periods, the changes in the volatilities and its associated variables are confined within each country. not much cross-border effect is discerned during the early periods. with the spread of the news, the global aspects of such changes in volatility start brewing, leading possibly into contagion. it is often hard to pinpoint when, and sometimes where, exactly the cross-border transmission starts. the global transmission, if any, resembles more of a gradual or an evolutionary process with bi-feedback than a sudden one-time transmission. the above process can not be fully captured by a simple or static measure of correlation. it needs a different type of analysis, i.e. one that is dynamic enough to account for the continuous changes in the market. we thus start with the assumption of time-varying correlations and resort to engle’s (2002) dynamic conditional correlations (dcc, henceforth). we believe ddc, particularly with grch(1,1), fits the transmission process of contagion very well. the application of dcc to contagion in general, and to the east asian financial markets in particular, is fairly recent. the existing literature on contagion includes one study (chiang et al., 2005) using dcc for the asian markets. though we also employ dcc, our approach is somewhat different from theirs. for example, they assume the same parameters for each pair of contagion source and target countries in the correlation process while we do let the parameters to differ across pairs of countries. furthermore, our approach for testing contagion is very different. they use regression method with dummy variables to examine if there are any significant increases in dcc. we use mean difference t-tests and median difference z-tests for this purpose. finally, the length of the period under our coverage is different (longer). the remaining parts of this paper are organized as follows. section 2 provides a brief review of the literature. section 3 presents the methodology that is employed. section 4 provides information on the data input and section 5 contains the empirical results. the last section offers the summary and conclusions. 2. prior literature contagion is variously defined (see, for instance, pericolo and sbracia, 2001). among the several definitions, the most commonly prevailed one is the existence of some degree of excess co-movement which cannot be explained by the the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 19 fundamentals. this type of contagion is often referred to as pure contagion (see kumar and persaud, 2002). apart from the variety of the definitions that could be forwarded as viable alternatives (see moser, 2003; pasquariello, 2007; castiglionesi, 2007), the measurement of contagion has also proved to go far beyond the simple static estimate of the correlation coefficients (see, for instance, diebold andyilmaz, 2007; forbes and rigobon, 2002; and boyer et al., 1999). for example, boyer et al. (1999) make some adjustments in the correlations to account for volatility. forbes and rigobon (2002), examining the 1994 mexican peso crisis and the 1997 asian financial crisis, estimate a set of cross-market correlation coefficients to measure the extent of co-movements between a contagion source country and a group of target countries. they correct the bias in the correlation coefficient that arises from the increased volatility during the turmoil period. their results indicate the presence of contagion when the correlation coefficients are not adjusted for volatility. this finding is reversed when the volatilityadjusted correlation coefficients are considered. under such adjustment, the co-movements among the source and the target countries do not increase significantly during the turmoil periods. forbes and rigobon’s interpretation is that the continued high level of market correlation is not contagion; it arises simply due to strong linkages among them. considering the complexities in measuring contagion, there are two main issues with forbes and rigobon’s (2002) approach that deserve reconsideration. first, they do not adjust volatility continuously. it has increasingly been accepted that time-varying volatility is one of the stylized facts of stock returns (see tse and tsui, 2002). second, their test results of co-movement difference between the stable and turmoil periods can be different for different lengths of the turmoil period (see, chiang et al., 2005). other complexities have also presented themselves. first, nearly all stock returns exhibit some degree of skewness and kurtosis (see, for instance, chiang, table 1, p. 38). this violates the assumption of normality that underlies some of the tests that are employed. second, identification of the break point to establish the beginning of the contagion period could carry some ambiguity. most prior research starts with pre-assigned break points. third, the period of study, i.e., the length of the overall, stable, and turmoil periods, could make a difference in the results (see, for instance, forbes and rigobon, 2002 and chiang, 2005). fourth, differences in time zones and operating hours have raised questions about the accuracy of contagion measurement. this has prompted the researchers to examine them separately. fifth, inclusion or exclusion of global variables has also expanded the dimension of contagion analysis. finally, the currency factor has posed an additional problem. some studies have used indices in local currencies, a practice that suffers from the lack of consistency in units of measurement. 3. methodology dcc(1,1)–garch(1,1) developed by engle (2002) and engle and sheppard (2001) are employed to examine the time-varying correlation coefficients. mean 20 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 ( ) ( )( ) ( ) ( )( ) = = ++= ++= t t tttttttt t t tttt vdrdvdrdn hhnl 1 111 1 log2log5.0 log2log5.0 difference t-test and median difference wilcoxon z-test are used to investigate whether there are significant differences in the estimated time-varying correlation coefficients between the stable and the turmoil periods. since the volatility is adjusted by the procedure, the time-varying correlation (or dynamic conditional correlation or dcc) does not have any bias from volatility. unlike the volatilityadjusted cross-market correlations employed in forbes and rigobon (2002), dcc-garch continuously adjusts the correlation for the time-varying volatility. hence, dcc provides a superior measure of correlation. estimation of the dynamic correlation coefficients follows three steps. the first step consists of a demeaning process (see engle and sheppard, 2001) whereby the residual returns are obtained. the regression model that we have employed for this process is: (1) where r t is the returns of local stock index and is the u.s. s&p 500 composite index. inclusion of the latter variable is to capture the effect of a global market factor. the existing literature supports a near consensus position that the u.s. equity markets have statistically significant influence on the asian markets. in the second step, the parameters in the variance models are estimated using the residual returns (ε t ) from the first step. a standard garch model is employed such that: ε t = d t v t ~ n(0, h t ) (2) where ε t is a k x 1 column vector of residual returns of r t , k is the number of countries considered, v t and is k x 1 a column vector of standardized residual returns. h t is a k x k matrix of time-varying variances. specifically, h t = d t r t d t (3) where r t is a k x k matrix of time-varying correlations. d t is a k x k diagonal matrix of time-varying standard deviations of residual returns. the variances are obtained with univariate garch (1,1) processes. specifically, (4) the log-likelihood function to determine the parameters in (4) and (6) is given below. r t-1 s&p & 0 1 1 2 1 s p t t t tr a a r a r ε− −= + + + 2 12 2 110 −− ++= ttt hbbbh ε the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 21 1 11 1 1 2 2 2 2 2 2 1 1 1 1 1 1 . t it t jt jtt it jt t it jt ijt t it jt t it t jt t it it t jt jt t it t jt e h v h ve e v v e v v e e e h v e h v e v e v = = = = 2 1 2 1 2 1 1 1 1.t it t it it it t ite v e h h e= = = ( ) ttttttttttttt vvddddd === 11112 , ( ) ( )( ) ( )( )2 1 1 1 1 2 0.5 log 2 2 log 0.5 log t t t t t t t t t t t t t t l n d d r r v v l l = = = + + + = + ( ) ( )( )21 1 0.5 log 2 2 log t t t t t t l n d d = = + + ( )( )12 1 0.5 log t t t t t t t t l r r v v = = + since , (5) where: (6) (7) as is shown above, log-likelihood function is separated into log-likelihood function of variances and that of correlations. the parameters of variances in l 1 are determined without simultaneous determinations of the correlation parameters by maximizing l 1 . in the third step, correlation coefficients are estimated. the correlation coefficients between stock index returns i and j at time t are defined as: where: the correlations ρ ijt constitute the correlation matrix r t of which diagonal elements are unity. let . then, (8) in order to parameterize the correlation coefficient ρ t , it is assumed that tq follows an autoregressive process. specifically, (9) where q is an unconditional correlation coefficient matrix. the unconditional correlations are determined in the second step and are used as [ ]1t t t tq e v v− ′= ( ){ } ( ){ } 1 1 2 2 t t t tr diag q q diag q − − = ( ) 1 1 11t t t tq q v v qα β α β− − −′= − − + + 22 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 ( )( )12 1 0.5 log t t t t t t l r rε ε− = ′= − +• predetermined values in this step1 (see engle and sheppard, 2001, p. 5). the parameters for the time-varying correlations are determined by maximizing the log-likelihood function l 2 . since v’ t v t does not involve the determination of the parameters, the log-likelihood function is reduced to (10) we implement the correlation model in (9) for each pair of contagion source and target countries to allow the parameters α and β to be different for each pair. in this regard, our approach is different from chiang et al. (2005) who keep these parameters constant across all country pairs. 4. data the east asian countries and their stock indices that are considered are hong kong (hang seng index), thailand (bangkok set index), south korea (korea se composite index), malaysia (kuala lumpur se index), singapore (singapore se index), taiwan (taiwan se weighted index), the philippines (philippines se composite index), and indonesia (jakarta se composite index). daily stock price indices for these exchanges are obtained from datastream. the period of the analysis is from january 1, 1996, through march 1, 2005. the starting date of january 1, 1996, is considered as the beginning of the stable period and is the same as in forbes and rigobon (2002). this date appears to be a suitable starting point for the asian countries since it is relatively distanced from the december 1994 mexican peso crisis. hence, estimates of the dynamic correlation coefficients will not be confounded with the effects of the mexican peso crisis. the ending date of the turmoil period is assumed to be december 30, 1998. given our choice of the break points (see below), it is a preferred date since it results in equal number of days in the stable and the turmoil periods. kaminsky, lyons, and schmukler (2000) provide evidence that mutual fund firms started to pull out their capital from the asian countries upon the outburst of the currency crisis. to capture the effect of such capital outflows requires the turmoil period be long enough to allow the co-movements among the financial markets occur. forbes and rigobon (2002) assume one month for the length of the turmoil period. chiang, et al. (2005) extend it and obtain different results. we have selected two dates that are often identified as the inception of the turmoil period. the first is july 2, 1997, when thailand baht was devalued. the second is october 17, 1997, when the hong kong stock market crashed. forbes and rigobon, among others, have selected the october date as well. 1 to provide stable and long term static measures, the unconditional correlations are computed using data from january 1, 1990, through december 30, 2003. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 23 5. empirical application in terms of considering further details in our analysis, we rely on the results of prior studies. for example, much effort has gone into a consideration of differences in time zones and the operating hours of the markets. at a given time zone, the opening hours among the asian markets differ by a maximum of three hours. the difference in the closing hours is higher since the operating market hours are unequal. for example, the philippines market is open only for two and a half hours (from 9:30 a.m. to 12:00 noon). forbes and rigobon (2002), for instance, average the returns over two consecutive days to overcome this problem. their results indicate no significant difference. we thus assume that a consideration of time-zone and operating-hour differences in our analyses is not going to have significant effects on the results. to visualize the computed dccs, we have graphed them in figures 1 through 3. figures 1 and 2 provide individual dcc plots for pair-wise countries when the contagion source countries are thailand and hong-kong, respectively. the break-point dates are represented by vertical dash lines. the increases in dccs beyond the break points in nearly most cases are obvious. the estimated dccs for the pairs of hong kong malaysia and hong kong indonesia do not appear to rise considerably, yet they exhibit large innovations over the turmoil period. figure 3 provides another insight into the overall contagion effects during the sample period. it shows the average of the estimated dynamic conditional correlations (dcc) for thailand and hong-kong as the contagion source countries. prior to calculating these averages, the dccs are normalized by dividing each country’s dcc by the value of its dcc on the day prior to the crisis break point. it is apparent from figure 3 that there are surges in the dccs. there are also some unusual sharp decreases in the dccs right after the start of the turmoil dates. these are, however, very immediate and short-lived. these quick and temporary decreases could be attributed to the need to quickly rebalance and adjust portfolios as explained by kaminsky, et al. (2000). table 1 show the estimation results of the mean, variance, and correlation models as given in relations (1), (4), and (9). all the estimates are statistically significant at the one percent or below. the statistical significance in this table is not indicated by asterisks, but rather by the p-values that are in parentheses under the estimates. as shown by the magnitude of the estimated parameter β, the dcc processes exhibit a high degree of persistency in general, except for thailand malaysia (0.888 in column 10) and to some extent, for hong kong malaysia and hong kong indonesia (0.900 and 0.902 in the very last column). additionally, our results show that the dcc processes for each pair have different innovation and persistency. therefore, it is evident that restricting the parameters α and β to be the same for all pairs of the countries may lead to different estimation results. chiang et al. (2005) restrict the correlation parameters α and β to be equal for all the pairs.2 their estimates are 0.006 for α and 0.989 for β. 2 see chiang et al. (2005), table 3, p. 41. 24 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 d c c ( t ha iland ho ng k ong ) -0 .1 0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6 0 .7 199 6 1 997 19 98 199 9 2 000 200 1 2002 20 03 d c c ( t ha ilan d k or ea ) 0 0 .05 0.1 0 .15 0.2 0 .25 0.3 0 .35 19 96 1 99 7 19 98 1 99 9 20 00 2 00 1 20 02 2 00 3 d c c (t h ailan d malay s ia) -0 .1 0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6 0 .7 199 6 1 997 19 98 199 9 2 000 200 1 2002 20 03 d c c (t h aila n d p hilip pin e s ) -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 1 9 96 19 9 7 1 99 8 1 9 99 20 0 0 2 00 1 2 0 0 2 2 00 3 d c c (t h ailan d s ing ap o re ) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1 99 6 1 9 97 19 9 8 1 99 9 2 0 00 20 0 1 2 00 2 2 0 03 d c c (t h a ilan d t a iw a n ) -0 .2 -0 .1 0 0 .1 0 .2 0 .3 0 .4 0 .5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 20 0 0 2 0 0 1 2 0 0 2 2 0 0 3 d c c (t h a ilan d in d o n e s ia ) 0 0 .0 5 0 .1 0 .1 5 0 .2 0 .2 5 0 .3 0 .3 5 0 .4 0 .4 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 20 0 0 2 0 0 1 2 0 0 2 2 0 0 3 figure 1: the estimated dynamic correlation coefficients (contagion source: thailand) this figure shows the estimated dynamic correlation coefficients (dcc) for each pair of the contagion source (thailand) and target country. the vertical dash line in each plot indicates july 2, 1997. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 25 d c c ( h ong k on g -t hailand) -0 .1 0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6 0 .7 19 96 199 7 1 998 19 99 200 0 2001 2 002 20 03 d c c (hong k ong k orea) -0 .1 0 0.1 0.2 0.3 0.4 0.5 0.6 1996 1997 1998 1999 2000 2001 2002 2 003 d c c (hong k ong malays ia) -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1996 1997 1998 1999 2000 2 001 2002 2 003 d c c ( h on g k on g p h ilipp in es ) 0 0 .05 0.1 0 .15 0.2 0 .25 0.3 0 .35 0.4 19 9 6 1 99 7 19 98 1 99 9 2 0 00 20 0 1 2 00 2 2 0 03 d c c (h ong k ong s in gap or e) 0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6 0 .7 0 .8 0 .9 1 99 6 199 7 19 98 1 99 9 200 0 20 01 2 00 2 200 3 d c c (hon g k o ng t aiw an) -0 .1 0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6 1996 19 97 1 998 1999 200 0 2 001 2002 200 3 d c c (ho ng k o n g ind on e s ia ) -0 .1 0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6 19 9 6 19 9 7 1 9 98 1 9 99 2 0 00 2 00 1 2 00 2 20 0 3 figure 2: the estimated dynamic correlation coefficients (contagion source: hong kong) this figure shows the estimated dynamic correlation coefficients (dcc) for each pair of the contagion source (hong kong) and target country. the vertical dash line in each plot indicates october 17, 1997. 26 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 730 this table provides the estimation results for the mean, variance, and correlation model as introduced in the methodology section. they are as follows. & 0 1 1 2 1 s p t t t tr a a r a r= + + + 2 2 0 1 1 2 1 1t t th b b b h= + + ( ) 1 1 11t t t tq q v v q= + + the correlation model is run for each pair of the contagion source (thailand and hong kong) and target countries. p-values are given in parentheses. represents the unconditional correlation coefficient in the matrix q .ρ table 1: estimation parameters of mean, variance, and correlation models parameter a 0 a 1 a 2 b 0 ρ10 4 b 1 b 2 thailand hong kong ρ α β ρ α β thailand 0.000 0.118 0.340 0.045 0.106 0.879 0.2923 0.019 0.966 (0.628) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) hong kong 0.000 -0.015 0.586 0.043 0.079 0.901 0.2923 0.019 0.966 (0.367) (0.346) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) korea 0.000 0.026 0.460 0.025 0.086 0.904 0.2035 0.005 0.992 0.2002 0.005 0.995 (0.578) (0.110) (0.000) (0.000) (0.000) (0.000) (0.000) (0.171) (0.000) (0.000) (0.000) (0.000) malaysia 0.000 0.072 0.360 0.079 0.101 0.889 0.3103 0.037 0.888 0.2918 0.038 0.900 (0.944) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) philippines 0.000 0.182 0.307 0.058 0.105 0.864 0.2251 0.015 0.968 0.1943 0.005 0.991 (0.925) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.003) (0.000) (0.000) (0.023) (0.000) singapore 0.000 0.105 0.422 0.058 0.130 0.835 0.5457 0.016 0.973 0.3628 0.045 0.910 (0.879) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) taiwan 0.000 0.033 0.361 0.054 0.065 0.919) 0.1479 0.024 0.948 0.128 0.009 0.991 (0.446) (0.042) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.000) (0.000) (0.004) (0.000) indonesia 0.000 0.191 0.260 0.016 0.115 0.891) 0.2463 0.009 0.985 0.221 0.034 0.902 (0.869) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.005) (0.000) (0.000) (0.000) (0.000) t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 730 27 table 2: tests of the mean and the median differences of the estimated dynamic correlation coefficients between the stable and turmoil periods this table shows the test results of the statistical differences of the means and the medians. the contagion source countries are thailand (panel a) and hong kong (panel b). the break points between the stable and turmoil periods are july 2, 1997, and october 17, 1997, respectively. the p-values of the t-tests show the statistical significance of the mean differences. the pvalues of wilcoxson z-tests represent the statistical significance of the median differences. panel a: contagion source is thailand (turmoil breakpoint: july 2, 1997) stable period turmoil period mean t test z test target n mean median std dev n mean median std dev increase (%) p value p value hong kong 392 0.2664 0.2458 0.1052 391 0.3316 0.3953 0.1543 24.5 0.0000 0.0000 korea 392 0.1153 0.1214 0.0334 391 0.1988 0.1929 0.0523 72.4 0.0000 0.0000 malaysia 392 0.2755 0.2708 0.0788 391 0.3018 0.3049 0.1041 9.5 0.0000 0.0000 philippines 392 0.1710 0.1696 0.0732 391 0.2385 0.2712 0.1213 39.5 0.0000 0.0000 singapore 392 0.3464 0.3385 0.0927 391 0.3673 0.4263 0.1526 6.0 0.0206 0.0000 taiwan 392 0.0655 0.0603 0.0492 391 0.1593 0.1603 0.0988 143.2 0.0000 0.0000 indonesia 392 0.2066 0.1897 0.0766 391 0.2780 0.2948 0.0827 34.6 0.0000 0.0000 panel b: contagion source is hong kong (turmoil breakpoint: october 17, 1997) thailand 491 0.2390 0.2286 0.1133 292 0.3998 0.4158 0.1086 67.3 0.0000 0.0000 korea 491 0.0326 0.0227 0.0307 292 0.1044 0.0828 0.0732 220.2 0.0000 0.0000 malaysia 491 0.3047 0.2947 0.0798 292 0.3320 0.339 0.1171 9.0 0.0000 0.0000 philippines 491 0.2139 0.2089 0.0451 292 0.3257 0.3262 0.0201 52.3 0.0005 0.0000 singapore 491 0.4793 0.4569 0.1235 292 0.5919 0.6014 0.0912 23.5 0.0000 0.0000 taiwan 491 0.0987 0.0996 0.0566 292 0.2692 0.2959 0.0829 172.7 0.0000 0.0000 indonesia 491 0.2418 0.2309 0.0855 292 0.2752 0.2666 0.0794 13.8 0.0000 0.0000 28 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 to check the existence of contagion, we employ t-tests for the mean difference and wilcoxon z-tests for the median difference. our test results on the contagion effects are shown in table 2. all the t-tests and the z-tests in the last two columns of this table are statistically significant at the one percent or below, thus rejecting the null hypothesis of no contagion. as represented by the significance of the p-values, the tests demonstrate the presence of contagion effects arising from the financial crisis. based on the increase in the dcc mean d c c ave ra ge (s o u rc e : t h a ila n d) -1 .0 0 .0 1 .0 2 .0 3 .0 4 .0 5 .0 6 .0 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 00 0 2 00 1 2 0 0 2 2 0 0 3 (panel a) 0 . 0 0 . 5 1 . 0 1 . 5 2 . 0 2 . 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 d c c a v e r a g e ( s o u r c e : h o n g k o n g ) (panel b) figure 3: averages of the estimated dynamic correlation coefficients this figure plots the averages of the estimated dynamic correlation coefficients for all the pair-wise countries. in panel a, the contagion source is thailand and in panel b it is hong kong. prior to averaging, each pair-wise dcc is normalized by dividing it by its value on the day prior to the crisis break point. the break-point dates are indicated by vertical dash lines. these dates are july 1, 1997, if the contagion source is thailand, and october 17, 1997, if the contagion source is hong kong. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 29 values in percentage term (see column 10 of table 2), the two countries that are most influenced by the contagion effects are taiwan if the contagion source country is thailand, and korea if the contagion source country is hong kong. 6. summary and conclusions consistent with prior literature, we have chosen two dates for financial distress in the east asian financial markets: july 2, 1997, the devaluation of thailand baht, and october 17, 1997, the hong kong stock market crash. we have provided a set of uniform results that suggest, overwhelmingly, contagion in the 1997 asian equity markets. our approach and findings contribute to the existing literature by addressing a few contentious points on contagion. first, our findings indicate the existence of contagion in the 1997 east asian financial markets as measured by the returns in the equity exchanges. second, our approach is less cumbersome and is statistically simpler in terms of procedure and data requirements. for example, there is no need to resort to other measures including dummy variables, nor is there a need to run further regressions to show the existence of contagion. third, we have provided a simple methodology to assist us in measuring and identifying contagion. this is done without resort to other measures, including dummy variables. under some conditions, it has been shown that dummy variables could be persistent and, thus, their interpretation could be subject to statistical shortcomings (see ferson et al., 2003). we have avoided this potential problem. obviously, consideration of other measures to add insight to the cause of contagion is highly warranted, but we do not recommend such consideration for the detection of contagion. finally, notwithstanding all of the above, what could still be the subject of further inquiry is the definition of contagion. we have opted for a statistically powerful and straightforward definition, i.e., statistical break in dynamic conditional correlation as measured by the shifts in the mean and the median of the computed dccs. further refinements and research along this line of analysis could be the subject of future research. author statement: this paper was submitted by ali parhizhari, ingersoll-rand professor, at the florida international university. it was reviewed anonymously, and then accepted. the authors wish to record their appreciation for the editorial assistance provided by the journal editors. the authors retain responsibility for any remaining errors. references bialkowski, jedrzej, m. t. bohl, and d. serwa, 2006. testing for financial spillovers in calm and turbulent periods, quarterly review of economics and finance 46/3, 397-412. 30 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 17-30 castiglionesi, fabio, 2007. financial contagion and the role of the central bank, journal of banking and finance 3/1, 81-101. corsetti, giancarlo, m. pericoli, and m. sbracia, 2005. some contagion, some interdependence: more pitfalls in tests of financial contagion, journal of international money and finance, 24/8, 1177-99 darbar, salim m. and partha deb, 1997. co-movements in international equity markets, journal of financial research 20/3, 305-322. diebold, francis and k. yilmaz, 2007. measuring financial asset return and volatility spillovers, with application to global equity markets, working paper, university of pennsylvania and nber. engle, robert f. and kevin sheppard, 2001. theoretical and empirical properties of dynamic conditional correlation multivariate garch, working paper, ucsd. engle, robert f., 2002. dynamic conditional correlation: a simple class of multivariate generalized autoregressive conditional heteroskedasticity models, journal of business and economic statistics 20, 339 – 350. ferson, w. e., s. sarkkissian, and t. t. simin, 2003. spurious regressions in financial economics, journal of finance 58/4, 1393-1413. forbes, kristin j. and roberto rigobon, 2002. no contagion, only interdependence: measuring stock market comovements, journal of finance 57, 2223 – 2261. kaminsky, graciela, richard lyons, and sergio schmukler, 2000. economic fragility, liquidity, and risk: the behavior of mutual funds during crisis, working paper. karolyi, g. andrew and rene m. stulz, 1996. why do markets move together? an investigation of u.s.-japan stock return comovements, journal of finance 51/3, 951-986. kumar, manmohan s. and a. persaud, 2002. pure contagion and investors’ shifting risk appetite: analytical issues and empirical evidence, international finance, 5/3, 401-36. moser, thomas, 2003. what is international financial contagion? international finance, 6/2, 157-78. pascual, antonio garcia, 2003. assessing european stock markets cointegration, economics letters 78 (2, feb), 197-203. pasquariello, paolo, 2007. imperfect competition, information heterogeneity, and financial contagion, review of financial studies 20/2, 391-426. pericolo, marcello and massimo sbracia, 2001. a primer on financial contagion, working paper, bank of italy. chiang, thomas c., bang nam jeon, and huimin li, 2005. dynamic correlation analysis of financial contagion: evidence from asian markets, working paper, drexel university. tse, y. k. and albert k. c. tsui, 2002. a multivariate generalized autoregressive conditional heteroscedasticity model with time-varying correlations, journal of business and economic statistics 20/3, 351-362. international journal of banking and finance 9-1-2008 east asian financial contagion under dcc-garch j. h. cho a. m. parhizgari recommended citation an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 21 an empirical investigation on credit card repayment pattern of academicians in malaysia kuah yoke chin faculty of business and finance, universiti tunku abdul rahman, malaysia wei chooi yi faculty of business and finance, universiti tunku abdul rahman, malaysia chin lai kwan faculty of business and finance, universiti tunku abdul rahman, malaysia chia mei si faculty of business and finance, universiti tunku abdul rahman, malaysia *corresponding author: kuahyc@utar.edu.my ________________________________________________________________ a r t i c l e i n f o _____________________________ article history: received 16 may 2018 revised 21 june 2018 accepted 27 june 2018 ____________________________ keywords: personal attitude, spending pattern, repayment pattern; gender, parenthood jel code: d14 a b s t r a c t _________________________________ the credit card market has witnessed tremendous growth resulting from a paradigm shift in technology. the genuine usage of credit cards as a form of convenience has diverged to overspending, irresponsibility, revolving liability and bankruptcy. this implies that possession of credit cards has brought about a change in users’ repayment behavior. thus, this study aims to determine academicians’ credit card repayment pattern in a private tertiary institution in malaysia. the dependent variable is the repayment pattern and the independent variables that serve as the attributes include personal attitude and spending pattern, while, gender and parenthood serve as the moderating variables. findings revealed that personal attitude and spending pattern were important attributes in determining the repayment pattern among the academicians. gender the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 21 22 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 1. introduction the usage of credit card as a medium of payment has evolved tremendously since their introduction in the united states in the early 20th century (ahmed, ismail, sohail, tabsh, & alias, 2010). credit cards provide an easy and convenient mode of payment to users. a credit card user is only required to present a plastic card loaded with a pre-approved amount to merchants upon transaction at the checkout counter to enjoy the benefits of goods and services. no cash payment is required at the point of purchase. the demand for credit cards accelerated when users began to realize such benefits credit cards offered. huebner, fuchs and ilic (2018) supported the credit card payment mechanism as a direct result of the development of a cashless society. the issuance of credit card was primarily intended to benefit both the merchant and consumer markets. merchants leverage on credit card usage to stimulate sales and to provide a facility to complete purchases, while consumers leverage on credit card usage to expand their purchasing power. the credit card issuer is practically advancing cash to the user upon transacting and the user is only required to settle the payment at a later date. such financing facility in the form of unsecured loans provided by the issuer is far easier to obtain by users compared to other financing means that require a lengthy and complicated process (elangkovan & ahmed, 2013). despite the convenience payment and financing flexibility that credit cards offer, users tend to misuse the cards. at the time of purchase, they often or are likely to underestimate the price of goods and portray a lack of awareness of the amount that have been charged to their cards, leading to overspending. at the time of repayment, users tend to pay only the minimum required amount, gradually accumulating a larger credit balance in their credit cards. the consequence of credit card usage in malaysia is apparent. according to bank negara malaysia (bnm) (2016), the total outstanding balances due from cardholders in malaysia amounted to rm35.77 million as of june 2016. however, and parenthood were found to play a moderating role in the repayment pattern. findings from this study is expected to facilitate the government and credit card companies to work towards nurturing a financially healthier and informed society, and also to reach out positively to the younger generations through the influence of the academicians an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 23 total purchases for the period was only rm10.26 million, representing 28.7% of the total outstanding balances. this meant that the remaining 71.3% were accumulated credit card debts from previous periods. in addition, a depressing fact was revealed that the total amount of bank loan applications by individuals for the purpose of credit card repayments had increased by 56.8% over a one-year period ending june 2016 from rm26.657 million to rm41.785 million (bnm, 2016). such increase signaled that credit card debts are rising. in contrast, users’ ability to repay credit card debts are thinning, leading them to seek bank loans as an alternative to resolve their debts. in addition, non-performing loans attributed to credit cards payments in malaysia are high. lydia (2017) in “the malaysian reserve” reported that the non-performing loan (npls) related to purchases via plastic cards was at rm473.3 million as at end of may 2017. such an amount has raised concerns on credit card traps. ahmed et al. (2010) stated that the easy availability of financing by credit cards has led to many adverse consequences such as compulsive spending and excessive debt. this is supported by omar, rahim, wel, and alam (2014) who added that users regarded the plastic card as a booster to self-esteem and tended to use it to satisfy their craving for material things, which resulted in financial distress and bankruptcy. the original intent of the credit card as a form of “convenience payment” has alluded to mean “overspending”, “irresponsibility”, “revolving liability” and “bankruptcy”. this indicated that the possession of credit cards has brought about a change in users’ repayment pattern. previous empirical studies on credit cards were predominantly focused on college students and working adults. limited studies were available on academicians. cebeci and sanli (2016) conducted a study concerning academicians’ attitude towards credit card usage on 158 academicians in a university in turkey in 2015. the study found that the trust on credit card had a significant and positive impact on credit card usage among the academicians, as the academicians did not have to adjust their expenses according to their budget when they used credit cards. in addition, the study also found that the usage of credit cards did not lead to overspending. the trust on credit card usage also did not cause irregular spending and excessive borrowing among the academicians. however, outcomes from the study by cebeci and sanli (2016) are unable to be generalized to other countries due to differences in legal, cultural, geographical and socio-economic structures between countries, industries and markets. in addition, the study focused on academicians’ credit card usage as their dependent variable, and not their repayment behavior. most of the past studies had been conducted on credit card usage behaviors in malaysia (ramayah, nasser, aizzat, & lim, 2002; ahmed et al., 2010; nga, yong & sellappan, 2011; jusoh & lin, 2012; rasiah & masuod, 2013). hence, this study attempts to fill research gaps by focusing on credit card repayment behaviors, with emphasis on academicians in a private tertiary institution. 24 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 attitude and leadership skills of academicians have a direct influence on the younger generation. in this respect, an academician is encouraged to portray a positive attitude and leadership skills in their coaching. according to nurul hudani, marof, and noor hisham (2013), self-management and social awareness are closely associated with the development of a positive attitude and leadership traits. this implied that academicians should possess a high level of self-management and social awareness. such positive attributes contribute to academicians’ ability to exercise discipline in their actions and a heightened awareness on the consequences of their actions. this study aims to provide evidence to explain the credit card repayment pattern among academicians in a private tertiary institution in malaysia. six research objectives are addressed in this study to investigate the relationship between personal attitude and spending pattern with the credit card repayment pattern of academicians. next, this study will investigate the relationship between the spending pattern and personal attitude of academicians with their credit card repayment pattern based on their gender and parenthood. 2. literature review the literature review in this study is divided into three parts. firstly, credit card repayment pattern as the dependent variable will be discussed. mchugh and ranyard (2012) indicated that repayment pattern varies among individuals based on income level, and attitude toward credit and debt. in addition, people who are future oriented will be motivated to maximize repayments in order to ameliorate the negative impact on credit decision (webley & nyhus, 2001). some tend to repay significantly more when they comprehend the long-term consequences (mchugh & ranyard, 2012). the characteristics of people who are willing to repay a higher amount are those that place value on achievement, self-direction, have a clear interpretation of borrowing and uses credit appropriately (wang, lu, & malhotra, 2011). leow and crook (2014) found that the self-employed and unemployed are at a higher risk of repayment delinquency compared to those who are employed. however, the ability to recover for employees may be lesser once they enter into repayment delinquency. thus, those who are frequently delinquent in repayments are more likely to go into delinquency again but not defaulting. this is due to the stable income that the employees enjoy while the self-employed and unemployed do not. however, the self-employed and unemployed are better in balancing their accounts and more able to stay out of defaulting. in 2017, credit card billings amounted to rm180 billion (bernama, 2017). according to the then deputy finance minister of malaysia, an average of 43.6% of total cardholders settled their credit card debts completely between january and june 2017. however, there was at least 5% of outstanding balance while 12.8% an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 25 did not even settle the minimum payment before or on the due date (tasnim, 2017). according to stewart (2009) and navarro-martinez, salisbury, lemon, stewart, matthews and harris (2011), the presence of minimum repayment information on a credit card statement reduces the consumer’s partial repayment decision. wang et al. (2011) defined revolving credit as three particular behaviors – repaying over the course of time, paying equal to or greater than the minimum required repayment and using up a revolving credit limit. furthermore, furnham and boo (2011) explained that the minimum repayment amount represents a number that can sway judgments and decisions for anchoring effect which influences the consumer’s repayment decision. secondly, personal attitude towards credit card spending and repayment will be discussed. the cashless payment mechanism has changed attitude toward the usage of credit cards throughout the world. the large and growing educated population in malaysia has influenced the acceptance of credit cards or plastic money. in 2017, malaysia hit 3.6 million main credit card holders with a rm36.9 billion outstanding balance (tasnim, 2017). initially, credit card usage was linked with luxury spending. however, it has now transformed to support daily expenses. convenience and the carry forward on payment have changed the consumer’s attitude toward credit card usage (ramayah et al., 2002). chan (1997) defined an active credit card user as an individual who spends with credit card more than 10 times per month. according to chien and devaney (2001), the more favorable an individual’s attitude toward credit spending, the higher the outstanding debt. consumer attitude toward credit card usage contribute to the increasing household outstanding debt (kish, 2006). lachance (2012) showed a positive relationship between attitude toward credit on education, the number of credit cards owned and knowledge of credit. kidwell and turrisi (2000) and hayhoe, leach, turner, bruin and lawrence (2000) found that the cognitive variable and affective credit attitude influenced the acquiring of credit card and its usage. people are willing to spend more when they pay with credit card, contributing to compulsive spending along with shorter decision-making time for item purchase (feinberg, 1986; shimp & moody, 2000; soman, 2001; awanis & chi cui, 2014). this is because cash payments represent immediate parting of money, which create the “pain of paying” and thus stimulates self-control on spending (chatterjee & rose, 2011; raghubir & srivastava, 2008). overvaluation of available income and undervaluation of past expenses lead to spending a greater amount than the credit limit allows (soman & cheema, 2002). in addition, low self-regulatory capabilities result in credit card misuse and indebtedness (chatterjee & rose, 2011). in a study by wong and lynn (2017), credit card payment method conjures up easy money which causes a momentary underestimated effort in income gaining thus lowering a perceived value of money. as a result, consumers are encouraged to spend easily as they are inspired by the “easy-money effect”. 26 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 based on the above arguments and past evidences, the first hypothesis of this study is as follows: hypothesis 1: there is a significant relationship between personal attitude and academicians’ credit card repayment pattern. thirdly, spending pattern and credit card repayment pattern will be discussed. related literature has acknowledged the effects of credit card spending on a nation’s socio-economy, personal finance and psychological health and well-being (chien & devaney, 2001; brown, taylor, & price, 2005). the cost associated with delinquency in payments can be far-reaching. the “buy now, pay later” stimulus that credit card offers has been attributed to massive credit build-up, resulting in card owners falling into deepening debt traps (livingstone & lunt, 1992; ausubel, 1997; yoo, 1998; silva & draut, 2004; stavins, 2006). economists and financial experts worldwide have signaled the alarming credit card debt traps as an onset of debt crisis at national level (chong, 2017; surane, 2017; bresiger, 2018). the multifunctional aspects of credit cards with their ability to purchase beyond affordability, to maintain a degree of anonymity in certain purchases and leverage on revolving provisions have led to an increase in credit card spending (klein, 1999). prelec and simester (2001) highlighted that the willingness-topay with credit card increases when consumers are instructed to use credit card rather than cash. in addition, feinberg (1986) posited that the consumer tends to spend a higher amount on a credit card purchase than a cash purchase for an identical purchase situation. soman (2001) concurred that credit card spending induces further purchases, as his study showed that consumers who made past payments with credit cards were more likely to spend on additional discretionary purchases. soll, keeney and larrick (2013) have also highlighted that consumers portrayed poorer recollection of their previous spending. more significantly, consumers experienced reduced levels of psychic pain when spending with credit cards in comparison with cash spending. these implied that credit card spending increased the likelihood to purchase, thus incurring a higher payable amount. thus the “buy more now, pay later” attitude is expected to lead to a behavioral change in debt repayment pattern among cardholders. according to bertaut and haliassos (date) and reiter (2009), the majority of us credit card holders revolve their credit card balance, even though they have substantial liquid and retirement assets. both the studies also suggest that credit card revolvers choose not to be debt-free for consumption smoothing, high switching cost from credit card usage to debit card usage or habitual reasons. kim and devaney (2001) examined the factors affecting outstanding credit card balances among debt revolvers in us. their findings proved that spending attitude (in particular on essentials and elective purchases) is significantly related to the likelihood of revolving credit card balances. in addition, increment in the an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 27 spending power of an individual, leveraging on higher credit limits available or a higher number of credit card ownership, are positively related to revolving balances. interestingly, they found that among the revolving credit card owners, education was positively related to the amount of outstanding credit card balances. on the contrary, sprenger and stavins (2008) showed that higher credit card spending had reduced the likelihood of cardholders revolving their balances. his study proved that revolvers do not only adopt, but also used debit more frequently in order to control their spending than convenience users. tan, yen and loke (2011) in their study on credit card revolvers in malaysia proved that consumer’s general financial/credit consumption tendencies are relevant in explaining both the probability and level of card debt. the study revealed that high financial commitments, high current account balances coupled with a bad debt history, will limit credit card spending ability, hence, leading to a more prudent repayment pattern. designated professions such as academicians in universities have greater access to revolving credit facilities as credit limits are normally set by income levels compared to undesignated professions which have very low or limited access to financing options (dewri, islam & saha, 2016). the influence of the plastic card has, not only, been associated to changes in consumer spending patterns but also associated to changes in debt repayment patterns (soll et al., 2013; soman, 2001; hamilton, & khan, 2001; zhao, zhao, & song, 2009). most of the literature examined looked at the influence of credit card usage in spending and/or repayment patterns. there are limited empirical findings on how credit card spending influences the repayment pattern of cardholders. this study aims to close the gap in literature. the spending pattern of academicians in a university would be analyzed in this study based on the above arguments and past evidences. thus, the second hypothesis of this study is as follows: hypothesis 2: there is a significant relationship between spending pattern and academicians’ credit card repayment pattern. apart from this, numerous studies have been undertaken in the past to differentiate usage behavior based on gender. mottola (2013) proved that more women were engaged in costly credit card behaviors than men in the us. the study via questionnaires showed that women, in comparison to men, were more likely to rollover credit card balances, more likely to pay the minimum amount on their cards, and more likely to be charged a late fee. in addition, women were less likely to compare rates when they signed up for credit cards. elangkovan and ahmed (2013) proved that women tend to overspend on unexpected and unnecessary expenses compared to men in malaysia. yet, this finding contradicted with the outcome by hussin, kassim, and jamal (2013) whose findings revealed that there were no significant differences between male and female consumers with regards to credit card spending and repayment in klang, malaysia. 28 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 however, cankaya, ucal and o’neil (2011) found that women in turkey understood better compared to men on the consequences of their credit card spending since they were keen to make full and on time payment rather than servicing minimum payment. turkish women spent mainly on food, clothing, cosmetics and household goods using credit cards. similar results were obtained from limbu, huhmann and xu (2012) who reported that women were far better than men in managing monthly outstanding balances, implying that women were more cautious in their credit card usage and finances. they reasoned that women had a lower tendency to take risks, thus they disassociated themselves from excessive credit card debts. similar results were obtained from previous studies by themba and tumedi (2012) who revealed that the youth and females had a negative attitude towards debt. this gender based attitude differed due to differences in culture, religious beliefs, earning capacity and credit availability (dewri et al., 2016). thus, understanding gender differences in credit card repayment has significant implications. therefore, this study has included gender as a moderating variable to differentiate the relationship of personal attitude and spending pattern on credit card repayment pattern between male and female academicians at universiti tunku abdul rahman (utar), kampar campus. thus, the third and fourth hypothesis of this study is as follows: hypothesis 3: there are significant differences between personal attitude and credit card repayment pattern based on gender. hypothesis 4: there are significant differences between spending pattern and credit card repayment pattern based on gender. the degree of family commitment of credit card holders in terms of child support, which is represented by parenthood in this study, is expected to influence credit card spending and repayment pattern. kinsey (1981) found that two-people households had more credit cards than single-person households, as two-people households tend to have higher expenses in their daily lives. besides that, godwin (1998) has proven that the larger the household size, the larger the household debt. chien and devaney (2001) also found that a larger-sized household was more likely to have a favourable attitude towards credit card usage and to have a higher outstanding credit card balance. similarly, bertaut and haliassos (2001) found that households with more children were more likely to have higher levels of usage leading to higher credit card debt. past studies have been conducted for over a decade and it has proven that more children meant higher daily household expenses. in this study, the researcher will revisit past studies by kinsey (1981), chien and devaney (2001), godwin (1998) and bertaut and haliassos (2001) taking into consideration parenthood as the moderating an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 29 variable to investigate the attributes that influence credit card repayment pattern. thus, the fifth and sixth hypothesis of this study is as follows: hypothesis 5: there are significant differences between personal attitude and credit card repayment pattern based on parenthood. hypothesis 6: there are significant differences between spending pattern and credit card repayment pattern based on parenthood. the theory of planned behavior (tpb) is a theory to describe the relationship between personal attitude and spending pattern with the credit card repayment pattern of academicians. this theory of planned behavior was developed by icek ajzen in 1991. this theory was developed to predict a person’s behavior through their intentions to improve on the predictive power of the theory of reasoned action (tra) by including perceived behavior control. previous studies such as by amin (2013) has extended the theory of tra to examine the acceptance factor of attitude, subjective norm and perceived financial cost within the context of islamic credit cards. amin (2012) has employed tra in his study to examine factor(s) towards islamic credit card usage intention. figure 1. conceptual framework as in the tpb theory, human behavior is influenced by behavioral beliefs, normative beliefs and control beliefs. the behavioral beliefs create a favorable 6 advertising campaigns and healthcare. previous studies such as by chudry, foxall and pallister (2011) examined attitudes toward borrowing from an education loan institution relative to the credit card institution by adapting the tpb in united kingdom. their findings revealed that attitude positively affected perceived behavioral control. similar studies which applied the theory of tpb was in a study by kennedy and wated (2011) which indicated that attitude positively influenced credit card usage intention. thus, in this study, personal attitude and spending pattern served as the independent variables, moderated by gender and parenthood to examine credit card repayment pattern among academicians in a private tertiary institution. the conceptual framework is shown in figure 1 as follows: ! personal attitude spending pattern credit card repayment approach "#$ "#% "#& gender "#' parent status "#( "#) figure 1. conceptual framework 3. methodology this exploratory study was conducted at universiti tunku abdul rahman (utar), kampar campus, perak, malaysia. the main purpose of this study was to determine the attributes of utar academicians in repaying their credit card spending. the population in this study consisted of 590 academicians from six different faculties, namely the faculty of engineering and green technology (fegt), faculty of information and communication technology (fict), faculty of arts and social science (fas), faculty of science (fs), faculty of business and finance (fbf) and institute of chinese studies (ics). stratified sampling technique was applied to determine the ratio of sample size to be obtained in each faculty. the relevant ratio via stratified sampling technique was applied to obtain data from each faculty in order to avoid response bias in this study. based on table 1, the approximate sample size would be 234 samples for a population of 600 academicians (krejcie & morgan, 1970). in addition, sekaran and bougie (2009) suggested that a sample size larger than 30 units but less than 500 units was appropriate for most research. thus, to ensure that the sample size was large enough, 250 questionnaires were administered to the academic staff across the faculties according to the ratio of sample obtained via the stratified sampling technique. 30 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 or unfavorable attitude towards the behavior (atb). on the other hand, the normative beliefs give rise to subjective norms (sn), while a rise in perceived behavioral control (pbc) is always a result of control beliefs (ajzen, 2006). this theory applies to studies on relations among beliefs, attitudes, behavioral intentions and behavior in various fields such as advertising, public relations, advertising campaigns and healthcare. previous studies such as by chudry, foxall and pallister (2011) examined attitudes toward borrowing from an education loan institution relative to the credit card institution by adapting the tpb in united kingdom. their findings revealed that attitude positively affected perceived behavioral control. similar studies which applied the theory of tpb was in a study by kennedy and wated (2011) which indicated that attitude positively influenced credit card usage intention. thus, in this study, personal attitude and spending pattern served as the independent variables, moderated by gender and parenthood to examine credit card repayment pattern among academicians in a private tertiary institution. the conceptual framework is shown in figure 1 as follows: 3. methodology this exploratory study was conducted at universiti tunku abdul rahman (utar), kampar campus, perak, malaysia. the main purpose of this study was to determine the attributes of utar academicians in repaying their credit card spending. the population in this study consisted of 590 academicians from six different faculties, namely the faculty of engineering and green technology (fegt), faculty of information and communication technology (fict), faculty of arts and social science (fas), faculty of science (fs), faculty of business and finance (fbf) and institute of chinese studies (ics). stratified sampling technique was applied to determine the ratio of sample size to be obtained in each faculty. the relevant ratio via stratified sampling technique was applied to obtain data from each faculty in order to avoid response bias in this study. based on table 1, the approximate sample size would be 234 samples for a population of 600 academicians (krejcie & morgan, 1970). in addition, sekaran and bougie (2009) suggested that a sample size larger than 30 units but less than 500 units was appropriate for most research. thus, to ensure that the sample size was large enough, 250 questionnaires were administered to the academic staff across the faculties according to the ratio of sample obtained via the stratified sampling technique. an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 31 table 1. determining sample size of a known population population sample size 500 217 550 226 600 234 650 242 700 248 source. krejcie and morgan (1970) 4. results all collected data were analyzed using statistical package for the social sciences (spss) version 23 and partial least squares structural equation model (plssem). the structural model was analyzed using smartpls 3.2 to perform plssem to test the hypothesized relationships among the constructs in the model depicted in figure 2. pls-sem is taken as the rigorous and most appropriate method to test the hypothesis in this study because it can handle samples with normality issues. in fact, pls-sem is a non-parametric approach, thus, distributional assumption is not required. in addition, pls-sem is able to achieve high levels of statistical power with small sample sizes (hair, black, babin & anderson, 2010). as shown in table 2, the highest number of respondents was from the faculty of business and finance (fbf) (41.9%). there were 127 males (55.5%) and 102 females (44.5%) in the sample. a majority of the respondents spent around rm501 to rm1000 on goods and services on a monthly basis using credit card. table 2. demographic profile of respondents frequency percentage (%) gender male 127 55.5 female 102 44.5 faculty fegt 30 13.1 fict 15 6.6 fas 35 15.3 (continued) 32 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 frequency percentage (%) fs 44 19.2 fbf 96 41.9 ics 9 3.9 average usage via credit card (monthly) rm100 – rm500 71 31.1 rm501 – rm1000 89 38.9 rm1001 – rm1500 30 13.1 do you have children? > rm1500 yes no 39 117 112 17.0 53.4 46.6 to ensure the credibility of the questionnaire and the results, the designed instruments went through the process of validity and reliability check. first, convergent validity was tested which is the degree to which multiple items that are used to measure the same concept are in agreement. as suggested by hair et al. (2010), factor loading, composite reliability and average variance extracted were the indicators used to access convergent validity. the loading for all items exceeded the recommended value of 0.6 (chin, gopal & salisbury, 1997). as shown in table 3, the composite reliability values depicted the degree to which the construct indicator showed that the latent construct ranged from 0.829 to 0.889, which was higher than the recommended value of 0.7 (hair et al., 2010). the average variance extracted, which reflected the overall amount of variance in the indicators accounted for by the latent constructs, were in the range of 0.533 to 0.652, which met the recommended value of 0.5 (hair et al., 2010). table 3. factor loading and reliability loading composite reliability average variance extracted cronbach’s alpha b4 0.630 0.829 0.555 0.721 b5 0.852 b6 0.856 b7 0.604 d1 0.845 0.882 0.652 0.825 d2 0.779 (continued) an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 33 loading composite reliability average variance extracted cronbach’s alpha d3 0.750 d4 0.850 e1 0.665 0.889 0.533 0.854 e4 0.721 e5 0.769 e6 0.726 e7 0.707 e8 0.724 e9 0.795 discriminant validity is the extent to which the measures do not reflect other variables and it is indicated by low correlations between variables by examining the comparison between average variance extracted (ave). based on the formula by fornell and larker (1981), it requires that the square root for each construct (ave) be higher than all its correlation with other constructs. as shown in table 4, the squared correlations for each construct were less than the square root of the average variance extracted, so it can be inferred that the constructs have a good level of validity. table 4. inter-construct correlation personal attitude repayment pattern spending pattern personal attitude 0.745 repayment pattern 0.600 0.730 spending pattern 0.505 0.730 0.807 once the reliability and the validity of the measures are assured, path coefficients can be reported based on the results of a pls structural model. table 5 shows the significance of the path coefficients. the t-value which can be compared with critical value forms the standard normal distribution to decide whether the coefficients are significantly different from zero. for instance, the critical value for a significance level of 5 percent, the probability of error is 1.96 (twotailed test). in this study, the results showed that personal attitude and spending pattern were significant attributes of the credit card repayment pattern of academicians. 34 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 figure 2. bootstrapping result from pls-sem table 5. summary of the structural model description hypothesis path coefficient t-value results personal attitude -> repayment pattern h1 0.310 3.884*** supported spending pattern-> repayment pattern h2 0.573 8.967*** supported note. all p-values are two-tailed, * significant at 0.05, *** significant at 0.001. besides this, a non-parametric multi-group analysis (pls-mga) method were conducted to see whether there was a significant difference in credit card repayment pattern between female and male academicians. a difference is significant if the p-value is smaller than 0.05. as seen from table 6, there were significant differences across the male and female groups at 0.05 significant level for personal attitude and spending pattern. 9 source: developed for research figure 2. bootstrapping result from pls-sem once the reliability and the validity of the measures are assured, path coefficients can be reported based on the results of a pls structural model. table 5 shows the significance of the path coefficients. the t-value which can be compared with critical value forms the standard normal distribution to decide whether the coefficients are significantly different from zero. for instance, the critical value for a significance level of 5 percent, the probability of error is 1.96 (two-tailed test). in this study, the results showed that personal attitude and spending pattern were significant attributes of the credit card repayment pattern of academicians. table 5. summary of the structural model description hypothesis path coefficient t-value results personal attitude -> repayment pattern h1 0.310 3.884*** supported spending pattern-> repayment pattern h2 0.573 8.967*** supported note: all p-values are two-tailed, * significant at 0.05, *** significant at 0.001. source: developed for research besides this, a non-parametric multi-group analysis (pls-mga) method were conducted to see whether there was a significant difference in credit card repayment pattern between female and male academicians. a difference is significant if the p-value is smaller than 0.05. as seen from table 6, there were significant differences across the male and female groups at 0.05 significant level for personal attitude and spending pattern. an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 35 table 6. pls-mga (difference in gender) description path coefficients (male) p-values path coefficients (female) p-values remarks personal attitude -> repayment pattern 0.262 0.020*** 0.419 0.000*** difference exist spending pattern-> repayment pattern 0.567 0.000*** 0.576 0.000*** difference exist note. all p-values are two-tailed, * significant at 0.05, *** significant at 0.001. this study also analyzed the categorical variable of parenthood as a multigroup analysis. table 7 shows that the results differed significantly across the parenthood groups at 0.05 significant levels under personal attitude and spending pattern. table 7. pls-mga (difference in parenthood) description path coefficients (with children) p-values path coefficients (without children) p-values remarks personal attitude -> repayment pattern 0.363 0.000*** 0.297 0.000*** difference exist spending pattern -> repayment pattern 0.577 0.000*** 0.575 0.000*** difference exist note. all p-values are two-tailed, * significant at 0.05, *** significant at 0.001. 5. discussion the results of this study revealed that all the independent variables had a direct relationship with the dependent variable. the findings revealed that personal attitude and spending pattern were important attributes in determining repayment pattern among academicians. this result is consistent with studies by lachance (2012) which showed a positive relationship between attitude toward credit to education, the number of credit cards owned and knowledge of credit. kidwell and turrisi (2000) and hayhoe et al. (2000) found that cognitive variable and affective credit attitude influenced the acquiring of credit cards and credit card usage. 36 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 on the other hand, the influence of spending pattern on the credit card repayment pattern of academicians was supported by chatterjee and rose (2011) as low self-regulatory capabilities end up with credit card misuse and indebtedness. people are willing to spend more when they are using credit cards, contributing to compulsive spending along with shorter decision-making time for item purchase (feinberg, 1986; shimp & moody, 2000; soman, 2001; awanis & chi cui, 2014). lastly, gender and parenthood differed among academicians at universiti tunku abdul rahman in terms of their repayment of credit cards. the findings indicated that male and female academicians revealed different credit card repayment patterns. furthermore, academicians with children had a different personal attitude and spending pattern in their credit card repayment behavior compared to those without children. 6. conclusion in conclusion, educators could play their part by giving advice and being role models to coach the younger generation to adopt a positive attitude and to sharpen their decision-making ability in particular concerning financial matters in this highly globalized society. the present study has two limitations. firstly, the sample of this study was relatively small and only focused on a private university. for future research, a nationwide study could be considered to secure a wider range of respondents from both private and public higher learning institutions. secondly, only two variables were included in this study, which could indicate that the findings may not be generalizable. future researchers could focus on other general attributes of credit card payment among academicians, for instance, attitude towards credit card debts and payment methods, inducement factors for acquiring and choice of credit cards. this study provides favorable implications to a number of interest groups on repayment studies. stakeholders would be interested to understand the personal attitude, spending pattern and repayment pattern of academicians. the outcome would assist cardholders to fathom the importance of their attitude towards managing money, credit and debt which could eventually contribute towards a healthier credit card repayment pattern. in addition, the above findings could provide some implications to banks and credit card companies. for instance, banks could publish or share relevant information to the public pertaining to the importance of credit card repayment and the consequences for failing to do so. in terms of the big picture, the findings from this study would enable the government to understand the correlation between credit card repayment and bankruptcy rates among credit card holders. campaigns could be conducted by the government to educate young consumers on the usage of credit cards an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 37 including tightening of regulations for new credit card applicants. outcome of this study would also enable credit card companies to review the requirements of issuing credit cards in contributing towards a financially healthier society. in light of the results, this study is also expected to positively reach out to the younger generation through the academicians. peer effect plays an important role in lowering repayment defaults (breza, 2012) and this study has evidenced that peers’ repayment behavior influences one’s repayment behavior significantly. therefore, educators and, academicians alike are encouraged to portray a clean and financial-distress-free image to the young and to foster positive peer-to-peer influences as the way forward to an informed society of low poverty. references ahmed, z. u., ismail, i., sohail, m. s., tabsh, i., & alias, h. (2010). malaysia consumers’ credit card usage behavior. asia pacific journal of marketing and logistics, 22(4), 528-544. ajzen, i. (1991). the theory of planned behavior. organizational behavior and human decision processes, 50(3), 179-211. ajzen, i. (2006). constructing a tpb questionnaire: conceptual and methodological considerations. retrieved from http://www-unix.oit. umass.edu/%7eaizen/ amin, h. (2012). patronage factors of malaysian local customers toward islamic credit cards. management research review, 35(6), 512-530. amin, h. (2013). factors influencing malaysian bank customers to choose islamic credit cards: empirical evidence from the tra model. journal of islamic marketing, 4(3), 245-263. ausubel, l. m. (1997). credit card defaults, credit card profits, and bankruptcy. am. banker. lj, 71, 249. awanis, s., & chi cui, c. (2014). consumer susceptibility to credit card misuse and indebtedness. asia pacific journal of marketing and logistics, 26(3), 408-429. bank negara malaysia. (2016). bnm catalogue monthly statistical bulletin june 2016. retrieved from http://www.bnm.gov.my/ index.php?ch=en_publication_catalogue&pg=en_publication_ msb&mth=6&yr=2016&lang=en. bank negara malaysia. (2016). bnm payment systems list of regulatees. retrieved from http://www.bnm.gov.my/?ch=ps&pg=ps_ regulatees#credit. bernama. (2017, october 10). maybank eyes rm2bil growth in card billings market share. retrieved from https://www.thestar.com.my/business/business news/2017/10/10/maybank-eyes-rm2bil-growth-in-card-billings-marketshare/. 38 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 bertaut, c. c., & haliassos, m. (2001). debt revolvers for self control. discussion paper. bertaut, c. c., haliassos, m., & reiter, m. (2009). credit card debt puzzles and debt revolvers for self control. review of finance, 13(4), 657-692. bresiger, g. (2018). the fed could trigger a personal debt crisis. new york post: retrieved from https://nypost.com/2018/02/03/the-fed-could-trigger-apersonal-debt-crisis/. breza, e. (2012). peer effects and loan repayment: evidence from the krishna default crisis. job market paper mit, 1. brown, s., taylor, k., & price, s. w. (2005). debt and distress: evaluating the psychological cost of credit. journal of economic psychology, 26(5), 642-663. cankaya, s., ucal, m., & o’neil, m. l. (2011). effects of gender on credit card usage among university students in turkey. african journal of business management, 5(22), 9023-9030. cebeci, i., & sanli, t. (2016). a research study on the attitude towards credit card usage of the academic staff at giresun university in turkey. global journal for research analysis, 5(4), 117-119. chan, r. (1997). demographic and attitudinal differences between active and inactive credit cardholders – the case of hong kong. international journal of bank marketing, 15(4), 117-25. chatterjee, p., & rose, r. l. (2011). do payment mechanisms change the way consumers perceive products? journal of consumer research, 38(6), 1129-1139. chien, y. w., & devaney, s. a. (2001). the effects of credit attitude and socioeconomic factors on credit card and installment debt. journal of consumer affairs, 35(1), 162-179. chin, w. w., gopal, a., & salisbury, w. d. (1997). advancing the theory of adaptive structuration: the development of a scale to measure faithfulness of appropriation. information systems research, 8(4), 342-367. chong, f. n. (2017). factors affecting credit card debt payment in malaysia. international journal of business and economics, 16(1), 21. chudry, f., foxall, g., & pallister, j. (2011). exploring attitudes and predicting intentions: profiling student debtors using an extended theory of planned behavior. journal of applied social psychology, 41, 119-149. doi:10.1111/j.1559-1816.2010.00705.x dewri, l. v., islam, m.r., & saha, n.k. (2016). behavioral analysis of credit card users in a developing country: a case of bangladesh. international journal of business and management, 11(4), 299. elangkovan, k., & ahmed, r. a. l. (2013). bankruptcy: a natural phenomena being attacked on malaysians. international journal of humanities and management sciences, 1(1), 74-76. an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 39 feinberg, r. a. (1986). credit cards as spending facilitating stimuli: a conditioning interpretation. journal of consumer research, 13(3), 348356. fornell, c., & larker, d. f. (1981). evaluating structural equation models with unobservable variables and measurement error. journal of marketing research, 39-50. furnham, a., & boo, h. c. (2011). a literature review of the anchoring effect. the journal of socio-economics, 40(1), 35-42. godwin, d. d. (1998). household debt quintiles: explaining changes 1983– 1989. journal of consumer affairs, 32(2), 369-393. hair, j. f., black, w. c., babin, b. j., & anderson, r. r. (2010). multivariate data analysis. prenticehall, upper saddle river, nj. hamilton, r., & khan, m. (2001). revolving credit card holders: who are they and how can they be identified? service industries journal, 21(3), 3748. hayhoe, c. r., leach, l. j., turner, p. r., bruin, m. j., & lawrence, f. c. (2000). differences in spending habits and credit use of college students. journal of consumer affairs, 34(1), 113-133. huebner, j., fuchs, k., & ilic, a. (2018). preventing over-spending: increasing salience of credit card payments through smartphone interventions. multikonferenz wirtschaftsinformatik. data driven x-turning data into value, 4. hussin, s. r., kassim, s., & jamal, n. (2013). credit card holders in malaysia: customer characteristics and credit card usage. international journal of economics and management, 7(1), 108-122. jusoh, z., & lin, y. l. (2012). personal financial knowledge and attitude towards credit card practices among working adults in malaysia. international journal of business and social science, 3(7),176-185. kaynak, e., & harcar, t. (2001). consumers’ attitudes and intentions towards credit card usage in an advanced developing country. journal of financial services marketing, 6(1), 24-39. kennedy, b. & wated, g. (2011). predicting credit card debt among college students: the attitude-behavior relation. journal of psychology & behavioral sciences, 22, 43-50. kidwell, b., & turrisi, r. (2000). a cognitive analysis of credit card acquisition and college student financial development. journal of college student development, 41(6), 589-98. kim, h., & devaney, s. a. (2001). the determinants of outstanding balances among credit card revolvers. journal of financial counseling and planning, 12(1), 67. kinsey, j. (1981). determinants of credit card accounts: an application of tobit analysis. journal of consumer research, 8(9), 172-182. 40 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 kish, a. (2006). perspectives on recent trends in consumer debt. federal reserve bank of philadelphia, philadelphia. klein, l. (1999). it’s in the cards: consumer credit and the american experience. greenwood publishing group. krejcie, r. v., & morgan, d. w. (1970). determining sample size for research activities. educ psychol meas. lachance, m. j. (2012). young adults’ attitudes towards credit. international journal of consumer studies, 36(5), 539-548. leow, m., & crook, j. (2014). intensity models and transition probabilities for credit card loan delinquencies. european journal of operational research, 236(2), 685-694. limbu, y. b., huhmann, b. a., & xu, b. (2012). are college students at greater risk of credit card abuse? age, gender, materialism and parental influence on consumer response to credit cards. journal of financial services marketing 17(2),148-162. livingstone, s. m., & lunt, p. k. (1992). predicting personal debt and debt repayment: psychological, social and economic determinants. journal of economic psychology, 13(1), 111-134. lydia, n. (2017). credit card spending on the rise. the malaysian reserve. retrieved from https://themalaysianreserve.com/2017/07/17/credit-cardspending-rise. mchugh, s., & ranyard, r. (2012). credit repayment decisions: the role of long-term consequence information, economic and psychological factors. review of behavioural finance, 4(2), 98-112. mottola, g. r. (2013). in our best interest: women, financial literacy, and credit card behavior. numeracy, 6(2), 4. navarro-martinez, d., salisbury, l. c., lemon, k. n., stewart, n., matthews, w. j., & harris, a. j. (2011). minimum required payment and supplemental information disclosure effects on consumer debt repayment decisions. journal of marketing research, 48(spl), s60-s77. nga, j. k. h., yong, l. h. l., & sellappan, r. (2011). the influence of image consciousness, materialism and compulsive spending on credit card use intentions among youth. young consumers, 12(3), 243-53. nurul hudani, m. n., marof, r. & noor hisham, m. n. (2013). the effects of relationship management, self-management and social awareness sub-scales on transformational leadership behaviour among educational leaders. akademika, 83(2&3), 3-11. omar, n. a., rahim, r. a., wel, c. a. c., & alam, s. s. (2014). compulsive buying and credit card misuse among credit card holders: the roles of selfesteem, materialism, impulsive buying and budget constraint. intangible capital, 10(1), 52-74. an empirical investigation on credit card repayment pattern of academicians in malaysia: 21-42 41 prelec, d., & simester, d. (2001). always leave home without it: a further investigation of the credit-card effect on willingness to pay. marketing letters, 12(1), 5-12. raghubir, p., & srivastava, j. (2008). monopoly money: the effect of payment coupling and form on spending behavior. journal of experimental psychology: applied, 14(3), 213. ramayah, t., nasser, n., aizzat, m., & lim, h. c. (2002). cardholders’ attitude and bank credit card usage in malaysia: an exploratory study. asian academy of management journal, 7(1), 75-102. rasiah, d., & masuod, s. (2013). the flexible alternative consumer financing in malaysia, credit cards. journal of social and development sciences, 4(3), 147-151. roberts, j. a., & jones, e. (2001). money attitudes, credit card use, and compulsive buying among american college students. journal of consumer affairs, 35(2), 213-240. sekaran, u., & bougie, r. (2009). research methods for business (5th ed.) john wiley & sons ltd. shimp, t. a., & moody, m. p. (2000). in search of a theoretical explanation for the credit card effect. journal of business research, 48(1), 17-23. silva, j., & draut, t. (2004). generation broke: the growth of debt among young americans. new york: demos: borrowing to make ends meet briefing paper, 2. soll, j. b., keeney, r. l., & larrick, r. p. (2013). consumer misunderstanding of credit card use, payments, and debt: causes and solutions. journal of public policy & marketing, 32(1), 66-81. soman, d. (2001). effects of payment mechanism on spending behavior: the role of rehearsal and immediacy of payments. journal of consumer research, 27(4), 460-474. soman, d., & cheema, a. (2002). the effect of credit on spending decisions: the role of the credit limit and credibility. marketing science, 21(1), 3253. sprenger, c., & stavins, j. (2008). credit card debt and payment use. federal reserve bank of boston working paper no. 08-2. stavins, j. (2000). credit card borrowing, delinquency, and personal bankruptcy. new england economic review, 15. stewart, n. (2009). the cost of anchoring on credit-card minimum repayments. psychological science, 20(1), 39-41. surane, j. (2017), u.s. credit-card debt surpasses record set at brink of crisis, bloomberg. retrieved from https://www.bloomberg.com/news/ articles/2017-08-07/u-s-credit-card-debt-surpasses-record-set-at-brinkof-crisis. tan, a. k., ten, s. t., & loke, t. j. (2011). credit card holders, convenience users and revolvers: a tobit model with binary selection and ordinal treatment. journal of applied economics, 14(2), 225-255. 42 the international journal of banking and finance, vol. 13. number 2, 2017: 21-42 tasnim, l. (2017, august 21). 3.6 million credit card holders have rm36.9 billion outstanding balance. new straits times. retrieved from https:// www.nst.com.my/news/nation/2017/08/270620/36-million-credit-cardholders-have-rm369-billion-outstanding-balance themba, g., & tumedi, c. b. (2012). credit card ownership and usage behavior in botswana. international journal of business administration, 3(6), 60-71. theory of planned behavior. (1991). value based management. retrieved from http://www.valuebasedmanagement.net/methods_ajzen_theory_ planned_behavior.html. wang, l., lu, w., & malhotra, n. k. (2011). demographics, attitude, personality and credit card features correlate with credit card debt: a view from china. journal of economic psychology, 32(1), 179-193. webley, p., & nyhus, e. k. (2001). life-cycle and dispositional routes into problem debt. british journal of psychology, 92(3), 423-446. wong, k. y., & lynn, m. (2017). the easy-money effect: credit card spending and hard-work reminders. journal of consumer marketing, 34(7), 541551. yoo, p. s. (1998). still charging: the growth of credit card debt between 1992 and 1995. federal reserve bank of st. louis review, (1), 19-27. zhao, y., zhao, y., & song, i. (2009). predicting new customers’ risk type in the credit card market. journal of marketing research, 46(4), 506-517. the international journal of banking and finance, vol. 15. number 2, 2020: 95-117 95 a b s t r a c t as an important part of the new environmental governance system in china, the policy effect of central environmental inspection has gained more attention. based on the data from heavily polluting listed companies in china, this paper examines the impact of central environmental inspection on corporate value by using an event study approach. the result of the study demonstrates that the central environmental inspection causes a general and significant negative impact on the corporate value of heavily polluting listed companies. more specifically, the market value of private firms and small-scale firms declined more than that of state-owned firms and big-scale firms. in addition, under the deterrent of the central environmental inspection, political connections have no longer become the effective way for polluting firms to evade strict environmental regulation. how to cite this article: tingting, x., & yong, w. (2020). the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed companies in china. international journal of banking and finance, 15(2), 95-117. https:// doi.org/10.32890/ijbf2020.15.2.5 the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed companies in china 1xie tingting peking university school of economics, beijing, china wang yong policy research center for environment and economy, ministry of ecology and environment, beijing china 1corresponding author: xietingting19@pku.edu.cn; wangyong_1228@163.com a r t i c l e i n f o article history: received 28 february 2020 revised 22 june 2020 accepted 29 june 2020 published 31 july 2020 jel code: g38, q58 keywords: central environmental inspection, event study, stock market, china 96 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 1. introduction in the space of a few years, china’s efforts to protect ecological environment have been greatly enhanced. taking the improvement of air quality as an example, the average concentration of pm10 in 338 prefecture-level cities had decreased by 22.7 percent in 2017 compared with 2013. while the average concentration of pm2.5 in beijing-tianjin-hebei region, yangtze river delta and pearl river delta decreased by 39.6 percent, 34.3 and 27.7 percent respectively. during the same period, the new environmental governance system was gradually taking shape in china. as the core institutional arrangement of environmental governance framework in the new era, the central environmental inspection which was implemented in 2016 directly promoted resolving a large number of long-standing environmental problems. and then it was defined as the regular function of the ministry of ecology and environment established in 2018 which means that it will play a more important role in china’s environmental protection. the fundamental purpose of central environmental inspection is to urge local governments to fulfill the responsibility of handling the environmental protection seriously and to promote the internalisation of environmental cost of polluting enterprises fully. from the perspective of financial market, the corporate environmental performance would become an important factor affecting investors’ decision-making under the strong promotion of central supervision. that is to say, if investors can respond positively to the environmental inspection of central government and then change their investment behaviours correspondingly to make a punitive reaction to the heavily polluting enterprises, it can be proved that central environmental inspection is really an effective policy that can drive the green transformation of enterprises through the path of financial market. the experiences of pollution control in united states and europe union both showed that the feedback of financial market on environmental policies, such as environment law enforcement and information disclosure, was the successful way to promote the internalisation of environmental cost. there are two types of literature focusing on the effect of environmental policies from the field of financial market. the first is to test the investors’ reaction to the environmental information disclosure. based on the us stock market, badrinath and bolster (1996) found that companies suffered an average market value loss of 0.43 percent after the disclosure of environmental penalty information. konar and cohen (1997) also believed that the environmental information disclosure was an effective mechanism to reduce pollution emission of polluting enterprises. when the environmental penalty information was released, companies who suffered a larger loss of market value reduced more pollution emission than other companies in the same industry. foulon et al.’s the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 97 companies in china: 95-117 study (2002) showed that the environmental information disclosure was better in improving the corporate environmental performance than the traditional legal regulation in canada. based on european and indian respectively, lundgren and olsson (2010), and gupta and goldar (2002) also found a similar mechanism on the impact of environmental disclosure policies on financial markets. however, most of the studies in china indicated that the response to environmental illegal information disclosure was weak in a-share market. in the study by xu et al. (2012), the environmental illegal information published by the ministry of environmental protection had no significant impact on the stock price of listed companies. wang and li (2013) found that a-share market had no significant punitive response to the environmental accidents and negative information that had not passed the environmental audit. using the case of pollution incident of zijin mining, shen et al. (2012) found that the response of a-share market to major environmental pollution accidents was weaker than that of the h-share market and a-share market cannot respond effectively to government penalties and environmental litigation. fang and guo (2018) believed that lower environmental violation cost was the fundamental reason for the failure of china’s environmental information disclosure policy because local governments preferred to relax environmental regulation to protect local economic growth and impose soft constraints on local enterprises. the second is the impact of macro environmental policies on the capital market. by using the event study method, ramiah et al. (2013) examined the impact of 19 environmental regulation policies on the stock market in australia from 2005 to 2011 and found that the australian stock market was most sensitive to the release of the carbon pollution reduction scheme. zhang and zhang (2017) evaluated the impact of the promulgation and implementation of china’s new environmental protection law on listed companies in heavy-polluted industries and found that the new law caused significant negative stock price shocks. from the literature review, the existing researches were mainly focused on examining the response of the financial market to corporate environmental information disclosure but less on the impact of macro-environmental policies on capital market. theye still lack in-depth empirical test of central environmental inspection which is not compatible with the actual progresses in china. taking the first round of central environmental inspection as the research object, we empirically examine the short-term impact of central environmental inspection on the corporate value of polluting listed companies using the event study method and the types of companies that are more sensitive to environmental inspection. the study shows that the implementation of central environmental inspection brings obvious negative shock effect on the corporate value of heavily polluted listed companies. and the negative impact becomes more and more significant with the increasing disclosure of central environmental inspection in various 98 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 media. besides, the negative impact of central environmental inspection appears to be more significant in private enterprises and relatively small enterprises. and under the deterrence of central supervision, political connection is no longer an effective way for enterprises to evade environmental regulation and the spillover effect appears in non-inspected provinces. the contribution of this study is mainly reflected in the following two aspects. firstly, existing literatures are mainly focusing on testing the impact of environmental law and regulation, environmental information disclosure and other relevant policies, but little research has been done on the impact of the recent implementation of the new environmental management system in china, lacking necessary evaluation of the effectiveness of central environmental inspection. therefore, this study complemented the existing literatures. secondly, the possible influencing mechanisms on capital market are combed and heterogeneous response of different listed companies is examined in this paper, which with a view of make a useful supplement to the relevant research. the remainder of this paper is structured as follows. the second section, the presents the institutional background and influencing mechanisms of the central environmental inspection. section three describes the research method and data source which were used in this study. section four presents the empirical results. meanwhile, section five reveals the results of heterogeneous effect. finally, section six concludes with a discussion on the outputs of the study. 2. institutional background and influencing mechanisms 2.1 institutional background weak enforcement was always regarded as a major factor affecting the effectiveness of environmental policy efforts in china. one part of this problem is that most works are carried out by local governments under china’s decentralised system of environmental governance. however, local governments are often more concerned with the economic growth rather than environmental protection. therefore, there is a strong incentive for the local governments who are the environmental policy executors to ignore the environmental violations for the sake of getting more fiscal revenue. although the new environmental protection law is called “the strictest environmental law” in recent chinese history after it was issued in 2015, its validity is still being questioned to some extent. the working arrangement of central environmental inspection was set up by china’s central government for the purpose of monitoring the local implementation of environmental laws and policies from 2016 which is an the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 99 companies in china: 95-117 important institutional innovation in the field of ecological civilisation system reform. in the last two years, the first round of inspection saw investigations conducted in 31 provinces and that had successfully solved more than 80,000 environmental problems. specifically, the first round of inspection received more than 135,000 complaints from the public in which it handled 29,000 cases and issued fines totaling 1.43 billion yuan. approximately, 18,448 officials were investigated and 18,199 were found responsible for some cases. in the name of the cpc central committee and the state council, central inspection was endowed with the higher authority and rigidity, emphasising the same responsibilities of cpc committees and governments at all levels in environmental protection. through the implementation of inspection, the targets of environmental supervision are shifted from sole enterprises to both local party committees and governments, making environmental supervision more deterrent than before. in the process of inspection, environmental supervision mainly targets provincial party committees, governments and their relevant departments and can directly sink into municipal party committees and governments or conduct in-depth investigation on some related enterprises if necessary. as shown in table 1, the specific situation of the first round of central environmental inspection was sorted out. on january 4, 2016, the first inspection group was stationed in hebei province to carry out the pilot work of supervision. after the pilot project, the central environmental inspection carried out four batches of supervision work in 2016 and 2017, covering the remaining 30 provinces. according to the work plan of environmental inspection, each batch of inspection lasts for about one month and involves seven or eight provinces in general. the whole inspection process was divided into three stages, each of which lasts for about 10 days. the first stage was to talk with the leaders of provincial cpc committees, governments and related departments to consult them on the relevant materials on local environmental protection work, to visit relevant departments and at the same time to receive the public complaints. the second stage was to conduct an investigation and verification aiming at the problems and clues sorted out during the first period. the purpose of this stage was to assess the seriousness of the problems and implement the allocation of responsibilities. the third stage was to summarize basic conclusions of inspection, present reporting framework and carry out targeted supplementary supervision. and then, about three months later, each environmental inspection group would provide feedback on the inspection results and corrective suggestions on supervised provinces which mainly include problems discovered and penalty opinions. 100 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 table 1. implementation of the first round of central environmental inspection batches period coverd region punishment results pilot project 2016.01.04-2016.02.04 hebei 2856 cases were ordered to be rectified, 123 persons were detained; 65 persons were inquired, 366 persons were held accountable. first batch 2016.07.12-2016.08.19 inner mongolia, heilongjiang, jiangsu, jiangsu, henan, guangxi, yunnan, ningxia 9617 cases were ordered to be rectified, 2659 cases were put on record and punished, 310 persons were detained; 2176 persons were inquired, 3287 persons were held accountable. second batch 2016.11.24-2016.12.30 beijing, shanghai, hubei, guangdong, chongqing, shaanxi and gansu 15 631 cases were ordered to be rectified, 6310 cases were put on record and punished, 265 persons were detained, 4666 persons were inquired, 3121 persons were held accountable. third batch 2017.04.24-2017.05.28 tianjin, shanxi, liaoning, anhui, fujian, hunan and guizhou 20 359 cases were ordered to be rectified, 8687 cases were put on record and punished, 405 persons were detained, 6657 persons were inquired, 4660 persons were held accountable. fourth batch 2017.08.07-2017.09.15 jilin, zhejiang, shandong, hainan, sichuan, tibet, qinghai and xinjiang 32 602 cases were ordered to be rectified, 10 806 cases were put on record and punished, 424 persons were detained, 4855 persons were inquired, 6471 persons were held accountable. 2.2 influencing mechanisms 2.2.1 information transmission as a highly authoritative body, the central environmental inspection can transmit more apparent signals for the determination of central government to strengthen environmental governance which will make investors to begin paying more attention to the environmental performance of listed companies. naturally, compared with other companies, the heavily-polluted enterprises will become the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 101 companies in china: 95-117 the risk aversion object for the potential investors. and although the main purpose of environmental supervision is to urge local governments to fulfil the environmental protection responsibility, its effect will eventually be transmitted to the production process of enterprises and greatly increase the environmental costs. under strict environmental supervision, enterprises need to reallocate part of theinvestment to pollution control, which will crowd out productive investment and lead to a decline in output to some extent. therefore, a sudden increase in the environmental cost caused by the central environmental inspection and its possible effect on corporate performance will become an important factor for investors to consider. in the short run, the central environmental supervision will undoubtedly send adverse news to the capital market and very likely trigger a negative reaction from the capital market, especially for polluting industries. besides, if the accurate information about some polluting companies can be transmitted effectively through the central environmental inspection, the negative reaction of the market value will be strengthened according to the different environmental cost borne by different enterprises. 2.2.2 deterrent effect under the powerful implementation of central environmental inspection, pollution behaviours of all kinds of enterprises and in particular, the local officials who did not act will be punished seriously. hence, for the local governments and polluters, the inspection action itself has a shocking effect. more critically, the local political connections, as an important competitive advantage of polluting enterprises, will be broken to an extent by the strict environmental supervision. the essence of the relationship between government and enterprise is the game between the administrative power and market power, of which political connection is the key factor. under the government-led economic development, some policy risks can be avoided through the political connection between enterprises and government. maung et al. (2015) found that state-owned enterprises have lower environmental taxes due to certain political connections. that is to say, political connections actually affect the implementation of local environmental policies. in fact, the main purpose of the central environmental inspection is to weaken the political connection between local governments and polluting enterprises and promote the implementation of the environmental regulation policies. 2.2.3 resource reallocation from the existing empirical results, the impact of pollution control on production performance shows obvious heterogeneity according to the different characteristics of enterprises. compared with other enterprises, listed companies have a better 102 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 performance in environmental information disclosure and environmental social responsibility. therefore, when facing central environmental supervision, listed companies may be less stressed than non-listed companies. . scattered, messy and highly polluting businesses are the main targets of the central environmental supervision. therefore, small companies in heavily polluting industries are usually under greater pressure, and more likely to reduce production or even be shut down. to a certain extent, due to the short-term crowding-out effect of environmental inspection on the output of smes, listed companies will be more competitive in the market because of their better environmental advantages. from this point of view, the central environmental supervision will bring about the reallocation of resources in the capital market. 3. research methodology 3.1 method we employed an event study approach to examine the reactions of the stock market to the central environmental inspection. event study method is widely used in relevant literature. the basic idea is to make sure whether there are significant abnormal returns of relevant listed companies before and after the occurrence of an event date. 3.1.1 event date to implement an event study, we first needed to identify a clear event date. according to the supervision process, each batch of the central environmental inspection involved two important dates; one was the date when the supervision groups were stationed, the other was the date when the supervision group provided feedback on specific opinions to the local goverment and also to the public synchronously. the latter reflects the inspected problems and corresponding processing results, such as the officials held accountable and the amount of the fine, which may have a direct deterrent effect. before the first batch of central environmental inspection, a pilot project was carried out in hebei province. but from the perspective of media exposure, public attention is relatively lower. in addition, as the central environmental inspection was a newly proposed mechanism, the specific impact on economy and enterprises may still be unknown to investors. after the pilot study period, investors may develop a certain understanding of the central environmental inspection. therefore, the first large-scale central environmental inspection may have a more significant impact on the stock market. for the second one, during the period of the third and fourth batch of central environmental inspection, the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 103 companies in china: 95-117 investors may already have obvious expectations after experiencing the first inspection. and also, the days of the inspection teams’ arrival are close to the feedback days of the previous batch which make it difficult to identify the effective impact of the event. because the first batch of central environmental inspection involved seven provinces, the dates of entry of each inspection group were different, though the dates were close. the date range of stationing was from july 12 to july 19, 2016 and the date range of feedback was from november 14 to november 23, 2016. the keyword search trend and information exposure were provided by the 360 search index (fig.1) the 360 search is a search engine which is commonly used in china as it can draw the public attention to the central environmental supervision events. in terms of trend index, the focus of the keywords of central environmental inspection and environmental inspection increased sharply on july 14, 2016 and july 15, 2016, respectively. in terms of exposure, the two keywords rose sharply on july 15. in the following month, there was still a continuous concern trend of central environmental supervision. but in september and october, after the end of the first batch of central environmental inspection, the public concern on the incident declined dramatically. until november 14, the inspection groups started to send feedback opinions to the inspected provinces and the public attention increased sharply again. thus, we used both the supervision groups station (july 12) and the supervision group feedback (november 14) in the first batch of central environmental supervision as the event dates. figure 1. trends in concern (above) and exposure (below) of central environmental inspection figure 1. trends in concern (above) and exposure (below) of central environmental inspection 3.1.2 event window event study requires the occurrence of events to be unpredictable. however, sometimes it is difficult to ensure that information is not disclosed in advance. and also, the station and feedback process of the inspection was gradual which would cause the attention of media and the public be lagged behind. from figure 1, it can be seen that the attention on environmental inspection was gradually increasing. therefore, according to the existing relevant research, four time windows [-5,5], [-5,10], [-10,10], [-10,20] are set to test the reaction of the stock market. 3.1.3 estimation window and calculation of car for the existing research, we take 120 days before the event day as the estimation window to avoid the possible impact of the event itself on the normal rate of return. capm model is used to estimate the normal return rate. ( )it ft i i mt ft itr r r r       （1） where itr is the daily rate of return of stock i on day t; mtr is the average rate of return in chinese stock market; ftr is the daily risk-free rate of return, which is usually assumed to be zero. itr and mtr , represented by the daily stock return rate of cash reinvestment and the comprehensive market return rate of cash reinvestment (the weighted average method of 104 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 3.1.2 event window event study requires the occurrence of events to be unpredictable. however, sometimes it is difficult to ensure that information is not disclosed in advance. and also, the station and feedback process of the inspection was gradual which would cause the attention of media and the public be lagged behind. from figure 1, it can be seen that the attention on environmental inspection was gradually increasing. therefore, according to the existing relevant research, four time windows [-5,5], [-5,10], [-10,10], [-10,20] are set to test the reaction of the stock market. 3.1.3 estimation window and calculation of car for the existing research, we take 120 days before the event day as the estimation window to avoid the possible impact of the event itself on the normal rate of return. capm model is used to estimate the normal return rate. where itr is the daily rate of return of stock i on day t; mtr is the average rate of return in chinese stock market; ftr is the daily risk-free rate of return, which is usually assumed to be zero. itr and mtr , represented by the daily stock return rate of cash reinvestment and the comprehensive market return rate of cash reinvestment (the weighted average method of circulation market value) respectively, are selected from csmar database. and is the error term. the abnormal return( itar ) and cumulative abnormal return( itcar ) are calculated respectively as below: where and are the constant term and regression coefficient of ols estimation in the event estimation window; 1 2( , )itcar t t is the cumulative abnormal return of listed company i in the event window 1 2( , )t t . 3.2 heterogeneity of market response to examine the heterogeneous reaction to the event, we further calculated the car by dividing the listed companies into different groups according to the enterprise characteristics and tested the significance of the difference. specifically, we considered the following grouping characteristics. figure 1. trends in concern (above) and exposure (below) of central environmental inspection 3.1.2 event window event study requires the occurrence of events to be unpredictable. however, sometimes it is difficult to ensure that information is not disclosed in advance. and also, the station and feedback process of the inspection was gradual which would cause the attention of media and the public be lagged behind. from figure 1, it can be seen that the attention on environmental inspection was gradually increasing. therefore, according to the existing relevant research, four time windows [-5,5], [-5,10], [-10,10], [-10,20] are set to test the reaction of the stock market. 3.1.3 estimation window and calculation of car for the existing research, we take 120 days before the event day as the estimation window to avoid the possible impact of the event itself on the normal rate of return. capm model is used to estimate the normal return rate. ( )it ft i i mt ft itr r r r       （1） where itr is the daily rate of return of stock i on day t; mtr is the average rate of return in chinese stock market; ftr is the daily risk-free rate of return, which is usually assumed to be zero. itr and mtr , represented by the daily stock return rate of cash reinvestment and the comprehensive market return rate of cash reinvestment (the weighted average method of (1) circulation market value) respectively, are selected from csmar database. and it is the error term. the abnormal return( itar ) and cumulative abnormal return( itcar ) are calculated respectively as below: ˆˆit it i i mtar r r    （2） 2 1 1 2 ˆˆ( , ) ( ) t t it it i i mt t t car t t r r       (3） where ˆi and ˆi are the constant term and regression coefficient of ols estimation in the event estimation window; 1 2( , )itcar t t is the cumulative abnormal return of listed company i in the event window 1 2( , )t t . 3.2 heterogeneity of market response to examine the heterogeneous reaction to the event, we further calculated the car by dividing the listed companies into different groups according to the enterprise characteristics and tested the significance of the difference. specifically, we considered the following grouping characteristics. 3.2.1 firm ownership there are great difference between state-owned enterprises and private enterprises in china. the differences are mainly manifested in the following aspects. the first is financing capacity. private enterprises usually face greater financing constraints. the second is political ties. stateowned enterprises are controlled by the governments. the natural political connections make the local governments likely show paternalism toward the state-owned enterprises and tend to relax the environmental regulation on the state-owned enterprises. the third is social responsibility. unlike the private enterprises, the state-owned enterprises not only pursue profits but also have a non-profit social function which determines that the state-owned enterprises need to take both economic and social responsibilities. in addition, the leaders of state-owned enterprises usually have administrative positions that may even be higher than the positions of the officials of local environmental protection departments. therefore, in the face of environmental regulation, the state-owned enterprises have stronger ability to resist. on the other hand, when the environmental protection becomes a national will, in order to meet the requirements of central government, the state-owned enterprises will have to show their nature of quasi-government to assume more social responsibility and improve the environmental behaviours. in this way, it can convey to the public the positive attitude toward the environmental protection and fulfil environmental responsibilities, alleviating the pressure from external public opinions. to examine the heterogeneous influence of central circulation market value) respectively, are selected from csmar database. and it is the error term. the abnormal return( itar ) and cumulative abnormal return( itcar ) are calculated respectively as below: ˆˆit it i i mtar r r    （2） 2 1 1 2 ˆˆ( , ) ( ) t t it it i i mt t t car t t r r       (3） where ˆi and ˆi are the constant term and regression coefficient of ols estimation in the event estimation window; 1 2( , )itcar t t is the cumulative abnormal return of listed company i in the event window 1 2( , )t t . 3.2 heterogeneity of market response to examine the heterogeneous reaction to the event, we further calculated the car by dividing the listed companies into different groups according to the enterprise characteristics and tested the significance of the difference. specifically, we considered the following grouping characteristics. 3.2.1 firm ownership there are great difference between state-owned enterprises and private enterprises in china. the differences are mainly manifested in the following aspects. the first is financing capacity. private enterprises usually face greater financing constraints. the second is political ties. stateowned enterprises are controlled by the governments. the natural political connections make the local governments likely show paternalism toward the state-owned enterprises and tend to relax the environmental regulation on the state-owned enterprises. the third is social responsibility. unlike the private enterprises, the state-owned enterprises not only pursue profits but also have a non-profit social function which determines that the state-owned enterprises need to take both economic and social responsibilities. in addition, the leaders of state-owned enterprises usually have administrative positions that may even be higher than the positions of the officials of local environmental protection departments. therefore, in the face of environmental regulation, the state-owned enterprises have stronger ability to resist. on the other hand, when the environmental protection becomes a national will, in order to meet the requirements of central government, the state-owned enterprises will have to show their nature of quasi-government to assume more social responsibility and improve the environmental behaviours. in this way, it can convey to the public the positive attitude toward the environmental protection and fulfil environmental responsibilities, alleviating the pressure from external public opinions. to examine the heterogeneous influence of central (2) circulation market value) respectively, are selected from csmar database. and it is the error term. the abnormal return( itar ) and cumulative abnormal return( itcar ) are calculated respectively as below: ˆˆit it i i mtar r r    （2） 2 1 1 2 ˆˆ( , ) ( ) t t it it i i mt t t car t t r r       (3） where ˆi and ˆi are the constant term and regression coefficient of ols estimation in the event estimation window; 1 2( , )itcar t t is the cumulative abnormal return of listed company i in the event window 1 2( , )t t . 3.2 heterogeneity of market response to examine the heterogeneous reaction to the event, we further calculated the car by dividing the listed companies into different groups according to the enterprise characteristics and tested the significance of the difference. specifically, we considered the following grouping characteristics. 3.2.1 firm ownership there are great difference between state-owned enterprises and private enterprises in china. the differences are mainly manifested in the following aspects. the first is financing capacity. private enterprises usually face greater financing constraints. the second is political ties. stateowned enterprises are controlled by the governments. the natural political connections make the local governments likely show paternalism toward the state-owned enterprises and tend to relax the environmental regulation on the state-owned enterprises. the third is social responsibility. unlike the private enterprises, the state-owned enterprises not only pursue profits but also have a non-profit social function which determines that the state-owned enterprises need to take both economic and social responsibilities. in addition, the leaders of state-owned enterprises usually have administrative positions that may even be higher than the positions of the officials of local environmental protection departments. therefore, in the face of environmental regulation, the state-owned enterprises have stronger ability to resist. on the other hand, when the environmental protection becomes a national will, in order to meet the requirements of central government, the state-owned enterprises will have to show their nature of quasi-government to assume more social responsibility and improve the environmental behaviours. in this way, it can convey to the public the positive attitude toward the environmental protection and fulfil environmental responsibilities, alleviating the pressure from external public opinions. to examine the heterogeneous influence of central (3) circulation market value) respectively, are selected from csmar database. and it is the error term. the abnormal return( itar ) and cumulative abnormal return( itcar ) are calculated respectively as below: ˆˆit it i i mtar r r    （2） 2 1 1 2 ˆˆ( , ) ( ) t t it it i i mt t t car t t r r       (3） where ˆi and ˆi are the constant term and regression coefficient of ols estimation in the event estimation window; 1 2( , )itcar t t is the cumulative abnormal return of listed company i in the event window 1 2( , )t t . 3.2 heterogeneity of market response to examine the heterogeneous reaction to the event, we further calculated the car by dividing the listed companies into different groups according to the enterprise characteristics and tested the significance of the difference. specifically, we considered the following grouping characteristics. 3.2.1 firm ownership there are great difference between state-owned enterprises and private enterprises in china. the differences are mainly manifested in the following aspects. the first is financing capacity. private enterprises usually face greater financing constraints. the second is political ties. stateowned enterprises are controlled by the governments. the natural political connections make the local governments likely show paternalism toward the state-owned enterprises and tend to relax the environmental regulation on the state-owned enterprises. the third is social responsibility. unlike the private enterprises, the state-owned enterprises not only pursue profits but also have a non-profit social function which determines that the state-owned enterprises need to take both economic and social responsibilities. in addition, the leaders of state-owned enterprises usually have administrative positions that may even be higher than the positions of the officials of local environmental protection departments. therefore, in the face of environmental regulation, the state-owned enterprises have stronger ability to resist. on the other hand, when the environmental protection becomes a national will, in order to meet the requirements of central government, the state-owned enterprises will have to show their nature of quasi-government to assume more social responsibility and improve the environmental behaviours. in this way, it can convey to the public the positive attitude toward the environmental protection and fulfil environmental responsibilities, alleviating the pressure from external public opinions. to examine the heterogeneous influence of central circulation market value) respectively, are selected from csmar database. and it is the error term. the abnormal return( itar ) and cumulative abnormal return( itcar ) are calculated respectively as below: ˆˆit it i i mtar r r    （2） 2 1 1 2 ˆˆ( , ) ( ) t t it it i i mt t t car t t r r       (3） where ˆi and ˆi are the constant term and regression coefficient of ols estimation in the event estimation window; 1 2( , )itcar t t is the cumulative abnormal return of listed company i in the event window 1 2( , )t t . 3.2 heterogeneity of market response to examine the heterogeneous reaction to the event, we further calculated the car by dividing the listed companies into different groups according to the enterprise characteristics and tested the significance of the difference. specifically, we considered the following grouping characteristics. 3.2.1 firm ownership there are great difference between state-owned enterprises and private enterprises in china. the differences are mainly manifested in the following aspects. the first is financing capacity. private enterprises usually face greater financing constraints. the second is political ties. stateowned enterprises are controlled by the governments. the natural political connections make the local governments likely show paternalism toward the state-owned enterprises and tend to relax the environmental regulation on the state-owned enterprises. the third is social responsibility. unlike the private enterprises, the state-owned enterprises not only pursue profits but also have a non-profit social function which determines that the state-owned enterprises need to take both economic and social responsibilities. in addition, the leaders of state-owned enterprises usually have administrative positions that may even be higher than the positions of the officials of local environmental protection departments. therefore, in the face of environmental regulation, the state-owned enterprises have stronger ability to resist. on the other hand, when the environmental protection becomes a national will, in order to meet the requirements of central government, the state-owned enterprises will have to show their nature of quasi-government to assume more social responsibility and improve the environmental behaviours. in this way, it can convey to the public the positive attitude toward the environmental protection and fulfil environmental responsibilities, alleviating the pressure from external public opinions. to examine the heterogeneous influence of central the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 105 companies in china: 95-117 3.2.1 firm ownership there are great difference between state-owned enterprises and private enterprises in china. the differences are mainly manifested in the following aspects. the first is financing capacity. private enterprises usually face greater financing constraints. the second is political ties. state-owned enterprises are controlled by the governments. the natural political connections make the local governments likely show paternalism toward the state-owned enterprises and tend to relax the environmental regulation on the state-owned enterprises. the third is social responsibility. unlike the private enterprises, the state-owned enterprises not only pursue profits but also have a non-profit social function which determines that the state-owned enterprises need to take both economic and social responsibilities. in addition, the leaders of state-owned enterprises usually have administrative positions that may even be higher than the positions of the officials of local environmental protection departments. therefore, in the face of environmental regulation, the state-owned enterprises have stronger ability to resist. on the other hand, when the environmental protection becomes a national will, in order to meet the requirements of central government, the stateowned enterprises will have to show their nature of quasi-government to assume more social responsibility and improve the environmental behaviours. in this way, it can convey to the public the positive attitude toward the environmental protection and fulfil environmental responsibilities, alleviating the pressure from external public opinions. to examine the heterogeneous influence of central environmental supervision on state-owned and non-state-owned enterprises, we constructed variable isoe , which equals to one for state-owned enterprise, and zero otherwise, according to the ownership of listed companies in csmar database. 3.2.2 political connection under the government-led economic development, some risks from the environmental regulation policies can be avoided by building the local political connections. compared with the natural political connections of state-owned enterprises, the private enterprises are more initiative and active in building their political connections. because political ties are helpful to relieve environmental policy pressure for enterprises, the disclosure quality of environmental information of private enterprises with high political ties is more sensitive to environmental regulations. so, the variable iconnect is built to consider the heterogeneous influence of the central environmental inspection on the enterprises with different political affiliates. in this study, iconnect is a dummy variable for political connections, which takes the value of one if a top management team member belongs to the governments and zero otherwise. 106 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 3.2.3 firm size because of the existence of scale economy, the emission reduction costs, especially the fixed costs, will be apportioned with the increase of enterprise scale, such as sales volume and lead to the decline of the long-term average cost curve. therefore, the relative emission reduction cost of large enterprises will be smaller. compared with the large-scale enterprises, the rising environmental costs will bring greater burden to small-scale enterprises, making the survival of small-scale enterprises more difficult. also, under the environmental regulation, the dominant position of large-scale enterprises will be further enhanced because of the increasing entry barriers. the inconsistency of raised environmental cost will bring different effect on enterprises of different sizes. we used the logarithm of the total assets of listed companies to reflect the firm size and examined the heterogeneity of the impact of central environmental supervision on different scale enterprises. 3.3 data source the data of listed companies used in this paper are from the china stock market and accounting research database (csmar). according to the guidelines for environmental information disclosure of listed companies issued by the ministry of environmental protection, 19 industries are defined as heavily polluting industries. at last, corresponding to launch event and feedback event of central environmental inspection, there are 790 and 814 listed companies in our research, which cover all polluting corporations listed on china’s stock exchanges. table 2 is the descriptive statistics of relevant variables. table 2. summary statistics of relevant variables variable obs mean std. dev. min max firm characteristics soe 813 0.4108 0.4923 0 1 size 813 22.3205 1.2768 18.6547 28.1788 connect 813 0.2940 0.4559 0 1 inspect 814 0.2236 0.4169 0 1 launch event of inspection car(-5,5) 792 -0.0035 0.0693 -0.3361 0.4019 car (-10,10) 790 -0.0195 0.1077 -0.5548 0.5760 car (-10,20) 790 -0.0401 0.1348 -0.6583 0.4543 feedback event of inspection car(-5,5) 811 -0.0047 0.0735 -0.2713 0.5144 car (-10,10) 814 -0.0242 0.0921 -0.4796 0.4499 car (-10,20) 811 -0.0443 0.1186 -0.5398 0.7937 the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 107 companies in china: 95-117 4. results 4.1 market responses to examine the market responses to the event, we first estimated a market model over a 120 estimation window ending 11days before the event date. a valueweighted average return of all stocks in our sample is adopted as the market return. we then calculated cumulative abnormal returns (cars) over an 11 day (-5, 5) event window centred on the start date and feedback date of the inspection and tested their statistical significance respectively. to establish the robustness of our results, we also considered two longer event windows, a 21-day window and a 31-day window. as shown in table 3, the average of the cars is negative and significantly different from zero. and compared with the start-up day event, we can see a greater negative response of the stock market of the feedback event. according to the cars in the event window of 11 days, the central environmental supervision results in a 0.35 and 0.47 percentage point decline of the market returns during the start-up and feedback event window respectively, but the results are not so significant. in the event window of 21 days, we can see a 1.95 and a 2.87 percentage point significant decline of market returns after the start date and feedback date of the supervision. the results show that with the increasing attention and exposure of the central environmental supervision, the impact on the market value of heavily polluted enterprises become more significant and persistent. using a natural experiment generated by the national specially monitored firms (nsmf) in china, zhang et al. (2018) found that central supervision significantly reduced the emission intensity of industrial enterprises. these results highlight the substantial room for improvement in chinese environmental regulations via central supervision. table 3. stock market reactions model average car inspection start inspection feedback car(-5, 5) -0.0035 -0.0047* (0.0025) (0.0026) car(-10, 10) -0.0195*** -0.0287*** (0.0038) (0.0018) car(-10, 20) -0.0401*** -0.0443*** (0.0048) (0.0042) note: ***, ** and * indicate significant levels of 1%, 5% and 10% respectively. in figure 2, taking 21 days as the event window, we reported the change trend of the car of heavily polluting listed companies. the results show that cars 108 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 of listed companies in heavy polluting industries dropped dramatically and the shock of the start-up day event was more obvious. however, the shocking effect of the feedback event became more and more significant only on the fifth day after the launch of the central environmental inspection. one possible reason is that the feedback is a sequential process. with the mass reporting of the feedback results in various news media, public attention increased quickly which induced investment behaviors to react quickly and lead to the decrease of returns of listed companies in heavyly polluting industries. in the whole event window (-10,10), the start and feedback of central environmental inspection made the listed companies in the heavyly polluting industries get 2% and 3% cars respectively. note: the event day on the left is the start date. the event day on the right is the feedback day. figure 2. time series chart of average cars 4.2 robustness checks the key of the event study is to identify the causal effect of central environmental inspection on the stock market. it is necessary to judge whether the change trend of cars after event day was really caused by the central environmental inspection. firstly, we tested whether the cars day by day before the event day was significant. if the cars was only significant after the event day, it could be proved that the result above was valid. secondly, we examined the result again excluding the listed companies that were likely to be affected by other related events in the event window. specifically, we deleted the sample with more than five days trading interval after the event day and the sample with dividend payment, changes in equity and major trading events happened during the event window. substantial room for improvement in chinese environmental regulations via central supervision. table 3. stock market reactions model average car inspection start inspection feedback car（-5, 5） -0.0035 -0.0047* （0.0025） （0.0026） car（-10, 10） -0.0195*** -0.0287*** （0.0038） （0.0018） car（-10, 20） -0.0401*** -0.0443*** （0.0048） （0.0042） note: ***, ** and * indicate significant levels of 1%, 5% and 10% respectively. in figure 2, taking 21 days as the event window, we reported the change trend of the car of heavily polluting listed companies. the results show that cars of listed companies in heavy polluting industries dropped dramatically and the shock of the start-up day event was more obvious. however, the shocking effect of the feedback event became more and more significant only on the fifth day after the launch of the central environmental inspection. one possible reason is that the feedback is a sequential process. with the mass reporting of the feedback results in various news media, public attention increased quickly which induced investment behaviors to react quickly and lead to the decrease of returns of listed companies in heavyly polluting industries. in the whole event window (-10,10), the start and feedback of central environmental inspection made the listed companies in the heavyly polluting industries get 2% and 3% cars respectively. note: the event day on the left is the start date. the event day on the right is the feedback day. figure 2. time series chart of average cars the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 109 companies in china: 95-117 in table 4, we presented the estimated timing results with the sample deleting the listed companies that may be affected by other important events during the event window. firstly, the shocked effects of start-up event and feedback event are still significantly negative, though the effect has become smaller. as shown in table 4, the cars are – 2.3% and 2.5% respectively in the event window (-10,10). we can also find that the cars are all insignificant within five days before the event date. thus, our estimated results are not disturbed obviously by other events, such as information disclosed in advance. the result in table 4 means that the central environmental inspection brings a significant negative market response. table 4. robustness of market responses event day inspection start inspection feedback coefficients standard deviation coefficients standard deviation -10 0.0018* (0.0009) -0.0044*** (0.0007) -9 -0.0016 (0.0012) -0.0029*** (0.0010) -8 -0.0035** (0.0015) -0.0013 (0.0011) -7 -0.0064*** (0.0016) 0.0031** (0.0014) -6 -0.0079*** (0.0019) -0.0008 (0.0015) -5 -0.0067*** (0.0022) -0.0004 (0.0017) -4 -0.0027 (0.0025) 0.0012 (0.0018) -3 0.0010 (0.0027) -0.0013 (0.0019) -2 0.0014 (0.0028) 0.0028 (0.0022) -1 -0.0024 (0.0032) 0.0012 (0.0023) 0 -0.0104*** (0.0034) 0.0019 (0.0025) 1 -0.0140*** (0.0034) -0.0033 (0.0025) 2 -0.0150*** (0.0033) -0.0014 (0.0024) 3 -0.0173*** (0.0034) -0.0033 (0.0024) 4 -0.0185*** (0.0036) -0.0053** (0.0026) 5 -0.0144*** (0.0036) -0.0027 (0.0027) 6 -0.0138*** (0.0036) -0.0065** (0.0029) 7 -0.0186*** (0.0037) -0.0090*** (0.0030) 8 -0.0207*** (0.0039) -0.0144*** (0.0030) 9 -0.0214*** (0.0041) -0.0177*** (0.0031) 10 -0.0231*** (0.0042) -0.0255*** (0.0031) n 618 741 note: ***, ** and * indicate significant levels of 1%, 5% and 10% respectively. 110 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 4.3 responses of different industries we calculated the average cars of different industries over the 21 days event window. in table 5, most of the industries made significant negative reactions to the first batch of central environmental inspection. and more bigger shock effects were observed in coal mining, ferrous metal mining, chemical industry and rubber industry. table 5. responses of different heavy pollution industries code trade name start-up event feedback event mean standard deviation mean standard deviation b06 coal mining and washing -0.0229*** (0.0005) -0.0384*** (0.0049) b07 oil and gas exploitation 0.0214*** (0.0007) 0.0012 (0.0041) b08 ferrous metal mine mining and processing -0.0787*** (0.0017) -0.0256*** (0.0095) b09 non-ferrous metal mine mining and processing 0.0370*** (0.0012) 0.0241*** (0.0038) c13 agricultural and sideline food processing -0.0425*** (0.0008) -0.0348*** (0.0034) c15 wine, beverage and refined tea manufacturing -0.0244*** (0.0009) -0.0605*** (0.0020) c17 textile -0.0541*** (0.0009) -0.0073*** (0.0026) c19 leather, feather and their products and shoemaking -0.0251*** (0.0005) 0.0044 (0.0079) c22 paper and paper products -0.0054*** (0.0010) -0.0399*** (0.0037) c25 petroleum processing, coking and nuclear fuel processing -0.0160*** (0.0010) -0.0215*** (0.0044) c26 chemical materials and chemicals -0.0565*** (0.0004) -0.0316*** (0.0015) c27 pharmaceutical manufacturing 0.0092*** (0.0003) -0.0408*** (0.0012) c28 chemical fiber manufacturing industry -0.0135*** (0.0010) 0.0041 (0.0035) c29 rubber and plastic products -0.0712*** (0.0007) -0.0283*** (0.0032) c30 non-metallic mineral products -0.0051*** (0.0005) -0.0064** (0.0029) c31 ferrous metal smelting and calendering -0.0085*** (0.0007) 0.0356*** (0.0031) c32 nonferrous metal smelting and calendering -0.0412*** (0.0006) -0.0258*** (0.0035) d44 electricity, thermal production and supply -0.0030 *** (0.0004) -0.0228*** (0.0016) note: ***, ** and * indicate significant levels of 1%, 5% and 10% respectively. the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 111 companies in china: 95-117 5. firm heterogeneity in order to investigate the heterogeneous impact of central environmental inspection on the market value of heavily polluted listed enterprises, we selected enterprise ownership, political connection and enterprise scale totally three comparable variables and tested the different shock effects. 5.1 ownership table 6 shows that t the central environmental inspection has a significant negative impact on the market value of private companies in different event windows. however, the start-up and the feedback event of the supervision both have no significant impact on state-owned enterprises in the 11 days and 21 days event window. only in the 31 days event window, the negative impact starts to be obvious. through comparing the results between state-owned and private companies, the negative impact of central environmental inspection on private companies is significantly higher than that of state-owned companies, with significant difference of 2.81 percent. on the one hand, the environmental performance of state-owned listed companies, such as environmental information disclosure, is relatively better than private companies. the study by cheng et al. (2017) showed that the corporate political connection can influence companies to more actively disclose environmental information but it can also mask political rent-seeking in the guise of protecting the environment. on the other hand, the ability for the state-owned companies to withstand policy risks is higher than that of the private companies which will bring different anticipation to the investors. table 6. different ownership companies start-up event feedback event [-5,5] [-10,10] [-10,20] [-5,5] [-10,10] [-10,20] private companies -0.0098*** -0.0338*** -0.0543*** -0.0073** -0.0306*** -0.0571*** (-2.93) (-5.80) (-8.21) (-2.10) (-7.07) (-10.24) state-owned companies 0.0054 -0.0080 -0.0192*** -0.0011 -0.0152*** -0.0291*** (1.52) (-1.38) (-2.82) (-0.29) (-3.17) (-4.71) difference -0.0152*** -0.0258*** -0.0351*** -0.0062 -0.0153** -0.0281*** (-3.05) (-3.05) (-3.61) (-1.19) (-2.34) (-3.33) note: double-tailed test and t-value in parentheses are shown in the table. ***, ** and * indicate significant levels of 1 percent, 5 percent and 10 percent respectively under t test. 112 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 5.2 political connections table 7 shows that environmental supervision has a significant negative impact on companies with and without political connection. that means that political affiliation is no longer the main factor affecting the environmental behaviours of polluting companies under the deterrence of central environmental inspection. even though the impact of central environmental inspection on political affiliated companies is relatively higher, the difference is not so obvious between the companies with and that without the political connections. the results are to a great extent related to the ousting of some local officials brought by the central environmental inspection. just as the study conducted by wang et al. (2018), in general, corporate investment expenditures of listed firms decline significantly after the ouster of the politicians, especially for non-soes relative to soes. table 7. companies with and without political connections start-up event feedback event [-5,5] [-10,10] [-10,20] [-5,5] [-10,10] [-10,20] with political connections -0.0031 -0.0190*** -0.0352*** -0.0021 -0.0218*** -0.0425*** (-1.06) (-4.10) (-6.31) (-0.66) (-5.53) (-8.53) without political connections -0.0048 -0.0347*** -0.0518*** -0.0112** -0.0301*** -0.0532*** (-1.02) (-3.79) (-5.40) (-2.64) (-5.42) (-6.95) difference 0.0017 0.0156* 0.0166 0.0091 0.0082 0.0107 (0.32) (1.67) (1.55) (1.60) (1.16) (1.17) note: double-tailed test and t-value in parentheses are shown in the table. ***, ** and * indicate significant levels of 1%, 5% and 10% respectively under t test. 5.30 firm size according to the total assets, companies are divided into large-scale and small-scale groups. specifically, companies below the median are identified as small-scale companies and those above the median are identified as large-scale companies. by comparison, the impact of central environmental inspection on the market value of small-scale companies is significantly greater than that of large-scale companies. as shown in table 8, the impact of the startup and the feedback event on the market value of large-scale companies is 6 and 5 percentage points smaller than that of the small-scale companies in the event window of 31 days. many studies have proved that environmental regulation brought a more adverse effect on small-scale enterprises, such as large firms can successfully lobby government and are less likely to exit a heavily regulated industry than small firms. the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 113 companies in china: 95-117 table 8. companies with different scales start-up event feedback event [-5,5] [-10,10] [-10,20] [-5,5] [-10,10] [-10,20] small-scale -0.0138*** -0.0353*** -0.0691*** -0.0094** -0.03357*** -0.0693*** (-3.84) (-5.67) (-9.67) (-2.54) (6.99) (-12.69) large-scale 0.0071** -0.0095* -0.0094 0.0003 -0.01547*** -0.0195*** (2.16) (-1.75) (-1.53) (0.08) (-3.55) (-3.19) difference -0.0035*** -0.0234*** -0.0597*** -0.0097* -0.01817*** -0.0498*** (-4.28) (-3.08) (-6.32) (-1.88) (-2.79) (-6.10) note: double-tailed test and t-value in parentheses are shown in the table. ***, ** and * indicate significant levels of 1 percent, 5 percent and 10 percent respectively under t test. 5.4 inspected and non-inspected provinces according to target provinces of the first batch of central environmental inspection, we divided the research sample into inspected and non-inspected provinces. the result shows that although the start and the feedback event of the central environmental inspection both exert a significant negative impact on the companies in inspected and non-inspected provinces, there is no significant difference between the two categories which illustrates that the supervision has a diffusive impact on other provinces (table 9). table 9. inspected and non-inspected provinces start-up event feedback event [-5,5] [-10,10] [-10,20] [-5,5] [-10,10] [-10,20] inspected provinces -0.0063 -0.0236*** -0.0443*** 0.0023 -0.0139* -0.0414*** (0.0049) (0.0081) (0.0104) (0.0053) (0.0071) (0.0082) non-inspected provinces -0.0028 -0.0183*** -0.0389*** -0.0068** -0.0271*** -0.0451*** (0.0028) (0.0044) (0.0054) (0.0029) (0.0036) (0.0048) difference 0.0035 0.0053 0.0054 -0.0090 -0.0133* -0.0037 (0.0059) (0.0092) (0.0115) (0.0062) (0.0077) (0.0100) note: double-tailed test and t-value in parentheses are shown in the table. ***, ** and * indicate significant levels of 1%, 5% and 10% respectively under t test. 114 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 6. conclusion by analysing the first batch of central environmental inspection in china, we estimated the short-term effect of policy events on the market value of heavypolluted listed companies. and to further demonstrate the heterogeneous micro effect of central environmental inspection, we investigated the group of listed companies that was more sensitive to the policy shocks. the results show that the central environmental inspection has a significant negative impact on listed companies in heavy polluting industries and the impact is more obvious in coal mining, ferrous metal mining, chemical, rubber and other industries. that is, the environmental supervision can urge the listed companies to improve environmental behaviours effectively to avoid the negative effect on the stock market. in addition, the central environmental inspection has been found to have a more significant negative effect on the market value of private companies and small-scale companies. at the same time, the implementation of central environmental inspection broken the political linkages between polluting enterprises and local governmentand aslo caused a nationwide diffusion effect. the results of this study confirm the real effect of central environmental inspection. firstly, the event has a significant negative impact on the market value of heavily polluted listed companies. the environmental policy signals are transmitted to the investors effectively through the implemention of central environmental inspection which can guide many investors to pay more attention to the environmental performance of listed companies and promote heavily polluted companies to concern about its own environmental performance in the stock market. secondly, the central environmental inspection breaks the role of political links in local environmental pollution and causes deterrent effect to some local governments that have shielded the polluting enterprises, weakening the partial behaviour of local governments and promoting the implementation of environmental policies. this output provides some empirical evidence to support the practical effect of the central environmental supervision and provide new evidence to verify the relationship between environmental supervision and market value. our research also has some limitations. for example, we only recognise the short-term impact of policy shocks on the market value of listed companies and cannot judge the long-term effect of environmental policies on corporate environmental management well. in addition, due to the limitations of the research sample, we cannot effectively see the impact of environmental policies on smes. these questions need to be further explored in future. 7. acknowledgments the authors are grateful for the financial support of the national social science foundation of china (project no. 19cjy029 ). the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 115 companies in china: 95-117 references badrinath, s. g., & bolster, p. j. (1996). the role of market forces in epaepa enforcement activity. journal of regulatory economics, 10(2), 165-181. bailey, j. b., & thomas, d. w. (2017). regulating away competition: tthe effect of regulation on entrepreneurship and employment. journal of regulatory economics, 52(3), 237-254. becker, r. a. (2011). local environmental regulation and plant-level productivity. ecological economics, 70(12), 2516-2522. bina, o. (2010). environmental governance in cchina: wweakness and potential from an environmental policy integration perspective. china review, 10(1), 207-239. cheng z., wang f., keung c., & bai, y. (2017). will corporate political connection influence the environmental information disclosure level? based on the panel data of a-shares from listed companies in sshanghai stock market. journal of business ethics, 143(1): , 209-221. china ecological environment status bulletin. retrieved may 20, 2019, from ministry of environmental protection of the people’s republic of china. (2018). china ecological environment status bulletin 2017. china reviews 1st round of central environmental protection supervision. retrieved may 20, 2019, from available online: https://chemlinked.com/ news/chemical-news/china-reviews-1st-round-centralenvironmentalprotection-supervision (accessed on 20 may 2019). fang y., & guo j. (2018). is the environmental violation disclosure policy effective in china: evidence from capital market reactions. economic research journal, 53(10), 158-174. jérôme foulon j., lanoie, p., & laplante, b. (2002). incentives for pollution control: regulation or information? journal of environmental economics and management, 44(1), 169-187. gupta, s., & goldar, b. (2005). do stock markets penalize environmentunfriendly behaviour: evidence from india. ecological economics, 52(1), 81-95. helland, e., & matsuno, m. (2003). pollution abatement as a barrier to entry. journal of regulatory economics, 24(2), 243-259.available online: http://www.mee.gov.cn/gkml/sthjbgw/qt/201805/t20180531_442212. htm (accessed on 20 may 2019). jaffe, a. b., peterson, s. r., & stavins, p. r. n. (1995). environmental regulation and the competitiveness of u.s. manufacturing: wwhat does the evidence tell us? journal of economic literature, 33(1), 132-163. konar, s., & cohen, m. a. (1997). information as regulation: tthe effect of community right to know laws on toxic emissions. journal of environmental economics and management, 32(1), 109-124. 116 the international journal of banking and finance, vol. 15, no 2, 2020 : 95-117 lanoie, p., ambec, s., cohen, m. a., & elgie, s. (2013). the porter hypothesis at 20: can environmental regulation enhance innovation and competitiveness?. review of environmental economics and policy, 7(1), 2-22. li, q., & feng, b. (2015). environmental regulation, political connection and the quality of environmental disclosure: empirical evidence based on the heavy pollution of listing corporation. economy and management, 29(04), 58-66. liming, z., fei, y., li, y., & guichuan, z. (2019). impact of political connections on corporate environmental performance: from a green development perspective. sustainability, 11(5), 1317. lu, d., fu, p., & yang, d. (2015). media type, media coverage and internal control quality of listed firms. accounting research, 41(4), 78-85. lundgren, t., & olsson, r. (2010). environmental incidents and firm value– international : eevidence using a multi-factor event study framework. applied financial economics, 20(16), 1293-1307. maung, m., wilson, c., & tang, x. (2016). political connections and industrial pollution: eevidence based on state ownership and environmental levies in china. journal of business ethics, 138(4), 649-659. miroshnychenko, i., barontini, r., & testa, f. (2017). green practices and financial performance: a global outlook. journal of cleaner production, 147(3), 340-351. poncet, s., steingress, w., & vandenbussche, h. (2016). financial constraints in cchina: firm-level evidence. social science electronic publishing, 21(3), 411-422. qian, a. m., & yu, z. (2017). government’s environmental regulation, officials’ promotion pressure and enterprises’ technological innovation. technology economics, 36(12), 11-22. ramiah, v., martin, b., & moosa, i. (2013). how does the stock market react to the announcement of green policies? journal of banking and finance, 37(5), 1747-1758. shen, h., ong-b.o, s., xie yue, x., & chen, z, r. (2012). environmental protection, corporate social responsibility and its market response: case study based on the environmental pollution incident of zijin mining group. china industrial economics, 32(1), 141-151. wang, f., xu, l., zhang, j., & shu, w. (2018). political connections, internal control and firm value: evidence from china’s anti-corruption campaign. journal of business research, 86(5), 53-67. wangyao, yw., & li, zheyuan, l. i. y. (2013). green-efficiency of stock market in cchina: stock markets reaction to 2003-2012 environmental events. finance and trade economics, 41(2), 37-48. the stock market’s reaction to strict environmental inspection: evidence from heavily polluting listed 117 companies in china: 95-117 xu, x. d., zeng, s. x., & tam, c. m. (2012). stock market’s reaction to disclosure of environmental violations: eevidence from china. journal of business ethics, 107(2), 227-237. zhang, b., chen, x., ,, & guo h. (2018). does central supervision enhance local environmental enforcement? quasi-experimental evidence from china. journal of public economics, 164(8), 7090. zhang, gen, -w.en, z., & wang -fzhang, wei. , z., (2017). profitability, environmental enforcement and the share price response to environmental law-based on the release of “the new environmental protection law. journal of guizhou university of finance and economics, 25(1), 59-69. zhang, l., ye, f., yang, l., & zhou, g. (2019). impact of political connections on corporate environmental performance: from a green development perspective. sustainability, 11(5), 1317. zhou, h., & yin, h. (2017). stock market reactions to environmental disclosures: nnew evidence from china. applied economics letters, 25(1), 1-4. the efficiency of non-bank financial intermediaries: empirical evidence from malaysia the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 149 the efficiency of non-bank financial intermediaries: empirical evidence from malaysia fadzlan sufian the university of malaysia and cimb bank berhad abstract this paper investigates the performance of malaysian non-bank financial institutions during the period of 2000-2004. several efficiency estimates of individual nbfis are evaluated using the non-parametric data envelopment analysis (dea) method. the findings suggest that during the period of study, scale inefficiency outweighs pure technical inefficiency in the malaysian nbfi sector. we find that the merchant banks have exhibited a higher, technical efficiency compared to their peers. the empirical findings suggest that scale efficiency tends to be more sensitive to the exclusion of risk factors, implying that potential economies of scale may be overestimated when risk factors are excluded. keywords: non-bank financial intermediaries, data envelopment analysis (dea), risk jel classification: g21, g28 1. introduction non-bank financial institutions (nbfis) play important dual roles in a financial system. they complement the role of commercial banks by filling in financial intermediation gaps by offering a range of products and services. they also compete with commercial banks, forcing the latter to be more efficient and responsive to their customers needs. nbfis’ state of development is usually a good indicator to the state of development of a country’s financial system as a whole. the importance of investigating the efficiency and productivity of malaysian nbfis could be best justified by the fact they play important roles in complementing the facilities offered by the commercial banks, as well as being key players in the development of the capital markets. as sophisticated and well-developed as capital markets are considered to be as the hallmark for a market-based economy worldwide, such a study of this nature is particularly important as the health and development of the capital market relies largely upon the performance of nbfis. hence, efficient and productive nbfis are expected to enhance the malaysian capital markets in its pursuit to move towards a full market based economy. ijbf 150 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 despite the significant, economic developments of the nbfi sector, studies that attempt to investigate this issue are relatively scarce. over the years, while there have been extensive literature examining the productivity and efficiency of banking industries in various countries, empirical works on nbfis’ productivity & efficiency are still in its infancy. to the best of our knowledge, there has been no microeconomic study performed with respect to nbfis. the study therefore aims to fill a demanding gap in that case. nevertheless, the study will also be the first to investigate the sources of nbfis’ productivity changes in developing economies. section 2 will provide a brief overview of the malaysian financial system with reviews of related studies. section 3 will outline the approaches to the measurement and estimation of efficiency change, while section 4 will discuss the results. naturally, section 5 will conclude the paper. 2. background and related literature the malaysian financial system can be broadly divided into the banking system and non-bank financial intermediaries. the banking system is the largest component, accounting for approximately 70 percent of the financial system’s total assets. the banking system can be further divided into three main groups, namely the commercial banks, financial companies, and merchant banks. the commercial banks are the main players in the banking system. they are the largest and most significant providers of funds in the banking system, enjoying the widest scope of permissible activities, those of which are able to engage in a full range of banking services. financial companies formed the second largest group of deposit taking institutions in malaysia. traditionally, financial companies specialize in consumption credit, comprising mainly of hire purchase financing, leasing, housing loans, block discounting, and secured personal loans. merchant banks emerged in the malaysian banking scene in 1970, marking an important milestone in the development of the financial system, alongside malaysian corporate development. they play a role in the short-term money market and capital raising activities such as financing, syndicating, corporate financing, and management advisory services that arrange for the issue and listing of shares, as well as managing portfolios. the malaysian financial system’s assets and liabilities continued to be highly concentrated at the commercial banking sector with total assets and liabilities amounting to rm 761,254.8 billion (or 3.05 times the national gdp at the end of 2004). prior to the asian financial crisis in 1997/98, financial companies’ assets and liabilities were seen increasing from only rm531 million (or 0.05 times the national gdp in 1970) to a high of rm 152.4 billion (or 0.77 times in 1997). the ratio however, has gradually declined to rm 123.6 billion (or 0.60 times in 1998) to rm 109,409.8 billion (or 0.52 times the gdp in 2000), before increasing again in year 2001, to reach a post crisis high of rm 141,911.0 billion (or 0.61 times the gdp in 2003). due to further consolidation in the malaysian financial sector, financial companies’ assets as a ratio of the national gdp declined again to reach a low of the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 151 0.27 times in 2004. as for the merchant banks, a similar trend is observed where their assets and liabilities (as a ratio of the national gdp) have been increasing since 1971, reaching a peak of rm 44.3 billion or 0.23 times gdp in 1997 (before the asian financial crisis). during the post crisis period, the merchant banks’ assets and liabilities continued to remain stable at 0.17 to 0.22 times the national gdp. a combination of both financial companies and merchant banks’ total assets reveal that the non-bank financial sector commanded approximately 22.8 percent of the banking system’s total assets and liabilities.1 table 1: assets of the financial system, 1960 – 2004 year commercial banks finance companies merchant banks rm million as a ratio of gdp rm million as a ratio of gdp rm million as a ratio of gdp 1960 1,231.9 0.21 n.a. n.a. n.a. n.a. 1970 4,460.2 0.38 531.0 0.05 19.6* 0.002 1980 32,186.1 0.63 5,635.4 0.13 2,228.7 0.05 1990 129,284.9 1.23 39,448.0 0.50 11,063.2 0.14 1995 295,460.0 1.77 91,892.0 0.55 27,062.0 0.16 1996 360,126.8 1.98 119,768.8 0.65 34,072.8 0.19 1997 480,248.1 2.46 152,386.8 0.77 44,300.0 0.23 1998 453,492.0 2.52 123,596.9 0.68 39,227.8 0.22 1999 482,738.3 2.50 116,438.0 0.60 39,184.0 0.20 2000 512,714.7 2.44 109,409.8 0.52 36,876.0 0.18 2001 529,735.5 2.51 121,811.1 0.58 41,025.2 0.19 2002 563,254.1 2.56 130,520.0 0.59 41,415.5 0.19 2003 629,975.3 2.71 141,911.0 0.61 44,103.6 0.19 2004 761,254.8 3.05 68,421.1 0.27 42,691.0 0.17 source: bank negara malaysia. *as at end 1971. the malaysian financial sector is currently facing a number of challenges such as frequent changes in technology required for modern banking, increasing competition, rising customer expectations, etc. hence, the efficiency and productivity issues have become a major area of concern for the banks’ management. in fact, productivity is an important criterion to measure the performance of banks in addition to profitability, financial, and operational efficiency. an efficient management of banking operations aimed at increasing the efficiency and productivity of the financial sector requires up to date knowledge. 1 the figure is at end-2003, prior to the consolidation of financial companies into their respective commercial banking parents. 152 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 a lot of research work has so far taken place concerning the views about the role of financial & banking developments in economic growth [mckinnon (1973); shaw (1973); rajan & zingales (1998); levine (2004); singh (2005)], as well as banking efficiency and productivity [(das & ghosh (2006); sufian (2007); and weill (2007)].2 similarly, some studies have been undertaken for measuring the productivity and efficiency of banks in malaysia [most notably, katib & matthews (2000) and okuda & hashimoto (2004)]. concerning our information, despite nfbis’ significance towards economic development, studies that attempt to investigate this issue are relatively scarce. over the years, while there have been extensive literature examining the productivity & efficiency of banking industries in various countries, empirical works on nbfis’ productivity & efficiency are still in its infancy. 3. methodology and data a non-parametric data envelopment analysis (dea) is employed with a variable return to scale assumption, measuring malaysian nbfis’ input-oriented technical efficiencies. dea involves constructing a non-parametric production frontier based on the actual input-output observations in the sample, relative to the measured efficiency of each firm in the sample (coelli, 1996). let us give a short description of the data envelopment analysis3. assume that there is data on k inputs and m outputs for each n nbfi. for the ith nbfi, these are represented by the vectors x i and y i , respectively. let us introduce the k x n input matrix, x, and the m x n output matrix, y. to measure the efficiency for each nbfi, we calculate a ratio of all inputs, such as (u’y i /v’x i ), where u is an m x 1 vector of output weights, and v is a k x 1 vector of input weights. to select optimal weights, we specify the following mathematical programming problem: min (u’y i /v’x i ), u,v u’y i /v’x i ≤1, j = 1, 2,…, n, u,v ≥ 0 (1) the above formulation has a problem of infinite solutions; therefore we impose the constraint v’x i = 1, which leads to: min (μ’y i ), μ,ϕ ϕ’x i = 1 μ’y i – ϕ’x j ≤0 j = 1, 2,…, n, μ,ϕ ≥ 0 (2) 2 see berger & humphrey (1997) for an excellent review. 3 good reference books on efficiency measures are coelli et al. (1998), cooper et al. (2000), and thanassoulis (2001). the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 153 where we change notation from u & v to μ & ϕ, respectively, in order to reflect transformations. using the duality in linear programming, an equivalent envelopment form of this problem can be derived: min θ, θ, l y i + yλ > 0 θx i xλ > 0 λ > 0 (3) where θ is a scalar representing the value of the efficiency score for the ith decision-making unit, which will range between 0 and 1. λ is a vector of n x 1 constants. the linear programming has to be solved n times, once for each nbfi in the sample. in order to calculate efficiency under the assumption of variable returns to scale, the convexity constraint (n1'λ=1) will be added to ensure that an inefficient nbfi is only compared against nbfis of similar size; thus providing the basis for measuring economies of scale within the dea concept. for the empirical analysis, all malaysian nbfis would be incorporated. the annual balance sheets and income statements used to construct the variables for the empirical analysis are sourced from published balance sheet information in the annual reports. due to scarce data from m & a activity, the final sample was an unbalanced panel sample of 92 nbfi observations. there are two main approaches that exist in banking theory literature to define the banking function: the production and intermediation approaches [sealey & lindley (1977)]. under the production approach, which was pioneered by benston (1965), a financial institution is defined as a producer of services for account holders. that is, they perform transactions on deposit accounts and process documents such as loans. the intermediation approach on the other hand, assumes that financial firms act as an intermediary between savers and borrowers, hypothesizing total loans and securities as outputs; whereas deposits with labor and physical capital are defined as inputs. for the purpose of this study, a variation of the intermediation approach or asset approach originally developed by sealey and lindley (1977) will be adopted in the definition of inputs and outputs used. given the sensitivity of efficiency estimates to the specification of outputs and inputs, we have estimated two alternative models. in dea model a, we model malaysian nbfis as multi-product firms, producing two outputs by employing two inputs. accordingly, total deposits (x1) include deposits from customers and other banks. fixed assets (x2) are used as input vectors to produce total loans (y1), which include loans to customers and other banks. investments (y2) include investment securities held for trading, investment securities available for sale (afs), and investment securities held to maturity. to assess the importance of risk and lending quality problems in explaining the efficiency of malaysian nbfis, following the approach by the likes of drake and hall (2003) and charnes et al. (1990), loan loss provisions (x3) is incorporated as an input variable in dea model b. 15 4 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 7/ 08 v ol . 5 . n um be r 2: 2 00 8: 1 49 -1 67 table 2: descriptive statistics for inputs and outputs the table presents summary statistics of the variables used to construct the efficiency frontier for both dea model a and dea model b over the period 20002004. the sample is divided into peer groups (i.e., merchant banks and financial companies). mb denotes merchant banks and fc denotes finance companies. 2000 (rmb) 2001 (rmb) 2002 (rmb) 2003 (rmb) 2004 (rmb) outputs mb fc mb fc mb fc mb fc mb fc total loans min 172.05 1,927.44 135.04 887.41 136.73 1,116.10 89.77 1,363.46 136.55 1408.4 mean 1,784.70 6,832.02 1,549.44 6,904.33 1,336.28 7,383.17 1,173.53 9,773.36 1,045.39 9,454.49 max 7,677.01 15,743.03 7,571.63 15,765.02 6,906.83 16,732.43 5,582.32 25,160.44 5,274.91 26,048.86 s.d 2,426.46 4,537.73 2,192.84 4,929.12 2,014.58 5,095.30 1,706.99 7,690.041 1,628.510 8,241.46 investments min 61.79 180.97 74.8 40.55 57.82 41.69 99.51 75.77 98.67 69.91 mean 1,710.31 1,473.56 1,530.41 1,116.91 1,655.44 818.38 2,085.44 966.80 2,058.65 797.97 max 5,525.08 3,416.52 4,985.66 2,800.68 5,999.55 1,730.22 8,023.00 2,454.12 6,558.26 2,317.31 s.d 1,945.90 1,220.12 1,858.97 1,063.46 2,035.57 599.90 2,503.18 991.46 1,974.16 910.38 inputs fixed assets min 0.84 21.68 0.25 5.31 0.32 6.71 0.10 6.54 0.06 2.02 mean 4.46 55.66 7.90 49.63 11.51 88.86 16.21 87.12 14.56 95.84 max 11.71 186.94 39.69 205.86 45.00 425.22 53.69 439.35 54.17 424.60 s.d 4.07 54.42 12.45 58.76 16.62 130.90 19.61 134.07 19.93 141.93 total deposits min 58.30 1,480.76 88.86 913.12 20.23 1,164.16 63.78 1,226.55 74.62 1,084.00 mean 2,331.99 7,145.82 1,906.52 6,514.09 1,555.06 7,445.49 1,660.76 8,306.54 2,003.80 7,903.50 max 8,110.02 14,546.27 8,853.50 13,928.60 5,356.46 16,025.89 5,302.27 19,609.19 5,929.86 20,411.79 s.d 2,543.59 4,183.57 2,596.72 4,757.44 1,676.94 5,590.04 1,676.48 6,506.15 1,796.95 7,057.08 loan loss provisions min 1.18 0.60 10.00 35.18 0.08 17.78 7.47 33.79 17.70 59.72 mean 55.7925 136.40 71.09 119.96 32.78 107.03 47.45 108.10 58.43 155.74 max 160.28 519.78 229.44 384.38 154.35 311.64 253.24 378.73 159.31 347.76 s.d 58.02 168.09 68.35 106.56 47.14 84.65 79.14 101.04 40.40 96.39 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 155 table 2 presents the summary statistics of the input and output variables used to construct the efficiency frontier. during the period of study, it is apparent that the financial companies were almost three times larger (in terms of asset size) and commanded higher market share in terms of loans & deposits, compared with their merchant bank peers. on the other hand, although the merchant banks were smaller, they seem to have produced a higher amount of investments with lower amounts of defaulted loans. the differences are further confirmed by a series of parametric (t-test) and non-parametric (kruskal-wallis and mann-whitney [wilcoxon ranksum] tests), which suggest that the differences in the mean are significant for all variables at the 1 per cent level of significance4. 4. results in this section, we will discuss the technical efficiency change (te) of the malaysian nbfi sector, measured by the data envelopment analysis (dea) method, along with its decomposition into pure technical efficiency (pte) and scale efficiency (se) components. with the existence of scale inefficiency, we will attempt to provide evidence on the nature of returns to scale of malaysian nbfi. the efficiency of malaysian nbfis was first examined by applying the dea method for each year under investigation by employing the traditional input-output variables. we extend the analysis to examine the merchant banks and financial companies’ efficiency results derived from an alternative model, which incorporates a non-discretionary, input variable. 4.1 efficieny of the malaysian nbfi sector table 3 presents the mean efficiency scores of the merchant banks for the years 2000 (panel a), 2001 (panel b), 2002 (panel c), 2003 (panel d), 2004 (panel e), and all years (panel f). the results from dea model a seems to suggest that the merchant banks’ mean technical efficiency has been on a declining trend during the earlier part of the studies, before increasing again during the latter years. the decomposition of overall efficiency into its pure technical and scale efficiency components suggest that the merchant banks have exhibited higher scale efficiency during 2000 and 2002. overall, the results imply that during the period of study, the merchant banks have been operating at the wrong scale of operations. during the period of study, the results from panel f of table 3 seem to suggest that the merchant banks have exhibited a mean technical efficiency of 69.6 percent, suggesting a mean input waste of 30.4 percent. in other words, the merchant banks could have produced the same amount of outputs by only using 69.6 percent of the amount of inputs it uses. from table 3 (panel f), it is also clear that scale inefficiency outweighs pure, technical inefficiency in determining the total technical inefficiency of the merchant banks. 4 investment is not significant in the case of the mann-whitney [wilcoxon rank-sum] and kruskal-wallis tests at any conventional levels. to conserve space, we do not report the results here. they are available from the authors upon request. 156 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 table 3: summary statistics of efficiency measures – merchant banks (dea model a) the table presents mean, minimum, maximum, and standard deviation of malaysian nbfis’ technical efficiency (te), its mutually exhaustive, pure technical efficiency (pte), and scale efficiency (se) components derived from dea model a (excluding the risk factor). panel a, b, c, d, and e shows the mean, minimum, maximum, and standard deviation of te, pte, and se of the merchant banks for the years 2000, 2001, 2002, 2003, and 2004, respectively. panel f presents the merchant banks mean, minimum, maximum, and standard deviation of te, pte, and se scores, respectively. the te, pte, and se scores are bounded between a minimum of 0 and a maximum of 1. efficiency measures mean minimum maximum std. dev. panel a: 2000 technical efficiency 0.908 0.443 1.000 0.193 pure technical efficiency 0.925 0.527 1.000 0.167 scale efficiency 0.974 0.841 1.000 0.056 panel b: 2001 technical efficiency 0.745 0.342 1.000 0.271 pure technical efficiency 0.897 0.547 1.000 0.180 scale efficiency 0.822 0.372 1.000 0.218 panel c: 2002 technical efficiency 0.750 0.216 1.000 0.327 pure technical efficiency 0.851 0.266 1.000 0.266 scale efficiency 0.861 0.438 1.000 0.222 panel d: 2003 technical efficiency 0.506 0.188 1.000 0.320 pure technical efficiency 0.894 0.429 1.000 0.201 scale efficiency 0.562 0.188 1.000 0.298 panel e: 2004 technical efficiency 0.582 0.331 1.000 0.209 pure technical efficiency 0.924 0.685 1.000 0.133 scale efficiency 0.636 0.386 1.000 0.226 panel f: merchant banks all years technical efficiency 0.696 0.188 1.000 0.295 pure technical efficiency 0.897 0.266 1.000 0.190 scale efficiency 0.770 0.188 1.000 0.258 table 4 presents mean efficiency scores of the finance companies for the years 2000 (panel a), 2001 (panel b), 2002 (panel c), 2003 (panel d), 2004 (panel e), and all years (panel f). similar to their merchant bank counterparts, the results from dea model a seem to suggest that the financial companies’ mean technical efficiency has been on a declining trend during the earlier part of the studies, before increasing during the latter years. the decomposition of technical efficiency into its pure technical and scale efficiency components suggest that scale inefficiency outweighs the pure technical inefficiency of the financial companies during all years. the results seem to suggest that the finance companies have exhibited a mean technical efficiency of 44.7 percent, which is lower compared to their merchant bank counterparts. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 157 likewise, the results suggest that the financial companies’ inefficiency was mainly due to scale, rather than pure technical albeit at a higher degree of 44.8 percent (merchant banks – 23.0 percent). the financial companies also seem to have exhibited a lower pure technical efficiency of 82.0 percent (merchant banks – 89.7 percent). overall, the results suggest that compared to their merchant bank counterparts, the financial companies were relatively managerially inefficient in controlling their operating costs and have been operating at a relatively less optimal scale of operations. table 4: summary statistics of efficiency measures – finance companies (dea model a) the table presents mean, minimum, maximum, and standard deviation of malaysian nbfis’ technical efficiency (te), its mutually exhaustive, pure technical efficiency (pte), and scale efficiency (se) components derived from dea model a (excluding the risk factor). panel a, b, c, d, and e shows the mean, minimum, maximum, and standard deviation of te, pte, and se of the finance companies for the years 2000, 2001, 2002, 2003, and 2004, respectively. panel f presents the finance companies’ mean, minimum, maximum, and standard deviation of te, pte, and se scores, respectively. the te, pte, and se scores are bounded between a minimum of 0 and a maximum of 1. efficiency measures mean minimum maximum std. dev. panel a: 2000 technical efficiency 0.538 0.350 1.000 0.216 pure technical efficiency 0.811 0.466 1.000 0.197 scale efficiency 0.679 0.399 1.000 0.228 panel b: 2001 technical efficiency 0.389 0.266 0.693 0.142 pure technical efficiency 0.807 0.491 1.000 0.219 scale efficiency 0.489 0.342 0.693 0.124 panel c: 2002 technical efficiency 0.248 0.058 0.589 0.149 pure technical efficiency 0.828 0.530 1.000 0.186 scale efficiency 0.300 0.092 0.589 0.155 panel d: 2003 technical efficiency 0.490 0.243 0.769 0.140 pure technical efficiency 0.822 0.440 1.000 0.199 scale efficiency 0.599 0.446 0.769 0.104 panel e: 2004 technical efficiency 0.625 0.296 0.974 0.188 pure technical efficiency 0.835 0.428 1.000 0.209 scale efficiency 0.758 0.540 0.974 0.169 panel f: finance companies all years technical efficiency 0.447 0.058 1.000 0.205 pure technical efficiency 0.820 0.428 1.000 0.193 scale efficiency 0.552 0.092 1.000 0.220 158 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 the findings are interesting in that although the merchant banks were small relative to their financial counterparts with having relatively limited operations, they seem to have exhibited higher efficiency levels. the findings support the divisibility theory, which holds that there will be no such operational advantage accruing to large nbfis if the technology is divisible. that is, small scale nbfis can produce financial services at costs per unit output comparable to those of large nbfis, suggesting no or possibly negative association between size and performance. this was made possible as advances in technology reduced the size and cost of automated equipment; thus, significantly enhancing small nbfis ability to purchase expensive technology, implying more divisibility in the banking industry’s technology (kolari & zardkoohi, 1987). since the dominant source of the total technical x(in) efficiency in the malaysian nbfi sector seems to be scale related, it is worth investigating the composition of the efficiency frontier. table 5 shows nbfis that lie on the efficiency frontier under dea model a. table 5: composition of production frontiers (dea model a) bank type 2000 2001 2002 2003 2004 count affin merchant bank mb irs drs drs drs drs 0 affin-acf finance fc drs drs drs drs drs 0 alliance finance fc drs drs drs drs 0 alliance merchant bank mb drs drs drs drs 0 arab-malaysian finance fc crs drs drs drs drs 1 arab-malaysian merchant bank mb crs drs drs drs drs 1 aseambankers mb crs drs crs drs drs 1 bumiputra-commerce finance fc drs drs drs drs drs 0 commerce international merchant bankers mb crs crs drs drs drs 2 eon finance fc drs drs drs drs 0 hong leong finance fc drs drs drs drs drs 0 malaysian international merchant bankers mb irs crs crs 2 mayban finance fc drs drs drs drs drs 0 public finance fc drs drs drs drs 0 public merchant bank mb crs crs drs drs 2 rhb delta finance fc drs drs drs drs 0 rhb sakura merchant bankers mb crs crs crs drs drs 3 southern finance fc drs drs drs drs drs 0 southern investment bank mb crs irs drs crs irs 2 utama merchant bank mb irs drs crs crs crs 3 number of nbfi n 6 4 5 2 1 note: crs – (constant returns to scale); drs – (decreasing returns to scale); irs – (increasing returns to scale); the nbfis corresponds to the shaded regions which have not been efficient in any year in the sample period (2001-2005) compared to the other nbfis in the sample; mb – merchant bank; fc – finance company international journal of banking and finance, vol. 5, iss. 2 [2008], art. 8 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 159 the composition of the efficiency frontier for dea model a suggests that the number of 100 percent efficient nbfis [operating at constant returns to scale (crs)], varies between one to six nbfis. during the period of study, the merchant banks seem to have dominated the efficiency frontier for dea model a. it is also clear from the results that two merchant banks, namely rhb sakura merchant bankers and utama merchant bank, have appeared the most times on the efficiency frontier. a total of eight merchant banks have appeared at least once on the efficiency frontier, while only two merchant banks have failed to make it to the frontier. on the other hand, the results seem to suggest that only one financial company has managed to make it to the frontier, while nine others have never made it to the efficiency frontier throughout the period of study. 2.2 non-performing loans and the gap between the two dea models having established the basic dea model, we now analyze the potential impact of risk and problem loans concerning the efficiency of malaysian nbfis. as indicated previously, these results are obtained by modifying the initial dea model to incorporate an additional, non-discretionary input variable, in the form of provisions of loans losses. in general, the findings seem to suggest that controlling for problem loans resulted in a higher mean technical efficiency of malaysian nbfis during all years5. in line with the findings by drake & hall (2003) and altunbas et al. (2000), the results seem to suggest that potential economies of scale may well be overestimated when risk factors are excluded. likewise, it is clear that the inclusion of loan loss provisions has resulted in a higher mean pure technical efficiency of malaysian nbfis6. the results support earlier findings by altunbas et al. (2000), who had suggested that the mean scale efficiency estimate is much more sensitive than the mean pure technical efficiency estimate to the exclusion of risk factors. we now turn to discuss the impact of the inclusion of loan loss provisions on the evolution of the merchant banks’ technical efficiency. the results from table 6 suggest that the inclusion of risk factors has resulted in a higher technical efficiency for merchant banks. it is also apparent that the inclusion of loan loss provisions has had a greater positive impact on the merchant banks’ scale efficiency. table 7 highlights the results for the financial companies. similar to their merchant bank counterparts, the results from table 7 suggest that the inclusion of risk factors has resulted in a higher technical efficiency for financial companies. likewise, it is also apparent that the inclusion of loan loss provisions has had a greater positive impact on the financial companies’ scale efficiency. with a closer look at the results, it seems that the magnitude of the increase in the financial companies’ pure technical and scale efficiency is higher compared to their merchant bank peers. a plausible reason is that during the period of study, the financial companies had a higher amount of defaulted loans compared to their peers. 5 except for the merchant banks during the year 2000. 6 except for the merchant banks during the year 2000. 160 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 the empirical findings clearly demonstrate the importance of risk in explaining financial institutions’ efficiency, in particular scale efficiency. if anything could be deduced from the results, the omission of risk factors may significantly overestimate financial institutions’ potential economies of scale, which could lead to bias conclusions and consequently, policy recommendations. the findings are particularly important for the malaysian policy makers in its quest to consolidate the banking system further to achieve greater economies of scale and efficiency. the malaysian government has always believed that such a move would result in larger institutions, which could withstand greater competition from foreign players, as well as any shocks to the financial system. as the actual potential economies of scale may significantly be lower than initially expected, policy makers should be table 6: summary statistics of efficiency measures – merchant banks (dea model b) the table presents mean, minimum, maximum, and standard deviation of malaysian nbfis’ technical efficiency (te), its mutually exhaustive, pure technical efficiency (pte), and scale efficiency (se) components derived from dea model b (inclusive of the risk factor). panel a, b, c, d, and e shows the mean, minimum, maximum, and standard deviation of te, pte, and se of the merchant banks for the years 2000, 2001, 2002, 2003, and 2004, respectively. panel f presents the merchant banks mean, minimum, maximum, and standard deviation of te, pte, and se scores, respectively. the te, pte, and se scores are bounded between a minimum of 0 and a maximum of 1. efficiency measures mean minimum maximum std. dev. panel a: 2000 technical efficiency 0.908 0.443 1.000 0.193 pure technical efficiency 0.926 0.532 1.000 0.165 scale efficiency 0.973 0.834 1.000 0.058 panel b: 2001 technical efficiency 0.818 0.437 1.000 0.219 pure technical efficiency 0.897 0.551 1.000 0.179 scale efficiency 0.903 0.683 1.000 0.119 panel c: 2002 technical efficiency 0.837 0.275 1.000 0.251 pure technical efficiency 0.914 0.492 1.000 0.159 scale efficiency 0.905 0.559 1.000 0.189 panel d: 2003 technical efficiency 0.885 0.504 1.000 0.204 pure technical efficiency 0.912 0.513 1.000 0.181 scale efficiency 0.963 0.793 1.000 0.085 panel e: 2004 technical efficiency 0.896 0.700 1.000 0.127 pure technical efficiency 0.950 0.857 1.000 0.048 scale efficiency 0.946 0.700 1.000 0.110 panel f: merchant banks all years technical efficiency 0.869 0.275 1.000 0.199 pure technical efficiency 0.929 0.492 1.000 0.151 scale efficiency 0.927 0.559 1.000 0.123 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 161 table 7: summary statistics of efficiency measures – finance companies (dea model b) the table presents mean, minimum, maximum, and standard deviation of malaysian nbfis’ technical efficiency (te), its mutually exhaustive, pure technical efficiency (pte), and scale efficiency (se) components derived from dea model b (inclusive of the risk factor). panel a, b, c, d, and e shows the mean, minimum, maximum, and standard deviation of te, pte, and se of the merchant banks for the years 2000, 2001, 2002, 2003, and 2004, respectively. panel f presents the merchant banks mean, minimum, maximum, and standard deviation of te, pte, and se scores, respectively. the te, pte, and se scores are bounded between a minimum of 0 and a maximum of 1. efficiency measures mean minimum maximum std. dev. panel a: 2000 technical efficiency 0.823 0.560 1.000 0.174 pure technical efficiency 0.902 0.561 1.000 0.162 scale efficiency 0.918 0.644 1.000 0.122 panel b: 2001 technical efficiency 0.799 0.511 1.000 0.173 pure technical efficiency 0.878 0.517 1.000 0.196 scale efficiency 0.918 0.747 1.000 0.094 panel c: 2002 technical efficiency 0.643 0.324 1.000 0.212 pure technical efficiency 0.860 0.533 1.000 0.181 scale efficiency 0.726 0.512 1.000 0.139 panel d: 2003 technical efficiency 0.801 0.554 1.000 0.157 pure technical efficiency 0.859 0.562 1.000 0.153 scale efficiency 0.940 0.752 1.000 0.085 panel e: 2004 technical efficiency 0.963 0.764 1.000 0.097 pure technical efficiency 0.982 0.769 1.000 0.098 scale efficiency 0.979 0.949 1.000 0.018 panel f: finance companies all years technical efficiency 0.795 0.324 1.000 0.189 pure technical efficiency 0.882 0.517 1.000 0.161 scale efficiency 0.900 0.512 1.000 0.130 more cautious in promoting mergers as a means in achieving greater efficiency by attaining better economies of scale. furthermore, most of the research conducted surrounding the explanation of bank or thrift industry failures had found that failing institutions carried a large proportion of non-performing loans in their books prior to failure [dermiguc-kunt (1989); whalen (1991); barr & siems (1994); berger & humphrey (1992); barr & siems (1994); and wheelock & wilson (1995)]. banks approaching failure tend to have low cost efficiency while experiencing high ratios of problem loans, as failing banks tend to be located far from the best practice frontiers. 162 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 table 8: composition of production frontiers (dea model b) bank type 2000 2001 2002 2003 2004 count affin merchant bank mb irs drs drs crs crs 2 affin-acf finance fc drs drs drs drs irs 0 alliance finance fc drs drs drs crs 1 alliance merchant bank mb drs drs irs irs 0 arab-malaysian finance fc crs drs drs drs drs 1 arab-malaysian merchant bank mb crs drs drs drs crs 2 aseambankers mb crs drs crs crs crs 4 bumiputra-commerce finance fc crs crs drs drs crs 3 commerce international merchant bankers mb crs crs crs crs drs 4 eon finance berhad fc crs drs drs drs 1 hong leong finance fc drs drs drs irs drs 0 malaysian international merchant bankers mb irs crs crs 2 mayban finance fc drs drs drs crs crs 2 public finance fc drs crs crs crs 3 public merchant bank mb crs crs crs crs 4 rhb delta finance fc irs drs crs crs 2 rhb sakura merchant bankers mb crs crs crs drs crs 4 southern finance fc drs drs drs drs irs 0 southern investment bank mb crs irs irs crs irs 2 utama merchant bank mb irs drs crs crs crs 3 number of nbfi n 8 6 7 9 10 note: crs – (constant returns to scale); drs – (decreasing returns to scale); irs – (increasing returns to scale); the nbfis corresponds to the shaded regions which have not been efficient in any year in the sample period (2001-2005) compared to the other nbfis in the sample; mb – merchant bank; fc – finance company next, the composition of the efficiency frontier and the nature of the returns to scale for dea model b are discussed. table 8 presents the results of the nature of returns to scale in the malaysian nbfi sector, derived from dea model b. unlike the results from dea model a, the composition of the efficiency frontier for dea model b suggests that the number of 100 percent efficient nbfis had increased substantially to between six and ten nbfis. the results from dea model b are very much similar to those from dea model a, where the merchant banks seem to have dominated the efficiency frontier. it is apparent from table 8 that the global leaders under dea model b have increased to four merchant banks, while there was only one merchant bank that failed to appear on the efficiency frontier throughout the period of study. unlike dea model a, the results from dea model b suggest that seven finance companies have managed to appear on the efficiency frontier, while there were only three finance companies that have never made it to the efficiency frontier throughout the period of study. t h e in tern a tio n a l jo u rn a l o f b a n kin g a n d f in a n ce, 2007/08 v ol. 5. n um ber 2: 2008: 149-167 163 table 9: summary of the null hypothesis tests of identical technologies between merchant banks and finance companies the table present results from the parametric (anova and t-test) and nonparametric (kolmogorov-smirnov, mann-whitney and kruskall-wallis) tests. the tests are performed to test the null hypothesis that domestic and foreign banks are drawn from the same population (environment). test methodology follows among others, aly et al. (1990), elyasiani and mehdian (1992), and isik and hassan (2002). *** indicate significant at the 5% level. test groups parametric test non-parametric test individual tests analysis of variance (anova) test t-test kolmogorov-smirnov [k-s] test mann-whitney [wilcoxon rank-sum] test kruskall-wallis equality of populations test hypotheses mean mb = mean fc distribution mb = distribution fc median mb = median fc test statistics f (prb > f) t (prb > t) k-s (prb > k-s) z (prb > z) χ2 (prb > χ2) panel a: 2000 te model a te model b pte model a pte model b se model a se model b 15.606*** 0.851 1.107 0.083 14.699*** 1.318 -3.950*** -0.922 -1.052 -0.289 -3.834*** -1.148 1.572*** 1.000 0.857 0.500 1.601*** 0.750 8.500*** 21.500 22.000 26.000 7.500*** 22.500 7.316*** 1.387 2.128 0.524 7.868*** 1.136 panel b: 2001 te model a te model b pte model a pte model b se model a se model b 0.634*** 0.045 0.040 0.051 17.639*** 0.095 -3.678*** -0.213 -0.996 -0.225 -4.200*** -0.308 1.342 0.671 0.671 0.447 1.565*** 0.447 11.500*** 44.000 35.000 44.000 12.000*** 47.000 8.541*** 0.211 1.541 0.283 8.314*** 0.053 16 4 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 7/ 08 v ol . 5 . n um be r 2: 2 00 8: 1 49 -1 67 panel c: 2002 te model a te model b pte model a pte model b se model a se model b 19.470*** 3.911 0.053 1.530 42.959*** 3.494 -4.412*** -1.978 -0.230 -1.237 -6.554*** -1.869 1.565*** 1.342 0.671 0.894 2.012*** 1.342 11.000*** 24.000 40.000 34.000 2.000*** 28.000 8.824*** 4.033*** 0.685 1.753 13.367*** 2.887 panel d: 2003 te model a te model b pte model a pte model b se model a se model b 0.021 0.866 0.614 0.489 0.140 0.034 -0.147 -0.753 -0.783 -0.699 0.374 -0.185 0.991 0.798 0.822 0.822 1.016 0.547 37.000 33.000 31.000 33.000 32.000 33.000 0.427 1.073 1.460 1.190 1.128 1.074 panel e: 2004 te model a te model b pte model a pte model b se model a se model b 0.198 0.168 1.149 0.816 1.564 2.158 0.445 0.409 -1.072 -0.903 1.250 1.469 0.657 0.457 0.572 0.514 0.915 0.686 32.500 35.000 28.000 33.000 24.000 36.000 0.114 0.012 0.743 0.150 1.333 0.000 (continued) the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 165 4.3 univariate results after examining the dea results, the issue of interest now is whether the two samples are drawn from the same population (i.e., whether the merchant banks and financial companies possess the same technology). the null hypothesis tested is that the merchant banks and financial companies are drawn from the same population or environment, having identical technologies. we have tested the null hypothesis by using a series of parametric (anova and t-test) and non-parametric [kolmogorovsmirnov, mann-whitney, (wilcoxon rank-sum), and kruskall-wallis] univariate tests. the results are presented in table 9. based on most of the results for dea model a, we failed to reject the null hypothesis at the 5 percent levels of significance that the merchant banks and the financial companies are drawn from the same population having identical technologies, while the results for dea model b failed to reject the null hypothesis during all years. this implies that there is no significant difference between the merchant banks and the financial companies’ technologies (frontiers); thus, it is appropriate to construct a combined frontier. furthermore, the results from the levene’s test for equality of variances do not reject the null hypothesis that the variances among the merchant banks and the financial companies are equal, implying that we can assume the variances between both groups to be equal. 5. conclusion the preferred, non-parametric data envelopment analysis (dea) methodology allowed us to distinguish between three different types of efficiency: technical, pure technical, and scale efficiencies. during the period of study, the results suggested that the malaysian merchant banks exhibited a mean technical efficiency of 69.6 percent, while the financial companies have exhibited a lower mean technical efficiency of 44.7 percent. overall, the results suggest that scale inefficiency dominates pure technical inefficiency effects in determining malaysian nbfis’ total technical inefficiency. the findings also seem to suggest that scale efficiency tends to be much more sensitive to the exclusion of risk factors, implying that potential economies of scale may be overestimated when risk factors are excluded. the empirical findings clearly demonstrate the importance of risk in explaining financial institutions’ efficiency, particularly scale efficiency. if anything could be deduced from the results, the exclusion of risk factors may significantly overestimate the financial institutions potential economies of scale, which could result in bias conclusions and policy recommendations. the findings are important for policy makers in its quest to consolidate the banking system further to achieve greater economies of scale and efficiency. as the actual potential economies of scale may significantly be lower than initially expected, policy makers should be more cautious in promoting mergers as a mean to achieve greater efficiency by attaining better economies of scale. author statement: fadzlan sufian is affiliated with the research department of a local bank, the cimb bank berhad and is a staff member of the the university of malaysia. e--mail: fadzlan14@gmail.com. 166 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 references altunbas, y., liu, m-h., molyneux, p., and seth, r. (2000). efficiency and risk in japanese banking, journal of banking and finance 24 (10): 1605-1628. aly, h.y., grabowski, r., pasurka, c. and rangan, n., (1990). technical, scale and allocative efficiencies in u.s. banking: an empirical investigation, review of economics and statistics 72 (2): 211-218. avkiran, n.k., (2002). productivity analysis in the service sector with data envelopment analysis. camira: n.k. avkiran. barr, r. and siems, t., (1994). predicting bank failure using dea to quantify management quality, working paper, federal reserve bank of dallas. benston, g.j., (1965). branch banking and economies of scale, the journal of finance 20 (2): 312-331. berger, a.n and humphrey, d.b., (1992). measurement and efficiency issues in commercial banking, in z.griliches, (eds.), measurement issues in the service sectors. national bureau of economic research: university of chicago press, 245-279. berger, a.n. and humphrey, d.b., (1997). efficiency of financial institutions: international survey and directions for future research, european journal of operational research 98 (2): 175-212. berger, a.n. and mester, l.j., (1997). inside the black box: what explains differences in the efficiencies of financial institutions, journal of banking and finance 21 (7): 895-947. charnes, a., cooper, w.w., huang, z.m., and sun, d.b., (1990). polyhedral cone – ratio dea models with an illustrative application to large commercial banks, journal of econometrics 46 (1-2): 73-91. coelli, t., (1996). a guide to deap version 2.1, cepa working paper 8/96, university of new england, armidale, australia. coelli, t., rao, d.s.p. and batesse, g.e., (1998). an introduction to efficiency and productivity analysis. boston, ma: kluwer academic publishers. cooper, w.w., seiford, l.m., and tone, k., (2000). data envelopment analysis. boston: kluwer academic publishers. das, a., and ghosh, s., (2006). financial deregulation and efficiency: an empirical analysis of indian banks during the post reform period, review of financial economics 15 (3): 193-221. dermiguc-kunt, a., (1989). deposit institutions failure: a review of the empirical literature, federal reserve bank of cleveland economic review 25 (4): 218. drake, l., and hall, m.j.b., (2003). efficiency in japanese banking: an empirical analysis, journal of banking and finance 27 (3): 891-917. elyasiani, e., and mehdian, s., (1992), productive efficiency performance of minority and non-minority owned banks: a non-parametric approach, journal of banking and finance 16 (5): 933-948. isik, i., and hassan, m.k., (2002). technical, scale and allocative efficiencies of turkish banking industry, journal of banking and finance 26 (4): 719-766. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 149-167 167 katib, m. n., and mathews, k., (2000). a non-parametric approach to efficiency measurement in the malaysian banking sector, the singapore economic review 44 (2): 89-114. kolari, j., and zardkoohi, a., (1987). bank costs, structure and performance. lexington books: usa. levine, r., (2004). finance and growth: theory, evidence & mechanism,in aghion, p. and durlauf, s., (eds), handbook of economic growth, amsterdam: northholland, pp. 81, in reforming corporate governance in southeast asia, by khai leong ho (2005), published by institute of southeast asian studies. mckinnon, p.i., (1973). money and capital in economic development, washington d.c., the banking institution. okuda, h., and hashimoto, h., (2004). estimating cost functions of malaysian commercial banks: the differential effects of size, location and ownership, asian economic journal 18 (3): 233-259. rajan r.g., and zingales, l., (1998). financial dependence and growth, american economic review 88 (2): 559-586. sealey, c., and lindley, j.t., (1977). inputs, outputs and a theory of production and cost at depository financial institutions, journal of finance 32 (4): 1251-1266. shaw, e.s., (1973). financial deepening in economic development, new york, oxford university press. singh, c., (2005). financial sector reforms and state of indian economy, indian journal of economics & business 4 (1): 88-133. sufian f., (2007). trends in the efficiency of singapore’s commercial banking groups: a nonstochastic frontier dea window analysis approach, international journal of productivity and performance management 56 (2): 99 – 136. thanassoulis, e., (2001). introduction to the theory and application of data envelopment analysis. kluwer academic publishers: boston. weill, l., (2007) is there a gap in bank efficiency between cee and western european countries? comparative economic studies 49 (1): 101–127. whalen, g., (1991). a proportional hazards model of bank failure: an examination of its usefulness as an early warning tool, federal reserve bank of cleveland economic review 27 (1): 21–31. wheelock, d.c., and wilson, p.w., (1995). explaining bank failures: deposit insurance, regulation and efficiency, review of economics and statistics 77 (4): 689-700. international journal of banking and finance 3-1-2008 the efficiency of non-bank financial intermediaries: empirical evidence from malaysia fadzlan sufian recommended citation market conditions and fund flows: evidence from hedge funds 1 the international journal of banking and finance, volume 10 (number 1), 2013: pages 1-33 market conditions and fund flows: evidence from hedge funds wen-hsiu chou (florida international university, usa), dongmin ke (kean university, usa) danielle xu (gonzaga university, usa/hanken school of economics, finland) ___________________________________________________________ abstract this paper investigates whether market conditions affect fund investor behavior in the hedge fund industry, especially the volatility in the up and down markets. using a sample of 5,254 individual hedge funds from january 1994 to december 2009, we find that hedge fund investors tend to invest less during up and down-volatile markets. they also adopt different investment strategies in these two market conditions. when market is calm and relatively predictable, there is almost no difference in their behaviors between up and down markets. we also find that smart money effect exists over both 3and 12-month periods under all market conditions except volatile markets. a further investigation suggests that the observed smart money effect is largely driven by hedge fund performance persistence, which is present and significant in quiet markets only. the findings are relevant to portfolio theories concerning investor recognition of upside and downside volatilities. keywords: hedge funds, flow sensitivities, smart money effect, performance persistence, time-varying performance, downside/upside volatility jel classification: g11, g23 ______________________________________________________________________ 1. introduction how investors allocate their assets to form portfolio according to risk, especially the risk in the up and down markets, is one of the topics under debate. on the one hand, practitioners use roy’s (1952) safety first principle to argue that investors only care about downside volatility. on the other hand, academic researchers claim that the risk is important in both up and down markets.1 although ang, chen, and xing (2006) have provided evidence to show that investors require a higher return to hold stocks with high downside beta than stocks with low downside beta, it is still not clear whether investors are only concerned about the downside volatility. it is possible that upside volatility also influences investors’ decision making. in this study, we separate market conditions into up, down, quiet, and volatile markets, and their combinations to investigate how market conditions affect fund investor behavior in the hedge 1 see rom and ferguson (1993,1994), kaplan and siegel (1994a,b), and nawrocki (1999). 2 fund industry.2 if upside volatility has no effect on investment behaviors, investors will have the same investment strategies across all market conditions except the down-volatile market; otherwise, investors may show different behaviors between volatile and quiet markets, and probably between up-volatile and down-volatile markets. we also examine whether market conditions have an impact on investors’ ability to identify outperforming hedge funds and hedge funds’ performance persistence. the findings are relevant to portfolio theories concerning investor recognition of upside and downside volatilities. studying the relation between hedge fund flows and market conditions allows for better understanding about investor’s behavior and attitude toward risk at the market level. during the 2008 financial crisis, most hedge fund investors fled out of the industry regardless of the fund performance.3 this seems to support roy’s principle that investors care about downside volatility and prefer safety first. as kelly (1997) documents, investors’ wealth is positively correlated with the probability of being smart money and negatively associated with the probability of being a noise trader. hedge fund investors are generally considered the most sophisticated investors on the market and possibly better informed, partly due to the unique characteristics of hedge fund investors. 4 in addition, they are likely equipped with sound financial knowledge. thus, it is reasonable to assume that hedge fund investors are not noise traders. if fund investors have certain preferences over market risk, their investment decisions will vary with market volatility. docking and koch (2005) show that market direction and volatility affect how investors respond to dividend change announcements. avramova et al. (2011) suggest that hedge fund investors will benefit by incorporating predictability which is based on macroeconomic variables into their investment decisions. among all macroeconomic variables, market volatility is of particular importance. hedge fund performance under each market condition also provides some insights into the debate. as one of the most sophisticated investment advisors, hedge fund managers possess the freedom of adopting flexible investment strategies. as a result, they can adjust fund betas more easily than their mutual fund peers, especially during volatile markets, thus producing better risk-adjusted returns for fund investors. however, hedge fund managers are unable to deal with volatility to provide reliable risk-adjusted returns. 5 some studies have documented that hedge fund performance is time varying and seems to 2 we refer a market characterized by both down and volatile as down-volatile market, both down and quiet as downquiet market, both up and volatile as up-volatile market and both up and quiet as up-quiet market, respectively. 3 for example, an article from new york times shows ‘the gilded age of hedge funds is losing its luster. ··· lose $180 billion during the last three months. investors, particularly wealthy individuals, are heading for the exits.’ (new york times, october 22, 2008) 4 by laws, hedge fund investors must be accredited investors who own a minimum net wealth of $1,000,000 or have a minimum consistent income of $200,000 in the past and future years. 5 for example, an article from the economist complains that ‘we (hedge fund managers) use to promise "absolute returns" but this pledge has proved impossible to honor. instead, we’re going to give you "risk-adjusted" returns or, failing that, "relative" returns. · · · it is also time to move on from the concept of delivering "alpha", international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 3 depend on general business cycles and market movements (for example, see fung and hsieh, 1999; edwards and caglayan, 2001; ennis and sebastian, 2003; and capocci, corhay, and hübner, 2005). when evaluating managerial skills, chen and liang (2007) find that self-reported market timing hedge funds generally possess timing abilities, but patterns of market timing behavior are significantly different in various market states. therefore, market conditions should play a role in the relation between hedge funds and fund investors, and fund investor behavior. using a comprehensive hedge fund sample from the january 1994 to december 2009 period that sees a number of significant equity market swings, we start our analysis by examining fund performance and fund flows independently under different market conditions. our primary investigation on hedge fund performance alone indicates that hedge funds only slightly outperform in up markets than in down markets, but generate much better returns in quiet markets than in volatile markets. when fund performance is examined with further breakdown of different market conditions, we find that hedge funds generate the worst risk-adjusted returns in down-volatile markets. hedge funds do not perform well in upvolatile markets either. this indicates that hedge funds may offer some ‘hedge’ protections in down markets to some extent, but they cannot offer the same protections in volatile markets. on the other hand, while there is very little difference in fund flows during up and down markets, hedge funds experience much higher flows in quiet markets than volatile markets. a further investigation with breakdown of market conditions suggests that hedge fund investors tend to invest much more during down-quiet markets while refrain much of their money during up-volatile and down-volatile markets. our results are consistent with previous findings in mutual fund industry that risk-averse fund investors reduce their investment when markets are expected to be more volatile (e.g., see luo, 2003; ederington and golubeva, 2009; ferson and kim, 2012) and in the general finance that investors request higher returns for bearing macroeconomic risk (savor and wilson, 2012). with initial findings that suggest an asymmetric impact of a particular market condition on both fund performance and new flows, we then turn to a more formal examination on the flow-performance relation under different market conditions while controlling for multiple non-performance related fund characteristics. next, we explore in greater detail flow sensitivities to fund past performance in various market states by using a piecewise linear regression model similar to the existing literature. compared to the mutual fund industry, behavior of hedge fund flows is usually more complicated and affected by many unique features of the hedge fund industry such as high minimum investments, lock-up periods, high-water mark the skill you’ve paid us such fat fees for. upon reflection, we have decided that we’re actually much better at giving you "smart beta".’ (the economist, february 18, 2012) 4 protection, use of leverage, and offshore operation. thus, we control for all these factors while investigating the impact of various market conditions on fund flow sensitivities. we find that prevailing market conditions significantly change the shape of flow-performance relation as well, even after controlling for multiple aforementioned cross-sectional fund characteristics.6 a generally concave flow-performance relationship is detected, but flow sensitivities change significantly when market conditions are examined separately. we further find that hedge fund flows are responsive to past performance asymmetrically. in particular, market conditions in general affect flow sensitivities among the worst and best performing funds. this observation is more evident during upand downvolatile markets. however, fund investors seem to adopt different investment strategies in these two market conditions they reduce investment in past winners in the best and worst fund performing groups when markets are up and volatile, and they are more sensitive to fund performance in the worst fund performing group during down-volatile markets. when market is quiet, there is almost no change in flow-performance relation between up and down markets. flow sensitivities to medium performers stay largely the same (and significant) throughout all market states. when we use a refined fund sample to explore the flow-performance relationship, the findings are generally similar that market volatility matters for investment behavior and both upside and downside volatilities influence investors’ decision making. we also examine whether hedge fund investors have the capability to identify outperforming hedge funds, and if so, whether such ability changes with market conditions. although the empirical evidence from the mutual fund industry is largely mixed, mutual fund investors seem to have the ability to identify funds that will subsequently outperform. 7 since hedge fund investors may behave differently, similar studies have been conducted to examine the smart-money effect in the hedge fund industry, and the empirical evidence is also mixed.8 our paper documents the existence of a general smart-money effect in the hedge fund industry conditional on the market conditions. except volatile markets (both up-volatile and down-volatile), we find that smart money effect exists in all other market conditions over both 3and 12-month investment horizons. this indicates that when market is calm and relatively predictable, fund investors are capable of identifying good managers and can benefit from good performance at least for a short period of time. market volatility does affect the ability of selecting good performers by hedge fund investors. 6 in unreported tables, we also add restriction periods, such as redemption period and required advanced notice period for redemptions, to control for the additional effect of share restrictions in the regressions. our results are robust. variable selections are detailed in the data section. 7 see, for example, gruber (1996), zheng (1999), sapp and tiwari (2004), frazzini and lamont (2008), keswani and stolin (2008), and ivković and weisbenner (2009). 8 see baquero and verbeek (2009), ding et al. (2009), ozik and sadka (2009), and wang and zheng (2008). international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 5 lastly, we investigate fund performance persistence over the same 3and 12-month horizons conditional on the market status. we find that hedge fund performance persists over both 3and 12month periods in all market conditions, with the level of persistence significantly strengthened during quiet markets, and weakened during volatile markets. both upside and downside volatilities affect performance persistence negatively. thus, fund performance in the volatile markets cannot be used to predict future hedge fund performance. this is consistent with busse’s (1999) findings in mutual fund literature. given that the pattern of performance persistence is matched with that of “smart money” effect under various market conditions, the smart money effect in the hedge fund industry is largely driven by fund performance persistence in quiet markets. although it is not clear whether upside market volatility is as important as downside volatility in explaining investor behaviors, our overall results provide evidence that market conditions play a role in the investment process, that market volatility affects not only investors’ ability to pick winner funds but also fund managers’ ability to deliver alpha, and that both upside and downside volatilities influence investment behaviors and hedge fund performance. thus, the strategy of chasing winners and smart money effect do not work during volatile markets, particularly due to the lack of persistency in fund performance. fund performance can be used as a good indicator of management ability, but it is no longer the case when market becomes volatile. our study makes several contributions to the literature. first, we complement previous studies to show that market conditions influence investor behavior and flow-performance relationship in the hedge fund industry. using mutual fund samples, luo (2003), ederington and golubeva (2009), and ferson and kim (2012) show that stock market volatility negatively affects equity fund flows. we show this observation also exists in the hedge fund industry. ferson and kim further indicate that fund flows predict future economic conditions. in our study, investors’ predictability only exists during the quiet markets. sirri and tufano (1998) show that lower search costs increase fund flows to performance sensitivity. huang, wei and yan (2007) explicitly indicate that lower participation costs make investors more sensitive to medium performance and less sensitive to high performance. ding et al. (2009) demonstrate that the flow-performance relation varies with share restrictions. we show that market conditions affect the flow-performance relationship among the high and low performance fund groups but have relatively little impact on the medium performance fund group. second, we demonstrate that the volatility in both up and down markets affects investment behaviors of hedge fund investors. similar to luo’s (2003) findings in the mutual fund industry, hedge fund investors put less money into hedge funds during upand down-volatile markets. they also change their investment strategies and become less sensitive to fund past performance in the up-volatile markets. although previous literature suggests that upside volatility may not be important (rom and ferguson, 6 1993,1994; and nawrocki, 1999), our results demonstrate that, like downside volatility, upside volatility does have a significant negative impact on the investors’ ability to pick outperforming hedge funds and fund managers’ ability to outperform benchmarks. the upside volatility also weakens hedge fund managers’ capability in providing better risk-adjusted returns. it is not clear how important upside volatility should play a role in investment; nevertheless, our results suggest that upside volatility affects investor behaviors. third, we add to a growing body of working papers that examine the existence of smart money effect in the hedge fund industry. ding et al. (2009) document a smart-money effect in the universe of hedge funds, while ozik and sadka (2009) argue that the smart-money effect predominantly stems from highflow-impact funds. on the other hand, wang and zheng (2008) find that the smart-money effect does not exist in the hedge fund industry, at least in the short run. similarly, baquero and verbeek (2009) do not find supporting evidence of smart money effect in their study either. our study differs from these papers on several dimensions. we show that market conditions may be the reason to cause the mixed findings. after we control for the market conditions, smart money exists when market is relatively calm and predictable. however, the smart money effect disappears when market becomes volatile. we further demonstrate the source of the smart money for hedge funds is likely from the performance persistence. agarwal and naik (2000) find that hedge fund performance persists at the quarterly horizon. our paper expands the literature to provide evidence to show that hedge funds can provide persistent performance at maximum of one year horizon after controlling for the market conditions. the rest of the paper is organized as follows. next section describes the data and methodology used in our analyses. section 3 discusses empirical findings of the flow-performance relationship under different market conditions, followed by smart-money tests in section 4. section 5 addresses the performance persistence and section 6 concludes. 2. data and methodology a. data the hedge fund sample used in this study comes from lipper-tass, one of the leading hedge fund research databases. 9 the complete database includes more than 9,000 hedge funds (both active and inactive), and reports detailed information about monthly fund returns, assets under management (aum), as well as other fund characteristics such as fund style, country of operation, minimum investment amount, management and incentive fees, and lockup period for individual hedge funds. our sample 9 we also use cisdm hedge fund database as our hedge fund sample. results are qualitatively similar. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 7 consists of all individual hedge funds in u.s. dollar denomination and with minimum 24 monthly return data. a total of 5,254 individual funds have met the criteria during the period of 1994 to 2009. as shown in the left panel of table 1, our final sample covers 12 investment styles including convertible arbitrage, dedicated short bias, emerging markets, equity market neutral, event driven, fixed income arbitrage, global macro, long/short equity hedge, managed futures, option strategy, multi-strategy, and others. among these investment styles, long/short equity hedge funds account for 37.5% (1,968) of the sample, followed by emerging markets, event driven, and managed futures funds, each accounting for about 10% of the sample. we further investigate countries that funds are domiciled, and find that funds in our sample are from 40 different countries. the top 13 countries are presented in the right panel of table 1, and it is worth noting that more than 90% (or 4,747 funds) are located in the following four countries: united states (38% of the sample), cayman islands (38%), british virgin islands (9%), and bermuda (6%). table 1: summary of hedge fund investment styles and domicile countries the table reports the summary of investment styles and domicile countries of individual hedge funds from lippertass database. the sample includes both live and graveyard funds which are in u.s. dollar denomination and have a minimum of 24 monthly return data available. funds of funds are excluded. sample period is from january 1994 to december 2009. investment style n % domicile country n % cum. % long/short equity hedge 1968 37.46% united states 1994 38.09% 38.09% emerging markets 541 10.30% cayman islands 1966 37.55% 75.64% event driven 540 10.28% british virgin islands 473 9.04% 84.68% managed futures 535 10.18% bermuda 314 6.00% 90.68% multi-strategy 402 7.65% bahamas 130 2.48% 93.16% global macro 319 6.07% ireland 83 1.59% 94.75% equity market neutral 318 6.05% guernsey 67 1.28% 96.03% fixed income arbitrage 207 3.94% luxembourg 57 1.09% 97.12% convertible arbitrage 191 3.64% canada 33 0.63% 97.75% other 172 3.27% netherlands antilles 27 0.52% 98.26% dedicated short bias 39 0.74% jersey 24 0.46% 98.72% options strategy 15 0.29% mauritius 16 0.31% 99.03% undefined 7 0.13% united kingdom 7 0.13% 99.16% total 5254 5235 table 2 reports summary statistics of multiple hedge fund characteristics over the whole sample period. fund characteristics include assets under management (aum, in millions of u.s. dollars), minimum investment amount (in millions of u.s. dollars), management fees (%), incentive fees (%), lockup period (in months), fund age (in years), and four dummy variables indicating whether a fund has a high-water mark, leverage, offshore operation, and manager’s personal capital invested in the fund. the high-water mark dummy is defined as taking the value of 1 if a fund imposes high-water mark protection for investors, and 0 otherwise. the leverage dummy is defined as taking the value of 1 if a fund is leveraged, and 0 otherwise. the offshore operation dummy takes the value of 1 if a fund is operated in a 8 country other than united states, and 0 otherwise. the personal capital dummy takes the value of 1 if a fund manager has personal capital invested in the fund assets, and 0 otherwise. these variables have been shown in the previous research to have an impact on both fund performance and flow-performance relation. goetzmann, ingersoll and ross (2003) find that high-water mark, return volatility, incentive contract, and management fees affect fund performance. agarwal, daniel and naik (2009) demonstrate that incentive contracts, managerial ownership and high-water mark are related to superior performance. aragon, liang and park (2008) find onshore and offshore hedge funds have different behaviors. fund size (aum) is included, as berk and green (2004) document decreasing returns to scale in fund industry.10 table 2: summary statistics of hedge fund characteristics the table reports summary statistics of hedge funds’ characteristics for the whole sample period from january 1994 to december 2009. fund characteristics include assets under management (aum, in $ million), minimum investment (in $ million), management fees (in %), incentive fees (in %), lockup period (in months), fund age (in years), and dummy variables indicating whether the fund imposes high-water mark provision, whether the fund is leveraged, whether the fund is operating outside the u.s. (offshore), and whether the fund has managers’ personal capital invested in the fund asset. the mean, standard deviation (sd), lower quartile (q1), median (median), and upper quartile (q3) of individual characteristics are reported for the whole sample period. variables mean sd q1 median q3 aum (mm) 125.15 368.67 9.42 31.74 101.75 minimum investment (mm) 1.11 14.15 0.10 0.50 1.00 management fees (%) 1.48 0.68 1.00 1.50 2.00 incentive fees (%) 18.29 5.72 20.00 20.00 20.00 lockup period (months) 3.56 6.56 0.00 0.00 6.00 age (years) 6.38 4.13 1.83 5.17 8.25 highwatermark (dummy) 0.66 0.47 0.00 1.00 1.00 leverage (dummy) 0.63 0.48 0.00 1.00 1.00 offshore (dummy) 0.62 0.49 0.00 1.00 1.00 personal capital (dummy) 0.33 0.47 0.00 0.00 1.00 over the whole sample period, the hedge fund industry has an average of $125.2 million assets under management and a standard deviation of $368.7 million in size. hedge funds on average require $1.11 million of minimum investment amounts, with a standard deviation of $14.2 million. hedge funds in general charge 1.48% of fund asset as expenses. the average lockup period is 3.6 months with a standard deviation of 6.6 months. 66% of the funds offer high-water mark protections for their investors, and the incentive fee charged by an average fund is 18.3% (with 5.7% standard deviation). a typical 20% incentive fee is charged by at least half of individual funds (from lower quartile q1 to upper quartile q3). the average age of hedge funds in our sample is 6.4 years, with standard deviation of 4.1 years. 63% of 10 we also have qualitatively the same results when removing funds that are less than $10m and excluding the first 12 months of fund returns (fung and hsieh, 2000). aggarwal and jorion (2010) demonstrate that the median backfill period is 480 days. fung and hsieh (2000) also report the median is 343 days for the backfill period. our adjustment should be able to minimize the backfill bias and omitted returns. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 9 funds are leveraged, and 62% of funds are operating outside the united states. in addition, about 33% of funds have their managers’ personal capital invested in the fund assets.11 two issues are related to our study. ding et al. (2009), and baba and goko (2009) argue that share restrictions affect the shape of flow-performance relation and hedge funds’ survival probability. share restrictions include lockup period, redemption period, advance notice period, subscription period, capacity constrained, open to investment, and onshore operation. another concern is the issue of the omitted time-invariant determinant of fund flows. since our purpose is to investigate whether market conditions affect flow-performance relation, in our later tests, we use lockup period and offshore operation in addition to lagged fund flow to control for share restrictions and some unobserved factors which may be related to fund flows but not in the regressions. this simplifies regression specifications and reports. 12 our baseline regression model reports a concave relation between fund flow and fund performance which is consistent with ding et al.’s findings in the presence of share restrictions. considering that investors may be risk-averse and that volatile funds have fewer restrictions as evidenced by ding et al., we also control for fund performance volatility. table 3 describes summary statistics of sample monthly raw returns (in %) and fund flows (in %) for the whole sample period and multiple subperiods. summary statistics of each measure include the sample mean, standard deviation (sd), lower quartile (q1), median, and upper quartile (q3). the monthly flow (in %) is defined as the monthly fund flow in dollar amount scaled by the previous month-end asset value. that is, 𝐹𝑙𝑜𝑤𝑖,𝑡 = 𝐴𝑈𝑀𝑖,𝑡−𝐴𝑈𝑀𝑖,𝑡−1×(1+𝑅𝑖,𝑡) 𝐴𝑈𝑀𝑖,𝑡−1 (1) where aumi,t represents the value of assets under management of fund i at month t, and ri,t is the return of fund i during month t. in our flow calculations, we notice that some funds report their aum quarterly or semiannually while reporting multiple zero monthly aums between two positive reporting values. these zero aum values are treated as missing when applying equation (1). in addition, some funds have reported the same aum for two consecutive months, a rare occasion that fund flow during a given month exactly offsets its asset growth in value. attributing such observations to possible stale reporting, we only 11 in unreported tables, we further investigate the sample by breakdowns of different market conditions, and by subperiods according to significant market events. all fund characteristics remain qualitatively the same. 12 in unreported tables, we also add restriction periods, such as redemption period, required advanced notice period for redemptions, as well as other share restrictions related variables to control for share restrictions in the regressions. results in general are qualitatively similar. however, most coefficient estimates in the regressions for these variables are insignificant. 10 include flows calculated from two consecutive aum amounts that are not identical. to ensure the quality of our data analysis employing fund flow variable, we further winsorize fund flows at 5% cutoff point to remove extremely large or small values that might be due to reporting errors.13 table 3: summary statistics of hedge fund monthly returns and flows the table reports summary statistics of hedge fund sample from tass database for the whole period and multiple subperiods from january 1994 to december 2009. in addition to the number of months (in parentheses) for each period, the table reports mean, standard deviation (sd), lower quartile (q1), median, and upper quartile (q3) of monthly fund raw returns (in %) and monthly fund flows (in %), respectively. monthly fund flow is defined as the change of monthly net assets scaled by the previous month-end asset value. multiple subperiods include yearly breakdown, up/down, quiet/volatile markets, and 4 subperiods sorted by combinations of up/down and quiet/volatile market conditions (that is, up-quiet, up-volatile, down-quiet, and downvolatile). up (down) market period is defined as the month when equity market return in excess of 1-month treasury-bill return is positive (negative). quiet (volatile) market is defined as the month when the average daily volatility of stock market in a month is less(more) than average daily volatility of the whole sample period. monthly return (%) monthly flow (%) mean sd q1 median q3 mean sd q1 median q3 whole sample 0.773 0.967 0.341 0.711 1.129 0.054 0.141 0.005 0.030 0.067 by year 1994 0.216 2.111 -0.706 0.272 0.914 0.059 0.191 -0.003 0.014 0.072 1995 1.750 2.478 0.747 1.442 2.483 0.054 0.177 -0.007 0.010 0.055 1996 1.635 2.044 0.826 1.483 2.321 0.068 0.184 -0.003 0.018 0.078 1997 1.548 2.068 0.721 1.368 2.233 0.079 0.213 0.001 0.025 0.094 1998 0.596 2.691 -0.294 0.748 1.760 0.053 0.231 -0.005 0.013 0.057 1999 2.391 3.685 0.559 1.529 3.661 0.054 0.208 -0.007 0.009 0.056 2000 0.824 2.829 -0.065 0.954 1.804 0.054 0.239 -0.006 0.012 0.059 2001 0.625 2.507 0.003 0.640 1.344 0.077 0.396 -0.004 0.014 0.064 2002 0.314 1.745 -0.402 0.347 1.060 0.063 0.263 -0.006 0.010 0.058 2003 1.608 2.176 0.477 1.198 2.287 0.080 0.371 -0.002 0.018 0.072 2004 0.786 1.270 0.219 0.638 1.201 0.072 0.259 -0.001 0.020 0.070 2005 0.849 1.538 0.224 0.644 1.173 0.046 0.249 -0.012 0.007 0.046 2006 1.149 1.509 0.527 0.963 1.489 0.049 0.292 -0.007 0.011 0.050 2007 1.082 2.098 0.240 0.851 1.643 0.043 0.183 -0.006 0.009 0.047 2008 -1.502 3.075 -2.803 -1.138 0.319 0.010 0.163 -0.026 -0.002 0.020 2009 1.432 5.220 0.037 1.008 2.541 -0.006 0.138 -0.030 -0.005 0.011 up market (118) 1.732 1.874 0.698 1.334 2.442 0.054 0.176 0.002 0.026 0.065 down market (74) -0.584 2.269 -1.481 -0.250 0.515 0.052 0.195 -0.001 0.021 0.060 quiet market (148) 1.220 1.286 0.568 1.010 1.633 0.060 0.150 0.005 0.032 0.073 volatile market (44) -0.683 2.921 -1.452 -0.204 0.666 0.040 0.227 -0.011 0.006 0.042 up-quiet (101) 1.843 2.012 0.778 1.441 2.535 0.055 0.143 0.003 0.027 0.068 up-volatile (17) 0.876 4.226 -0.540 0.735 2.320 0.038 0.316 -0.014 0.002 0.036 down-quiet (47) 0.006 2.069 -0.749 0.146 0.750 0.067 0.227 0.001 0.025 0.072 down-volatile (27) -1.505 4.183 -3.044 -0.836 0.561 0.039 0.257 -0.011 0.006 0.042 as shown in the table, hedge fund managers generate an average monthly return of 0.77% (or 9.28% per annum) during the january 1994 to december 2009 period, with a standard deviation of 0.97%. in contrast, the u.s. equity market as a whole delivers an average monthly return of 0.74% (or 8.86% per 13 we also winsorize our data at 10% level and the results remain qualitatively the same. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 11 annum) during the same period, with a much larger standard deviation of 4.66%.14 the comparison suggests that hedge funds perform slightly better than the equity market, but with much smaller variations. when inspected on the yearly basis, performance of individual funds peaked to an average monthly return of 2.39% (or 28.69% per annum) in 1999 with a standard deviation of 3.69%, and plunged to a -1.50% average monthly return (or -18.0% per annum) in 2008 with a standard deviation of 3.08%. on the other hand, the hedge fund industry has an average monthly net growth of 0.05% in assets under management, with standard deviation of 0.14% during the sample period. the monthly flow reached to a high of 0.08% in 2003, and a low of -0.01% in 2009, after the financial crisis stormed the hedge fund industry. to better understand the swinging year-by-year patterns of returns and flows, we take a closer look at the subperiods of our sample. that is, we divide our sample according to the following market conditions: up, down, quiet, and volatile markets. here an up (down) market period is defined as the month when equity market return in excess of 1-month treasury-bill return is positive (negative).15 a quiet (volatile) market period is defined as the month when the average daily equity market return volatility is less (more) than average volatility of the whole sample period.16 summary statistics on the subperiods suggest that individual hedge funds generate much better raw returns when stock market rallies. in particular, hedge funds generate an average of 1.73% monthly return over the 118 up-market months, while producing a negative average return (-0.58%) during the 74 months of down markets. interestingly, over both markets individual hedge funds have attracted very similar levels of average monthly flows, with flows in up markets (0.054%) slightly higher than that in down markets (0.052%). in contrast, quiet and volatile markets present a very similar pattern in returns like up and down markets, but different in fund flows. hedge fund managers on average seem to perform better during quiet markets than volatile markets. over the 148 quiet-market months, an average of 1.22% monthly return is generated by hedge fund managers, much better than the negative average return (-0.68%) over the 44 volatile market months. on the other hand, an average hedge fund attracts more monthly flows in quiet markets (0.06%) than in volatile markets (0.04%). the difference of fund flows between quiet and volatile markets is significant at conventional levels. 14 we use the crsp value-weighted market return, including all dividend distributions, as the proxy of equity market returns. 15 in an alternative setting, we define up (down) market period as the month when equity market return in excess of 1-month treasury-bill return exceeds the previous market excess return by more (less) than 1%. the summary statistics are very similar. 16 the volatile market condition dummy used in our study is defined according to the realized market volatility during the sample period. we also looked at an alternative definition according to the cboe’s vix index value, or implied volatility. our results remain qualitatively the same. 12 to further explore this issue, we sort our sample by both up/down and quiet/volatile market conditions, thus dividing the sample period into the following 4 subperiods: up-quiet (101 months), upvolatile (17 months), down-quiet (47 months), and down-volatile market periods (27 months). as shown in the table, hedge fund managers generate the best raw returns in up-quiet markets (1.84% monthly, or 22.1% per annum), while the worst during down-volatile periods (-1.51% monthly, or -18.1% per annum), a return difference of 3.35% per month (or 40.2% per annum). at the same time, hedge funds receive the highest flows during down-quiet market months (0.067%) and the lowest during up-volatile months (0.038%). the fund flows for up-volatile markets are even lower than down-volatile markets. it is interesting to note that hedge funds actually earn an average return of 0.88% in the up-volatile markets (the second highest return among all market conditions). this indicates that hedge fund investors may care about upside volatility. if upside volatility is not important for investors, we should not see a significant change of fund flows in the table between up-quiet and up-volatile markets or down-quiet and up-volatile markets. our preliminary results demonstrate that hedge fund investors have different behaviors in each market condition. they especially invest less when markets are volatile. such a notable difference in hedge fund returns and flows over different subperiods deserves a more detailed investigation of the flowperformance relationship, which we will actively pursue in the subsequent sections. b. hedge fund performance in addition to raw monthly returns reported by the funds, we measure fund performance using riskadjusted returns from an augmented eight asset-based style factor model based on fung and hsieh (2001, 2004). the risk-adjusted performance of individual hedge funds is evaluated using the 8-factor model as follows: 𝑟𝑖,𝑡 = 𝛼𝑖 + 𝛽𝑖,1 × 𝑃𝑇𝐹𝑆𝐵𝐷𝑡 + 𝛽𝑖,2 × 𝑃𝑇𝐹𝑆𝐹𝑋𝑡+ 𝛽𝑖,3 × 𝑃𝑇𝐹𝑆𝐶𝑂𝑀𝑡+ 𝛽𝑖,4 × 𝑆&𝑃𝑡 + 𝛽𝑖,5 × 𝑆𝑖𝑧𝑒𝑆𝑝𝑟𝑒𝑎𝑑𝑡 +𝛽𝑖,6 × 𝐵𝑜𝑛𝑑𝑀𝑎𝑟𝑘𝑒𝑡𝑡 + 𝛽𝑖,7 × 𝐶𝑟𝑒𝑑𝑖𝑡𝑆𝑝𝑟𝑒𝑎𝑑𝑡 + 𝛽𝑖,8 × 𝑀𝑆𝐶𝐼𝐸𝑀𝑡 + 𝜀𝑖,𝑡 (2) where ri,t is fund i’s return at time t in excess of the risk free rate, ptfsbd, ptfsfx, and ptfscom are trend-following risk factors constructed by fung and hsieh (2001) to measure trend-following strategies in bonds, foreign currencies, and commodities, respectively.17 s&p and sizespread are equity-oriented risk factors, measured by s&p 500 market index monthly return and russell 2000 index monthly return in 17 the trend-following factors are obtained from the link: http://faculty.fuqua.duke.edu/dah7/datalibrary/tf-fac.xls. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 http://faculty.fuqua.duke.edu/ 13 excess of s&p 500 return, respectively. bondmarket and creditspread are bond-oriented risk factors, measured by monthly change in the 10-year treasury constant maturity yield and monthly change in the moody’s baa yield less 10-year treasury constant maturity yield, respectively. msciem is the msci emerging market index monthly total return, and is included into the model given the fact that more than 10% of the funds in our sample are involved with emerging markets investments, thus exposed to the emerging market risk. 𝜀𝑖,𝑡 is the error term, uncorrelated with these eight factors. βi,1 ... βi,8 are risk factor loadings for fund i. as a result, αi represents the risk-adjusted performance for fund i. table 4: risk-adjusted fund performance and market conditions the table reports the cross-sectional average of monthly risk-adjusted returns for the whole sample period and by different market states. risk-adjusted returns (α, in %) are estimated from the following augmented 8-factor model: 𝑟𝑖,𝑡 = 𝛼𝑖 + 𝛽𝑖,1 × 𝑃𝑇𝐹𝑆𝐵𝐷𝑡 + 𝛽𝑖,2 × 𝑃𝑇𝐹𝑆𝐹𝑋𝑡+ 𝛽𝑖,3 × 𝑃𝑇𝐹𝑆𝐶𝑂𝑀𝑡+ 𝛽𝑖,4 × 𝑆&𝑃𝑡 + 𝛽𝑖,5 × 𝑆𝑖𝑧𝑒𝑆𝑝𝑟𝑒𝑎𝑑𝑡 +𝛽𝑖,6 × 𝐵𝑜𝑛𝑑𝑀𝑎𝑟𝑘𝑒𝑡𝑡 + 𝛽𝑖,7 × 𝐶𝑟𝑒𝑑𝑖𝑡𝑆𝑝𝑟𝑒𝑎𝑑𝑡 + 𝛽𝑖,8 × 𝑀𝑆𝐶𝐼𝐸𝑀𝑡 + 𝜀𝑖,𝑡 where rt is fund i’s return at time t in excess of the risk free rate, ptfsbd, ptfsfx, and ptfscom are trendfollowing risk factors constructed by fung and hsieh (2001) to measure trend-following strategies in bonds, foreign currencies, and commodities, respectively. s&p is the s&p 500 market index monthly return. sizespread is the russell 2000 index monthly return in excess of s&p 500 return. bondmarket is the monthly change in the 10-year treasury constant maturity yield. creditspread is the monthly change in the moody’s baa yield less 10-year treasury constant maturity yield. msciem is the msci emerging market index monthly total return. 𝜀𝑖,𝑡 is the error term, uncorrelated with these seven factors. for each market state, mean, median, and standard deviation of fund α are reported, as well as the fraction of funds with significant (5% level) positive and negative α values. see table 3 for definitions of various market conditions. sample period is from january 1994 to december 2009. markets mean t-stat median std α >0 (%) α <0 (%) whole sample 0.43 1.22 0.37 1.17 25.80 2.13 up market 0.48 0.62 0.43 1.70 12.64 5.50 down market 0.48 0.93 0.43 1.25 17.06 4.89 quiet market 0.53 1.43 0.42 1.29 24.56 5.31 volatile market 0.33 -0.24 0.27 1.67 3.93 29.63 up-quiet 0.63 0.60 0.40 2.15 6.72 16.92 up-volatile 0.70 0.77 0.53 2.11 2.09 61.97 down-quiet 0.61 1.06 0.48 1.53 18.03 11.50 down-volatile -0.09 0.25 0.00 2.43 0.24 75.06 table 4 reports the cross-sectional monthly average alphas for the whole sample period and in various market states. as shown in the table, an average hedge fund yields positive monthly alpha of 0.43% (or 5.16% per annum) with median value of 0.37% (or 4.44% per annum). although statistically insignificant, the positive monthly risk-adjusted return generated by the hedge fund industry is in sharp contrast to the mutual fund industry, which generates average alphas very close to zero. over the whole sample period, 25.8% of hedge funds deliver positive alphas that are statistically significant at the 5% level, while only 14 2.1% of funds yield significantly (at 5% level) negative alphas. the evidence from the whole sample period seems consistent with the notion that hedge funds deliver alpha regardless market conditions. interesting observations emerge when fund performance is examined in different market conditions separately. although hedge funds generate almost identical positive alphas of 0.48% (or 5.76% per annum) in both up and down markets, about 17% of funds generate significantly positive alphas in the down market, compared to 12% in the up market. the evidence of identical performance in both markets with more funds generating superior performance and fewer funds generating inferior performance in down markets tends to suggest that hedge funds can offer good protection for investors when the market is sliding. a very different pattern is revealed, however, when the fund performance is investigated in quiet and volatile markets separately. the average alpha of 0.53% (or 6.36% per annum) generated in quiet markets is much higher than the mean alpha of 0.33% (or 3.96% per annum) in volatile markets, a difference of 0.20% per month (or 2.4% per annum). furthermore, the percent of funds that deliver significant positive alphas in quiet markets (24.6%) is much higher than in volatile markets (3.9%), while poorly performing funds in volatile markets (29.6%) substantially outnumber those in quiet markets (5.3%). since volatile markets seem to have a more negative effect on fund performance than down markets, we then turn to a more detailed investigation using combinations of various market conditions. the further breakdown by a combination of up/down and quiet/volatile market conditions indeed offers some more interesting observations. the best performance of hedge funds occurs in up-volatile markets with a mean of 0.70%, followed by up-quiet markets (0.63%), down-quiet markets (0.61%), and down-volatile markets (-0.09%). it is notable that individual hedge funds generate similar positive riskadjusted performance in the first three market conditions, while the worst and the only negative average performance comes from down-volatile markets. in particular, about 75% of funds produce significantly negative alphas in down-volatile markets, while only 0.24% of funds generate significantly positive alphas in the same market. up-volatile markets have the second worst fund performance in our sample period, with 62% of funds generating significantly negative alphas, and only 2.1% generating significantly positive alphas. in contrast, down-quiet markets seem to observe the most reliable hedge fund performance among individual hedge funds, with 18% of funds generating significantly positive alphas, and 11.5% generating significantly negative alphas. our findings tend to suggest that hedge fund managers in general cannot beat the market and offer effective protections when the market is in extreme conditions (such as down-volatile), and that the downside risk protection observed in the down markets mainly comes from down-quiet periods. overall international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 15 results from tables 3 and 4 seem to support roy’s (1952) safety first principle. during down-volatile markets, hedge funds generate unreliable risk-adjusted returns and the fund flows are lower. however, the same observation also applies to up-volatile markets. in the up-volatile markets, hedge funds could not offer reliable performance for their investors and fund investors restrain their investment from hedge funds (the lowest fund flows) even though the average of risk-adjusted returns across all hedge funds in this market condition is the highest among all market conditions. 3. flow sensitivities to past performance summary statistics from the previous section indicate very different patterns for both fund performance and flows under different market conditions. reduced fund flows during upand down-volatile markets are the evidence indicating that volatility in the up and down markets matters for investors. however, it is not clear whether investors perceive the same between upside and downside volatilities while hedge funds receive similar fund flows in both market conditions. it is possible that upside and downside volatilities have different effects on the investors’ behavior. in this section, we examine the relationship between hedge fund flows and past performance in various market conditions using a piecewise linear regression model to offer more details about how fund investors change their investment strategies pertaining to market conditions or they just reduce investment when markets become volatile. previous studies document that fund investors learn about fund management’s ability from past fund performance (e.g., sirri and tufano, 1998; and berk and green, 2004). this is because fund investors cannot directly observe fund managers’ skill, and managers’ ability and investment strategy should not change during a certain time period. thus, fund investors first try to evaluate funds’ management quality from operating performance, consider current situation, estimate possible future outcomes and then, invest their money (lynch and musto, 2003; and berk and green, 2004). how fund investors interpret fund past performance and relate it to the future fund return and what the risk and investment costs could be are essential to the flow-performance relationship. therefore, using the flow-performance sensitivity to investigate fund investors’ behavior offers better understanding about some subtle yet important attitudes that investors have toward overall volatility, and each of upside and downside volatilities than using simple relation between market conditions and fund flows. we investigate the flow-performance sensitivities in a piecewise linear regression model setting similar to sirri and tufano (1998). for a given month t, we first assign fractional performance rank 16 (rankt) from 0 to 1 based on the past performance of individual funds, and then define three performance groups (low, mid, and high) as follows:18 lowt = min(1/3,rankt) midt = min(1/3,rankt −lowt) (3) hight = min(1/3,rankt −lowt −midt) here the fund performance is measured by either average raw returns or risk-adjusted returns using the augmented 8-factor model over past 12 months. fund flow is then regressed against lagged performance ranks to examine flow sensitivities at different levels of performance. since our main interest is on the impact of a particular market condition on the flow-performance relationship, we interact with performance rank on various market condition dummy variables to capture this effect. the general piecewise linear regression model is thus formulated as follows: 𝐹𝑙𝑜𝑤𝑖,𝑡 = 𝑎0 + 𝑏1 × 𝐿𝑜𝑤𝑖,𝑡−1 + 𝑏2 × 𝑀𝑖𝑑𝑖,𝑡−1 + 𝑏3 × 𝐻𝑖𝑔ℎ𝑖,𝑡−1 +𝑏4 × 𝐿𝑜𝑤𝑖,𝑡−1 × 𝑀𝑎𝑟𝑘𝑒𝑡𝐷𝑢𝑚𝑚𝑦𝑡−1 + 𝑏5 × 𝑀𝑖𝑑𝑖,𝑡−1 × 𝑀𝑎𝑟𝑘𝑒𝑡𝐷𝑢𝑚𝑚𝑦𝑡−1 +𝑏6 × 𝐻𝑖𝑔ℎ𝑖,𝑡−1 × 𝑀𝑎𝑟𝑘𝑒𝑡𝐷𝑢𝑚𝑚𝑦𝑡−1 + 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑡−1 + 𝜀𝑖,𝑡 (4) one of the following dummy variables is used to replace the marketdummy variable in equation (4): up, volatile, up-quiet, up-volatile, down-quiet, and down-volatile. for a given month t, the dummy variable up takes value of 1 if the equity market return in excess of 1-month treasury-bill return is positive, and 0 otherwise. the dummy variable volatile takes value of 1 if the average daily equity market return volatility in month t is more than average volatility of the whole sample period, and 0 otherwise.19 compared to the up and volatile dummy variables that describe general market movements and volatility, the latter four dummy variables capture market conditions in a much greater scope. for instance, the dummy variable up-quiet is defined as taking value of 1 if the equity market in month t is both up and quiet (or when up=1 and volatile=0), and 0 otherwise. the estimated coefficients on the interaction terms (b4, b5, b6) thus capture the impact of a specific market condition on the flowperformance relationship when that market condition is present (that is, marketdummy=1), while the estimated coefficients on low, mid, and high (b1, b2, b3) represent the slope of the flow-performance 18 we have also used the original methodology developed by sirri and tufano (1998), and obtained similar results using the following fractional performance rank definitions of low, mid, and high: lowt = min(0.2, rankt); midt = min(0.6, rankt − lowt); and hight = rankt − lowt − midt. 19 we have tried alternative definitions of up and volatile dummies, and results remain qualitatively the same. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 17 relationship over their range of sensitivity when a specific market condition is not present (that is, marketdummy=0). we also include multiple non-performance related fund characteristics as control variables in equation (4). the monthly dummies and style dummies are also included in the regression to control for monthly effect and different fund investment styles respectively. empirical results estimated from equation (4) are reported in table 5, where panel a reports results associated with past raw returns, and panel b with risk-adjusted returns (8-factor alphas). among seven models reported, model 1 (m1) serves as the baseline regression model that groups all market conditions together, while models m2-m7 report the regression results that correspond to individual market conditions described by the six market dummy variables defined above respectively. a. flow and past performance the baseline piecewise regression model m1 in table 5 suggests a concave relationship between hedge fund flow and past performance. hedge fund investors as a group are positively responding to fund past performance, although the level of responsiveness is different at different levels of past performance. this effect is captured by coefficient estimates of different performance rank groups. in particular, estimated coefficients (in %) of (low, mid, high) are (0.026, 0.045, 0.005) in the case of raw returns, and (0.022, 0.023, 0.005) in the case of risk-adjusted returns, respectively. except the group with top risk-adjusted returns, all other groups have statistically significant coefficients at 1% level. this is consistent with goetzmann, ingersoll, and ross (2003) who find a concave flow-performance relation during the 19901995 period and ding et al.(2009) who find a concave relation between fund flow and fund performance when share restrictions are considered. consistent with most existing literature, we find multiple fund characteristics included in the model as controls have a significant impact on future fund flows. investors tend to favor funds with higher management fees, larger amount of fund flows during the past year, proposing high-water mark, and managers’ personal capital invested in the fund. these characteristics are generally perceived as positive signals of good managerial skills and better manager incentives, and therefore attracting more flows into the fund. on the other hand, fund size has a significant negative impact on fund flows. this may indicate investors’ general concern on the diminishing returns to scale in the hedge fund industry, which is in line with the observation by berk and green (2004) in the mutual fund industry. in addition, new fund flows tend to shy away from funds with high past performance volatility and with minimum investment amount requirement. furthermore, u.s.-based funds and proposing lockup period seem to have a marginal 18 positive effect on the flows, while investors are not significantly affected by the fund age and its leverage status. table 5: impact of different market conditions on flow sensitivity to past performance the table reports pooled piecewise ols regression results of fund flow in a given month on lagged fund performance over the past 12 months, their interactions with market condition dummies and lagged fund characteristics. fund performance is measured as either average raw returns (panel a) or risk-adjusted returns using eight-factor model (panel b) over the past 12 months. to estimate piecewise regressions, we assign monthly fractional ranks from 0 to 1 based on their past performance. the bottom group (low) is defined as min(rank,1/3); the middle group (mid) is defined as min(1/3, rank-low) and high group is defined as (rank-low-mid). for a given month, the dummy variable up takes value of 1 if the equity market return in excess of one-month treasury-bill return is positive, and 0 otherwise. the dummy volatile (vol) takes value of 1 for the month when the average daily equity market return volatility is more than average volatility of the whole sample period, and 0 otherwise. the lagged fund characteristics include performance volatility over the past 12 months (volatility), management fees, natural logarithm of assets under management (fund size), natural logarithm of (1+age) (fund age), fund flows over the past 12 months (lagged flow), lockup period (lockup), minimum investment, incentive fees, off shore dummy (offshore), fund manager’s personal investment in the funds (personal capital), high-water mark dummy (watermarked) and leverage dummy (leveraged). style dummies are also included to control for hedge funds’ investment styles. t-statistics reported along with the coefficients are adjusted for heteroskedasticity and autocorrelations. adjusted r2s are in percent. the sample period is from 1994 to 2009. panel a: raw returns ass performance measure m1 m2 m3 m4 m5 m6 m7 coef. t-stat coef. t-stat coef. t-stat coef. t-stat coef. t-stat coef. t-stat coef. t-stat low 0.026 7.81 0.047 11.27 0.032 9.38 0.021 5.26 0.037 10.86 0.019 4.59 0.019 5.69 mid 0.045 21.20 0.043 15.32 0.043 18.00 0.047 18.63 0.043 19.27 0.044 16.41 0.045 20.35 high 0.005 4.87 0.009 5.93 0.007 5.49 0.005 4.16 0.007 6.11 0.003 2.22 0.005 4.02 upxlow -0.045 -8.29 upxmid 0.004 0.94 upxhigh -0.007 -3.54 volxlow -0.034 -9.58 volxmid 0.008 1.65 volxhigh -0.006 -2.32 up-quietxlow 0.014 2.10 up-quietxmid -0.008 -1.84 up-quietxhigh -0.001 -0.29 up-volxlow -0.091 -12.55 up-volxmid 0.015 2.58 up-volxhigh -0.012 -3.83 down-quiet xlow 0.015 2.36 down-quiet xmid 0.001 0.17 down-quiet xhigh 0.005 2.42 down-vol xlow 0.064 8.30 down-vol xmid -0.007 -1.18 down-vol xhigh 0.004 1.18 volatility(%) -0.059 -7.02 -0.057 -6.88 -0.075 -9.21 -0.059 -7.07 -0.064 -7.71 -0.059 -7.11 -0.053 -6.37 mgmt. fees 0.017 4.10 0.017 4.12 0.013 3.06 0.017 4.11 0.018 4.29 0.017 4.13 0.017 4.07 fund size(%) -0.189 -12.59 -0.188 -12.52 -0.209 -13.75 -0.19 -12.60 -0.189 -12.62 -0.189 -12.57 -0.189 -12.55 fund age(%) 0.016 0.28 0.016 0.28 -0.159 -2.78 0.017 0.29 0.02 0.35 0.016 0.28 0.016 0.28 lagged flow 0.489 54.40 0.490 54.53 0.515 57.70 0.489 54.38 0.489 54.48 0.489 54.39 0.491 54.64 lockup 0.002 1.94 0.002 1.90 0.001 1.04 0.002 1.94 0.002 1.90 0.002 1.92 0.002 1.98 min. investment -0.006 -2.24 -0.006 -2.18 -0.007 -2.64 -0.006 -2.24 -0.005 -2.05 -0.006 -2.21 -0.006 -2.26 incentive fees(%) -0.020 -0.37 -0.019 -0.36 -0.034 -0.62 -0.019 -0.36 -0.015 -0.29 -0.019 -0.35 -0.023 -0.42 offshore(%) -0.074 -1.46 -0.073 -1.44 -0.138 -2.69 -0.074 -1.46 -0.074 -1.48 -0.075 -1.49 -0.072 -1.42 personal capital(%) 0.128 2.73 0.129 2.76 0.209 4.45 0.129 2.74 0.126 2.69 0.128 2.73 0.13 2.77 watermarked(%) 0.178 3.32 0.176 3.27 -0.005 -0.09 0.179 3.34 0.181 3.37 0.178 3.31 0.177 3.31 leveraged(%) 0.018 0.37 0.019 0.37 0.034 0.67 0.018 0.36 0.018 0.35 0.018 0.37 0.018 0.35 adjusted r2 13.84 14.01 11.21 13.85 14.19 13.86 14.01 international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 19 table 5 – continued: impact of different market conditions on flow sensitivity to past performance panel b: risk adjusted returns ass performance measure m1 m2 m3 m4 m5 m6 m7 coef. t-stat coef. t-stat coef. t-stat coef. t-stat coef. t-stat coef. t-stat coef. t-stat low 0.022 6.71 0.043 10.09 0.026 6.90 0.021 5.20 0.031 9.08 0.018 4.27 0.016 4.60 mid 0.023 9.21 0.021 6.00 0.022 7.57 0.025 8.08 0.022 7.91 0.023 6.99 0.024 9.09 high 0.005 1.51 0.010 2.27 0.011 2.79 0.004 0.96 0.010 2.83 0.001 0.17 0.004 1.11 upxlow -0.043 -7.74 upxmid 0.006 1.13 upxhigh -0.011 -1.76 volxlow -0.017 -2.27 volxmid 0.006 1.06 volxhigh -0.019 -2.69 up-quietxlow 0.004 0.55 up-quietxmid -0.005 -1.00 up-quietxhigh 0.003 0.53 up-volxlow -0.076 -10.23 up-volxmid 0.016 2.16 up-volxhigh -0.028 -3.12 down-quiet xlow 0.009 1.37 down-quiet xmid 0.000 0.05 down-quiet xhigh 0.011 1.77 down-vol xlow 0.069 8.81 down-vol xmid -0.009 -1.14 down-vol xhigh -0.002 -0.16 volatility(%) -0.016 -2.57 -0.015 -2.40 -0.020 -3.17 -0.016 -2.55 -0.022 -3.49 -0.017 -2.79 -0.010 -1.53 mgmt. fees 0.020 4.78 0.020 4.79 0.020 4.86 0.020 4.78 0.021 4.99 0.020 4.81 0.019 4.66 fund size(%) -0.171 -11.39 -0.170 -11.33 -0.172 -11.44 -0.171 -11.39 -0.172 -11.42 -0.171 -11.39 -0.170 -11.30 fund age(%) 0.038 0.64 0.036 0.61 0.037 0.63 0.038 0.64 0.036 0.61 0.038 0.64 0.036 0.61 lagged flow 0.526 58.72 0.527 58.85 0.526 58.68 0.526 58.72 0.526 58.86 0.526 58.72 0.527 58.92 lockup 0.003 2.43 0.003 2.42 0.003 2.43 0.003 2.43 0.003 2.40 0.003 2.42 0.003 2.46 min. investment -0.004 -1.66 -0.004 -1.65 -0.004 -1.62 -0.004 -1.67 -0.004 -1.56 -0.004 -1.65 -0.004 -1.68 incentive fees(%) -0.012 -0.23 -0.013 -0.23 -0.011 -0.21 -0.012 -0.23 -0.010 -0.18 -0.012 -0.22 -0.016 -0.30 offshore(%) -0.100 -1.97 -0.101 -1.98 -0.103 -2.02 -0.100 -1.97 -0.104 -2.04 -0.101 -1.99 -0.099 -1.95 personal capital(%) 0.145 3.07 0.147 3.11 0.144 3.05 0.145 3.07 0.145 3.07 0.145 3.06 0.146 3.10 watermarked(%) 0.209 3.87 0.207 3.83 0.210 3.88 0.209 3.87 0.207 3.82 0.209 3.86 0.210 3.88 leveraged(%) -0.016 -0.31 -0.016 -0.31 -0.017 -0.33 -0.016 -0.32 -0.016 -0.32 -0.016 -0.31 -0.014 -0.28 adjusted r2 12.66 12.82 12.68 12.66 12.95 12.68 12.84 b. flow sensitivities under different market conditions m2 and m3 of table 5 investigate how a general market condition such as up/down or quiet/volatile affects the fund flow-performance sensitivities. results indicate a strong and asymmetric impact of different markets on hedge fund flow-performance sensitivities. m2 tests whether up markets have a significantly different impact on flow sensitivities to past performance from down markets. market dummy up is thus introduced into the model and interact with different performance ranks to capture the effect. we find that among three estimated coefficients of interaction terms, up × low and up × high are significantly negative in both panels, indicating significantly reduced flow-sensitivities during the up markets for bottomand top-tier performing groups. in particular, compared to down markets (up dummy=0), up markets (up dummy=1) show that flow 20 sensitivities to bottom performance are significantly reduced by about 96%, from 0.047 to 0.002 in the case of raw returns (and by about 100%, from 0.043 to 0 in the case of risk-adjusted returns). similarly, average flow sensitivities to top performance during up markets drop by about 78%, from 0.009 to 0.002 when measured by raw returns (or by 100% from 0.010 to -0.001 when measured by risk-adjusted returns). as a result, estimated coefficients (in %) of (low, mid, high) change from (0.047, 0.043, 0.009) in down markets (up dummy=0) to (0.002, 0.047, 0.002) in up markets when fund performance is measured by raw returns, or from (0.043, 0.021, 0.010) in down markets to (0.000, 0.027, -0.001) in up markets when measured by alphas. in other words, a typical concave flow-performance relationship is only observed during down markets, while flows turn completely insensitive to past losers and winners during up markets for hedge funds in the high and low performance ranges. at the same time, up and down market conditions do not affect flow sensitivities to funds with medium performance, which stay significantly positive all times. in m3, the market dummy volatile replaces the up dummy in m2 and interacts with performance ranks to test whether volatile and quiet market conditions affect the flow-performance sensitivities differently. an asymmetric flow-performance relationship is again observed when market is divided into quiet and volatile conditions. similar to up/down markets, flow-performance sensitivities are not much affected by specific market conditions for medium performance funds. funds in both bottom and top performing groups, however, have much flatter flow sensitivities during volatile markets (volatile dummy=1), while quiet markets (volatile dummy=0) are associated with a generally concave flowperformance relationship. in particular, for (low, mid, high) performance ranges, flow sensitivities are significantly altered from (0.032, 0.043, 0.007) in quiet markets to (-0.002, 0.051, 0.001) in volatile markets when measured by raw returns, and from (0.026, 0.022, 0.011) in quiet markets to (0.009, 0.028, -0.008) in volatile markets when measured by risk-adjusted returns. results from m2 and m3 tend to suggest that the general concave flow-performance relationship observed in the baseline model (m1) is largely driven by either down or quiet market conditions. more importantly, our findings reflect significant changes of investor sentiment toward hedge funds with extreme performance under different market conditions. during volatile markets, it seems intuitive that hedge fund investors become more cautious and refrain from their investment activities when there is significant uncertainty in predicting the future market movements, hence reducing the sensitivities to funds with extreme performance. on the other hand, it is interesting to note that during down markets hedge fund investors appear to be more sensitive to past fund performance than they do during up markets. while both volatile and down markets are usually considered undesirable market conditions, our evidence suggests very different patterns of flow sensitivity in the two markets. in light of this observation, we international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 21 further break down the market conditions and investigate the impact of a specific market condition on flow-performance pattern in a greater detail. results from m4-m7 confirm a generally strong but asymmetric impact of a particular market condition on flow sensitivities toward funds with extreme performance. both performance measures show similar patterns, with slightly weaker statistical significance in the case of risk-adjusted returns as the proxy for fund performance. except up-volatile markets that significantly affect flow sensitivities to mid-range performance funds, all other market conditions usually have strong impacts, if any, on low and high performance groups only. more specifically, up-volatile markets are associated with significantly reduced flow sensitivities for both bottom and top performance groups, while all other markets are with increased sensitivities. as a result, net flow sensitivities to low performance are 0.035, -0.054, 0.034, and 0.083 when measured by raw returns (or 0.025, -0.045, 0.027, and 0.085 when measured by risk-adjusted returns) for up-quiet, up-volatile, down-quiet, and down-volatile markets separately. the net flow sensitivities to high performance funds are 0.004, -0.005, 0.008, and 0.009 (or 0.007, -0.018, 0.012, and 0.002 when measured by risk-adjusted returns) for up-quiet, up-volatile, down-quiet, and down-volatile markets respectively. for low performance fund groups, it is worthwhile to note that among the four market conditions, up-volatile markets witness the least flow sensitivity to fund performance with coefficient estimate of the interaction term up-volatile × low of -0.091 (t-stat=-12.55) in the case of raw returns, and -0.076 (t stat=-10.23) in the case of 8-factor alphas. this suggests an overall negative flow sensitivity to the low performance funds during up-volatile markets (-0.054 in the raw return case, and -0.045 in the riskadjusted return case), indicating that past losers even attract more new funds than past winners. on the other hand, the flow-performance sensitivity is most significant in down-volatile markets, indicating that investors invest more past winners in the low performing fund group. for high performance fund groups, however, flow sensitivities are most positively affected during down-quiet markets and most negatively affected during up-volatile markets. when fund performance is measured by risk-adjusted returns, the “chasing past winners” behavior of fund investors is strongest during down-quiet markets, with estimated coefficient of the down-quiet × high of 0.011 (t-stat = 1.77), resulting a net flow sensitivity to high performance funds of 0.012. the most negative impact toward flow-performance relationship is detected during up-volatile markets, with coefficient of interaction term of -0.028 (t−stat= -3.12), resulting a net flow sensitivity of -0.018. on the other hand, up-quiet and down-volatile markets seem to have little influence on the flow sensitivities toward past winners. 22 our overall results indicate changes of investment behaviors from investors between up and down markets, and between quiet and volatile markets are mainly driven by the up-volatile and down-volatile market conditions. this observation is clearer when using risk-adjusted returns as the performance measure. if markets are quiet, hedge fund investors have strong incentive to chase fund performance and flow-performance relation in general remains similar between up and down markets. therefore, when markets are relatively calm and predictable, market direction has almost no impact on the investment behaviors of hedge fund investors. however, investors become less sensitive to past performance during up-volatile markets, but become highly sensitive to past performance during down-volatile markets. thus, both upside and downside volatilities matter for hedge fund investors. combining results from summary statistics, we find that hedge fund investors not only reduce investment during both upand down-volatile markets but also demonstrate different behaviors under these two markets. c. robustness check we also apply stricter rules to adjust sample funds to mitigate issues related to the times in which investment decisions are taken with the market conditions prevailing at those times, the match of the frequency and length of the hedge fund investing process, and a short sample for up-volatile times. it is arguable that dealing frequencies and share restrictions may affect our findings. moreover, there is a relatively shorter period of 17 months in the up-volatile market. in this study, we assume that hedge fund investors are smart. thus, they have already incorporated all these restrictions into the consideration when they make their investment decisions. as ferson and kim (2012) document, investors may face some frictions when they make their investment decisions even though fund flows can predict future economic conditions. it is of interest to explore how these frictions could alter our findings. thus, we repeat our tests using a refined sample that is supposed to have fewer problems with these issues. our new sample includes hedge funds with (1) both redemption and subscription frequencies are one month or below (daily, monthly, fortnightly, semi-monthly and weekly); (2) the notice period is between none and one month (30 days); (3) no other share restrictions are applied. we also divide months into halves or terciles according to their monthly market volatility and define the top half or tercile as volatile markets. the numbers of months for up-volatile and down-volatile markets are now 43 and 45, respectively, when grouping months in halves, and 32 and 26, respectively, when using the months in the top tercile as volatile markets.20 20 the results are not sensitive to grouping methods for months. the complete tables are available upon request. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 23 results are generally similar to the ones in table 5. hedge fund investors have different behaviors between up-volatile and down-volatile markets. they also demonstrate different investment strategies between quiet and volatile markets although the change of investment behaviors is marginal significant (t-stat= -1.78). however, we notice that there are some changes. the flow-performance relation in the model 1 for the best performing funds becomes significant while the shape is still concave. we further find that the changes of investment behaviors mainly occur in the fund group with the worst performance hedge fund investors become insensitive to fund performance in the up-volatile markets and are more sensitive to fund performance in the down-volatile markets. when comparing new results with table 5, restrictions with hedge fund investing seem to strengthen the behavior patterns in table 5. although share restrictions may affect the investment decisions, assessing the overall net impact of share restrictions on investment behaviors is beyond the scope of this study. our purpose in this paper is to show that upside and downside volatilities both affect investment behaviors and to explore whether upand down-volatile markets have different impacts on the investors’ investment strategies. 4. smart money under different market conditions given the asymmetric flow-performance sensitivities observed under different market conditions from previous subsection, it is of interest to examine whether hedge fund investors have an ability to predict future fund performance and how this ability varies with market conditions. if hedge fund investors have the ability to predict future fund performance, their change of investment behaviors found in the previous section should offer some meaningful implications. therefore, it is worthwhile to investigate whether fund investors can benefit from doing so, or whether the money is “smart” in the hedge fund industry. we follow the standard approach in the literature to test the smart money effect. for each given month t, we first sort all funds according to their past 3-month flows into quintiles, and denote them from the highest (group 1) to the lowest (group 5). we then track the average value-weighted portfolio returns over 3-month and 12-month period within each flow quintile respectively, and test the statistical significance for returns. the return spreads between the highest and lowest flow quintile portfolios are also examined. the tests are performed for the whole sample period, four market conditions (up, down, quiet, and volatile), and the further market condition breakdowns (up-quiet, up-volatile, down-quiet, and down-volatile markets). table 6 reports the test results, with panel a reporting the flow performance for a 3-month period, and panel b for a 12-month period. for brevity, the performance measures reported in both panels are the augmented 8-factor alphas, although very similar results are obtained when using buyand-hold raw returns.21 21 results using buy-and-hold returns are available upon request. 24 table 6: smart money performance of hedge fund flows the table reports average risk-adjusted returns (α) on all hedge funds within each flow quintile for a period of three months (panel a) or 12 months (panel b), and performance difference between top and bottom quintiles (h-l). at the beginning of each month, hedge funds are sorted into five fund flow quintiles, according to the past three-month fund net flow. quintile 1 (high) consists of hedge funds with the highest fund flow. all months are further divided into subperiods according to the market conditions. market conditions include up, down, quiet, volatile markets and their combinations. see table 2 for definitions of various market conditions. risk-adjusted returns are evaluated by using the augmented eight-factor model described in table 4. the sample period is from 1994 to 2009. flow rank α t-stat α t-stat α t-stat α t-stat panel a: performance in three months overall 1-high 0.588 7.46 2 0.499 6.23 3 0.505 6.08 4 0.462 5.39 5-low 0.366 4.45 h-l 0.222 3.73 up down quiet volatile 1-high 0.628 7.59 0.575 7.07 0.589 7.33 0.758 4.36 2 0.537 6.37 0.508 6.02 0.510 6.00 0.624 3.50 3 0.555 6.32 0.487 5.71 0.521 5.85 0.417 2.83 4 0.483 5.29 0.434 4.93 0.463 5.16 0.516 3.09 5-low 0.400 4.47 0.327 3.92 0.347 4.06 0.591 3.53 h-l 0.228 3.44 0.248 3.79 0.242 3.71 0.167 1.30 up-quiet up-volatile down-quiet down-volatile 1-high 0.572 6.53 0.760 3.77 0.624 7.51 0.637 2.44 2 0.495 5.62 0.567 2.88 0.525 6.00 0.570 2.24 3 0.548 5.76 0.327 2.03 0.541 5.84 0.318 1.38 4 0.442 4.72 0.473 2.47 0.450 4.84 0.438 1.95 5-low 0.317 3.61 0.465 2.32 0.339 3.85 0.657 2.99 h-l 0.255 3.53 0.295 1.97 0.285 4.10 -0.019 -0.09 panel b: performance in twelve months overall 1-high 0.558 7.35 2 0.470 5.96 3 0.477 6.51 4 0.442 5.57 5-low 0.399 4.98 h-l 0.158 3.13 up down quiet volatile 1-high 0.586 7.65 0.557 7.29 0.572 7.48 0.576 4.69 2 0.479 5.90 0.473 5.94 0.470 5.96 0.400 3.11 3 0.478 6.56 0.476 6.36 0.479 6.53 0.386 3.68 4 0.437 5.57 0.433 5.29 0.425 5.33 0.430 3.85 5-low 0.422 5.16 0.396 4.91 0.385 4.77 0.481 4.08 h-l 0.164 3.19 0.161 3.02 0.187 3.59 0.095 1.14 up-quiet up-volatile down-quiet down-volatile 1-high 0.573 7.15 0.523 4.08 0.575 7.67 0.522 3.54 2 0.477 5.83 0.429 2.92 0.478 6.07 0.377 2.61 3 0.475 6.47 0.434 3.53 0.487 6.5 0.325 2.81 4 0.422 5.41 0.397 2.98 0.421 5.08 0.415 3.45 5-low 0.407 4.96 0.455 3.32 0.384 4.71 0.466 3.76 h-l 0.166 3.09 0.069 0.78 0.191 3.57 0.056 0.53 panel a indicates the existence of a strong smart money effect among hedge fund investors over a 3-month period. using the whole sample period (overall), we find a generally decreasing flow performance along with the descending flow quintile order, with the highest flow quintile significantly outperforming the lowest flow quintile. when market conditions are divided into either up/down or international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 25 quiet/volatile cases, the performance pattern of each flow quintile in up, down, and quiet markets is very similar to the overall pattern. it is interesting, however, to note that the smart money effect is much weaker or even disappearing during volatile markets with a typical u-shaped pattern in performance. that is, fund flow performance is first decreasing and then increasing with the flow quintile rank after medium group, causing a much smaller (and insignificant) performance gap between the top and bottom flow quintiles. when market conditions are further divided into combinations of up/down and quite/volatile markets, it is worth noting that down-quiet markets witness the strongest smart money effect over the 3-month window (0.29% per month or 3.42 per annum, with t−stat of 4.10 at the 1% significance level). the ushaped pattern of fund performance across flow quintiles during volatile markets is particularly magnified in down-volatile markets. the bottom flow quintile even slightly outperforms the top quintile, resulting in a negative but insignificant performance difference between two groups. the up-volatile markets present a similar but less extreme u-shaped pattern, resulting in a slightly weaker yet still statistically significant smart money effect (with top-bottom performance difference of 0.295% and t−stat = 1.97). this observation suggests that during volatile markets, especially down-volatile markets, future fund performance becomes more uncertain and harder to predict, thus introducing a lot of noise into investors’ investment decisions. as a result, funds that experienced the highest level of flows do not necessarily outperform those with the lowest level of flows subsequently. panel b of table 6 tracks flow performance over a longer period (12 months), and the smart money effect over the whole sample period is still strong and significant. compared to the findings in panel a, all flow quintiles except the bottom group have lower returns, and that performance of each quintile tends to converge. the bottom flow quintile has slightly improved performance, resulting in a narrowed but still significant difference between the top and bottom quintiles. the breakdowns of market conditions also show patterns very similar to their counterparts in panel a, with smaller difference between the top and bottom flow quintiles. again, the strongest smart money effect is detected in down-quiet markets (0.19% per month or 2.29% per annum, with t−stat of 3.57 at the 1% significance level), and volatile markets still stand out as a special case with a unique u-shaped pattern of flow performance compared to other market conditions. during turbulent market periods, funds with the highest flows perform better than those with the lowest flows over the 12-month horizon, but the difference is insignificant, indicating no smart money effect for new investments made. such a pattern is further confirmed in both up-volatile and downvolatile markets, where the smart money effect completely disappears over the 12-month window. 26 instead, the bottom flow quintile generates higher returns than all the middle three quintiles in all volatile markets. therefore, hedge fund flows cannot predict future fund performance when market is turbulent and volatile, while the observed smart money effect in general is largely driven by quiet market conditions where fund performance is more stable and predictable. in summary, table 6 suggests that various market conditions play a significant role in investors’ investment decisions and affect subsequent performance of new fund flows asymmetrically. in particular, volatile markets have a much weaker or disappearing smart money effect compared to stable and quiet markets, while up and down markets have very similar patterns of smart money effect, which is also comparable to the overall case. we further observe that first, hedge fund investors have the best ability to predict future fund performance in down-quiet markets and have the worst ability in down-volatile markets; and second, investors have the second best ability in the up-quiet markets and the second worst ability in the up-volatile markets. thus, upside volatility also reduces hedge fund investors’ ability to pick outperforming funds. since fund investors usually use fund past performance to judge fund managers’ skill, it is interesting to examine whether past performance cannot be used to predict future fund performance during volatile markets as busse (1999) suggests. if this is the case, we will find that fund performance does not persist in volatile markets. in the next section, we continue to investigate performance persistence for hedge funds under various market conditions. 5. performance persistence given a smart money effect detected in the overall market but with a very different pattern during volatile markets, it is interesting to investigate whether this effect is associated with the fund performance persistence over the same windows. performance persistence is well documented in the hedge fund literature. for example, agarwal and naik (2000) have documented maximum fund persistence at a quarterly horizon. bares, gibson, and gyger (2003) also detect fund performance persistence over shortterm holding periods such as one to three months, but rapidly vanishes with longer holding periods. however, none of the studies have examined the performance persistence under a given market condition. in this section, we investigate whether fund performance persists under different market conditions, and if so, whether it contributes to the observed smart money effect in table 6. a pooled regression approach is employed to examine the hedge fund performance persistence. namely, we use the following regression model to perform the analysis: 𝛼𝑖,𝑡+𝑛 = 𝑎0 + 𝑏1 × 𝛼𝑖,(𝑡−12,𝑡−1) + 𝑏2 × 𝐹𝑙𝑜𝑤𝑖,(𝑡−12,𝑡−1) + 𝑏3 × 𝑀𝑎𝑟𝑘𝑒𝑡𝐷𝑢𝑚𝑚𝑦𝑖 × 𝛼𝑖,(𝑡−12,𝑡−1) +𝑏4 × 𝑀𝑎𝑟𝑘𝑒𝑡𝐷𝑢𝑚𝑚𝑦𝑖 × 𝐹𝑙𝑜𝑤𝑖,(𝑡−12,𝑡−1) + 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠(𝑡−12,𝑡−1) + 𝜀𝑖,𝑡 (5) international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 27 here αi,t+n is fund i’s performance over the subsequent n (n =3 or 12) months. αi,(t−12,t−1) is lagged fund performance during the past 12 months. flowi,(t−12,t−1) is the average monthly fund flow during the past year. for brevity, the fund performance is evaluated using the risk-adjusted return estimated from the augmented 8-factor model.22 to estimate the risk-adjusted return α, factor loadings of the 8-factor model are first estimated using a 24-month window of (t-11, t+12) for any given month t. with parameter estimates, the one-month α is then calculated as the difference between the realized excess return and the expected return estimated from the model. past fund performance (αi,(t-12,t-1)) is measured as the average monthly risk-adjusted return from the 8-factor model over the past 12 months. other lagged fund characteristics included in the model as control variables are volatility of fund performance, management fees, incentive fees, fund size, fund age, and dummy variables to control for the fund’s lockup period proposition, minimum investment amount requirement, high-water mark protection, leverage, offshore operation, and investment of manager’s personal capital in the fund. in addition, 12 style dummies are included to control for different investment styles. table 7 provides results for fund performance persistence tests, with panel a reporting results for a 3month test, and panel b for a 12-month test, separately. similar to table 5, model 1 (m1) serves as the baseline regression model that groups all market conditions together without including any market dummy variables, while models m2-m7 report regression results that correspond to individual market conditions respectively. the reported t-statistics are adjusted for the heteroskedasticity and autocorrelations. for brevity, we only report the coefficients on the main variables of interest. as shown in m1, the baseline model that does not include any market dummies suggests a clear pattern of strong persistence in hedge fund performance over both three and twelve months past fund performance has a significant positive impact on future performance over these two periods even after controlling for various hedge fund characteristics. past fund flows, as one of control variables, however, has a significant negative impact on subsequent fund performance. thus, net fund inflows disrupt fund mangers’ ability to provide consistent fund performance and this negative effect can even outweigh the benefit of fund performance persistence. m2 investigates whether up and down markets have a significant impact on fund performance persistence by interacting with the up market dummy on both past fund performance and lagged fund flows respectively. both panels report negative but insignificant coefficient estimates of interaction terms up × α(t-12,t-1) and up × flow(t-12,t-1). results suggest a negligible additional impact of up or down market on fund 22 we also use the raw returns as the fund performance measure, and results are qualitatively the same. 28 table 7: performance persistence of hedge funds under different market conditions the table reports the pooled ols regression results of fund performance over the subsequent 3 months (panel a) and 12 months (panel b) on lagged 12-month fund performance and fund flow, their interactions with market condition dummies and lagged fund characteristics. fund performance is measured as risk-adjusted returns (estimated abnormal return). to calculate the estimated risk-adjusted return, the parameters of the eight-factor model are estimated first using window of (t-11, t+12). with the parameter estimates, the one-month abnormal return is calculated as the difference between real excess return and expected return from the model. past fund performance (α(t−12,t−1)) is measured as the average monthly risk-adjusted returns from the eight-factor model over the past 12 months. see table 2 for detailed information of market conditions. in addition to lagged fund performance and flows, we also include the following lagged fund characteristics as control variables (not reported): fund performance volatility over the past 12 months, management fees, incentive fees, assets under management, fund age, and dummies that indicate the fund’s lockup period proposition, minimum investment requirement, offshore operation, manager’s personal investment in the fund, high-water mark protection, and whether the fund is leveraged. style dummies are also included to control for hedge funds’ investment styles. t-statistics are reported along with the coefficients after adjusted for heteroskedasticity and autocorrelations. adjusted r2s are in percent. the sample period is from 1994 to 2009. m1 m2 m3 m4 m5 m6 m7 coef. t-stat coef. t-stat coef. t-stat coef. t-stat coef. t-stat coef. t-stat coef. t-stat panel a: performance persistence in three months α(t−12,t−1) 0.220 26.93 0.222 20.22 0.248 26.82 0.202 20.45 0.235 26.97 0.204 19.26 0.229 26.68 flow(t-12,t-1) -1.154 -2.51 -0.871 -1.42 -0.944 -1.88 -0.924 -1.92 -1.264 -2.24 -1.191 -2.50 -1.364 -2.25 up x α(t−12,t−1) -0.005 -0.32 vol x α(t−12,t−1) -0.119 -6.42 up-quiet x α(t−12,t−1) 0.053 3.22 up-vol x α(t−12,t−1) -0.113 -4.91 down-quiet x α(t−12,t−1) 0.038 2.42 down-vol x α(t−12,t−1) -0.090 -3.46 up x lagflow -0.601 -0.70 vol x lagflow -1.213 -1.08 up-quiet x lagflow 0.265 0.30 up-vol x lagflow -2.318 -1.63 down-quiet x lagflow 0.477 0.55 down-vol x lagflow 0.468 0.29 control variables yes yes yes yes yes yes yes adjusted r2 7.24 7.23 7.36 7.32 7.26 7.26 7.25 panel b: performance persistence in twelve months α(t−12,t−1) 0.176 32.19 0.184 25.05 0.204 32.99 0.161 24.36 0.193 33.01 0.156 22.07 0.184 31.99 flow(t-12,t-1) -0.945 -3.07 -0.762 -1.86 -1.053 -3.14 -0.981 -3.04 -0.758 -2.00 -1.005 -3.15 -1.015 -2.50 up x α(t−12,t−1) -0.018 -1.70 vol x α(t−12,t−1) -0.120 -9.68 up-quiet x α(t−12,t−1) 0.044 4.00 up-vol x α(t−12,t−1) -0.125 -8.09 down-quiet x α(t−12,t−1) 0.046 4.40 down-vol x α(t−12,t−1) -0.078 -4.45 up x lagflow -0.384 -0.66 vol x lagflow 0.559 0.74 up-quiet x lagflow -0.538 -0.90 up-vol x lagflow 0.291 0.31 down-quiet x lagflow 0.173 0.30 down-vol x lagflow 0.765 0.71 control variables yes yes yes yes yes yes yes adjusted r2 9.84 9.85 10.10 10.02 9.88 9.89 9.89 international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 29 persistence in 3and 12-month test windows. however, a very different pattern emerges when volatile market dummy replaces up dummy and interacts with fund past performance and lagged flows respectively (m3). while the coefficient estimate of interaction term volatile × flow(t-12,t-1) stays insignificant, the coefficient of volatile × α(t-12,t-1) is significantly negative in both panels. this indicates that the level of fund performance persistence is significantly reduced during volatile markets (volatile dummy=1) compared to quiet markets (volatile dummy=0). results are consistent with our early conjecture that fund managers usually perform better when market is relatively calm and stable, and with predictable future movements, which will be reflected on the persistence of fund performance in subsequent months. in contrast, managers have very limited capability in generating consistently superior returns during volatile markets as market swings vibrantly without clear directions, giving a hard time for managers to produce persistent performance. results from m2 and m3 largely match the observations in the “smart money” test in the previous section. while investors generally possess good ability in selecting successful funds, the flow performance in up markets is not significantly different from that in down markets. on the other hand, flow performance is much better during quiet markets than during volatile markets. combining the fund performance persistence test with smart money test, we can see that the flow performance is largely driven by the fund performance persistence. thus, the manager talent and relevant investment strategies could carry forward from the past during quiet markets for one year (lynch and musto, 2003; and berk and green, 2004). even though investors can be successful in picking good funds during volatile market periods, the significantly lower performance persistence across the hedge fund industry eliminates the “smart money” effect commonly observed in other market conditions. to further differentiate the impact of a specific market condition on the fund performance persistence, we follow a similar approach in the previous sections and include one of the following four market dummy variables in the model to interact with lagged fund performance as well as lagged flow respectively: up-quiet, up-volatile, down-quiet, and down-volatile. models m4 m7 in table 7 report test results. distinct patterns of performance persistence emerge after the further breakdown of market conditions. in all four models and both testing windows, coefficients of the interaction terms between market dummies and past performance are significant. in particular, during up-quiet and down-quiet markets fund performance persistence is much stronger than other market conditions. in contrast, during upvolatile and down-volatile markets performance persistence is significantly reduced. this observation is consistent with our findings in m2 and m3, where market volatility has a negative impact on performance persistence but there is no difference between up and down markets. therefore, volatile markets are the 30 primary market condition that affects the fund performance and their persistence in a negative way, while quiet markets are where investors see better fund performance and much stronger performance persistence. comparing to the smart money tests in the four market breakdowns in the previous table, results from model m4 m7 offer consistent evidence as well. due to significantly reduced performance persistence during up-volatile and down-volatile markets, the smart money effect is much weaker or even disappearing in these markets. the observation implies that if hedge fund investors tend to weigh more on past performance to make investment decisions during volatile markets, they are not better off by doing so. in contrast, the smart money effect is much stronger when performance is more persistent during upquiet and down-quiet markets. thus, the smart money effect largely depends on the performance persistence in a specific market condition. in sum, we provide evidence to show that market conditions matter for hedge fund performance and fund performance persistence. during volatile markets (both up-volatile and down-volatile markets), past fund performance cannot be used to predict future fund performance and most hedge funds also underperform. this is consistent with busse’s (1999) finding in mutual fund literature. combining tests in this section and in the previous section, our results imply that down-volatile markets may have a greater negative effect than up-volatile markets. however, upside volatility still hurts fund investors’ ability to pick funds, reduces fund managers’ ability to provide better performance, and influences fund investors to change their investment strategies. our overall results demonstrate that both downside and upside volatilities affect investment behaviors. therefore, upside volatility should not be just neglected. upside and downside volatilities may be considered separately when developing portfolio theories. 6. conclusion in this paper we explore how market conditions affect hedge fund performance and fund investors’ behavior to shed light on whether investors only respond to downside volatility. specifically, we focus on how market conditions affect relations between current fund flows and past fund performance, between current fund flows and future fund performance, and between past fund performance and future fund performance. as previous literature documents, fund investors learn about fund managers’ unobserved ability by using fund past performance to make their investment decisions. therefore, how investors interpret fund performance and how investors relate this learning to future fund performance are essential to the flowperformance relationship and the change of investment behaviors reflect investors’ attitude toward some international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 31 unobserved factors. market risk or uncertainty is one of the most important factors when investors form their portfolio holdings. recent study suggests that incorporating predictability based on macroeconomic variables can significantly improve hedge fund investor performance. among all macroeconomic variables, market volatility is one of important factors for evaluating hedge fund performance. thus, market conditions should have different effects on the fund performance and the level of risk. the varying fund performances and risk will cause investors to show different behaviors. we classify market conditions as up, down, quiet, volatile, and combinations of up-quiet, up-volatile, down-quiet, and down-volatile markets. using a sample of 5,254 individual hedge funds from january 1994 to december 2009, we find that fund flows and fund performance in the hedge fund industry vary with market conditions. market conditions play an important role on the fund performance and fund investors’ behavior. we further find that hedge fund investors chase past performance in a highly asymmetric fashion under different market conditions. they invest more in past winners in down-quiet markets but avoid investing in the past winners in the best and worst performance fund groups in the upvolatile markets. we also document the existence of “smart money” effect over both 3and 12-month periods under all market conditions except volatile markets (both upand down-volatile markets). we further provide evidence to show that the observed smart money effect is largely associated with fund performance persistence. fund performance is strongly persistent during quiet markets, leading to a significant smart money effect. in contrast, volatile markets significantly reduce the level of persistence in both 3and 12-month investment horizons, which contribute to a marginal to disappearing smart money effect. our overall results indicate that hedge fund investors have different behaviors between quiet and volatile markets, between each of upand down-volatile markets and quiet markets, and between upand down-volatile markets. although downside volatility may have a stronger negative impact on the fund performance, fund performance persistence and smart money effect, both upside and downside volatilities do provide a drag on investors’ ability to pick funds and fund managers’ ability to generate better returns. our results provide evidence to show that market conditions matter to hedge fund investors. ___________________________________________________________ authors information: chou: college of business administration, florida international university, miami, fl 33199; wchou@fiu.edu; (305)348-0533. ke: department of graduate management studies, nathan weiss graduate college, kean university, union, nj 07083; dke@kean.edu; (908)737-5967. xu: school of business administration, gonzaga university, spokane, wa 99258 and hanken school of economics, helsinki, finland; xu@jepson.gonzaga.edu; (509) 313-7030. we thank timo korkeamaki, qinghai wang and seminar participants at 2010 northeast business and economics association meetings, 2011 financial management association meetings, florida international university and hanken school of economics for their helpful comments and suggestions. all remaining errors are our own. 32 references agarwal, vikas and narayan y. naik, (2000). multi-period performance persistence analysis of hedge funds, journal of financial and quantitative analysis 35, 327-342. agarwal, vikas, naveen d. daniel, and narayan y. naik, (2009). role of managerial incentives and discretion in hedge fund performance, journal of finance 64, 2221-2256. ang, andrew, joseph chen, and yuhang xing, (2006). downside risk, review of financial studies 19, 1191-1239. aragon, george o., bing liang, and hyuna park, (2008). onshore and offshore hedge funds: are they twins? working paper. avramova, doron, robert kosowski, narayan y. naik, and melvyn teo, (2011). hedge funds, managerial skill, and macroeconomic variables, journal of financial economics 99, 672-692. baba, naohiko, and hiromichi goko, (2009). hedge fund survival: non-normal returns, capital outflows, and liquidity, journal of financial research 32, 71-93. baquero, guillermo, and marno verbeek, (2009). a portrait of hedge fund investors: flows, performance and smart money, working paper, erasmus university. bares, pierre-antoine, rajna gibson, and sebastien gyger, (2003). performance in the hedge funds industry: an analysis of short and long-term persistence, journal of alternative investments 6, 25-41. berk, jonathan b., and richard c. green, (2004). mutual fund flows and performance in rational markets, journal of political economy 112, 1269-1295. busse, jeffrey a., (1999). volatility timing in mutual funds: evidence from daily returns, review of financial studies 12, 1009-1041. capocci, daniel, albert corhay, and georges hübner, (2005). hedge fund performance and persistence in bull and bear markets, european journal of finance 11, 361-392. chen, yong, and bing liang, (2007). do market timing hedge funds time the market?, journal of financial and quantitative analysis 42, 827-856. ding, bill, mila getmansky, bing liang, and russ wermers, (2009). share restrictions and investor flows in the hedge fund industry, working paper. docking, diane scott, and paul d. koch, (2005). sensitivity of investor reaction to market direction and volatility: dividend change announcements, journal of financial research 28, 21-40. ederington, louis, evgenia golubeva, (2009). evidence on investor behavior from aggregate stock mutual fund flows, working paper, university of oklahoma. edwards, franklin, and mustafa onur caglayan, (2001). hedge fund and commodity fund investment styles in bull and bear markets, journal of portfolio management 27, 97-108. ennis, richard m., and michael d. sebastian, (2003). a critical look at the case for hedge funds, journal of portfolio management 29, 103-112. ferson, wayne e., and min s. kim, (2012). the factor structure of mutual fund flows, international journal of portfolio analysis and management 1, 112-143. frazzini, andrea, and owen lamont, (2008). dumb money: mutual fund flows and the cross-section of stock returns, journal of financial economics 88, 299-322. fung, william, and david a. hsieh, (1999). a primer on hedge funds, journal of empirical finance 6, 309-331. fung, william, and david a. hsieh, (2000). performance characteristics of hedge funds and commodity funds: natural vs. spurious biases, journal of financial and quantitative analysis 35, 291-307. fung, william, and david a hsieh, (2001). the risk in hedge fund strategies: theory and evidence from trend followers, review of financial studies 14, 313-341. fung, william, and david a hsieh, (2004). hedge fund benchmarks: a risk based approach, financial analyst journal, 60, 65-80. goetzmann, william n., jonathan e. ingersoll, and stephen a. ross, (2003). high-water marks and hedge fund management contracts, journal of finance 58, 1685-1718. international journal of banking and finance, vol. 10, iss. 1 [2013], art. 2 33 gruber, martin j., (1996). another puzzle: the growth in actively managed mutual funds, journal of finance 51, 783-810. huang, jennifer, kelsey d. wei, and hong yan, (2007). participation costs and the sensitivity of fund flows to past performance, journal of finance 62, 1273-1311. ivkovi ć, zoran, and scott weisbenner, (2009). individual investor mutual fund flows, journal of financial economics 92, 223-237. kaplan, paul d., and laurence b. siegel, (1994a). portfolio theory is alive and well, journal of investing 3, 18-23. kaplan, paul d., and laurence b. siegel, (1994b). portfolio theory is alive and well, journal of investing 3, 45-46. kelly, morgan, (1997). do noise traders influence stock prices?, journal of money, credit, and banking 29, 351-363. keswani, aneel, and david stolin, (2008). dollar-weighted returns to stock investors: a new look at the evidence, working paper, city university london. luo, dengpan, (2003). market volatility and mutual fund cash flows, yale icf working paper. lynch, anthony w., and david k. musto, (2003). how investors interpret past fund returns, journal of finance 58, 2033-2058. nawrocki, david, (1999). a brief history of downside risk measures, journal of investing 3, 9-25. ozik, gideon and ronnie sadka, (2009). smart money or smart about money? evidence from hedge funds, working paper, boston college. rom, brian m., and kathleen w. ferguson, (1993). post-modern portfolio theory comes of age, journal of investing 3, 11-17. rom, brian m., and kathleen w. ferguson, (1994). portfolio theory is alive and well: a response, journal of investing 3, 24-44. roy, arthur d., (1952). safety first and the holding of assets, econometrica 20, 431-449. sapp, travis, and ashish tiwari, (2004). does stock return momentum explain the “smart money” effect?, journal of finance 59, 2605-2622. savor, pavel, and mungo wilson, (2012). how much do investors care about macroeconomic risk? evidence from scheduled economic announcements, journal of financial and quantitative analysis, forthcoming. sirri, erik r., and peter tufano, (1998). costly search and mutual fund flows, journal of finance 53, 1589-1622. wang, ashley, and lu zheng, (2008). aggregate hedge fund flows and asset returns, working paper. zheng, lu, (1999). is money smart? a study of mutual fund investors’ fund selection ability, journal of finance 54, 901-933. the dynamics of credit quality and implications for the pricing of small business loans the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 31 the dynamics of credit quality and implications for the pricing of small business loans adrian m. cowan and charles d. cowan st. mary’s university, san antonio and the university of alabama, birmingham abstract we analyze the implications of the dynamics of credit scores for small businesses; strategies for banks to maximize revenues; and basel ii minimum capital requirements on loan pricing for loans to small firms that do not have access to capital markets. relating dynamic changes in the competitive environment to pricing decisions also provides a contribution to the literature. a theoretical model is developed to investigate the differences between relationship and transactional lending to small businesses in the context of these factors in the banking industry. the model demonstrates that in highly competitive markets, each type of lender occupies overlapping spaces, and can be simultaneously attractive to different types of borrowers if banks take advantage of their knowledge of the dynamics. keywords: credit scoring, loan pricing, small business lending, basel ii jel classification: g21, d81, l14 1. introduction banks that compete for small business loans follow two basic operational strategies: relationship and transactional lending. boot and thakor (2000) had argued that the degree to which a bank will engage in relationship lending relative to transaction lending is a function of technology, competition, and regulation, among other factors. however, their argument had only considered a single time period. analysis of a longer “relationship” shows that the degree to which banks can and will engage in relationship lending is broader than first thought. the dynamics of small business credit worthiness also has implications for the impact on the small business loan markets. we consider how the dynamics impact the competitive loan pricing of small business loans. such themes we consider are how these dynamics may impact loan pricing, whether small businesses strategically target banks to obtain the best value possible, or vice versa, the analyzing the behavior of banks targeting specific groups of borrowers. the goal of our analyses is to investigate these questions and the implications of the dynamics of credit ijbf 32 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 scores for both the banks and their small business customers. a model that captures the complexity of credit scoring on small business loan pricing when considering the dynamics surrounding the market and the changing characteristics of borrowers will be introduced as well, further justifying our claims. section 3 discusses the competitive environment in which banks must compete with one another, while in section 4, we develop our basic model with corresponding results of this model in section 5 for individual points in time. we present our extended results in sections 6 and 7 where relationship bankers are shown to be able to compete effectively at several levels with each other and with transactional lenders. section 8 provides the empirical and policy implications of our model, while section 9 concludes our work. 2. the competitive environment for small business loans 2.1 credit scoring small business credit scoring (sbcs) has been in use in lending for more than a decade. the implementation of sbcs differs between banks, yet many banks use some form of credit score, whether purchased from a third party or computed from the borrower’s data. the recognition that business owner characteristics outperform business characteristics as predictors for very small commercial loan performance supported the application of credit scores to small firms with minimal or no credit history (see mester, 1997). additionally, third party vendors of sbcs, which began in 1995 with fair isaac’s small business scoring service, further extended the availability of this technology for banks with limited resources for analysis. as banks extract a signal from a credit score, the informational asymmetry between banks as described by sharpe (1990) is greatly reduced. despite sbcs, informational asymmetries persist between borrowers and lenders (diamond, 1984). during the lending process, banks may integrate credit scores in several ways. whereas some banks use the credit score exclusively in the lending decision, others adopt a mixture of credit scoring with relationship lending. other banks have not integrated the use of sbcs at all and continue to use only relationship lending with a more traditional underwriting approach. in addition, the approach may differ within the same bank by loan size, business sector, or region. alternative implementation strategies are found in the empirical literature: akhavein, frame, and white (2005); frame, srinivasan, and woosley (2001); berger and udell (2002); cole, goldberg, and white (2004); cowan and cowan (2006). the different approaches introduce a complicated framework with the analyses focusing upon a single time period. blöchlinger and leippold (2006) had presented a model of loan pricing within the context of internal credit rating systems. to the best of our knowledge, this is the only theoretical paper that examines credit scoring in the context of loan pricing. the focus of that paper was on the development of internal ratings models and the relationship between the lending decision and the quality of that model. 2.2 basel ii during the progression of sbcs development, the basel committee has been updating capital requirements on small and medium size enterprise loans to more the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 33 closely reflect the risk inherent in these loans. basel ii provides several alternative approaches to the measurement of capital requirements based on credit risk. the three approaches for credit risk involve the standardized approach and two separate internal ratings based (irb) approaches (i.e., foundation irb and advanced irb). the two irb approaches differ primarily by the expansion of parameters estimated by banks in the advanced approach. at a minimum under the foundation irb, banks estimate the probability of default. there has been a good deal of discussion in the industry regarding the competitive advantage offered to those banks that are able to adopt the irb, and in particular, a-irb. it is expected that the adoption of the irb approach of basel ii will lead to a reduction in the minimum capital requirement for small business lending: see repullo and suarez (2004) and berger (2004). repullo and suarez (2004) had presented a theoretical model predicting that loan pricing is impacted through the reduction of the capital requirements with the irb approach when moving from basel i to basel ii. their model suggests that low risk, small businesses benefit by obtaining loans from banks adopting the irb approach, while high-risk firms would benefit by obtaining loans from banks that do not adopt this advanced approach. however, their model also predicts that banks will maintain the minimum capital requirement without holding excess capital. this is important because the degree to which banks adjust their levels of capital impacts the total cost of capital. any competitive advantage achieved by banks adopting the a-irb approach depends critically on the amount capital ratios fall relative to the reduction in required capital, as noted by berger (2004). a theoretical model developed by allen, caaretti, and marquez (2006) suggests that minimum capital requirements are not binding constraints. 3. relationship versus transaction-based lending small businesses in the united states typically depend on financial institutions for their funding: i.e., petersen and rajan (1994) and berger and udell (1995). the importance of the bank is particularly significant for the types of small businesses that are the focus of this paper, such as those small businesses with no access to the capital markets. the literature identifies transaction and relationship-based lending as the two basic approaches that banks select for purposes of lending to small businesses. boot and thakor (2000) had argued that the degree to which banks engage in relationship lending relative to transaction lending is a function of technology, competition, and regulation, among other factors. we further the discussion by investigating the dynamics of lending to small businesses, as well as the impact of changes in regulation on the competitive pricing of relational and transactional loans. 3.1 information asymmetries informational asymmetries provide the basis for several theories of financial intermediation, whereby banks are at a competitive advantage to other lenders in the development of information about borrowers as in diamond (1984, 1991) and 34 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 ramakrishnan and thakor (1984). this information asymmetry is used to further motivate the development of models with regard to relationship lending and loan pricing. sharpe (1990) had highlighted that informational asymmetry arises between banks. in contrast with the current paper, he had assumed that the borrower quality remains constant. we allow the signal to change over time in the lending process as a function of the outcomes from earlier time periods. our analytical construct to do this is similar to diamond (1991), but differs significantly in the issues addressed. he had focused on borrower choices and classes of credit that encompass many borrowers, whereas we aim to focus on bank alternatives, providing individual credit scores for each firm. 3.2 loan pricing there have been numerous contributions regarding loan pricing for small business loans within the context of relationship lending. some models predict that the loan interest rate will decline over the duration of the banking relationship. in an examination of the long-term role of collateral, boot and thakor (1994) had presented a model of optimal contracting, where subsequent to borrower project success, the firm enjoys an unsecured, below market rate in every subsequent period. however, an interesting aspect of this model is that the results do not depend on learning. petersen and rajan (1995) had analyzed loan pricing for small businesses in competitive markets relative to concentrated markets where banks have market power. they had found that market power allows banks to engage in the intertemporal smoothing of rates charged to firms. as a result, although their model predicts that interest rates decline as firms age while the uncertainty regarding the firm is resolved, this decline is faster in competitive markets than in concentrated markets. these models and results are grounded in contracting theory. other models predict that interest rates increase over time. within these models, banks subsidize borrowers in the early years only to extract information rents in later years. accordingly, sharpe (1990) had presented a theory of informational asymmetry suggesting that banks exhibit an ex-post monopoly power, despite the fact that they are ex-ante competitive. banks can use informational advantages to extract rents. greenbaum, kanatas, and venezia (1989) had offered a unique model that is motivated by the bank’s need to predict loan demand for funding purposes rather than on the creditworthiness of the borrower. the conflicting results likely reflect the difficulty with incorporating the complexities of the lender borrower relationship in a theoretical model. the credit quality of the borrower changes over the duration of the relationship, which may lead to changes in loan rates charged to small business customers. our model predicts that rates will increase or decrease as a function of both bank costs (monitoring) and the credit quality of the borrower. 3.3 loan value rajan (1992) had offered a model where the borrower is influenced to continue by the bank; because of anticipated returns, the borrower exerts more or less effort in the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 35 achieving success in the project examined. our paper differs significantly from rajan (1992) in that we assume the quality of the borrower drives the pricing of the loan, since transactional lenders cannot anticipate the outcome of a specific project. additionally, rather than assuming differential or optimal levels of effort for the borrower, we examine levels of investment that the bank can make in the relationship, leading to improved probabilities of success for the firm. we also assume that there is an added value to relationship lending that is achieved through monitoring. we use the boot and thakor (2000) definition of a relationship loan as one that permits the bank to use its expertise in improving the borrower’s project payoff. monitoring in our model encapsulates all aspects of relationship development and maintenance. whereas previous models incorporate a cost for monitoring, they fail to incorporate a value for monitoring. we include a variable that is representative of the highly correlated nature between the likelihood of success in the borrower’s project vis-a-vis to the likelihood of loan repayment for the bank. according to diamond (1996), the value of monitoring to the bank is the expected savings in financial distress costs. the bank can use the threat of liquidation to obtain as much as possible in repayment from the borrower. this value can be added through a change in firm behavior (reduced moral hazard) or the informal feedback mechanism provided by a banker with expertise in the industry. the level of monitoring can be optimized to maximize the bank’s return, trading off additional costs in originating the loan relative to increasing the anticipated return. in a recent survey conducted on behalf of the small business administration, cowan and cowan (2006) had found that a significant number of banks that lend to small businesses continue to rely exclusively on traditional relationship lending. there is an anecdotal evidence available in the industry that some banks that have gone to transactional lending have reverted to relational lending for small business, both to reduce risk and to increase returns to the institution. 3.4 bank competition boot and thakor (2000) had presented a model of bank competition, which serves as a suggestive framework for relationship lending in light of increased bank competition within the capital markets. we build on the work of boot and thakor (2000) in that we assume banks are not limited in their lending strategy. however, our paper differs from boot and thakor (2000) in several important ways. we do not examine the impact of competition from the capital markets. as noted above, small businesses do not have access to the capital markets, and thus, banks competing for this line of business compete exclusively with other financial intermediaries. we also allow for relationship and transaction aspects to influence the same borrower, whereas boot and thakor (2000) had assumed that an individual loan is either solely transaction or relationship based. when competition is limited to inter-bank competition in their model, it leads to the increased importance of relationship lending to banks, but with a reduction in the added value of each loan. because we integrate the quality of the borrower into our model, we find that the value of the relationship to both the bank and to the borrower increases inversely 36 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 with the quality of the borrower. there seemingly is no incentive for relationship lending to higher credit quality borrowers, although this is only true for one time period. over multiple time periods, there can be a value of relationship lending to most higher credit quality borrowers. it would only be those borrowers with extremely high credit quality who would not be a target for relationship bankers. thus, we provide an important, new result contrary to what has been published thus far in the literature by incorporating changes in the quality of the borrower over time. we propose a model in which banks start with specific information about firms. over any given period of time, banks tend to “learn” as information changes or becomes readily available, while the quality of firms fluctuates over time as reward or punishment for good or bad decisions. a key result of our research is that there is a role for relationship banking even in the absence of informational asymmetries between competitors. 3.5 basel ii we also extend the literature by incorporating relationship and transaction based lending aspects when analyzing the potential impact of basel ii. consistent with repullo and suarez (2004), we assume that banks take advantage of potential capital reductions. however, we do not require that banks hold minimum capital, but simply that they take advantage of reductions available from basel ii relative to their current levels of capital. we go beyond repullo and suarez (2004) by extending beyond the limits of two risk categories to multiple categories of risk that is consistent with practice. however, the focus of repullo and suarez (2004) with that of this paper are quite different. this paper focuses on the impact of dynamic changes in credit quality in which implications for basel ii are secondary, but important. in contrast, repullo and suarez (2004) had focused on loan pricing changes in the limited context of the impact of the changes from basel i to basel ii. therefore, the impact of regulatory change was the primary focus of that paper. 4. the model 4.1 general framework the participants in the model are commercial banks, borrowers, and depositors. commercial banks invest in illiquid, small business loans. we follow boot and thakor’s (2000) definition of a relationship loan whereby the bank has the potential to improve the borrower’s project payoff. a choice parameter designates how banks elect the degree of investment in relationship lending. whereas boot and thakor (2000) had allowed banks to participate in the project payoff through an investment in specialization, we simply allow banks to invest in monitoring without making it more specific to the choice of industries. our model improves the likelihood of repayment, but the project payoff goes solely to the small business. there is a limited amount of historical information available for these firms that makes underwriting difficult. the firms of course, are constrained in raising capital due to size limitation. we exclude from consideration those high growth the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 37 firms that seek equity financing, such as venture capital for early business phases, rather focusing on those loans obtained from financial institutions. in this paper, we examine the differential impact of lending strategies based on firm quality; the benefits of monitoring; and the differing costs among banks associated with obtaining information. there is a very significant difference in our results from those of earlier authors. with small businesses, the bank is learning about the quality of the borrower rather than assessing only the quality of the project. earlier papers hypothesize a quality rating for a “project”. in transactional lending, the lender only obtains a quality score for the borrower without a detailed investigation of the project. the relational lender for a small business loan similarly finds out about the quality of the borrower and may not examine the quality of the project as closely at the beginning of the project. 4.2 model details consider an economy in which firms are endowed with a project and seek external financing from a bank. a firm shops multiple banks in the initial time period, both transactional and relational. relationship loans are those in which the bank can affect the project payoff of the borrower. in contrast, transaction lending is a pure funding transaction in which the project payoff is unaffected by the bank’s involvement. two types of financial intermediaries can facilitate the financing of the projects: those banks that use credit scoring only and those that use relationship banking only. many banks use both approaches and can decide to pursue a transactional or relationship strategy on a firm-by-firm basis. for these banks, the bank chooses only the strategy that benefits the bank. from the perspective of the borrower, the borrower sees only one rate and arrangement offered. the loan is either transactional or relational, and will differ in implementation because of different monitoring requirements. all banks are perfectly competitive ex-ante; thus, we ignore intertemporal discounting. for small businesses, banks have only two methods to evaluate the quality of a firm: a credit score or underwriting. banks observe the credit quality measure prior to making a loan. relationship lenders collect information on a firm in a traditional manner for the purposes of underwriting the loan. in time period 0, a bank starts by collecting information on a firm and on the principals in the firm. the bank calculates an internal score for a firm and uses this information to make the lending decision. relationship lending involves a monitoring cost that declines over the length of the relationship. transactional lenders use credit scores to make lending decisions. with credit scoring, there are two possibilities for rating the quality of a business. one is for a bank to use the score purchased from a credit bureau. banks with limited resources may not have the ability to customize the information they receive, but still may benefit from the use of a credit score that is indicative of a small business. another possibility is to combine the credit score with the internal score as part of a customized scoring system tuned to the particular population of firms that the bank considers as its customer base. 38 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 due to the nature of credit scoring, all transactional lenders purchase exactly the same credit score for a particular firm. when determining the quality of the borrower, underwriters use the same information, so the relational score computed by a relational bank is assumed to be the same as the credit score. this is somewhat of a restrictive assumption, as it does not allow for noise in the system (to relax this assumption, we would have to model how noise in the system introduces a new source of risk for the banks, both systemic and random). this is beyond the scope of the current paper, but raises interesting questions for the future. 4.3 information observed over time we assume that during time period t, the quality of a firm, f, is observed as θ f,t . the quality measure is a value with a range [0,1], characteristic of an individual firm. for all firms, the distribution of quality scores is f(θ). when considering the initial quality of a firm, a small business does not spring into life. rather, it is the culmination of preparatory work and operations funded through private sources. time zero is the first time that a small business seeks outside funding. how well a firm has prepared until this time determines the quality measure for the firm at time zero, θ f,0 . a firm’s quality changes according to its performance over time, depending upon the success or failure of projects. a successful project builds up the firm; an unsuccessful one weakens the firm. a series of unsuccessful projects weakens the firm to the point where it can no longer borrow and must go out of business. the quality of a firm at time t is a function of its quality in time t-1 and the +/change as the firm signals success or failure. in general, the firm’s quality changes as: θ f,t+1 = h(θ f,t ,+) (1) the only requirement on h(o) is that θ f,t+1 > θ f,t if the project is a success, θ f,t+1 < θ f,t if the project is a failure, and h(o) is monotonic and invertible. we assume that an outcome at t is independent of outcomes in previous time periods. quality measures from third parties or computed by banks are assumed to range along a fixed range that can be transformed linearly to the [0,1] range, preserving the relative variability and shape of the distribution of predictions. transformed, these quality measures become comparable scores sold by a credit bureau or created by a bank that engages in relational lending. 4.4 costs of obtaining information costs associated with each method of obtaining a score differ. for transactional lenders, bureau-based scores are relatively inexpensive, with some minor fixed costs associated with collecting information for purposes of due diligence. there are no ongoing costs such as monitoring. this means that the bank abandons having an impact on the success of a project and that the cost of obtaining information about a firm recurs every time the firm applies for a loan. this cost is c t = t for all t, where the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 39 t is the cost of generating an account that includes the cost of obtaining the credit score. for a relational bank, there is a similar cost of initial information plus a cost of monitoring the relationship. the greater there is an investment in monitoring, the greater the likelihood of a repayment. over time, as the lender gathers more information and more knowledge about a firm, the monitoring costs decline asymptotically to a lower bound. if the relationship is interrupted, the cost of obtaining initial information recurs. the cost of establishing a relationship to collect information in determining the viability of a project with an ongoing cost of monitoring is c r,t = m t . the value m t declines over time, ∂m t /∂t<0, but the decline lessens in each subsequent time period so that ∂2m t /∂t2<0. a function of the initial expenditure m 0 is m t where relational costs are greater than transactional costs for earlier time periods, but can be less than transactional costs at later time periods if the borrower stays with the same lender for a sufficiently longer period of time. in summary, c t = t and c r,t = m t (2) 4.5 probabilities of success and recovery the quality score is an indicator of repayment. with a suitable transformation, any credit score can be transformed to be on the range [0,1]. without loss of generality, we use the quality score as a probability of repayment. information regarding a business’s successes and how likely it is to repay is captured in the credit score that a bank purchases from a credit bureau. since this is what a credit bureau provides, why would a bank take on the greater cost of developing and maintaining a relationship with a small business? the answer would appear to be in the tradeoff between the costs of monitoring versus the improved likelihood of repayment. the likelihood of repayment can be improved through monitoring in relational loans. as had boot and thakor (2000), we too had modified the likelihood of repayment by increasing the likelihood of project success. with monitoring, the likelihood of repayment becomes p f,t = θ f,t + γ[m 0 ] * (1-θ f,t ) (3) where p f,t is the likelihood of success of the project (and repayment of the loan), and γ[m 0 ] is a factor, denoting the proportion of the likelihood of failure that will be converted to an additional likelihood of success, as a result of monitoring. the proportion of increase in equation (2) is denoted as γ[m 0 ]. this factor is a function only of m 0 , the amount spent by the lender on monitoring in the first period when the loan is initiated. economies of scale realized, m t declines uniformly, but γ[m 0 ] remains constant even though spending goes down. if this were not the case, then the value of monitoring would decline over time, even though the 40 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 lender is learning more about the borrower in each subsequent period. we require that 0<γ[m 0 ]<1, and diminishing returns as an expenditure increases, or ∂γ/∂m 0 >0, ∂2γ/∂m 0 2<0. for the remainder of this paper, the terms γ[m 0 ] and γ are synonymous with one another. 4.6 pricing of loans boot and thakor (1994) have a pricing model for collateralized lending, but they present a pricing model for business loans with no recovery if the loan defaults. sharpe (1990) has a similar model with no provision for collateral, and the probability of repayment is a function of the success of the project, not of the firm. many small business loans are collateralized or guaranteed by owners. berger and udell (1996) had found that small business loans administered by large banking organizations tend to have collateral pledged less often than banks with assets less than $1 billion. we develop a pricing model that incorporates a recovery if the loan defaults. we assume that the borrower takes a $l loan to invest for one time period. if the project is successful, the project yields $l*ψ to the borrower. the probability of repayment is a function of the quality of the borrower and the degree of monitoring of the project. the bank’s zero excess profit loan rate is established by setting the bank’s gross cost of funds to the gross expected return on the loan, as shown in the following: (1+r m )l+c j,t = p f,t *(1+r c,t )l+(1-p f,t )*β*r (4) where r m = the cost of funds, r c,t = the rate charged by lender type c at time t, l = the amount loaned, β = the recovery rate for the collateral r, p f,t = θ f,t + γ[m 0 ] * (1-θ f,t ) = the probability of repayment, and = for transactional lenders. the rate that would be charged to the borrower from loan source j at time t is then: (5) this rate has three components: the cost of funds rate; the per dollar cost of originating & monitoring the loan; and the loss if collateral recovery doesn’t fully cover the loan amount. these components relate to the final rate charged to the borrower in several ways. the rate to the borrower increases if either the cost of funds to the bank increases or the cost of originating and monitoring the loan increases. conversely, the rate to the borrower decreases if the recovery rate on the collateral increases or the likelihood of the project success increases. note that 0]m[ 0 ≡γ ( ) tf, tf, tj, m tc, p l r 1p-1 l c r r ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ β−++ = the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 41 r t,k for transactional lenders is the same for all k for fixed credit quality, so for the remainder of this description r t,k = r t for all k. similarly, r c,k is for relational lenders. since the r t,k are all identified as constant and subscripted by “t”, for the remainder of this description, r c,k = r k will be used. the numeric subscript will always be for relational lenders, while the letter subscript “t” will always be for transactional lenders. we will now investigate the differences between transactional and relational lenders that involve initial lending to a firm, lending to a firm after the initial relationship is established, and strategies for profit maximization by the banks. 5. basic findings and relationships at each point in time this paper focuses on dynamic change. we need to derive some basic results for each time period. using these relationships, we can then examine long-term versus short-term decision-making. remember that relationship and transactional lenders observe the same measure of credit quality. the relationship bank offers a different rate than the transaction bank because the costs of lending differ between the two; resultant being that the relational bank offers a greater likelihood of success to the firm (services offered by the relational lender improve the probability of a borrower’s success, so it offers a loan rate commensurate with the improved likelihood of success). in this section, we summarize relationships and show that it is possible for both the borrower and the relational lender to maximize returns. 5.1 borrower’s choice the borrower will only see the two rates (relational and transactional) being offered by the banks. no borrower will be willing to borrow if the rate being charged is greater than the rate of return that the borrower expects to achieve, denoted as ψ. assuming both loan rates are less than ψ, the borrower chooses the loan with the greater expected payoff. the expected payoff to the borrower is the likelihood of success multiplied by the net return: for a relational loan (6) for a transactional loan (7) the net return to the borrower is greater for a relational loan than for a transactional loan if improvement in the likelihood of success is sufficiently enhanced. even though the borrower is paying more for the loan, the net return will be greater if the probability of success sufficiently increases. this is true for all borrowers with credit scores less than θu,t: l * )r(1-)(1 * 1*]m[ tr,tf,0tf, l * )r(1-)(1 * ttf, 42 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 figure 1: minimum acceptable quality score for the borrower who expects to achieve at least an 8% rate of return and quality score breakeven for choice between relational and transactional lender the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 43 (8) a borrower with a credit score less than θ u,t will choose the relational rate over the transactional rate, even if the relational rate is higher because the relation improves the likelihood of a positive net return (see figure 1). this calculation is repeated for each time period for declining monitoring costs. there will be a sequence of values, θ u,0 <θ u,1 <θ u,2 <… where later values of θ u,t are increasing because costs are declining. a similar sequence arises for borrowers if they do not expect to achieve a rate of return ψ greater than the value of the loan. for a transactional loan, this means that ψ must be greater than r t,t . for a relational loan, ψ must be greater than r r,t , which occurs at (9) for first-time borrowers with credit quality below θ l,0 cannot afford to take on the loan since the project return won’t cover the expense of the loan. as stated above, there will be a sequence of values, θ l,0 <θ l,1 <θ l,2 <… where later values of θ l,0 are decreasing because costs are declining. if the rate on the loan exceeds the expected rate of return, no borrower will accept a loan, meaning that there is a minimum quality score needed for borrowers to accept a loan that they can afford. similarly, for a single point in time, there is a unique breakeven point where the value of a loan is the same for the borrower who can choose between a transactional and relational lender. 5.2 profit maximization for relational bankers attracting new borrowers the relational banker has an opportunity to maximize his return, both by improving the probability of success to his borrowers and also by taking business away from the transactional banker. the increased investment in monitoring leads to improving the likelihood of success with diminishing returns as the expenditure increases. however, interest rates offered by the relational banker also increase as the cost of monitoring increases. there is a gain at the upper and lower end of the range where the relational banker has an advantage as the improvement in the probability of repayment improves, but there is also a loss at each end as the interest rate increases. the greatest return for relational lenders comes from balancing gains and losses in market share with expected returns for individual borrowers. we define γ 0 as the expected return for the relational banker from the first loans made. we integrate over the range where borrowers will accept a relational loan. for successful projects, the return to the bank is 1+r r,0 . for projects that do not succeed, the bank recovers some fraction (β) of the collateral as a proportion of the loan value. ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ + ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ++⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ++ = γ −= l r β-1ψ l r β-1 l t rl r β-1 l m r j where, ]m[ j 1θ m t m t 0 t tu, ( ) ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ +ψ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ +ψ++ = l r βγ-1 l r βγl r β l m r θ t m tl, 44 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 (10) where θ u,0 = was defined as the highest quality score for which a borrower would accept a relational loan, above θ u,0 = the borrower would take a transactional loan, θ l,0 = the lowest quality score for which a borrower would accept a relational loan: note: if in other words, the gain offered by the bank has to be greater than the increase in cost to the borrower relative to his net return when borrowing from a transactional lender. the relational banker can control the level of investment made, m 0 , in monitoring the loan. the expected return can be maximized by taking the derivative of γ with respect to the initial investment. for a single time period, we have the following: (11) with setting the above equation (11) equal to zero and solving for m 0 gives us the investment that the relational bank can make while maximizing its return. unless the density at θ l,0 is significantly less than the density at θ u,0 , we would expect that the second term in (11) to be negative; thus, we can find a positive m 0 that satisfies the required conditions, but only for one period. 6. excess profits from retained borrowers in this section and the next, we present three key ways in which a relational lender can realize excess or additional profits. in some instances, these gains come from adding borrowers that the lender might not have otherwise taken on. in other ( ) ( ){ }( ) { } dθ l r βθ-1γ)1(r1θ-1γθθf r,0 θ θ 0 u,0 l,0 ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ −+++=γ ∫ l r β l t l m 1 -1θ t tu, +ψ = γ ( ) ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ +ψ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ +ψ++ = l r βγ-1 l r βγl r β l m r θ t m tl, tl,tu, θθ > ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ +ψ − > l t rl t l m γ m t ( ) ( )[ ] ( ) ( ) ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ∂ ∂ ∂ ∂ ⎭ ⎬ ⎫ ⎩ ⎨ ⎧ +++= ∂ ∂ 0 l,0 l,0 0 u,0 u,0 0 ml,0u,0 0 0 m θ θfm θ θf l m r1θf-θf l 1 m γ ( ) l r βγ-1 l γ-1 l m r m γ m θ and l r βγ l θ l t l m m γ m θ 2 0 m 0 0 l,0 2 0 0 0 u,0 ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ +ψ +⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ψ+ ∂ ∂ = ∂ ∂ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ +ψ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ∂ ∂ = ∂ ∂ the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 45 instances, these gains come from being able to charge more for services because of a temporary, dominant position. having considered how long a borrower will stay with a lender, we now consider how the relational lender can increase his profits through this longer-term relationship. in section 5 we considered only those normal profits that would be obtained by optimizing an investment in one time period. for a constant return per dollar loaned, the more loans made the better if they could be taken away from the transactional lenders. but these profits reflect only the transfer of profits that would occur from a transactional to relational lender. once a relational lender has a relationship, it is also possible to accrue monopoly profits over time. 6.1 acquired and retained borrowers a transactional borrower who fails on a project may be better off with a relational loan in a subsequent time period. this is easily seen in figure 2 where the borrower who fails on their first project has a decline in their credit score (the failure uses resources which leads to a decline in borrower quality). thus, some borrowers who were transactional borrowers will become relational in subsequent time periods. conversely, a successful relational borrower may have a sufficient increase in quality to find that he would have been better off with a transactional loan, but will stay with the relational lender if the cost reduction for the transactional loan is less than the decline in cost in the relational loan. this is seen in figure 3, where the borrower’s quality increase would lead them to consider a transactional loan, but the decline in the cost of a second relational loan (because the monitoring costs decline) leads the borrower to stay with the relationship. if the borrower fails in the first project, at a future time period, the quality of the borrower is diminished. some borrowers who had a sufficiently high quality to obtain a low rate, transactional loan might suffer a decline of quality after a failure that would require them to seek a relational loan because the relational loan may now have a relative pricing advantage. if the borrower succeeds on the first project, at the next time period, the quality of the borrower is improved. some borrowers who had obtained a relational loan will now have a better quality. however, the relational cost for monitoring will decline. as a result, for these borrowers, the relational loan will still be more advantageous. from equation 1, we have θ f,t+1 =h(θ f,t ,+). dropping the notation for the firm, “f”, we have simply that for a one period time change that there is an increase (decrease) in the quality score if the firm succeeds (fails). in time period 1, we have θ 1 =h(θ 0 ); conversely, we can invert the relationship to find what the score would have been when the borrower first applies for the loan if we know the current quality score: h-1(θ 1 )=θ 0 . proposition one: banks making relational loans will capture borrowers who had transactional loans whose credit quality declines, and will retain some of the borrowers who would have improved to a transactional loan because of declining monitoring costs. 46 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 figure 2: declining quality, transactional borrowers who move to relational borrowing 0.00 0.02 0.04 0.06 0.08 0.10 0.5 0.6 0.7 0.8 0.9 1.0 quality score q u alit y sco re, t ime 0 q u alit y sco re, t ime 0 q u alit y sco re, t ime 1 bo rro w ers w h o se q u alit y sco re d eclin es 0.5 1.0 u , 0 _ ( )−,_h u , 0 1u , 0_ q u alit y sco re relatio n sh ip bo rro w ers t ran sactio n al bo rro w ers b o rro w e rs w ith t ra n s a c tio n a l l o a n s a t t im e 0 w h o t a k e a r e la tio n a l l o a n a t t im e 1 b o rro w e rs w ith r e la tio n a l l o a n s a t t im e 0 w h o c a n b e c h a rg e d a r e la tio n a l t im e 0 r a te in s u b s e q u e n t p e rio d s l , 0 _ 0.00 0.02 0.04 0.06 0.08 0.10 0.5 0.6 0.7 0.8 0.9 1.0 quality score q u alit y sco re, t ime 0 q u alit y sco re, t ime 0 q u alit y sco re, t ime 1 bo rro w ers w h o se q u alit y sco re d eclin es 0.5 1.0 ( )−,h u , 0 1u,0 q u alit y sco re relatio n ship bo rrowers t ransactional bo rro w ers l,0 θ θ θ u,0θ borrowers with relational loans at time 0 who can be charged a relational time 0 rate in subsequent periods borrowers with transactional loans at time 0 who take a relational loan at time 1 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 47 figure 3: successful relational borrowers who (do not) move to transactional borrowing 0.00 0.02 0.04 0.06 0.08 0.10 0.5 0.6 0.7 0.8 0.9 1.0 q uality score 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0.5 0.6 0.7 0.8 0.9 1 1.0 .9 .7 .8 q u alit y sco re, t ime 0 bo rro w ers w h o se q u alit y sco re imp ro v es .4 .5 .6 relatio n sh ip bo rro w ers t ran sactio n al bo rro w ers q u alit y sco re, t im e 1 q u ality sco re u , 0_ u , 1 _ r e la t io n a l b o rro w e rs a t t im e 0 w h o s e q u a lit y s c o re i m p ro v e s , b u t d o n o t t a k e a t ra n s a c tio n a l l o a n a t t im e 1 b e c a u s e o f d e c lin in g r e la t io n a l c o s t s u , 0 _ ( )+,_h u , 0 1l , 0_ b o rro w e rs w it h r e la t io n a l l o a n s a t t im e 0 w h o c a n b e c h a rg e d a r e la t io n a l t im e 0 r a t e in s u b s e q u e n t p e rio d s 0.00 0.02 0.04 0.06 0.08 0.10 0.5 0.6 0.7 0.8 0.9 1.0 q uality score 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0.5 0.6 0.7 0.8 0.9 1 1.0 .9 .7 .8 q u alit y sco re, t ime 0 bo rrowers w h o se q u alit y sco re imp ro v es .4 .5 .6 relatio n sh ip bo rro w ers t ran sactio n al bo rro w ers q u alit y sco re, t im e 1 q u ality sco re 1 ( )+,h 1l u, 0θ u,θ,0θ u, 0θ u, 0θ borrowers with relational loans at time 0 who can be charged a relational time 0 rate in subsequent periods relational borrowers at time 0 whose quality score improves, but do not take a transactional loan at time 1 because of declining relational costs 48 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 under the model proposed here, a borrower who fails defaults on their loan, and so they begin the borrowing process anew if they can find a new lender. one can imagine that their quality score would decline (precipitously!) if they default. consequently, they would need to find new collateral and a new lender to be able to continue any endeavor. the proportion of borrowers that will switch from transactional to relational is: (12) these borrowers, since they were in the position of potentially switching between transactional lenders, will now find it more advantageous to obtain a relational loan (see figure 2). the proportion of borrowers that would have switched from relational to transactional, yet won’t, due to declining monitoring costs is (see figure 3): (13) 6.2 additional profits on retained borrowers a second result in this section addresses the profitability of the loans for relational lenders. because of competition, transactional lenders never make more than ordinary profits through profit that is structured into fees (the cost of originating the loan). these are also minimized because of competition with other transactional lenders. the same is true for relational lenders in the initial period since relational lenders compete with one another and with transactional lenders for new business. however, once a relationship is established, competition exists only between the one relational lender who holds the relationship and the transactional lenders. there is an opportunity to realize extra profits from borrowers that relational lenders retain. relational lenders can charge a higher rate (up to the rate charged in the initial period) for borrowers retained from the initial period. borrowers who were successful in the first period have improved quality scores. some of the borrowers with improved quality scores will not improve sufficiently to attract comparable offers from transactional lenders. therefore, this subset of borrowers will still only attract relational lenders, and only one relational lender will have a competitive advantage over all the other relational lenders. this one lender can charge any rate up to the rate charged by other relational lenders and make an additional profit on the transaction. if there is a cost to the borrower to shop lenders (the borrowers time to search out a new lender), then the borrower will stay with the same lender and the lender can charge the same rate as in the first loan. monopoly profits are calculated as the difference in rates multiplied by the loan amount over the set of relevant quality measures. the profit is the large gap bounded below and above by time 0 and 1 relational curves. on the left & right, the minimum quality is defined by the borrower’s expected rate of return and the breakeven point between relational and transactional lending at time 0. the expected monopoly profit in time period 1 that can be realized is: ( ) ( ) ( )∫∫ −− −= ),θ(h θ ),θ(h θ u,0 -1 u,0 u,0 -1 u,0 df(x)x-1γ1df(x)θ|failp ( ) ( )∫∫ ++ += u,0 u,0 1u,0 u,0 1θ ),θ(h θ ),θ(h df(x)x)-γ(1xdf(x)θ|succeedp the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 49 proposition two: banks can make excess profits on some relational loans because of retention of borrowers whose cost to monitor has declined in subsequent periods. (14) note that this result is for those borrowers whose project succeeds. if they fail to repay, they’re assumed to default, and if they plan to borrow again, they would have to start with a new lender. the same result is seen for those borrowers retained in successive time periods until the borrower defaults or finally transitions to a transactional lender. at each successive time there are fewer borrowers (since some will have defaulted or transitioned), but greater profits are to be had from these borrowers. however, there are greater monopoly profits that can be gained by considering longer time periods and other strategies that would affect the boundaries given in the above equation (14). the next section addresses possible strategies that would expand the monopoly profits that can be realized over longer time horizons. 7. more aggressive profit seeking now, we consider the dynamic nature of credit quality and how this dynamism affects borrowers and lenders. the lender has no control over how much of an increase or decrease one observes in the quality of a borrower (other than the assistance given by relational lenders). we assume that it changes exogenously. however, we noted earlier that the relational lender chooses how much of an investment it makes in the improvement of the project’s likelihood of success, and the relational lender controls the prices set. the relational lender can also offset first period losses with expected profits in subsequent time periods and take on borrowers that may be profitable only if retained over more than one period. recognizing that there is a possibility to attain profits even greater than described in the last section, the lender can attempt two strategies that will increase profits. 7.1 borrowers who otherwise would not obtain a loan some borrowers cannot obtain a loan as the rate they would be charged would be greater than their anticipated rate of return. these borrowers would be only those with the lowest quality scores. the lender knows that of this set, some borrowers would be successful, and in the next period these borrowers would be charged a substantially lower rate. it may be possible for the bank to make profitable loans to these borrowers (who might otherwise not get a loan) by engaging in tradeoff pricing: lowering the price of the loan to the rate of return in the first period and compensating in the pricing of loans for this group in subsequent periods. proposition 3: it is possible for a relational bank to absorb an expected one time loss for low quality borrowers in the initial period and compensate for the loss with [ ] [ ] ( ) (x)dfx-1xm|)x,(hrm|)x,(hr(p2)profitsadditional ),θ(h ),(θh 010 u,0 -1 l,0 1∫ + + ⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ γ+ ⎭ ⎬ ⎫ ⎩ ⎨ ⎧ +−+= 50 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 an expected gain from borrowers in later times for a net gain. to do this, the bank must extend loans below the minimum quality level set down to a new minimum quality level, . proof: for one time period, a bank’s expected net gain is calculated as: (15) note that profits have to be calculated separately across those firms that succeeded and those that failed, since the regions of integration will differ. the calculation over multiple time periods is presented later, but equation (15) demonstrates that a profit is possible over one time period (see figure 4). relational banks have regions of potential monopoly gains because the cost of lending declines after the initial loan, due to a decrease in the cost of monitoring. these banks can also offer loans at a loss in the first period to borrowers with very low quality or to borrowers who would otherwise obtain a loan from a transactional bank, making up the loss in subsequent periods because of the potential for monopoly pricing gains. the first term in equation (15) is the gain from the second loan to successful borrowers, described in proposition two as the additional amount obtained by charging up to the price of a first loan as costs decline. the second term is the interest loss from the first loan for successful borrowers, while the third term is the loss on the collateral for borrowers who default. to find the new lower bound for θ, namely θ l, π , one can take the derivative with respect to θ l, π of equation (15). before pursuing this, there are other factors to address that will impact this calculation. 7.2 borderline borrowers of quality a similar strategy can be pursued for better quality borrowers, those near the boundary going from relational to transactional lending. by considering the potential gains in future periods, one may be able to extend loans to a broader group with higher quality. again, figure 4 displays this result. this strategy can be more risky in that it relies on transactional lenders not cutting profits and dropping prices for their loans. this would require a shift for all transactional lenders under the assumption of perfect competition, where we will assume that transactional lenders do not respond with a profit cut. proposition 4: it is possible for a relational bank to absorb an expected one-time loss for borrowers above the breakeven point in the initial period by compensating ( )( )∫ ∫ ∫ π π π ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎥⎦ ⎤ ⎢⎣ ⎡ ⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ +⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ψ −⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ +⎥ ⎦ ⎤ ⎢ ⎣ ⎡ = + + 0l, l, 0l, l, 0l, -1 l, 1θ θ 0 θ θ 00 ),θ(h ),(θh 010 (x)dfx-1γ-1 l r β-1 -(x)dfx)-γ(1x-)m|r(x (x)dfx)-γ(1x}m|r{h(x)-}m|)x(h{r(p3)profitsadditional the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 51 figure 4: regions of potential monopoly pricing θ θ 52 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 for the loss with an expected gain from those borrowers in later times for a net gain. to do this, the bank must extend loans above the quality level set in equation (8) up to a new upper quality level, θ u, π . proof: for one time period, this net gain is calculated as: (16) the first term in equation (16) is the gain from the second loan to successful borrowers. the second term is the interest loss from the first loan for successful borrowers, while the third term is the loss on the collateral for borrowers who default. as in proposition three, to find the new lower bound for θ, namely θ u, π , one can take the derivative with respect to θ u, π of equation (16). again, before pursuing this, there are other factors to address that will impact this calculation. 7.3 length of stay with a relational banker to examine how long borrowers stay with a lender once a relationship has been established, we compute an expected length of a relationship predicated on a single assumption. we assume that once the relationship has been established, the borrower stays with the relationship lender because that lender offers a lower rate than any other lender. the only time to leave a relational lender would be when the borrower defaults or when the borrower’s quality measure has increased to the point where the borrower can find other forms of financing, such as a transition to a transactional loan. the bulk of these calculations are presented in the appendix, but a summary is presented in this section so that we can have a measure of how long a borrower is expected to maintain a relationship, which is key to planning purposes for relationship bankers. section 4.3 described how quality measures were updated over time. to compute the expected length of a relationship with a bank, we make a simplifying assumption. we observe that increasing credit scores can only move a certain distance from their current values to improved credit scores, so a first order approximation to equation (1), θ f,t+1 =h(θ f,t+1 ,+) , is given by: θ f,t+1 = θ f,k + α (1-θ f,k ) if project successful (17) = θ f,k + αθ f,k if project successful where 0 < α < 1. u, 0u, u, 0u, u, -1 0u, 10 00 },{h },{h 01 (x)dfx-1-1 l r -1 (x)dfx)-(1x t} |r{x -}m |r{x (x)dfx)-(1x }m | r{h(x) -t} | r{h(x) (p4) profits additional the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 53 if there were no anticipated improvements due to the relationship, at each point in time, we would have the expectation that the quality of the borrower would stay the same (some borrowers are successful, some borrowers fail, and the quality on average stays the same). thus, from (17) we can compute that e(θ k+1 ) = θ k if project successful, implying that a borrower with a transactional loan would always hover around their current quality score. but relational borrowers always have a better chance of success due to the value of the relationship. this means that over time, the relational borrower will be expected to have their credit quality become better because of the value of the relationship. we find that the approximate average time a borrower with credit score θ 0 at time zero can be expected to stay with a relational bank would be: (18) in this case, φ is the expected amount of time that a borrower will remain with a relational lender, allowing for occasional project failures so that the borrowers credit score both increases and decreases over time. 7.4 strategies for relational lenders the final part of this paper examines strategies that a bank can undertake to maximize its profit making as part of the loan process. as described in the last several sections, there are multiple sources of such profits, termed extra-ordinary profits in this paper. again, assuming perfect competition, transactional banks force each other to offer essentially the same rates to borrowers of the same quality. relational banks, having higher costs, can only compete with transactional banks by offering more value to the borrower through monitoring, increasing the likelihood of success for a single project. at some point, the added cost outweighs the benefit, particularly to borrowers with a high probability of success in their projects. and under perfect competition, relational banks offer the same rates among themselves to new borrowers of the same quality. the leverage that the relational bank has is the ability to capitalize on the unique advantage it has with a borrower once a loan has been made. costs of monitoring decline after the initial loan has been made, and this gives the lender a competitive advantage over all other lenders in establishing pricing. profits come from three different components captured over multiple time periods. as noted in the two previous sections, the calculation starts with losses taken in the initial period with two small sets of borrowers. at t = 1, 2, 3, …, φ we can then expect to make monopoly profits in these two regions plus the monopoly profits for borrowers that were naturally relational borrowers without implementation of these strategies. at time zero, we take a loss, as shown in (19), to be offset by a gain, shown in (20). * 1ln -1ln -1ln 0u,0 54 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 (19) (20) the profit function, net profit (np) = expected gain – expected loss, is in itself a function of the choices θ l, π ,θ u, π , and m 0 (the latter of course, relies on how it impacts γ[m 0 ]. it is also a function of the number of time periods used in the calculation (in other words, the number of periods used by the relational bank for its time horizon in planning). a simpler view of the expected loss and gain comes from recognizing that the range of integration extends across θ l, π to θ u, π , with subtractions for double counting. proposition 5: profit maximization, maximization of np = expected gain – expected loss, follows through the solution of: ∂np/∂θ l, π = 0,∂np/∂θ u, π = 0, and ∂np/∂m 0 = 0 (21) subject to the appropriate second order conditions. iterative solutions are necessary to find θ l, π ,θ u, π , and m 0 . 7.5 unchanging borrower quality sharpe and several others employ a particular model where the quality of the borrower is unchanging, which is a special case of the model described here. if we restrict the model to this special case, we can find two of the necessary parameters, θ l, π and θ u, π directly, noting some interesting characteristics. it is not possible to solve directly for the remaining parameter, m 0 , but the form of the equation that u, 0u, u, 0u, 0l, l, 0l, l, 000 000 (x)dfx-1 1 l r -1 (x)dfx)-(1x t} |r{x -}m |r{x (x)dfx-1 1 l r -1 (x)dfx)-(1x -)m |r(x loss expected },{h },{h 0 1k k kk ),(h ),(h 0 1k k k 0 k ),(h ),(h 0 1k k k 0 k 1k u, k0u, k0l, kl, ku,0 kl,0 k(x)dfx)-(1x m | )(x,hr t | )(x,hr (x)df x)-(1x m|)x,(hrm|)x,(hr (x)dfx-1x m|)x,(hrm|)x,(hr profits expected the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 55 would be used to find m 0 is also instructive. derivation of these parameters is given in the appendix to this paper. proposition 6: when borrower quality is unchanging, the optimal lower end of the borrowing range depends strictly on the relationship of the cost of lending to the expected rate of return for the borrower, adjusted by the expected improvement in the likelihood of success due to monitoring. the optimal lower bound quality score, θ l, π , for lending is the value that satisfies: (22) equation (22) has all values r k determined at θl, π, so the solution of (22) is not as direct as it appears, but rather, requires an iterative solution. there are some very intuitive results that come from equation (22). the average cost of monitoring is declining, so that if the time horizon is extended sufficiently far, it might be possible that the average rate that would be charged under perfect competition. if all lenders had the same information, it would decline sufficiently so that the expected loss from the first period is equal to the expected gain for a large number of time periods. proposition 7: when borrower quality is unchanging, the optimal upper end of the borrowing range depends strictly upon the relationship of the relative difference in the cost of monitoring to the cost of a transactional loan, adjusted by the expected improvement in the likelihood of success due to monitoring. when borrower quality is unchanging, the upper bound, θ u, π , to which relational lenders should extend credit by taking a loss in the first period, is the value that satisfies: (23) equation (23) has all values r k and r r determined at θ u, π , so the solution of (23) is not as direct as it appears, but rather, requires an iterative solution. equation (23) shows that the upper bound for the relational lenders is a function of the average monitoring costs over time periods. since the average monitoring cost can be reduced over many time periods, the upper bound is increased, and a larger share of the borrowers can be taken if the bank expects a longer time horizon. note that neither of these values is dependent upon the probability distribution of the quality measure. each is a function of the tradeoff between the relative cost k l,l, 0k kl, 00l, -1r -1-1 r where rr ~r-1-1 k u,u, 0k ku, tt0u, -1r -1-1 r where rr ~rr-1-1 56 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 of making the loan with the expected improvement that the lender can provide to the borrower. even though we can summarize how to solve explicitly for these two values, a similar looking, closed form solution for the value m 0 eludes us because of the complexity of the function. this is so despite the fact that we have fixed the quality of the borrower to never change. 8. empirical and policy implications there are numerous empirical and policy implications for our model. each is discussed below. informational monopoly rents no longer strictly accrue to relational lenders for all borrowers, as noted by sharpe (1990). high credit quality firms may obtain better loan rates by switching to a transactional lender. this result is consistent with the results of repullo & suarez (2004). with the extension of our model to multiple time periods, we are also able to show that there is no incentive for high quality firms that begin with a transactional lender to ever transfer to a relational lender, given the higher borrowing rates based upon higher monitoring costs. therefore, they have to be captured by the relational bankers by taking greater risk with these borrowers in early time periods. numerous studies have shown that larger banking organizations offer greatly reduced interest rates to small businesses relative to their community bank counterparts, i.e., berger and udell (1996); berger, rosen, and udell (2001); and carter, mcnulty, and verbrugge (2004). interest rates charged to small firms are a direct function of costs of lending and the quality of the borrower. thus, banks incorporate risk based pricing. survey evidence of risk based pricing by banks is provided by cowan and cowan (2006). this pricing leads to the possibility for either increases or decreases in interest rates for borrowers over time. a key assumption in our model is that monitoring adds value. relational lenders have the possibility of impacting the probability of firm success. thus, they should experience fewer difficulties given the same credit grade with transactional lenders, all else held equal. evidence for the validity of this assumption is apparent in the community investment literature. in addition, there is a significant amount of anecdotal evidence from small business lenders, suggesting that practitioners believe this assumption to hold. however, with the exception of boot and thakor (2000), the value of monitoring is not considered in the previous literature. our model suggests that higher risk credits will gravitate toward banks that engage in relationship lending. berger (2004) had argued that this is, in fact, the current state of the industry. our model indicates that relationship banking affords lenders with opportunities to strategically price relationship loans. loans that are subsidized initially may be profitable with relationship longevity. empirical support for such a result is provided by petersen and rajan (1995), as well as cetorelli nd gambera (2001). this point is crucial in understanding the typical view, that as higher quality borrowers may benefit most from basel ii, this may not also be true if relational banks compete and offer other advantages in taking these borrowers from transactional lenders. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 57 there are several ways to test our model predictions with the data. the ability to test our propositions to a large degree requires information that is not publicly available. we specifically need to identify the operational characteristics of banks related to credit scoring, which is not publicly available information. nevertheless, both case studies and survey methodologies provide opportunities to investigate our predictions. for example, using survey data from 1977 through 1989, berlin and mester (1998) find that bank loan pricing practices have historically been too insensitive to credit risk. if our propositions hold, then a more recent dataset should exhibit greater sensitivity to credit risk by banks in general. 9. conclusion in summary, we provided a theoretical model of small business lending within a competitive environment that incorporates technological and regulatory changes. borrower quality and lending costs interact to attract borrowers to banks that exhibit specific modes of lending (i.e., transactional versus relationship). our model suggests that interest rates offered to borrowers may either increase or decrease over time, which is intuitively appealing. based on the assumption that banks add value through monitoring, we show that banks may pursue a broader range of borrowers due to reduced risk. in our model, the improvement offered by the bank depends on how much the bank is willing to invest in the relationship. in general, we find that there is a unique market niche for both transactional and relationship lenders, but a grey area in between that depends on the level of investment the relationship lender wishes to make in the relationship. we also show that banks that take advantage of internal risk based models are able to reduce small business loan rates. the proposed model differs from previous theoretical models in several significant ways. we look at information about firms that is available to banks and the costs associated with obtaining that information. whereas previous models assume that banks have no information about businesses in the first time period, our model assumes that banks initially have some information specific to a firm. additionally, we assume that the quality of a firm changes over time due to the success of projects. information about a firm develops over a longer time frame than two periods, especially since the quality of the firm is changing. banks strategically consider the impact of maintaining a long-term relationship with a firm. although our model is yet to be tested to a large degree, our predictions appear to be consistent with the trends reflected in the small business data. for example, small business credit card competition is driving down loan rates at the large institutions that offer these cards. several extensions to this paper are possible, eventually providing for a richer model. in addition, this model restricts a lender to a single type of lending; i.e., either relationship or transaction. it is clear that lenders might adopt a strategy that incorporates elements of both relationship lending and transactional lending, depending of course, upon the quality of the borrower. a natural extension is thus adding a third type of competitive lender that uses quality thresholds in determining the degree of monitoring utilized. 58 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 author statement: adrian m. cowan is affiliated with the bill greehey school of business at the st. mary’s university e-mail: acowan@stmarytx.edu. charles d. cowan’s affiliation is with the university of alabama at birmingham c.cowan@ analyticfocus.com. references akhavein, j., frame, w.s., and white, l j., (2005). the diffusion of financial innovation: an examination of the adoption of small business credit scoring by large banking organizations. journal of business, 28, 577-596. allen, f., caaretti, e., and marquez, r., (2006). credit market competition and capital regulation. working paper, federal reserve board, finance and economics discussion series, 2006-11. berger, a. n., (2004). potential competitive effects of basel ii on banks in sme credit markets in the united states. working paper, federal reserve board, finance and economics discussion series, 2004-12. berger, a.n., rosen, r.j., and udell, g.f., (2001). does market size structure affect competition? the case of small business lending. working paper, federal reserve board. berger, a.n., and udell, g.f., (1995). relationship lending and lines of credit in small firm finance. journal of business, 68, 351-381. berger, a.n., and udell, g.f., (1996). universal banking and the future of small business lending. a. saunders and i. walter ed: financial system design: the case for universal banking. irwin publishing, burr ridge, il, 559-627. berger, a.n., and udell, g.f., (2002). small business credit availability and relationship lending: the importance of bank organizational structure. economic journal, 112, 32-53. berlin, m., and mester, l.j., (1998). on the profitability and cost of relationship lending. journal of banking and finance, 22, 873-897. blöchlinger, a., and leippold, m., (2006). economic benefit of powerful credit scoring. journal of banking and finance, 30, 851-873. boot, a.w.a., and thakor, a.v., (1994). moral hazard and secured lending in an infinitely repeated credit market game. international economic review, 35, 899-920. boot, a.w.a., and thakor, a.v., (2000). can relationship banking survive competition? journal of finance, 55, 679-713. carter, d.a., mcnulty, j.e., and verbrugge, j.a., (2004). do small banks have an advantage in lending? an examination of risk-adjusted yields on business loans at large and small banks. journal of financial services research, 25, 233-252. cetorelli, n., and gambera, m., (2001). banking market structure, financial dependence and growth: international evidence from industry data. journal of finance, 56, 617-648. cole, r.a., goldberg, l.g., and white, l.j., (2004). cookie cutter vs. character: the microstructure of small business lending by large and small banks. journal of financial and quantitative analysis, 39, 227-251. the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 59 cowan, c. d., and cowan, a. m., (2006). a survey based assessment of financial institution use of credit scoring for small business lending. research study, banking and financial economic research, small business association office of advocacy. diamond, d.w., (1984). financial intermediation and delegated monitoring. review of economic studies, 51, 393-414. diamond, d.w., (1991). monitoring and reputation: the choice between bank loans and directly placed debt. journal of political economy, 99, 689-721. diamond, d.w., (1996). financial intermediation as delegated monitoring: a simple example. federal reserve bank of richmond economic quarterly, 82, 51-66. frame, w.s., srinivasan, a., and woosley, l., (2001). the effect of credit scoring on small business lending. journal of money, credit, and banking, 33, 813825. greenbaum, s.i., kanatas, g., and venezia, i., (1989). equilibrium loan pricing under the bank-client relationship. journal of banking and finance, 221235. mester, l.j., (1997). what’s the point of credit scoring. federal reserve bank of philadelphia business review, 3-16. petersen, m.a., and rajan, r.g., (1994). the benefits of lending relationships: evidence from small business data. the journal of finance, 49, 3-37. petersen, m.a., and rajan, r.g., (1995). the effect of credit market competition on lending relationships. quarterly journal of economics, 110, 407-443. rajan, r.g., (1992). insiders and outsiders: the choice between informed and arm’s-length debt. the journal of finance, 47, 1367-1400. ramakrishnan, s., and thakor, a.v., (1984). information reliability and a theory of financial intermediation. review of economic studies, 51, 415-432. repullo, r., and suarez, j., (2004). loan pricing under basel capital requirements. journal of financial intermediation, 13, 496-521. sharpe, s.a., (1990). asymmetric information, bank lending, and implicit contracts: a stylized model of customer relationships. the journal of finance, 45, 1069-1087. 60 the international journal of banking and finance, 2007/08 vol. 5. number 2: 2008: 31-60 appendix: profit maximization parameters when quality measures do not change a1: a summary of model components tf, the observed quality of a firm f in time period t, 10 tf, f cumulative distribution of tf, over all firms applying for loans, f = 1, . . . , n ordered by credit quality ,h tf,1tf, ,h tf,1tf, quality at time t+1 is function of quality at time t and success or failure at time t; + for a success; – for a failure tr mc costs for relational lending for t = 0,1,2,3,… ; tm is the expenditure by the bank to monitor a firm; 1-tt mm tc r costs for transactional lending for t = 0,1,2,3,… , tmt for all t ]m[ 0 the proportion of improvement to a project that can be applied to increase the likelihood of repay; 0]m[ 0 for transactional lenders. tf,0tf,tf, 1*]m[p probability that borrower will repay rate of return to borrower for a project that is funded mr cost of funds to the bank j,cr , tk,t rr , kk,c rr interest rate to be charged to the borrower by bank type c in time period k, abbreviated for transactional and relational lenders l amount of the loan r amount of collateral underlying the loan recovery rate on collateral proportional increase \ decrease in quality score after project success \ failure expected amount of time that a borrower will remain with a relational lender international journal of banking and finance 3-1-2008 the dynamics of credit quality and implications for the pricing of small business loans adrian m. cowan charles d. cowan recommended citation empirical determinants of us equity flows to developed countries: does valuation matter? the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 49 empirical determinants of us equity flows to developed countries: does valuation matter? joseph j. french university of northern colorado, united states of america abstract this paper explores a new panel data set on us gross cross-border equity flows to 20 industrialized nations combined with measures of market valuation for the period of 1977-2005. empirical evidence of imperfect integration across world equity markets indicates that valuation matters. consistent with relative value trading as a determinant of equity flow patterns, we find that quity flows decrease sharply with host-country market valuations. this paper also finds that equity flows increase sharply with us equity market valuations. the findings of this research suggest that american investors are informed about both domestic markets and foreign markets. peripheral findings of this research confirm the findings of other researches, but this research is based on a longer sample period. consistent with existing literature, there is a negative influence of interest rates spreads, and information asymmetries on cross-border trades in equities. keywords: equity flows, cross-border portfolio investment, international markets jel classification: f21, g15, g14, 016, g11 1. introduction and motivation the past two and a half decades have been characterized by a dramatic increase in international capital mobility. in 1975, gross cross-border transactions in bond and equity flows for the us residents were equivalent to only 4 percent of gdp. this share increased to 100 percent in the early 1990s and has continued to increase to 245 percent at the turn of the current century. furthermore, a growing percentage of these portfolio flows consists of equity (hau and rey, 2006). however, empirically, there are few established results on the determinants of equity flows between nations (portes and rey, 2005). ijbf 50 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 empirical work has been stymied by data problems, imperfect mobility of capital, and behavior of international investors which is contrary to established theory. empirical agreement has also been difficult to reach because different researchers focus on different time-periods and different sets of countries. for example, calvo, leiderman and reinhart (1994) focus on the role of external (push) and internal (pull) factors as potential determinants of foreign investment using a cross section of developing nations. they found that low interest rates in the us played an important role in accounting for the revival of equity flows to these nations in the 1990’s. using data on bilateral portfolio equity flows from a set of 14 industrialized countries during 1989-1996, portes and rey (2005) find evidence that imperfection in the international credit markets and variables that proxy information asymmetries have the greatest influence on cross-border equity flows. this paper investigates the factors motivating cross-border equity flows to developed countries from large us investors. the traditional literature on the empirical determinants of equity flows have not paid particular attention to the overall role that equity markets play in shaping long-term portfolio investment decisions among a cross-section of developed nations. equity market variables have been considered in the portfolio approach to modeling capital flows, but this literature is concerned with addressing the issues of the lead and lag relationships between equity flows and prices. in the traditional literature on the determinants of equity flows for a broad cross-section of countries, the role of equity market valuation has only been considered peripherally. the effects of diversification on cross-border equity flows have been dealt with extensively in literature (stulz, 1999, griffen, nardari, and stulz, 2004, rey and hau, 2006). in the portfolio approach to understanding equity flows, authors are primarily concerned about modeling the linkages between equity flows and equity markets. the focus of this research in on the long-term empirical determinants of equity flows using traditional capital flows equations and a cross section of countries. most theoretical models and empirical studies have correctly assumed that cross-border equity flows are the outcome of home investors attempting to optimally diversify and the resulting equity flows reflect the behavior of investors adjusting portfolio weights. the influence of equity market valuation on cross-border equity flows to developed nations has not been considered in the traditional capital flow literature. however, in reality relative value trading is common, the us offshore funds directory (1999) lists several dozen hedge funds that use ‘pair trading’ as one of their principle equity investment strategies. this finding indicates that relative market valuations are important, to one subgroup of american investors. there is theoretical support that longer term us investors may have an advantage over hedge funds in relative value trading. shleifer (2000) and brunner and nagel (2004) suggest that longer-term investors have an advantage over short-term hedge funds. for example, suppose a hedge fund manager sells the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 51 an overvalued stock short. if overpricing increases before it reverses, she may be obligated to close the position due to the margin requirements or the agency problem, suggested by brunnermeir and nagel (2004), which shortens the horizon period. the fact that hedge funds are engaged in relative value trading implies that there is incentive for longer-term us investors to also be engaged in such strategies. further evidence that valuation may be an important factor in understanding why equity flows between nations is evidenced by empirical findings of ‘siamese twin’ stocks, which shows that the same stock can trade at different prices in different markets. froot and dabora (1999) find that royal dutch and shell transport have often not been priced in line with their relative claims on cash flows. in the early 1990’s, the two companies merged with an agreement that entitles royal dutch and shell to split the two entities’ combined cash flows on a 60:40 basis. this agreement was in place until the two firms officially merged in 2005. theoretically, royal dutch should have been priced at 60:40=1.5 times the value of shell. however, the stock price was observed to vary between 36:40=0.9 and 66:40=1.65 times the value of shell. since royal dutch and shell trade in possibly the two best functioning financial markets (usa and uk) and other ‘twin’ shares display similar behavior (froot and dabora 1999), it is likely that relative mispricing of corporate equities across international capital markets is common. the royal dutch shell experiment ended with the final combination of the two entities. there is a dearth of studies of the influence of equity market valuation on gross equity flows. the majority of the literature in this area (see stulz, 1999, rey and hau, 2006, and bohn and tesar, 1996) has assumed cross-border equity flows occur in an integrated and efficient international capital market and are result of portfolio allocation decisions. this paper steps back from this assumption and introduces the possibility that equity flows may be motivated by equity market valuation in an international capital market that is not completely integrated or efficient. as the royal dutch shell case, coupled with the relative value trading patterns of large hedge funds illustrates, the world market is likely not completely integrated. the primary objective of this research is to consider the implications of stock market valuation as a determinant of cross-border equity flows. specifically, this paper considers the role of equity market valuation on cross-border equity flows between the us and 20 industrial countries. the questions this research addresses are: how does source country valuation influence long term patterns of equity flows and how does host country valuations influence long term patterns of equity flows. in addition, this paper contributes to the literature in nature of the data. this paper presents results that span almost three decades, for 21 nations, and controls for fundamental determinants of equity flows found in previous literature. the relatively long time horizon of this data allows for testing of other fundamental determinants of equity flows without some of the time horizon problems in other literature. to test whether relative valuations help to explain equity flows between countries, methodology that helps to determine the presence of a valuation 52 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 effect. this paper analyzes how equity flows depend on host and source country stock market valuations. the key econometric issue is to determine whether the correlation between equity flows and stock market valuation is due to relative value trading or, alternatively from traditional determinants of equity flows. this paper begins with a general empirical methodology is applied to the main sample, which merges the treasury international capital (tic) data on equity flows and the extended international stock market valuation and returns data assembled by kenneth french. the merged sample spans 1977-2005, and includes observations on 21 countries. the analysis involves panel regressions of equity flows on source and host country stock market valuations measures. it is found that equity flows are very strongly negatively related to the average market-to-book, price-to-earning, priceto-cash earnings, and dividend yield ratios of publicly traded firms in the host country. furthermore, host country valuations have as strong or stronger effect than essentially any other determinant of equity flows considered. at the same time, equity flows are strongly positively influenced by us market valuations. this suggests that high us valuation encourages long-term reallocation abroad. this research also documents consistent with other literature on the empirical determinants of equity flows: 1) that proxies for information asymmetries are negatively related to equity flows, 2) that as interest rate spreads increase (i.e. foreign interest rate above us rates) equity flows decrease, 3) that equity flows are negatively related to tax rates of host countries. the remainder of this paper is structured as follows: section 2 outlines the methodology, section 3 discusses and describes my data, section 4 presents the results, and section 5 concludes this paper. 2. methodology panel data methodology is used to test the effects of mispricing as an empirical determinant of equity flows. this methodology is common in traditional capital flow literature. panel data methodology is employed and is appropriate for several reasons. first, panel data techniques solves or at least reduces some of the problems by associated with few degrees of freedom by increasing the data points. second, it is the appropriate estimation technique to alleviate the effects of omitted time-invariant variables that are correlated with explanatory variables. third, panel structure recognizes that each country can have its own country specific effects, which can be correlated or uncorrelated with some or all of the explanatory variables. fourth, panel data estimation method is among the most efficient techniques to analyze the impact of a common set of factors across diverse country groupings (greene, 2003 and calvo, leiderman and reinhart, 1994). country level stock market valuation ratios and returns to proxy valuation. because of the problem with model misspecification, particularly the difficulty is to individually identify the effect of market valuation from the effects of other factors on equity flows, because of these considerations; several competing models are estimated (baker, foley and wrugler, 2007). assume that the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 53 equity flows from the us (indexed with i) to host country (index with j, give time t) is a function of the following: f ijt = (δ it , δ jt ,ø it ,ø jt ) (1) where δ it is the degree of overvaluation in country i at time t and ø it ,ø jt represent vectors of control variables, for example past returns (bohn and tesar, 1996), interest rates spreads (chulan et al, 1998), country dummy variables, and information variables (portes and rey, 2005). this research hypothesizes that that controlling for the other determinants of equity flows that flows should decrease with the degree of valuation in host country or δ it < 0, and increase with the degree of valuation in the source country or δ jt > 0. to empirically test the above hypotheses, this research develops the relationship between valuation and gross equity flows assuming that expected returns are a function of commonly used valuation ratios: )/,/,/()( itititit pdepbmfre = (2) where m it / b is the book to market ratio for country i at time t, p it / e is the earning to price ratio for country i at time t and d / p it is the dividend to price ratio for country i at time t. the determinants of differential stock returns are stable over time, and the forecasting power of fama and french types of models are surprisingly high. (haugen, 1996). the variables used to valuation are common in literature, it has been found that market-to-book value serves as a rough proxy of underlying fundamentals; a low market-to-book suggests that the country’s stock market is undervalued (kothari and shanken, 1997). fama and french (1998) find that market-to-book is inversely related to future equity returns for international stocks and basu (1983) and fama and french (1992) find similar for us stocks. additionally, kothari and shanken (1997) find that aggregate market-to-book is negatively related to subsequent returns. they also find that in some time periods dividend-to-price outperforms book to market. additionally, the common use of price to earnings ratios by practitioners (graham and harvey, 2001) and the findings of chuhan, claessens and mamingi’s (1998) argues for the consideration of the earnings to price ratio. according to kathari and shanken (1997) one view of the predictive power of financial ratios reflect the degree to which the market is overvalued (high mb or pe) or undervalued (low mb or pe) at a given point in history. in the case of overvaluation, for example, future returns (and hence true expected returns) will be low insofar as the overvaluation is likely to be corrected over time. they find that overwhelming evidence that returns are forecasted by bm and dividend yields, casting doubt on the efficient market hypothesis. several different models are considered in order to get the most consistent and efficient results possible. the starting point will be a pooled ordinary least squares regression. consider the following general panel regression framework (modified from green, 2003): 54 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 f ijt = z i α + x it β + ε it (3) where f ijt is a scalar dependent variables, observed for country i at time t, x it is a k dimensional vector of data that varies overtime, and z i is a vector of data that varies across countries, but is constant over time. one could consider this a country effect. the first model considered is pooled regression. pooled regression considers z i to be observable for all countries, and common estimates of parameters should be found through ordinary least squares on pooled data. the problem is that if z i is partially unobservable and if the unobservable portion is correlated with x it , then the parameter estimates will be biased and inconsistent. in order to correct this potential problem fixed effects models are also considered. fixed effects model assumes that z i α = α i , or estimates country specific intercepts that do not vary over time to capture unobserved heterogeneity (greene, 2003). in order to determine whether fixed effects are appropriate i will estimate the following lm test for group effects. under the null hypothesis α = α i for all countries. under the alternative hypothesis intercepts vary from country to country. if one fails to reject the null hypothesis then the appropriate efficient estimator is pooled ols. the f statistic for this test is calculated as: (4) if the null hypothesis is rejected, then fixed effects model could be the appropriate method or potentially another class of panel data models may be appropriate. a random effects model assumes that unobserved heterogeneity is uncorrelated with x it and models z i α = α + u i , where u i is an individual specific disturbance that is drawn once and is not allowed to change over time. so random effects allow for differing intercepts across individuals, but the variation is the result of a draw from a random distribution. the appropriate estimation technique crucially depends on the nature of the latent variable. in order to test whether fixed effects or random effects is appropriate hausman wald style test (hausman and taylor, 1981) are estimated. under the null hypothesis βre βfe = 0, this implies that both estimation techniques are consistent. if the test fails to reject the null hypothesis then, while both estimation techniques are consistent random effects model will give more efficient parameter estimates. the hausman test statistic is calculated in the following manner (greene, 2003): (5) which is distributed chi-squared with degrees of freedom equal to the number of parameters in the b fe coefficient vector. rejection of the null hypothesis rejects that the random effects model holds. the following model is estimated to see whether valuation ratios better explain cross-border flows than lagged returns (i.e. return chasing), by including lagged returns as an additional variable. knnt r n rr knntnf fe pooledfe −− − − − =−−− 2 22 1 1 )( ),1( [ ] )(()()( 1' fererefefere varvarh ββββββ −−−= − the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 55 (6) if it is found that b 1 < 0 this does not prove the hypothesis and b 2 < 0. as market-to-book, earnings-to-price, or dividend to price may be a good proxy for δ in the above regressions do not control for other factors, which may influence equity flows. for example, some theories link interest rate differentials, industrial production, tax rates, country of legal origin, and exchange rates with equity flows and these fundamentals may be correlated with the stock market. this will result in the betas above being bias estimators. however, market to book, price to earnings and dividend to price ratios are exchange rate invariant and may be a good proxy for δ. then the following panel regressions will be run to determine the basic relationship between valuation and equity flows to control for other factors directly. f ijt = α i + β 1 (m it / b) + β 2 (m jt / b) + β 4 x i + ε (7) where x represents a vector of control variables and f ijt represents inflows as a percentage of initial stock. model using cash earning to price, price-toearnings, and dividend yield are also estimated. if cross-border equity flows are influenced by equity market valuation then β 1 < 0 and β 2 > 0. in addition, the marginal difference in valuation to induce flows will be lower for more developed economies, because the amount of friction is smaller. to control for this a variable to proxy time varying information asymmetry and institutional development is included. additionally, country level fixed effect should control for these concerns. 3. data and summary statistics portfolio flows are distinguished from other international capital flows by the degree the flows can be reversed. some clarification and definitions may be useful. capital flows are generally broken into three flows: direct foreign investment (fdi), bond flows, and equity flows. fdi flows are distinguished from other international capital flows by the degree to which the investor owns or controls the firms. fdi is typically defined as the direct or indirect ownership or control by a single domestic entity of at least ten percent of the voting securities of an incorporated foreign business firm or the equivalent in an unincorporated enterprise. bond flows represent flows from the us to foreign bond markets for portfolio reasons. similarly, equity flows used in this study represent flows from us investors to foreign equity markets for portfolio reasons. the source for the equity flows used in this study is from u.s department of the treasury (tic). data from the u.s department of treasury is the most comprehensive source of publicly available data for cross-border equity flows (tesar and warner, 1994). tic is the appropriate data set to test the longer term influences of equity market valuation, because the data taken from reports are mandatory and are ijtitijt jtit ijt rbdisb b m b b m bf εα +++ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ +⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ += − )()(log 14321 56 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 filed by banks, securities dealers, investors, and other entities in the u.s., who deal directly with foreign residents in purchases and sales of long-term securities (equities and debt issues with an original maturity of more than one year) issued by u.s. or foreign-based firms. the data reflect only those transactions between u.s. residents and counterparties located outside the united states. flows are calculated from a foreign perspective (i.e. non-u.s resident). hence, inflows to country i would be from the us minus outflows from country i to the us. the data span is 1977 to 2005 and include observations in which 20 countries are the host of equity flows out of the us. the series are reasonably complete, and they have been collected on a consistent basis over time. equity flows as percentage of initial equity position are measured; this is consistent with tesar and werner (1995), chuhan et al (1998) as: (8) where the us is the source country and j is the host. scaling by initial position renders the equity flow measure more comparable across countries. while scaling is not important in regressions where country fixed effects are included, in regressions where legal origin is included, it is preferable not to use country fixed effects, because time constant variables drop out of fixed effect estimations, so this scaling is appropriate. additionally, since small initial positions can lead to outliers in this measure, the flow variable is winsorized at +100 percent. stock market valuation and return are obtained data from kenneth french’s website. this data includes annual observations of the capitalization-weighted market-to-book, dividend to price, cash earning to price, equity to price, and stock market returns in both dollars and local currency for 20 countries for the period of 1975 to 2005 for most countries. the countries included are: austria, australia, belgium, canada, finland, france, germany, hong kong, ireland, italy, japan, malaysia, netherlands, new zealand, norway, singapore, spain, sweden, switzerland, and the united kingdom. kenneth french’s data was constructed using msci, crsp and compustat data. fama and french (1998) claim that the construction used does not suffer from survivor bias. the raw data are from morgan stanley’s capital international perspectives (msci). the set of firms whose data is used to construct country-level returns and valuation ratios is essentially the set of firms included in morgan stanley’s stock index for that country. these tend to be large firms, and for a typical country cover roughly 80 percent of the domestic stock market capitalization. control variables are gathered from several sources. the real exchange rate is calculated from nominal exchange rates and price indices from the imf international financial statistics (ifs). exchange rates are indexed with the us dollar exchange rate in 1995 set to one in each country. real exchange rate is included to capture the increase in productivity over a given period, (cavlo et al, 1994), gross domestic product in current dollars are from the world bank’s world development indicators. statutory corporate income taxes, representing jusa t jusa t usajt position flow f → − → = 1 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 57 the maximum marginal statutory corporate tax rates in that country in the given year, are from the world tax database maintained by the office of tax policy research at the university of michigan. tax rates proxy the attractiveness of the business environment in a country and one would expect that higher tax rates in foreign country would discourage equity flows; additionally desai, foley and hines (2004) find that us companies move equity toward low-tax locations. distance has been widely used as a proxy for information asymmetry (portes and rey, 2005). a variable termed ‘relative distance’, which is the average distance (in nautical miles) from the capital city of a particular country to washington, d.c., this distance, is then weighted by gdp of the foreign country. the gdp weights capture the negative relationship between size and information asymmetry. this time-varying proxy for information asymmetry is similar to alfaro,kalemli-ozcan and volosovych (2006). inflation rates, industrial production, and interest rates are taken from the ifs database. treasury yield or call money yield for are used to measure interest rate series. interest rate yield spreads are calculated as us interest rate minus host country interest rate. a positive yield spread would indicate that us interest rates are higher than host country interest rates. la porta, lopezde-silanes, shleifer, and vishney (1997, 1998) emphasize the importance of the historical legal origins in shaping the current financial environment (i.e. attractiveness for portfolio investment). they examine the effect of legal origin on the laws governing investor protection, the enforcement of these laws, and the extent of concentration of firm ownership across countries. most countries legal rules, either through colonialism, conquest, or outright borrowing, can be traced to four distinct european legal systems: english common law, french civil law, german civil law, and scandinavian civil law. these legal origin variables have been adopted as exogenous determinants of institutional quality, in particular financial markets and institutions (beck, demirguc-kunt and levine, 2002). to investigate (and control for) whether legal origins have a direct effect on equity inflows by adding legal origin dummies as additional right hand side variables in regressions without fixed effects. summary statistics for the transaction flow data, valuation ratios, and country characteristics are given in table 1. table 1: summary statistics means, medians, standard deviations, and extreme values for equity flows, stock market valuation ratios and returns are reported in the table below. rlc is the annual country return in local currency from kenneth french’s website, rusd is the annual country return in usd, m/b is the book to market ratio, p/e is the earnings to price ratio, p/ce is the cash earnings to price ratio, yld is the dividend yield. flow data is from the us treasury department tic data, iflow is equity inflow (i.e. from the usa into country i), oflow is equity outflow (i.e. out of country i to usa), and nflow is net equity flow (inflow-outflow). industrial production, interest rates, and real exchange rates are from imf international financial statistics. tax rates are from the world tax database maintained by the office of tax policy research at the university of michigan. distance scaled by gdp is calcu58 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 lated using number of nautical miles divided by gdp in current dollars. gdp in current dollars is from world bank’s world development indicators. panel a summarizes equity flow data, panel b reports valuation ratios and returns and panel c summarizes country controls and characteristics. n mean median sd min max panel a: equity flows inflow (us to foreign) 494 30995.47 6107 87026.70 0 676079 outflow (foreign to us) 494 32398.73 6625.50 88954.18 0 704559 net flow (inflowoutflow) 494 1596.74 277.50 6268.155 -46134 38493 panel b: stock market valuations and returns m/b 494 1.84 1.70 0.90 0.37 9.84 p/e 494 16.05 14.88 7.93 3.89 63.69 p/ce 494 8.35 7.74 4.26 1.30 38.76 dividend yield (%) 494 3.16 2.77 1.914 0.43 14.93 return (usd) 494 17.13 15.83 27.27 -47.33 135.8 return (local currency) 494 16.23 16.37 24.85 -38.91 121.01 panel c: country characteristics and controls industrial production 494 83.80 85.62 16.61 25.32 123.11 distance (scaled by gdp) 494 9.45 5.12 11.94 0.30 76.89 implied change in foreign exchange (%) 494 6.68 1.00 24.40 -124.00 100.00 real foreign exchange (1995) 494 106.38 103.81 20.65 57.73 179.45 tax (%) 494 34.13 35.00 10.58 8.50 52.00 cpi (base=1995) 494 81.21 86.84 20.38 19.92 113.34 interest rate (%) 494 6.77 5.52 4.36 0.07 19.80 uk legal origin 494 0.39 french legal origin 494 0.42 german legal origin 494 0.05 scandinavia legal origin 494 0.14 *in regression models, log of cpi, industrial production, real foreign exchange and distance are used. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 59 the total number of observations is 494, some countries lack full data sets (i.e. some early years of equity flow data are unavailable) for ireland, malaysia, new zealand, belgium, austria, and finland. portfolio equity investment grew rapidly over the period. the mean of net flows for the us in the sample is positive, consistent with the idea that home bias is declining. in this annual data, the net equity flows are small by comparison with gross inflows and outflows. average equity return for the 21 countries in this study for the sample of 1977-2005 was 17.13 percent in us dollar terms or 16.23 percent in local currency terms. this return compares to the historical market return for small us firms. the country with the highest return in given year in my sample period was italy, with its equity markets up 135 percent in 1985, while the worst return was reported by -47 percent in hong kong in 1982. the best average return for the 21 countries in my sample was 1985 with an average return of 56 percent and the worst year for world markets included in my sample was 2001 where markets lost on average of 12.58 percent. the highest median return in was 21.76 percent in hong kong, and the lowest median return occurred in austria at 5.47percent. the average top marginal tax rate for the 21 countries was about 34 percent with the minimum of 8.5 percent occurring in switzerland for the period of 20012005. the maximum tax rate of 56 percent occurred in germany from 1977 to 1987. of the sample about 39 percent was british legal origin, 42 percent were french legal origin, 4.6 percent were germanic, with the balance of observations being scandinavian in legal origin. 4. results a ‘stripped’ down model is estimated first, to establish the basic correlation between equity flows and stock market valuation ratios. the dependent variable is inflow as a percentage of initial stock and the explanatory variables are the source countries valuation ratios and us valuation ratios. table 2.2 presents the results of my initial estimations for the entire sample. pooled ols, fixed effects and random effects models are estimated, parameter estimates for all three models are similar and with stable signs. in order to estimate the most efficient model, f-tests for fixed effects are performed and if able to reject the null hypothesis, then hausman tests for random effects are estimated, under the null hypothesis both the fixed and random effects model are consistent, but the random effects model is more efficient, rejection of the null hypothesis implies that the fixed effects model is most appropriate. under the null hypothesis, there is no correlation between the repressors and the residuals. the underlying idea of the hausman test is to compare two sets of estimates, one of which is consistent under both the null and the alternative and another, which is consistent only under the null hypothesis. a large difference between the two sets of estimates is taken as evidence in favor of the alternative hypothesis, or in this case, the fixed effects model. 60 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 4 966 table 2: equity flows and stock market valuations (full sample): regressions of equity flows as a percentage of initial position into host country on source and host country market-to-book, price-to-earnings, cash earnings-to-price and dividend yields. white heteroskedastic robust t-statistics are and p-values are reported. variable coef t-stat p-value coef t-stat p-value coef t-stat p-value coef t-stat p-value m/b (for) -2.10 -3.11 0.00 m/b(usa) 5.97 3.92 0.00 p/e (for) -0.22 -1.71 0.09 p/e (usa) 0.51 3.37 0.00 p/ce (for) -0.54 -11.58 0.00 p/ce (usa) 0.68 11.42 0.00 dividend yield (for) -0.49 -3.15 0.00 dividend yield (usa) 2.62 11.56 0.00 country fixed fixed random fixed year no no no no n 494 494 494 494 r-squared 0.16 0.17 0.15 0.14 country fixed effects are included if the f-test for fixed effects rejects the null and the hausman test for random effects also rejects the null. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 61 pooled ols, fixed effects and random effects models are estimated, parameter estimates for all three models are similar and with stable signs. in order to estimate the most efficient model, f-tests for fixed effects are performed and if able to reject the null hypothesis, then hausman tests for random effects are estimated, under the null hypothesis both the fixed and random effects model are consistent, but the random effects model is more efficient, rejection of the null hypothesis implies that the fixed effects model is most appropriate. under the null hypothesis, there is no correlation between the repressors and the residuals. the underlying idea of the hausman test is to compare two sets of estimates, one of which is consistent under both the null and the alternative and another, which is consistent only under the null hypothesis. a large difference between the two sets of estimates is taken as evidence in favor of the alternative hypothesis, or in this case, the fixed effect model. this procedure is followed for all panel estimates, except when time invariant parameters are included, then pooled ols are estimated. p-values and t-statistics in all tables are derived using white hereroskedasticity robust standard errors. when the form of heteroskedasticity is not known, it may not be possible to obtain efficient estimates of the parameters using weighted least squares. ols provides consistent parameter estimates in the presence of heteroskedasticity, but the usual ols standard errors will be incorrect and should not be used for inference. white (1980) has derived a heteroskedasticity consistent covariance matrix estimator, which provides correct estimates of the coefficient covariance in the presence of heteroskedasticity of unknown form. table 2 reports preliminary results of the basic relationships between proxies for equity market valuations and equity flows. the results shed light on the influences of market valuations as long-term empirical determinants of equity flows. the preliminary results indicate that the effect of stock market valuation is two sided. high source country stock market valuations appear to spur outward equity flows and low host country valuation seems to attract inward equity flows, this finding is consistent with chuhan et al (1998) who also find a negative sign on price-to-earnings ratio for both asia and latin america. the results are both statistically and economically significant. a one-unit change in the market to book ratio of the host country leads to a 2 percent decrease in equity flows from the us. the relative wealth effect is substantially stronger for changes in the us with a one-unit increase in market to book ratio leading to an increase in equity flows abroad of 6 percent. the results of the relationship between price-to-earnings ratio and cashearning-to-price ratios are also negative and all significant at the 10 percent level, they too also appear to be economically significant with an one unit increase in the price-to-earnings ratio leading to a decline in the growth of equity flows of about a quarter of a percent. the results of the relationship between equity flows and valuation are consistent when dividend yield is used to proxy valuation. increases in dividend yields in the host country leads to a decrease us equity flows. a potential explanation for this relationship, for the case of dividend yield, is that us institutional investors may avoid high dividend paying markets 62 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 in an effort to avoid increased exchange rate risk or hedging activities. however, the positive significant coefficient on us dividend yield further supports the idea that as domestic wealth increases more equity is funneled abroad. if the results on the effects of source country valuation ratios were identified from only cross sectional variation, they would raise concerns. for example, the measured effects of the source valuation ratios might merely reflect the effects of countrylevel differences in accounting conventions (joos and lang, 1994, ball, kothari, and robin, 2000). to address such concerns, in unreported results, regressions country-by-country were estimated and then averaged the coefficients to try to isolate the pure time component. the results were very similar. additionally, the fixed effects estimator will also alleviate these problems (greene, 2003). the next step of the analysis to attempt to control for other factors of that influences equity flows that may be correlated with valuation ratios in order to reduce omitted variable bias and test the stability of the initial results. table 3 presents the results of regressions of market valuation proxies and control variables. the first regression, controls for ‘return chasing behavior’, widely documented in literature as a short-term determinant of equity flows. bohn and tesar (1996) coined the phrase ‘return chasing’, it is generally proxied in empirical work as a positive relationship between lagged returns and equity flows. the second variable controlled for in the first model of table 3 is the log of distance scaled by gdp following alfaro et al (2005). 1 the previous results showing that valuation ratios are important determinants of long-term equity flows remain significant both statistically and economically. the coefficient estimates are not significantly altered and the signs remain consistent. similar to portes and rey (2005), no evidence of returns chasing is uncovered in the full data set. this could be in large part due to the nature of my empirical methodology, while the use of annual data is common for the determination of long-run factors that influence capital flows, to capture the dynamic relationship between variables as suggested by bohn and tesar (1996) more frequent observations are required. distance, which was recently used by portes et al (2005) to proxy information asymmetries, is found to be negative and statistically significant. this result is consistent with a large literature, which hypothesizes that information asymmetries lead to exaggeration of the home bias puzzle. the interpretation for the negative coefficient on my time varying measure of distance is logical. as the distance between nations shrinks or the size of the economy grows, information asymmetries decline and more equity flows to these countries. several additional models are estimated including more variables that have been found to be significant in literature to see how the results are were influenced. when more controls are included, the positive coefficient on the valuation ratio for the us becomes insignificant. this indicates that my initial strong results for source country valuation ratios are suspect. however, the negative relationship between valuation of foreign markets and equity flows remains robust, using either price-to-earnings or market-to-book. 1 see bohn and tesar (1996) and froot et al. (2001). t h e in tern a tio n a l jo u rn a l o f b a n kin g a n d f in a n ce, 2008/09 v ol. 6. n um ber 1: 2008: 49-66 63 table 3: equity flows and stock market valuations (full sample): regressions of equity flows as a percentage of initial position into host country on source and host countries market-to-book, price-to-earnings and controls. white heteroskedastic robust t-statistics are and p-values are reported. variable coef t-stat p-value coef t-stat p-value coef t-stat p-value coef t-stat p-value mb (for) -2.04 -9.64 0.00 -2.05 -10.51 0.00 -1.89 -8.38 0.00 mb (usa) 3.44 4.68 0.00 0.75 1.01 0.31 2.90 4.09 0.00 p/e (for) -0.18 -6.49 0.00 p/e (usa) -0.08 -1.02 0.31 p/ce (for) p/ce (usa) dividend yield (for) dividend yield (usa) return (t-1) 0.00 -0.82 0.41 industrial production (log) 4.68 5.23 0.00 5.72 5.95 0.00 4.27 3.71 0.00 distance (log) -2.10 -4.00 0.00 -1.07 -1.57 0.11 -1.06 -1.83 0.07 -1.36 -2.47 0.01 cpi (log) implied change in forex interest rate -0.36 -6.01 0.00 -0.44 -6.75 0.00 -0.23 -4.09 0.00 tax rate -0.06 -2.13 0.03 -0.05 -1.77 0.08 -0.08 -4.11 0.00 uk legal origin 0.64 1.30 0.19 french legal origin -0.76 -1.49 0.14 german legal origin -0.59 -0.69 0.49 country fixed fixed fixed pooled ols year no no no no n 494 494 494 494 r-squared 0.20 0.25 0.23 0.19 country fixed effects are included if the f-test for fixed effects rejects the null and the hausman test for random effects also rejects the null. pooled ols is estimated when time-invariant repressors are included. 64 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 additional controls are also added to capture the relative productivity of the economy, interest rate spreads, and tax rates. table 3 reports that when the spread between us interest rates and foreign interest rates increase, investors tends to decrease flows as a percentage of initial positions from the us to host nations. this is consistent with the finding of calo, leiderman and reinhart (1993) who document that reduction in interest rate spread in argentina lead to a sharp increase in capital flows. one explanation for this phenomenon is that increases in foreign interest rates will lead to depreciation of currency and therefore subject the us investors to increased interest rate risk. alternatively, the fact that flows are negatively related to interest rate spreads can be explained, as simply us investors tend to stay at home when interest rates are relatively high. a third explanation is that high us interest rates may have decreased americans wealth, and therefore decreased their risk tolerance, causing them to rebalance away from foreign equities. table 3 also reports a significant negative coefficient on tax rates. this is consistent with the findings of densi et al (2002) that capital flows to environments with lower tax rates, and is also consistent with the findings that increases in taxes reduces expected returns. a panel regression that includes dummy variables for legal origin was also estimated. several papers have found that legal origin proxies institutional structure and investiblity (beck et al, 2002). however, in the full sample no statistically significant relationship is present. the signs on the legal origin variables are, however, consistent with other literature that finds that british origin indicates strong institutional structure, whereas french and german legal origin have a negative effect on flows. to summarize, the association between country level stock valuations and equity flows in mostly developed countries are studied. a strong new fact about equity flows is documented: there is a very strong negative link between host country stock market valuations and equity flows. indeed, the effect of host country valuations is almost as strong, in statistical terms, as any other determinants of equity flows included in this study. this paper is the first, to my knowledge, to document the role of valuation in developed markets as a determinant of equity flows and indicates that large us portfolio allocation decisions are negatively related to high valuations in host countries. 5. conclusions the majority of theories of equity flows assume that the world capital markets are informationally efficient and integrated. however, various lines of empirical evidence suggest that country-level shocks to investor optimism or risk aversion, combined with information asymmetry, sometimes cause the same capital asset to sell for different prices in different locations. these observations suggest that valuation may be an important determinant of cross-border equity flows. this research discusses and empirically evaluates the effects of us and host country valuation as a determinant of equity flows from the us to mainly developed nations. to provide a large-sample test, country and year variation the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 65 in stock market valuations, realized returns, and country controls are exploited. the results are consistent with the view that equity flows increase from the us to abroad when us valuations are high, indicating a sort of ‘wealth effect’ of equity flows. additionally, host country valuations are strongly and consistently negatively related to flows. this indicates that us portfolio investors seek ‘undervalued’ equity markets and increase flows to these markets as valuations decline. several findings consistent with literature are also documented, the negative role of information asymmetries on equity flows, the positive influence of productivity on equity flows, the fact that as us interest rates are high the american investors stay at home, and the negative influence of taxes on equity investment abroad. in conclusion, while the results of this research certainly do not find that other explanations of the determinants of equity flows are unimportant, they do appear to indicate that equity market valuations are an important piece of the puzzle in understanding the behavior of cross-border equity flows. author statement: corresponding author is joseph french, monfort college of business, box 128, greely, co 80639. tel:970.351.1226. email: joseph.french@ unco.edu. this paper was reviewed by one of the editors, and by anonymous reviewer before being edited to conform to the format of the journal. the author wishes to thank the editorial office, and the reviewer while acknowledging responsibility for remaining errors. references alfaro, l., kalemli-ozcan, s., and volosovych, v., (2005). capital flows in a globalized world: the role of policies and institutions, nber working papers 11696, national bureau of economic research, inc. baker, malcolm, p., foley, c., fritz, and wurgler, jeffrey, a., (2007). multinationals as arbitrageurs? the effect of valuations on foreign direct investment, working paper. formerly nber working paper series, no.10449, forthcoming in the review of financial studies basu, saniov, (1983). the relationship between earnings’ yield, market value and return for nyse common stocks: further evidence, journal of financial economics 12: 129-56. beck, t., demirguc-kunt, and a. levine., r., (2002), law and finance: why does legal origin matter, worldbank working paper. bohn, h., and tesar, l., (1996). us equity investment in foreign markets: portfolio rebalancing or return chasing? american economic review 86 (2): 77–81. brunnermeier, markus, and stefan nagel, (2004). hedge funds and the technology bubble, journal of finance 59: 2013-2040 calvo, g., a., leiderman, l., and reinhart, c.,m., (1993). capital inflows and real exchange rate appreciation in latin america: the role of external factors, imf staff papers, washington, dc. 66 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 49-66 chuhan, p., s. claessen, and n. mamigni, (1998). equity and bond flows to latin america and asia: the role of global and country factors, journal of development economics 55: 439-463 desai, mihir, c. fritz foley, and james r. hines jr., (2004). foreign direct investment in would of multiple taxes, journal of public economics 88: 2727-2744. fama, eugene f., and kenneth r. french, (1992), the cross-section of expected stock returns, journal of finance 47, 427-465. fama, eugene f., and kenneth r. french, (1998). value versus growth: the international evidence, journal of finance 53: 1975-1999. froot, kenneth a., and emil m. dabora, (1999), how are stock prices affected by the location of trade? journal of financial economics 53: 189-216. greene, w., (2003), econometric analysis, prentice hall, 5th edition griffin j., f. nardari, and r. stulz, (2004), daily cross-border equity flows: pushed or pulled?, the review of economics and statistics, 2004, vol. 86 (issue 3): 641-657. hau, h., and h. rey, (2006): exchange rate, equity prices and capital flows, review of financial studies, 2006, vol. 19 (issue 1): 273-317 hausman, a., jerry, and taylor e. william, (1980). panel data and unobserved individual effects, economtrica, 49: 66-87. kothari, s.p., and jay shanken, (1997). book-to-market, dividend yield, and expected market returns: a time-series analysis, journal of financial economics 44: 169-203. la porta, rafael, josef lakonishok, andrei shleifer, and robert vishny, (1997). good news for value stocks: further evidence on market efficiency, journal of finance 52: 859-874. pontiff, j., and l.d. schall, (1998). book-to-market ratios as predictors of market returns, journal of financial economics 49: 141-160 portes r., and h. rey, (2005). the determinants of cross-border equity flows, journal of international economics, 65: 269-296. shleifer, andrei, (2000). inefficient markets: an introduction to behavioral finance, oxford university press, oxford. stulz, r, (1999). globalization of equity markets and the cost of capital, nber working papers 7021. tesar, l., and i., werner, (1994). international equity transactions and u.s. portfolio choice. in: frankel, j. (ed.), the internationalization of equity markets, university of chicago press, chicago, il, pp. 185–215. the us offshore funds directory, (1999), new york. international journal of banking and finance 9-1-2008 empirical determinants of us equity flows to developed countries: does valuation matter? joseph j. french recommended citation the impact of manipulation in internet stock message boards 1 the international journal of banking and finance, volume 8 (number 4) 2011: pages 1-18 the impact of manipulation in internet stock message boards jean-yves delort, bavani arunasalam, henry leung and maria milosavljevic google, capital markets crc ltd, university of sydney and australian government office, canberra ____________________________________________________________ abstract internet message boards are often used to spread information in order to manipulate financial markets. although this hypothesis is supported by many cases reported in the literature and in the media, the real impact of manipulation in online forums on financial markets remains an open question. this paper is on the effect of manipulation in internet stock message boards on financial markets by employing a unique corpus of moderated messages to investigate market manipulation. internet message board administrators use the process of moderation to restrict market manipulation. we find that manual supervision of stock message boards by moderators does not effectively protect internet users against manipulation. by focusing on messages that have been moderated as manipulative due to ramping, we show ramping is positively related to market returns, volatility and volume. stocks with higher turnover, lower price level, lower market capitalization and higher volatility are more common targets of ramping. key words: ramping, market manipulation, internet stock message boards, event study jel classification: d82, g14, g22 _____________________________________________ 1. introduction internet stock message boards provide an effective medium for investors to communicate, to disseminate and discover new information. however, the value and integrity of internet stock message boards are often criticized and investors are warned not to trade on the information provided (fraser, 2007). in order to address these issues, regulators often dictate that message board providers conform to specific guidelines. 1 nevertheless, people continue to use internet forums for their own purposes, and financial surveillance analysts 1 for example, in australia, see australian securities & investments commission (asic) regulatory guide 162, http://www.asic.gov.au. delort et al.: the impact of manipulation in internet stock message boards javascript:popup('contact.cgi?popup=yes&window=contact&context=ijbf&u=1373505&article=1077&for=editor') javascript:popup('contact.cgi?popup=yes&window=contact&context=ijbf&u=1377257&article=1077&for=editor') javascript:popup('contact.cgi?popup=yes&window=contact&context=ijbf&u=1377259&article=1077&for=editor') javascript:popup('contact.cgi?popup=yes&window=contact&context=ijbf&u=1383078&article=1077&for=editor') http://www.asic.gov.au/ 2 at regulators and exchanges are known to use message boards for investigating suspicious trading activities. hotcopper 2 is the most popular internet stock message board in australia and forms the focus of our study. this site operates under a strict code of conduct which prohibits unethical or illegal use of the forum. administrators moderate postings which do not conform to the guidelines, and can revoke access for users who consistently disregard the rules. moderated posts are labeled with the reason for moderation, for example ramping, flaming or profanity. these posts serve to proxy manipulative behaviour. using an event study methodology, we investigate whether ramping is related to stock price volatility, trade volume and returns. further, we study the firm characteristics targeted by rampers. first we find that manual moderation of messages cannot effectively prevent ramping to take place in internet stock messages. this result is supported by the high number of posts that are moderated due to ramping every day, which indicates that ramping is a common phenomenon. in addition, we find that the lead time to moderation is approximately 8 hours, allowing sufficient time for these posts to impact the market. our results also demonstrate that ramping is positively related to market returns, volatility and volume. firms with higher turnover, lower price level, lower market capitalization and higher volatility also receive a higher proportion of ramping posts suggesting that they are more common targets of ramping. the remainder of this study is organized as follows. in section 2 we present an overview of previous works. this serves as motivation for the development of our hypotheses in section 3. in section 4, we describe the data and empirical methodology used in our analysis. the results are reported and discussed in section 5, and section 6 concludes by providing some areas for further work. 2. literature review stock market manipulation is generally defined as the creation and exploitation of arbitrage 2 http://www.hotcopper.com.au. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 1 http://www.hotcopper.com.au/ 3 opportunities (zigrand, 2006). manipulation of stock prices may occur through direct trading strategies or indirectly via the dissemination of distorted price sensitive information. the former may be observed when trades are intentionally placed in the wrong direction or short term losses are being undertaken to move prices in the desired direction (chakraborty and ylmaz, 2004). the latter may occur through mediums such as investment blogs, spam emails and online stock message forums (antweiler and frank, 2004; das and sisk, 2005; aggarwal and wu, 2006). however, the detection of both types of manipulative behaviour remains difficult due to the lack of proxy for the occurrence of manipulation. the impact of trade strategy manipulation can be observed in the market response to such strategies. for example, aggarwal and wu (2006) demonstrate that stock market manipulation cases pursued by the u.s. securities and exchange commission (sec) during 1990 through october 2001 were associated with greater stock volatility, greater liquidity and higher returns. other evidence of stock market reaction to manipulation is documented by hillion and suominen (2004), who highlight a general rise in volatility, trading volume and one-minute returns in the closing price of stocks in paris bourse during january and april, 1995. further, the proportion of partially hidden orders increased during this period. evidence of manipulation has also been found in certain types of derivatives markets such as the futures market. it is shown that uninformed investors earn positive profits by creating a futures position and simultaneously trading in the spot market. as a result, arbitrage free profit opportunities are exploited to derive profitable cash settlement at the time of delivery (kumar and seppi, 1992; merrick et al., 2005). additional evidence of trade strategy manipulation include those found by guo et al. (2008), who uses low earnings quality firms to proxy for firms‟ intention to manipulate. they find that these firms tend to use stock splits to manipulate equity values before acquisition announcements. on the other hand, eom et al. (2009) find that investors strategically place spoofing intraday orders in the korean exchange. spoofing orders have little chance of being executed, and are used to mislead other traders of an order book imbalance and thereby influence the direction of stock price movement. they conclude that stocks targeted for manipulation had higher return volatility, lower market capitalization, lower price level, and lower managerial transparency. finally, large traders delort et al.: the impact of manipulation in internet stock message boards 4 have also been found to carry out manipulation due to its scale of operation and its ability to better sequence and time trades (gastineau and jarrow, 1991; allen et al., 2006). manipulative behaviour involving the intentional misinterpretation and dissemination of price sensitive information is more difficult to document. bagnoli and lipman (1988) address the problem of market manipulation through the announcement of a takeover bid. they develop a model where the manipulator takes a position, announces a takeover bid, and then unwinds his position. in their attempt to elucidate market manipulation from the 1600s to the internet technology frauds of today, leinweber and madhavan (2001) show that, through technology, information can be quickly disseminated to influence investor perception on particular stocks. stock spam emails and posting on internet message boards are two major avenues manipulators employ to affect investor sentiment because of their global reach and high degree of accessibility to investors. in particular, significant reactions of trading volume and returns have been shown to respond to spam campaigns (bohme and holz, 2006). hanke and hauser (2008) then extends the works of leinweber and madhavan (2001) and bohme and holz (2006) to provide corroborating evidence that stock spam e-mails significantly and positively impact volatility and intraday spread. similar evidence is noted by hu et al. (2009), who found that pump and dump email campaigns show a statistically significant decline in stock price from the peak spam day to the following day. evidence of online impact of message board posting on stock activity is well documented in prior literature (antweiler and frank (2004); das and sisk (2005); aggarwal and wu (2006); cook and lu (2009)). antweiler and frank (2004) use computational linguistics methods to analyze the effect of message sentiment (buy, sell or hold) in forums to determine whether the information content of posts influence the stock market. they find that stock messages help predict volatility and disagreement between messages to be associated with increased trading volume. however, they find no correlation of message sentiments with the direction of returns. by improving sample selection and removing noise caused by program generated sentiments, cook and lu (2009) find that the bullishness of board messages positively and significantly predict abnormal stock return up to 2 days ahead. when taking poster‟s credibility into account, they also find that the board messages‟ predictive power over stock returns becomes much stronger in terms of both economic magnitude and significance. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 1 5 these studies demonstrate stock markets are likely to be sensitive to internet message board activities. moreover, many anecdotal evidences of market manipulation in internet stock message boards have been reported in the financial literature and in the media (leinweber and madhavan, 2001; harris, 2002). however, the phenomenon has not been investigated through a rigorous approach. to the best of our knowledge, this work provides the first quantitative analysis of market manipulation in internet stock message boards. 3. hypotheses development this research employs moderated postings unique to the australian hotcopper online message forum. improving upon prior literature such as das and sisk (2005), chemmanur and yan (2009) and brown and cliff (2004), hotcopper moderated postings allow us to proxy for manipulative behaviour directly. this is a unique opportunity available to us. stock ramping through forums occurs when a ramper recommends a stock and highlights its huge potential to rise in share price in the very near future. the ramper buys a large quantity of a stock, and posts information to influence the masses into buying the stock with the aim to drive up the share price. by implying that readers of these posts are in possession of privileged information such as insider knowledge of an impending takeover offer rampers seek to persuade the gullible into purchasing a particular stock. if a significant enough number of easily-led individuals invest in the touted stock, a ramper can “ramp up” the share price so that he or she could sell their shares at a quick profit. ramping can be both up or down. stock message board administrators use the moderation process to restrict the manipulations of stock prices through ramping posts as a service to internet users. however, the wide accessibility of internet technology and the high speed at which investors absorb news may diminish the effect of moderation. therefore we want to test the following hypothesis: hypothesis 1: manual moderation of stock message boards, as a monitoring strategy, does not protect users against manipulation. the direct measure of ramping through hotcopper moderated posts allow us to investigate the impact of ramping on stock market activity. it has been shown in the delort et al.: the impact of manipulation in internet stock message boards 6 literature that well advertised stocks are related to higher stock returns (das and sisk, 2005; chemmanur and yan, 2009; brown and cliff, 2004). based on the investor attention hypothesis proposed in chemmanur and yan (2009), which implies that high levels of stock advertising is associated with larger stock returns, stocks heavily promoted to investors via ramping postings may also generate higher returns. one possible explanation proposed by barber and odean (2008) may be investors‟ need to select stocks out of the many available the ones that attract their attention will be the ones most likely to be included in their investment portfolio. furthermore, extreme news announcements have been shown to produce high trade volume (tetlock, 2007). similarly, ramping posts, as a type of extreme news announcements, may potentially have the same effect. both stock spam e-mails and messages in online message forums rely on the internet technology to increase its reach to investors (leinweber and madhavan, 2001). drawing upon prior works (bohme and holz, 2006; hanke and hauser, 2008; hu et al., 2009) on the impact of stock spam e-mails on trade volatility and intraday spread, we hypothesize that manipulators may employ online message forum ramping to influence investor sentiments in a similar fashion. if ramping posts does indeed have an impact on investor sentiment prior to moderation, we may form the following hypothesis to test the relation between moderated ramping posts and stock market response: hypothesis 2: ramping in internet stock message boards affects stock price volatility, volume and returns. finally, we attempt to describe the characteristics of rampers‟ target firm stocks. since manipulation is more effective when there is greater uncertainty about the value of a firm, we expect that stocks with a higher return volatility would attract more ramping attempts (eom et al., 2009). firm size has also been shown to be significantly positively associated with the level of disclosure (alsaeed, 2006). this can be explained by the fact that larger publicly listed firms receive higher analyst coverage. on the other hand, smaller firms receive less coverage and thus are more susceptible to price movements driven by private information release. to that end, we expect smaller firms to be more prone to manipulation through ramping. hanke and hauser (2008) report that liquidity is a significant characteristic of companies targeted by spammers. the lower the liquidity, the larger the price impact. in addition, the impact of spamming on trade volume is markedly higher for international journal of banking and finance, vol. 8, iss. 4 [2011], art. 1 7 low-turnover stocks. the reaction of high-turnover stocks to spamming is generally lower because it takes a greater number of participants to move prices. to similar effect, we expect rampers to target low-turnover stocks (low liquidity). therefore, in order to investigate the characteristics of firms targeted by rampers, we form the following hypothesis. hypothesis 3: stocks targeted for manipulation have higher return volatility, lower market capitalization, lower price level, and lower turnover. 4. data and methodology 4.1. data the data for this study comes from hotcopper, australia‟s largest internet stock message board. the time period runs from january 2008 to december 2008 inclusive. messages were downloaded from the hotcopper website using software written by the authors, and we restricted our collection of data to only include messages that have a ticker symbol field representing an asx listed company. each message contains the following fields: date, time, author, ticker symbol and content. in total, the data set contains 1,146,223 messages for 1,825 firms listed on the australian securities exchange (asx). an interesting feature of the hotcopper forum is that the moderated posts are labeled with the main reason of their moderation. in accordance with australian law regarding all information available in the securities market, hotcopper users are expected to comply with a set of strict usage guidelines. 3 messages that do not comply may be moderated by a volunteer or by an administrator. 4 such messages are not removed from the forum, but their content is replaced by a message, which contains the following information: moderation time, moderation type, and a comment specifying the reasons for moderation. the full set of moderation types with their respective percentage composition are listed in table 1. each type is a simple phrase that describes the primary reason for moderation. some examples of textual comments are provided below: 3 user posting guidelines, http://www.hotcopper.com.au/postingguidelines/. 4 list of moderators, http://www.hotcopper.com.au/forum_modlist.asp. delort et al.: the impact of manipulation in internet stock message boards http://www.hotcopper.com.au/postingguidelines/ http://www.hotcopper.com.au/forum 8  all you seem to be doing is plaguing threads with your down-ramping on non-held stocks. find something else to do with your time instead of wasting others. snide remarks on stocks are not considered helpful or useful posts.  rumors aren‟t required thanks. they usually lead to misinformation on a public forum. no more like this, thanks.  this post is being moderated because of the unsubstantiated information particularly “every deal done falls through”. this just isn‟t true. please refrain from flaming other posters too. in addition to the hotcopper data, we obtained the daily closing prices, high and low intraday prices and volume data for all companies listed on the asx in 2008 from reuters. we excluded all firms that had fewer than 50 posts during 2008 from our data set. our final sample size consists of 1,083,913 posts for 938 firms. note that results presented in table 1 are based on the final sample size. table 1: moderation types found on hotcopper in 2008 moderation cause total total % other 6,379 45.1% flaming 3,583 25.3% ramping 1,519 10.7% profanity 1,207 8.5% spamming 389 2.8% defamatory 351 2.5% unknown reason 159 1.1% advertising 152 1.1% unlicensed advice 72 0.5% duplicate 73 0.5% insider trading 73 0.5% copyright 63 0.4% sexist 43 0.3% racist 43 0.3% blasphemous 33 0.2% total 14,139 4.2. methodology we apply an event study methodology to analyze the market impact of ramping in international journal of banking and finance, vol. 8, iss. 4 [2011], art. 1 9 internet stock message boards. we define a trade event for a certain stock as a day on which the market is open for trading in that stock. we further define a posting event for a certain stock as a day on which at least one message has been posted in the message board discussing this stock. similarly, we define a moderation event for a certain stock as a day on which at least one message discussing this stock has been moderated. finally, we define a ramping event for a certain stock as a day on which at least one message has been moderated because of an attempt to ramp this stock. the performance of a stock is measured in terms of the dependent variables defined by hanke and hauser (2008) and also used by hu et al. (2009), that is, stock return, volatility, intraday volatility and turnover. their definitions of these variables are shown in figure 1. figure 1: dependent variables defined by hanke and hauser (2008) as discussed in section 4.1, hotcopper posts may be moderated due to several reasons, but each moderated post can only be labeled with one of the moderation types given in table 1. this restriction and the close similarity between some of the moderation types may result in the ramping posts being labeled differently. for example, a post which could be labeled as both „unlicensed advice‟ and „ramping‟ may only be labeled as „unlicensed advice‟. posts that are moderated due to multiple reasons are often labeled as „other‟, which explains the high percentage of moderation under „other‟ (45.1%) as given delort et al.: the impact of manipulation in internet stock message boards 10 in table 1. we found a significant number of ambiguous moderated posts in which the moderators assign one label but mention several causes for moderation in the comment field. since our focus is on investigating the impact of ramping, it is important that we incorporate all posts that could potentially be seen as ramping. as a result, we include posts that mention the keyword „ramp‟ or „unlicensed advice‟ in the moderator‟s comment as an additional explanatory variable in our regression analysis. figure 2: regression model for the impact study the regression equation for this impact study is given in figure 2. the market and volume are used as added control variables. we also included in our model a dummy variable that takes the value 1 if a company announcement for stock i was released on day t and 0 otherwise. we include this in our model in order to control for the impact of company announcements on the performance of stocks. as outlined in section 3, particular types of firms are more likely to be targeted for manipulation than others. we use the regression model shown in figure 3 in order to investigate the characteristics of the firms targeted by rampers. in this regression, the dependent variable is the probability for a post to be moderated because of ramping, which we take as a proxy for the probability for the stock of being ramped. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 1 11 figure 3: regression model for determinants of target firms 5. results and discussion 5.1. descriptive statistics moderation activity: we define moderation delay as the difference between publication time and moderation time for a particular post on an internet message board. in 2008, for the 938 stocks included in the sample, the mean moderation delay was 7.95 hours (σ = 15.93). in addition, 40.78 per cent of moderated posts were moderated after two hours. this is sufficient time for ramping on forums to be effective, which supports hypothesis 1. we found that in general, inappropriate posts tend to be published outside of normal trading hours. this is demonstrated by the graph shown in figure 4, which represents the proportion of moderated posts and with respect to the hour of day. what is most interesting here is the daily pattern of posts, which are moderated because of ramping. the graph identifies quite clearly that ramping is more active during trading hours, which is quite different to the general pattern for moderated posts. we believe that the reason for this is that ramping messages created outside of normal trading hours are less likely to be effective since they are more likely to be moderated before market opening. this figure plots the proportions of moderated posts and ramping posts divided by the delort et al.: the impact of manipulation in internet stock message boards 12 proportion of posts with respect to the hour of the day. figure 4: ramping versus moderation activities dependent variables: table 2 presents descriptive statistics for our sample. in 2008, for the 938 stocks included in the sample, we have 238,252 trade events, 100,695 posting events, 4,870 moderation events and 768 ramping events. the number of ramping events reported is only a lower bound to the actual number of ramping events. with the increasingly high volume of messages that are posted on internet stock forums, it is likely that some of the inappropriate posts will not be discovered in the manual moderation process. however, these posts are not taken into account in our analysis. for example, the following unmoderated posts could be considered as ramping:  i heard a rumor that the plant is still being repaired and may be operating again later today.  i am focusing on oil stocks now, not affected by all this, and also i heard a rumor even general motors now admit our oil has run out on what we can produce in one day and now in decline, which is good news to the oil stocks. table 2 shows that ramping events have the highest mean for returns, volatility, intraday volatility and volume which supports hypothesis 2. ramping events have the lowest international journal of banking and finance, vol. 8, iss. 4 [2011], art. 1 13 mean price of the four types of events. table 2: descriptive statistics it indicates that the “pump and dump” interest is the highest for small-priced firms. interestingly, the mean returns for panel a, b and c are negative. a possible explanation could be the beginning of the global financial crisis that hit the market in the last quarter of 2008. however, the mean return for ramping events is positive, which shows that ramping co-occurs with price increase. in sub-section 5.3 we further investigate the impact of ramping on the four dependent variables taking into account the firm characteristics. this table presents the mean, median and standard deviation for the dependent variables namely, return (ret), volatility (sig), intraday volatility (spr) and turnover (vol). panel a presents the statistics for all the firms across all the days for january 2008 through december 2008. panels b, c and d present the same statistics only for days with at least one post, days with at least one moderated post and days with at least one ramping post respectively. impact of manipulation in ims: table 3 shows the results of the regression analysis for the model given in figure 2. the results show that the independent variables volume, market, the release of company announcements and ramping are significantly correlated to return, volatility, intra-day volatility and volume. the correlation coefficients are delort et al.: the impact of manipulation in internet stock message boards 14 positive for the four variables, but the confidence is stronger for volatility, intra-day volatility and volume. insider trading is also is also significantly correlated to volume, intra-day volatility and return. table 3: impact study results we previously introduced the dummy variable „ramping in comment‟ which corresponds to moderated events where at least one moderation comment contain the word “ramp” or “unlicensed advice”. the variable is also correlated to the two volatility measures, and volume to a lower extent. for volume we find a positive correlation with ramping, insider trading, other, advertising, flaming and profanity. other is a general category that is used for posts moderated because of miscellaneous or mixed reasons (and so can include ramping, flaming and profanity). our results are limited by the fact that we only have closing prices and do not currently have access to intra-day prices. this results in an inability to distinguish between different moderated posts, that is, we cannot identify the impact of a specific ramping post on the market. indeed, a ramping posting cannot be measured independently of other moderated posts (such as profanity), which occur on the same day. hence, we cannot accurately identify whether a specific ramping post has been effective in its cause. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 1 15 table 4: determinants of target firms determinants of target firms: in this section, we present our findings in investigating the characteristics of firms, which are the most common targets for ramping in forums. since some of the independent variables are significantly correlated, we implement both simple and multiple regressions in order to minimize the multicollinearity problem. in the simple regressions (regressions (1) through (4)), stocks with higher turnover, lower price level and higher volatility tend to receive a higher proportion of ramping posts. market capitalization is also negatively correlated to the proportion of ramping posts but with a low significant value. regression (5), which controls for multicollinearity, confirms that firms with higher turnover, lower price level, lower market capitalization and higher volatility receive a higher proportion of ramping posts. however, the correlation is highly significant only for volume. 6. conclusions and directions for future work to the best of our knowledge, this work provides the first quantitative analysis of market manipulation of internet stock message boards. employing a unique body of moderated messages, we are able to identify messages that have been moderated as manipulative due to ramping and examined its relationship with market response. we find that manual moderation of internet stock message boards does not effectively protect users against manipulation. this is consistent with our expectation that rampers in forums are difficult to identify even for moderators. the possible reasons could be due to the fact that ramping delort et al.: the impact of manipulation in internet stock message boards 16 can be a long progressive process, by which a ramper inconspicuously disseminates his ideas but attempts to hide the motivation to ramp. our experiment also shows that ramping is positively and significantly related to market returns, volatility and volume. in addition, our results demonstrate that stocks with higher turnover, lower price level, lower market capitalization and higher volatility are more common targets of ramping. our proxy for ramping is not perfect. future works may include the development of techniques to detect well-concealed smart manipulators amongst posts not yet reviewed by moderators. computational linguistics techniques could be employed to detect ramping and to refine our proxy for ramping. as discussed in section 5, the number of posts moderated due to ramping is only a lower bound to the actual number of ramping posts. to identify ramping posts that have not been labeled by moderators, we need to examine the content of all unmoderated posts. we plan to build a text classifier to automatically identify posts with inappropriate content. applying this classifier to unmoderated posts will enable us to discover posts that should have been moderated due to ramping and subsequently to refine our proxy for ramping. the following two sample unmoderated posts shows that an improvement can be made to detect the poster‟s intention to ramp.  “i heard a rumor that the plant is still being repaired and may be operating again later today. does anyone know how to contact the fellow from atmos that is mentioned in this article?”  “i‟m focusing on oil stocks now, not affected by all this, and also i heard a rumor even general motors now admit our oil has run out on what we can produce in one day and now in decline, which is good news to the oil stocks.” author information: jean-yves delort, the submitting author is a software engineer at google and was a staff member in the department of computing, macquarie university, sydney, when this work was conducted. contact e-mail: jydelort@google.com. bavani arunasalam is a staff member at the capital markets crc limited, sydney, nsw 2001, australia. henry leung is a staff member at the faculty of economics and business, university of sydney, nsw 2006, australia. maria milosavljevic is a specialist in advanced analytics working for the australian government in canberra. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 1 javascript:popup('contact.cgi?popup=yes&window=contact&context=ijbf&u=1373505&article=1077&for=editor') mailto:jydelort@google.com javascript:popup('contact.cgi?popup=yes&window=contact&context=ijbf&u=1377257&article=1077&for=editor') javascript:popup('contact.cgi?popup=yes&window=contact&context=ijbf&u=1377257&article=1077&for=editor') javascript:popup('contact.cgi?popup=yes&window=contact&context=ijbf&u=1377259&article=1077&for=editor') javascript:popup('contact.cgi?popup=yes&window=contact&context=ijbf&u=1383078&article=1077&for=editor') 17 references aggarwal, r. k., wu, g., (2006). stock market manipulations. journal of business 79 (4): 1915–1953. allen, f., litov, l., mei, j., (2006). large investors, price manipulation, and limits to arbitrage: an anatomy of market corners. review of finance 10: 645–693. alsaeed, k., (2006). the association between firm-specific characteristics and disclosure: the case of saudi arabia. managerial auditing journal 21 (5): 476–496. antweiler, w., frank, m. z., (2004). is all that talk just noise? the information content of internet stock message boards. the journal of finance 59 (3): 1259–1294. mark bagnoli, barton l. lipman, (1988). successful takeovers without exclusion, review of financial studies, oxford university press for society for financial studies 1 (1): 89-110. barber, b. m., odean, t., (2008). all that glitters: the effect of attention and news on the buying behavior of individual and institutional investors. review of financial studies 21 (2): 785–818. bohme, r., holz, t., (2006). the effect of stock spam on financial markets. brown, g. w., cliff, m. t., 2004. investor sentiment and the near-term stock market. journal of empirical finance 11: 1–27. chakraborty, a., ylmaz, b., (2004). manipulation in market order models. journal of financial markets 7: 187–206. cook, d. o., lu, x., (2009). noise, information, and rumors: internet board messages affect stock returns. working paper. das, s., sisk, j., (2005). financial communities. journal of portfolio management 31 (4): 112–123. eom, k. s., lee, e. j., park, k. s., (2009). microstructure-based manipulation: strategic behavior and performance of spoofing traders. ssrn elibrary. fraser, j., (2007). the mysterious world of stock forums. compareshares.com.au. gastineau, g. l., jarrow, r. a., (1991). large-trader impact and market regulation. financial analysts journal 47: 40–72. guo, s., liu, m. h., song, w., (2008). stock splits as a manipulation tool: evidence from mergers and acquisitions. journal of financial management (winter): 695-712. hanke, m., hauser, f., february (2008). on the effects of stock spam e-mails. journal of financial markets 11 (1): 57–83. harris, l., (2002). trading and exchanges: market microstructure for practitioners. oxford university press, usa. hillion, p., suominen, m., (2004). the manipulation of closing prices. journal of financial markets 7: 351–375. hu, b., mcinish, t., zeng, l., (2009). the can-spam act of 2003 and stock spam emails. financial services review 18 (1): 87–104. kumar, p., seppi, d. j., (1992). futures manipulation with cash settlement. the journal of finance 47 (4): 1485–1502. leinweber, d. j., madhavan, a. n., (2001). three hundred years of stock market manipulations. the journal of investing 10 (2): 7–16. merrick, j. j., naik, n. y., yadav, p. k., (2005). strategic trading behavior and price distortion in a manipulated market: anatomy of a squeeze. journal of financial economics 77: 171–218. delort et al.: the impact of manipulation in internet stock message boards http://ideas.repec.org/a/oup/rfinst/v1y1988i1p89-110.html http://ideas.repec.org/s/oup/rfinst.html http://ideas.repec.org/s/oup/rfinst.html 18 tetlock, p. c., (2007). giving content to investor sentiment: the role of media in the stock market. the journal of finance 62 (3): 1139–1168. zigrand, j. p., (2006). endogenous market integration, manipulation and limits to arbitrage. journal of mathematical economics 42: 301–314. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 1 effectiveness of central bank independence via its internal characteristics 19 the international journal of banking and finance, volume 8 (number 2) 2011: pages 19-39 effectiveness of central bank independence via its internal characteristics carroll h. griffin fontbonne university, united states ___________________________________________________________ abstract this study examines central bank independence in developing countries of latin america and asia as well as selected developed countries. many countries around the world, both developed and developing, have accepted the idea of central bank independence over the last several decades, so central banks have autonomy. a majority of studies has examined primarily the impact of central bank independence on inflation as promoting the theoretical benefits of a more stable and prosperous macroeconomic environment. however, there is only now sufficient data to empirically determine whether these claims are true. this research attempts to answer why developing economies with an informal sector resort to inflationary measures to finance their activities; how does a government induce an agent to choose the formal economy. in the trade-off between inflation and reserve requirements, the optimal policy is maximum inflation and minimum reserve requirements as increasing the steady-state utility of an optimizing agent. also agents prefer the informal economy if policy relies on a maximum reserve requirement. keywords: central banks and their policies, international financial markets, monetary policy, asia, central bank independence, financial markets, inflation, latin america jel classification: g22 _____________________________________________ 1. introduction this paper addresses the unintended consequences on the informal and formal sectors of an economy from the regulations on bank reserve and inflation financing observed in several developing countries. the term central bank independence (cbi) can be broadly defined as griffin: central bank independence 20 the degree of freedom of the central bank to pursue monetary policy without interference from political considerations (sirivedhin and hataiseree, 2000). in other words, independence implies that a nation’s central bank can follow a trajectory outside the political realm and does not have to abide by the government’s request to follow a certain monetary policy, such as printing money to pay for federal deficits (gruben and welch, 1993). since being first written about in the late 1980s, many countries have come to adopt this policy and governments have come to recognize this as standard procedure. from the late 1980s to the mid-1990s, countries as diverse as malaysia, new zealand, chile, mexico, argentina, spain, france and brazil granted their respective central banks more autonomy from the government (cukierman, 1994). however, despite this proliferation, highly independent central banks in many countries of the developing world are relatively new phenomena (boylan 2001). the idea of granting a nation’s central bank autonomy from the government, although a relatively simple one at first glance, is powerful in the sense that, if a central bank is independent in its decision making abilities, the government in power at a given time will be unable to turn back monetary decisions that have been made or, in a similar vein, force actions to be undertaken. thus, in theory, an independent and autonomous central bank should lead to a more stable economic environment in the respective country (maxfield, 1997). it represents a prominent transformation, because it not only removes a key aspect of economic decision making from political control but has kept many governmental powers from overturning growth-oriented central bank policies (boylan 2001). the primary motive behind the granting of such autonomy appears to be the belief that an independent and sovereign central bank would have more power to diminish inflation and price volatility and thusly, promote economic growth. according to cukierman (1994), there is widespread consensus now that central banks should be independent because it helps to achieve price stability. this belief stems not only from monetary theory but also from empirical studies on the subject, both in developed and developing countries. the traditional argument for central bank autonomy is that the power to spend money (the government) should be separate from the power to print money (the central bank). although there has been much study done on the theoretical rationale behind it, few empirical studies, other than those examining the effects of inflation, have been conducted to international journal of banking and finance, vol. 8, iss. 2 [2011], art. 2 21 assess accurately whether central bank independence actually leads to desirable macroeconomic outcomes. that is, although many countries have granted autonomy to their respective central bank with the hope of not only taming inflation and inflation volatility, but also achieving a broader spectrum of macroeconomic goals, such as low unemployment, sustained gdp growth, increased investment, less volatile interest rates and further financial market liberalization, many of these hopes have not been proven empirically, especially in the developing world. the focus of this study will be to examine a select group of countries in three defined regions (asia, latin america, and the developed west) and identify whether macroeconomic variables such as inflation, unemployment, interest rates, exchange rates, direct investment, exports, savings rates and economic growth have responded positively since the granting of central bank autonomy. some of these variables have been examined indirectly in previous studies, but many of the results, in particular regarding developing countries, have been inconclusive. thus, one of the primary contributions of this study is that it will directly examine the effects of central bank independence on these principal macroeconomic variables over a longer term period (since 1960 in most cases). we analyze both the preand postcentral bank independence periods. another important contribution is that it also recognizes the heteroscedastic (variance in errors) nature of the countries’ time series data, so, when necessary, utilizes a different regression model, that of generalized autoregressive conditional heteroskedasticity (i.e. garch). once correlation is established, causality (which is a stronger relationship) will be examined as well. another of the unique contributions of this paper is that it will examine the phenomenon of financial crises as they relate to central bank independence, including the current global crisis of 2008-09. it will not only examine the effects that financial crises had on countries, but will also examine countries such as mexico that experienced this type of crisis both before and after the central bank was made independent in order to determine what effect (if any) this independence had on the lessening of the severity of a given crisis. it is important to determine what impact, if any, central bank independence has had on lessening the severity of crises, as it is perhaps during these types of extraordinary circumstances that an independent central bank is most needed. griffin: central bank independence 22 the rest of the paper has four sections. section 2 provides a brief literature review on cbi while section 3 is about the methodology to be applied in this study. the results are presented and discussed in section 4 which is followed by section 5 with concluding comments. 2. literature on central bank independence 2.1 central bank autonomy: an overview motivation for autonomy: among both developed and developing countries, there has been significant controversy regarding central bank independence. however, despite this debate, during 1990 and 1995, over 30 countries including five from latin america granted independence to their respective central banks. this is due to most countries’ central banks and finance ministries understanding and accepting that high inflation is detrimental to growth and that price stability is one of the central banks’ primary functions (maxfield, 1997). central bank reforms have been focused in several particular areas. one is the rewriting of legal statutes regarding central bank operations and relations with other branches of the government. another is the issue of central bank accountability, and the third is the issue of increased central bank transparency (crowe and meade, 2008). overall, one of the main factors motivating central bank reforms and the granting of autonomy by their respective country governments has been the newfound belief that central bank independence may be an institutional mechanism that leads to price stability, thus increasing the credibility of policymakers’ commitments to stable prices (diana et al., 2004). furthermore, an independent central bank has the freedom to implement monetary and exchange rate policy in order to achieve its objective. additionally, a clear definition and prioritization of the objectives of the central bank enhances its credibility both domestically and internationally (gutiérrez, 2003). however, it should be noted that central bank autonomy is in many cases part of a broader package of reforms that many times includes judicial reform, privatization, removal of trade barriers, etc. (acemoglu et al., 2008). international journal of banking and finance, vol. 8, iss. 2 [2011], art. 2 23 2.2 early history and global events although the idea of central bank independence may seem new, perhaps it is not. independent central banks have been around since the interwar period of the earlier part of the 20-th century, although not much was studied about them nor were the independent central banks of the time considered exceptional. beginning in the postwar era of the early 1950s, more began to be written about them, and the idea had its supporters and detractors (forder, 2005). however, it was only in 1988 that alesina stated, “independent central banks have been associated with a lower average inflation rate and may have been responsible for reducing politically induced volatility of monetary policy and inflation” and that the idea began to be accepted as common policy (alesina, 1988). one of the early success stories was that of new zealand, which granted its central bank autonomy in 1989 and achieved a considerable decrease in inflation rates (carlstrom and fuerst, 2006). by 1997, with the granting of independence to the bank of england, the transition could be considered complete and alesina’s suggestion just ten years prior could be considered an established finding. there are several specific reasons that the concept of central bank independence only began to take hold beginning in the late 1980s and into the 1990s. first, the highly independent central banks such as the german bundesbank and the swiss national bank produced very good long-term results in achieving price stability. their success in controlling inflation for relatively long periods of time gave initial impetus to other countries to grant autonomy to their own central banks. second, the rampant inflation of the 1970s in many countries around the world led many political and monetary policy leaders to search for a solution to the ravaging effects of high inflation. the episodes of hyperinflation in latin america in the 1980s (greater than 1000 per cent per year in many cases) caused these countries to look for mechanisms that would quell the recurrence of this monetary malady, and a more independent central bank was seen as a viable alternative. third, the desire of the breakaway republics of the former soviet union to transform their economies into market style economies also led to further interest in the granting of autonomy to respective country central banks (cukierman, 1994). griffin: central bank independence 24 2.3 legal vs. operational independence the conclusion of many empirical studies from the 1990s and early 2000s was that inflation is negatively related to central bank independence in both developed and developing countries. however, for central bank independence to be effective, it must not only be independent in the legal (de jure) sense, but also in the operational (de facto) sense as well. there is perhaps not a great distinction between the two in most developed countries, but in many developing countries the difference can be quite large. as long as central bank independence is only legal (formalized), and not operational in scope, the expected benefits will not occur. in fact, it may actually be counterproductive (cukierman, 1994). operational independence implies that central banks are empowered, among other things, to control their own interest rates, and thus, those of commercial banks in the respective country. the differences between de jure and de facto independence is great in countries where institutions and rule of law is weak (carstens and jácome 2005). therefore, one explanation for the weak empirical relationship between central bank independence and inflation is that the legal measure of independence fails to capture operational independence, particularly in developing countries where the rule of law may be weaker (maxfield, 1997). in reality, some central banks are more independent than what is implied by law (such as the bank of brazil), while others are less independent than what legal independence would imply (such as in venezuela). therefore, legal provisions may not be consistent with de facto independence (jácome and vazquez, 2005). for example, although legal indices are important, other aspects of central bank independence such as informal arrangements, tradition and culture of monetary stability are also influential in the setting and carrying out of monetary policy. therefore, legal independence is not the same as actual independence in many cases for at least two reasons. first, central bank laws are incomplete. second, even if scope of authority and procedures are described explicitly in the law, actual practice may be differing (eijffinger et al., 1996). moreover, although the degree of legal independence is normally dictated by the central bank laws in a given country, the level of conservativeness can generally be chosen by the central bank. thus, a lack of legal independence can be compensated for by selecting international journal of banking and finance, vol. 8, iss. 2 [2011], art. 2 25 more conservative central bank governors and other directors. a conservative central bank governor will generally place more weight on inflation stabilization (price stability) than society. in other words, a conservative central bank governor leads to a lower inflationary bias, which is optimal for society (eijffinger and hoeberichts, 1998). the implementation of independent central banks constitutes an institutional reform that could add creditability to macroeconomic policy. this measure would at least partially isolate the monetary and exchange rate policy from partisan short-term political battles. however, as the mexican crisis of 1994-5 clearly indicates, the formal independence of a nation’s central bank is not enough to avoid monetary crisis. it is imperative that appropriate monetary policies also be implemented and that it also has political legitimacy. one implication of this is that a central bank should have not only legal independence but also operational independence. for example, as the venezuelan experience of april 1994 illustrates, when the majority of the central bank board members resigned under political pressure from the executive branch, where there are extreme differences of opinion among the members of the central bank and the executive branch, purely legal independence of the central bank may not be sufficient (edwards, 1995). however, it seems that the likelihood of independence as well as the degree of independence rises as the differences in political parties and ideologies rise (as was the case in chile in the 1980s). the rationale for this is that there is an incentive for both political parties to have an independent, neutral central bank that will pursue policies that are roughly in the middle between the desires of the two parties (alesina, 1988). in arnone et al. (2006), the authors examine an updated index of central bank autonomy and apply the index to a set of developed, developing and emerging market countries. the results indicate an increase in central bank autonomy between 1992 and 2003, especially for developing countries. in most cases, this increase is achieved over three stages, from the initial stage in which the foundation for autonomy is laid to the second stage in which operational autonomy develops to the final stage in which further autonomy in terms of policy formulation and leadership is achieved. griffin: central bank independence 26 2.4 transparency and corruption transparency of central banks has been defined as the degree of genuine understanding by the public of the monetary policy process and policy decisions (winkler, 2000). it is the extent to which central banks disclose information that is related to the policymaking process (eijffinger and geraats, 2004). there are several general categories of central bank transparencymonetary policy, political, economic, procedural, policy and operational transparency. to express the concept in model form, consider the following equations. first, assume that the central bank maximizes the objective function: w = -½α (π – τ) 2 – ½(1α) (y – k) 2 (1) where π equals inflation, τ is the central bank’s inflation target, α equals the weight placed on inflation stabilization, y equals the output gap, and k is the output gap target. further assume: π = π e + y + s (2) where π e equals expected inflation, y again equals the output gap, and s is a supply shock. further assume that the aggregate demand equation is expressed by: y = -r + d (3) where r equals the real interest rate (the central bank’s policy instrument) and d is a demand shock. if equations 2 and 3 are then substituted into equation 1, the following may be obtained: r = α(π e –τ) – (1α)k + αsa + da (4) where r now reflects both the central bank’s preferences and the macroeconomic shocks it foresees. if the central bank releases its forecasts for inflation and output, the private sector can then identify any anticipated macroeconomic shocks and can deduce the central bank’s policy preferences using equation 4. thus, intentions can be inferred from policy actions (geraats, 2009). one of the major benefits of increasing the autonomy of central banks is that its actions become more transparent to those observing them. for example, according to poole et al. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 2 27 (2008: p 418), “the more regular and straightforward are federal reserve actions and communications, the more the market will focus on fundamentals and the less will it be distracted by incorrect interpretations and speculations.” on the other hand, central bank opacity (or lack of transparency) regarding policy decision making causes the policy interest rate to be a noisier signal of the central bank’s intentions, thus leading the private sector to be less attuned to these intentions. additionally, if this is the case, financial markets are likely to put disproportionate weight on public information to coordinate their actions (geraats, 2009). compared with past decades, there is substantially more transparency now in central bank operations relating to monetary policy. to illustrate, in most central banks price stability is by far the most important objective with all other objectives (such as employment, growth, etc.) being secondary to varying degrees (as was the case with the bank of england charter of 1997). this represents a vast increase in transparency over previous years because for much of the 20-th century central banks operated with no clear indication of priorities (cukierman and gerlach, 2003). although central banks have traditionally been associated with secrecy, as a central bank becomes more independent, it needs to be more accountable for its actions (fraser, 1994). the main benefits of transparency, according to theoretical models of recent years, is that it enhances the credibility, reputation and flexibility of monetary policy, which should ultimately lead to lower nominal interest rates. however, the actual change in interest rates is dependent upon the degree of actual transparency (geraats et al., 2006). in recent years, a number of central banks have moved toward greater transparency regarding their objectives, procedures, rationales, models and data. additionally, transparency has shown to be greater in countries with more stable and developed political systems and more developed financial markets (dincer and eichengreen, 2007). central banks with inflation targeting regimes in place have experienced the largest expansion in transparency along with countries that suffered high inflation in the 1990s. it also appears that countries with higher monetary policy transparency tend to experience less subsequent inflation (geraats, 2009). griffin: central bank independence 28 2.5 measuring central bank independence as has been previously stated, there are many nuances of central bank autonomy and independence. therefore, reliable procedures for measuring the degree of sovereignty must be found in order to gauge the impact of independence or dependence of a respective country’s central bank. cukierman et al. (1993) note the difficulty in quantifying central bank independence, and how legal vs. operational independence can vary, especially in developing countries. in their study, among other things, the authors use the turnover rate of central bank governors as a proxy for actual independence in developing countries, while legal independence is considered an adequate proxy for independence in developed countries. the proxy of central bank governor turnover has since been used in many studies, such as carstens and jácome (2005), who found that the average rate in latin america is 0.4 (meaning that the average central bank governor remains in office around 2.5 years). this high rate of turnover suggests that latin american central banks still remain vulnerable to political pressures. a central bank is considered more politically independent the longer the governor’s term in office (jácome and vazquez, 2005). several other measures of central bank independence have been proposed in the literature, such as the 1991 grilli, masciandaro and tabellini (gmt) index, which was one of the first formulated indices for central bank independence (which measures political and economic autonomy), and the original 1992 cukierman index, which is based on 16 detailed criteria of central bank independence. however, some more recent authors such as crowe and meade (2008) use the cukierman, webb and neyapti (the modified cukierman index) measure described in part above because it is very comprehensive and includes developing as well as developed countries. the index has four components relating to appointment procedures (including a proxy for central bank governor turnover), resolution of conflict, the use of explicit policy targets, and rules limiting lending to the government and is an updated version of the original cukierman index. 3. hypotheses and methodology we formulate the methodology and hypotheses for this study in this section. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 2 29 h1: there is stronger evidence of cbi effectiveness among countries operating under a higher level of operational independence. h2: there is stronger evidence of cbi effectiveness among countries operating under a higher level of central bank transparency. h3: there is a beneficial signaling effect present with the announcement of central bank independence. methodology will include a variety of econometric techniques to test for volatility, correlation and causation with results organized by individual hypothesis. macroeconomic data from q1:1960 to q1:2009 will be taken entirely from the international monetary fund (considered a reputable, neutral, global financial institution) in order to maintain consistency of measurements, data collection processes and to avoid biases of individual country central banks or national data collection agencies. the following macroeconomic variables will be used as dependent variables to measure the impact of central bank independence and outside phenomena on a country’s growth and stability and are measured on a quarterly basis. for econometric reasons, the first iteration (first difference) of each variable will be used.  inflation (as measured through changes in the consumer price index)  unemployment rates (%)  gross domestic product volume index independent variables will be used to examine the impact of the nuances (political and economic) of central bank independence and related phenomena on the previously mentioned dependent variables. dummy variables are used frequently to classify data into mutually exclusive categories (gujarati 2003) and will be used in the following equations to represent, among other things, the effects of country development, foreign exchange rate regimes, the occurrence of financial crises, the degree of international inflation, and the presence of legal cbi. griffin: central bank independence 30 regression analysis is one of the most useful econometric techniques in practice. according to gujarati (2003): regression analysis is concerned with the study of the dependence of one variable, the dependent variable, on one or more other variables, the explanatory variables, with a view to estimating and/or predicting the mean or average value of the former in terms of the known or fixed (in repeated sampling) values of the latter (p. 18). the basic regression in this study will take the following form: y = α + β1cbi1 + β2 cbi2 + β3 cbi3 + … + ε (5) the specific regression models given below is an expansion of the above basic model using the dependent variables, which are the objectives of cbi (namely inflation, employment, and gdp growth) and the central bank independence variables as in: inflation rate = α – β1development dummy – β2modified cukierman index β3gmt index β4foreign exchange rate regime dummy + β5financial crisis dummy + β6international inflation dummy – β7legal central bank independence dummy + β8central bank governor turnover – β9it dummy + β10cons – β11taxrev + ε (6) unemployment % = α – β1development dummy – β2modified cukierman index β3 gmt index β4 foreign exchange rate regime dummy + β5financial crisis dummy + β6international inflation dummy – β7legal central bank independence dummy + β8central bank governor turnover – β9it dummy + β10cons – β11taxrev + ε (7) gdp volume index = α + β1development dummy + β2modified cukierman index + β3 gmt index + β4foreign exchange rate regime dummy – β5financial crisis dummy – β6international inflation dummy + β7legal central bank independence dummy – β8central bank governor turnover + β9it dummy β10cons + β11taxrev + ε (8) we use ordinary least squares multivariate regression using garch to test for correlation and relationships among the variables in question. regressions will be done across time using time series over the test time period under consideration (quarterly data for almost 50 years with more than 190 observations for most selected cases). the goodness of fit ratio, or r 2 will be used to gauge how well each dependent variable is explained by the independent variables. the durbin-watson test will also be used for autocorrelation. a comparison of the r 2 and durbin-watson test will be used as verification that the regression is not spurious (as mentioned above, if the durbin-watson results are quantitatively higher than the r 2 , this is a sign that the regression is not spurious, or nonsensical). only results that are at least at the 10 international journal of banking and finance, vol. 8, iss. 2 [2011], art. 2 31 per cent statistical significance level will be listed throughout the results section and all will be listed with coefficients. 4. results for the following hypotheses, a representative country for each level will be examined using inflation, gdp volume and unemployment rates to ascertain to what degree these variables are influenced by the internal central bank independence characteristics. h1: there is strong evidence of cbi effectiveness among countries operating under a higher level of operational independence. the benefit of cbi in terms of stabilizing inflation is mixed. it is noteworthy that in the medium-high sample, legal cbi has a positive coefficient but mci (operational independence) has a negative relationship, illustrating the difference that can (and many times do) exist between operational and legal cbi. also noteworthy are the positive coefficients for international inflation in several subsamples, indicating that the relationship the world inflation levels can have on a particular country’s inflation. inflation table 5.1a: inflation results for operational independence level r 2 dw significant variables (coefficients) high (u.s.) 0.38 2.17 cons (0.49), deficit (0.52), forward (0.20) medium high (portugal) 0.46 1.80 mci (-6.92), lcbi (2.02), it (-2.41), ii (-0.68), dd (-2.78) medium (thailand) 0.11 1.29 ii (.96) medium low (china) 0.19 0.96 dd(22.74), fxd (4.53) low (japan) 0.36 2.17 dd(-1.58), fcd (.42), ii (1.51), lcbi (-1.05), mci (6.70), taxrev (-0.06) b. gdp growth griffin: central bank independence 32 table 5.1b; gdp volume index results for operational independence level r 2 dw significant variables (coefficients) high (u.s.) 0.51 2.22 cons (0.35), ii (-1.07), surplus (0.19), ocbi (8.57), forward (0.26) medium high (portugal) 0.07 2.97 fcd (-0.77), taxrev (0.12) medium (thailand) 0.28 1.72 ii (1.38), dd (11.10), surplus (-0.06) low (japan) 0.01 2.60 none according to the above statistics, there is little evidence to indicate that the degree of operational independence of a central bank is in any way affecting the level of gdp growth. c) unemployment results for unemployment in this regard are also unclear. not only are explained variation very low, the parameter estimates too do not provide strong evidence to support cbi promotes economic growth. table 5.1c: unemployment results: operational independence level r 2 dw significant variables (coefficients) high (u.s.) 0.44 2.80 cons (-.40), surplus (-0.42) medium high (portugal) 0.13 2.03 cons (0.10), taxrev (0.05) medium (thailand) 0.04 2.15 fcd (0.56) medium low (china) 0.10 2.13 fxd (-.20) low (japan) 0.13 2.27 dd (-0.95), gmt (0.13), lcbi (0.40), mci (-1.77) international journal of banking and finance, vol. 8, iss. 2 [2011], art. 2 33 h2: there is stronger evidence of cbi effectiveness among countries operating under a higher level of central bank transparency. a) inflation table 5.2a: inflation results and central bank transparency level r 2 d w significant variables (coefficients) high (new zealand) 0.32 1.1 6 fxd (0.99), gmt (-0.16) medium (brazil) 0.46 1.0 6 gmt (2.36), ii (9.11) low (mexico ) 0.66 1.1 6 cbgov (-0.01), dd (-3.63), ii (0.06) from the above analysis regarding inflation, it would appear that the most benefit (legally and operationally) from cbi comes at high levels of transparency. at lower levels, there is no clear benefit from cbi. b) gdp growth from the statistics in the table 5.2b, it appears that there is little relationship between cb transparency and gdp growth. this is at all levels of transparency. table 5.2b: gdp volume index results and central bank transparency level r 2 dw significant variables (coefficients) high (new zealand) 0.07 2.45 fcd (-1.19) medium (brazil) not available low (mexico) 0.12 3.40 dd (22.60), fcd (-2.88), surplus (-1.16) c) unemployment griffin: central bank independence 34 table 5.2c: unemployment results and central bank transparency level adj. r 2 d w significant variables (coefficients) high (new zealand) 0.08 2.3 3 none medium (brazil) 0.10 2.0 6 none low (mexico) 0.14 2.9 2 surplus (0.12) there appears to be no relationship between central bank transparency and unemployment levels as could be verified from the results in table 5.2c. h3: there is a beneficial signaling effect present with the announcement of central bank independence. regression statistics will be examined using event study format for the three target variables as dependent variables: inflation, gdp volume index, and unemployment for the five years immediately prior to each respective country’s central bank independence and the five years immediately afterwards, for a total test window of ten years. this is done in order to determine if there is any type of announcement effect (any positive effect just from news being disseminated in the market). one representative country from each region will be examined. each event study set of results will be compared to the results for the total time period. table 5.3a: inflation results: event study vs. total time period region event study--significant variables (coefficients) total time period—significant variables (coefficients) asia (japan) none dd(-1.58), fcd (.42), ii (1.51), lcbi (1.05), mci (6.70), taxrev (-0.06) latin america (mexico ) lcbi (18.12), gmt (-2.65), cbgov (51.62), ii (2.66) cbgov (-0.01), dd (-3.63), ii (0.06) develop ed west (u.s.) none cons (0.49), deficit (0.52), forward (0.20) international journal of banking and finance, vol. 8, iss. 2 [2011], art. 2 35 it is apparent from the statistics in table 5.3a that there is very little beneficial signaling effect regarding cbi on inflation. for mexico, where the legal cbi variable appears significant, it is actually positively correlated with inflation instead of the theory-suggested negative sign. in the other regions, lcbi is not significant at all. however, operational cbi is negatively related to inflation in mexico, perhaps indicating a slight increase in de facto power on the part of the respective central bank to curb inflation. next, gdp volume is examined. table 5.3b: gdp volume index results: event study vs. total time period region event study--significant variables (coefficients) total time period—significant variables (coefficients) asia (japan) mci (10.38) none latin america (mexico) none dd (22.60), fcd (-2.88), surplus (-1.16) develope d west (u.s.) none cons (.35), ii (-1.07), surplus (0.19), ocbi (-8.57), forward (0.26) in asia, there is a positive relationship between operational independence as expressed through mci and an increase in the gdp volume index. therefore, there is some evidence to support the possibility of an announcement type effect for this region. table 5.3c: unemployment results, event study vs. total time period region event study--significant variables (coefficients) total time period—significant variables (coefficients) asia (japan) lcbi (0.61) dd (-0.95), gmt (0.13), lcbi (0.40), mci (-1.77) latin america (mexico) fcd (0.45) surplus (0.12) developed west (u.s.) none cons (-0.40), surplus (-0.42) next, unemployment is examined. as can be seen, there seems to be no announcement effect for unemployment. however, it is of note that the financial crisis variable is positively correlated with unemployment in latin america during this time period. griffin: central bank independence 36 as can be seen, there seems to be no announcement effect for unemployment. however, it is of note that the financial crisis variable is positively correlated with unemployment in latin america during this time period. 6. conclusion the following table 6 is a summary of the findings from this research using an extensive data set on several countries. table 6: summary of findings on central bank independence 1. on inflation, there is some evidence that a higher operationally independent central bank is associated with lowering inflation. for the other two variables, there is little evidence that independence promotes growth or employment. 2. there is some correlation between lower levels of inflation and high central bank transparency. however, there is no evidence growth or employment is promoted. 3. there is some evidence for a cbi signaling effect that spurs economic growth through an increase in operational cbi, but little support for growth or for employment. the idea of central bank independence has been at the forefront of financial market liberalization since the late 1980s and is still considered quite relevant (de haan et al., 2008). many countries around the world, both developed and developing, have granted autonomy to their central banks in order to achieve monetary and fiscal objectives. there are several general motives put forth in the literature as to what macroeconomic and political events drove countries to grant this authority. however, the results in the early literature on the subject, with some exceptions, seemed to come to a common conclusion that an increase in central bank independence would lead to a lower inflation rates. however, much of the early work was done in the 1990s before many of the developing and developed country central banks had achieved independence. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 2 37 this study is an attempt to update and expand upon empirical studies using a current lengthier and richer database, available more appropriate methodology and an increased perspective of the potential gains from central bank independence. one of the main objectives of this study has been to examine the nuances of operational vs. legal central bank independence. given the fact that now sufficient post-independence data over 10-15 years or more are available, it is now possible to quantify the potential benefits of central bank independence by examining countries that achieved cbi in the 1990s. the verdict from this study is that there is still more work needed to verify if central bank independence is correlated with not just inflation but also growth and employment. author information: dr carroll howard griffin is a staff member in eckelkamp college of global business and professional studies, fontbonne university, 6800 wydown blvd., st. louis mo 63105, united states: e-mail: chgriffin@hotmail.com. references acemoglu, d., johnson, s., querubin, p., & robinson, j. (2008). when does policy reform work? the case of central bank independence. nber working paper 14033. alesina, a. (1988). macroeconomics and politics. nber macroeconomics annual, 1988. (cambridge, ma: mit press). arnone, m., laurens, b., & segalotto, j. (2006). measures of central bank autonomy: empirical evidence for oecds, developing, and emerging market economies. imf working paper, 06/228. boylan, d. (2001). defusing democracy: central bank autonomy and the transition from authoritarian rule. carlstrom, c., & fuerst, t. (2006, september). central bank independence: the key to price stability? federal reserve bank of cleveland. carstens, a., & jácome, l. (2005). latin american central bank reform: progress and challenges. imf working paper. wp/05/114. crowe, c., & meade, e. (2008). central bank independence and transparency: evolution and effectiveness. imf working paper. wp/08/119. griffin: central bank independence mailto:chgriffin@hotmail.com 38 cukierman, a., (1994, november). central bank independence and monetary control. the economic journal, 104: 1437-1448. cukierman, a., and gerlach, s., (2003). the inflation bias revisited: theory and some international evidence. the manchester school, 71(5): 541565. cukierman, a., kalaitzidakis, p., summers, l., and webb, s., (1993). central bank independence, growth, investment, and real rates. carnegie-rochester conference series on public policy, 39: 95-140. cukierman, a., webb, s., and neyapti, b., (1992). measuring the autonomy of central banks and its effects on policy outcomes. the world bank economic review, 6: 353398. de haan, j., masciandaro, d., and quintyn, m. (2008)., does central bank independence still matter? european journal of political economy 24: 717-721. diana, g., and sidiropoulos, m., (2004). central bank independence, speed of disinflation and the sacrifice ratio. open economic review 15: 385-402. dincer, n., and eichengreen, b., (2007, march). central bank transparency: where, why and with what effects? national bureau of economic research. working paper no. 13003. edwards, s., (1995). crisis and reform in latin america: from despair to hope. the world bank. oxford university press. eijffinger, s., and hoeberichts, m., (1998). the trade-off between central bank independence and conservativeness. oxford economic papers, 50: 397-411. eijffinger, s., & geraats, p. (2004). how transparent are central banks? unpublished manuscript. eijffinger, s., van rooij, m., & schaling, e. (1996). central bank independence: a panel data approach. public choice, 89(1-2): 163-182. forder, j. (2005). why is central bank independence so widely approved? journal of economic issues, 39(4): 843-858. fraser, b. w., (1994, december). central bank independence: what does it mean? reserve bank of australia bulletin, 1-8. geraats, p., (2009, february). trends in monetary policy transparency. cesifo. working paper no. 2584. geraats, p., eijffinger, s., and van der cruijsen, c.. (2006, february). does central bank transparency reduce interest rates? centre for economic policy research. discussion paper no. 5526. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 2 39 grilli, v., masciandaro, d., and tabellini, g., (1991). political and monetary institutions and public financial policies in the industrial countries. economic policy, 13: 341-392. gruben, w., and welch, j., (1993). interpreting central bank independence in mexico. federal reserve bank of dallas. gujarati, d. (2003). basic econometrics. (fourth edition). mcgraw-hill publishers. gutiérrez, e., (2003). inflation performance and constitutional central bank independence: evidence from latin america and the caribbean. imf working paper. wp/03/53. international monetary fund. www.imf.org. jácome, l. and vázquez, f., (2005). any link between legal central bank independence and inflation? evidence from latin america and the caribbean. imf working paper. wp/05/75. maxfield, s., (1997). gatekeepers of growth: the international political economy of central banking in developing countries. princeton, nj: princeton university press. poole, w., (2008, july/august). the importance of being predictable. review. federal reserve bank of st. louis, 415-419. sirivedhin, t., and hataiseree, r., (2000, january). central bank independence: a thai perspective. seanza advisors’ meeting. winkler, b., (2000). which kind of transparency? on the need for clarity in monetary policymaking. ecb working paper, 26. griffin: central bank independence http://www.imf.org/ present and future value formulae for uneven cash flows based on performance of a business 67 the international journal of banking and finance, volume 8 (number 2), 2011: pages 68-73 present and future value formulae for uneven cash flow based on performance of a business ameha tefera tessema construction and business bank, ethiopia ___________________________________________________________ abstract computational methods for present and future value calculations are difficult when the firm's cash flow income is uneven. the firm’s decision to invest or borrow based on uneven cash flows needs a simple method to arrive at the present value of uneven cash flows. this paper is an attempt to simplify the current tedious calculations. a methodology to quickly estimate the present/future value of uneven cash flows is invaluable to practitioners in the banking and financial industries. keywords: uneven cash flow, present value formula, future value formula, performance rate, rate of growth, uneven cash flow income, economic value added jel classification: c02 _____________________________________________ 1. introduction money has different values at different time periods, which is the foundation of time value of money calculations. the literature on this topic is by now historic. among the reasons often cited are: time preference, inflation premium, risk premium, and productivity. compounding and discounting techniques are commonly employed to calculate the time value of money. the present and future annuity formulas developed through such technique are, by definition, for equal amount of periodic cash flows. they have the following shortcomings, particularly in light of a banking institution: they do not easily allow for the calculation of cash flows that grow according to the project’s performance, particularly when such performance is variable. tessema: uneven cash flows 68 they assume constant cash flows throughout the project’s life, which is unrealistic. a project whose performance does not change according to the market situations throughout the project life may be out of tune with the market. they do not consider future inflation rates, which could also be variable. inflation is assumed to occur at a constant rate. since inflation reduces the purchasing power of money, considering the future inflation effect on project cash flows will reduce the firm’s unexpected loss. they do not consider flexibility of deposits in banks or flexibility of payments by creditors to the banks. again, the high possibility of uneven cash flows. this paper attempts to resolve the above problems through offering a simpler method to compute uneven cash flows. 2. business aspects of uneven cash flows we link the sum of the uneven cash flows, either for investment or for repayments of borrowed loans, to the performance of the firm. a new business has a small amount of cash at initial stage. eventually, after covering the establishment cost, the liquidity amount on hand grows accordingly. the growth of the firm’s cash flow will then depend on its performance. when performance improves, its profits and cash flows increase. as result, the depositor can deposit the excess amount above his/her business consumption. this excess may increase from period to period according to the business performance at a later stage of profitability. traditional finance measures of firm’s performance are profits, earnings per share (eps), return on investment (roi), free cash flows (fcf), capital productivity (kp), labor productivity (lp), and return on capital employed (roce). each ignores cash and cost of capital so as to generate the target profit (brealey and myers, 2003: 414). a more complete measure of performance of a business firm is economic value added (eva) which provides the money value created for investor in a given period of time by weighting the profit generated against the cost of capital employed (firer, ross, westerfield and jordan, 2007). international journal of banking and finance, vol. 8, iss. 2 [2011], art. 4 69 since eva considers the amount of capital invested, the return earned on capital and the cost of capital (wacc) – reflecting the risk adjusted required rate of return it is thought to have all the characteristics of a complete measure though it is valid only for short period of time. furthermore, since eva is a measure of both performance and value, it is assumed to be a method to determine the value created, above the required return, for the shareholders. the firm creates wealth for the shareholders when the revenue of the firm exceeds the cost of doing business and the cost of capital. a business firm creates value for its shareholders on the basis of positive eva rather than simply making accounting profits. the positive magnitude of eva indicates the business firm is improving its net cash return on invested capital. the incremental eva from year to year will result in a change in the market value of the firm (damodaran, 2001). accounting information about the firm’s performance for a single accounting period is valuable. a manager with good knowledge about the firm’s performance may predict future performance of the firm on the basis of the past financial statements with some degree of accuracy. if records of past trend are available, performance rates can be determined to project the cash flows of the firm investment and of the firm bank loan repayments (alexander hamilton institute, 1998: 64-65 and 331-332). since the firm's performance rate is often assumed to be progressive, the cash flow of the firm can be assumed to grow from period to period. performance rate is a percentage by which the current performance of a firm (in this case, eva) grows from the previous period. 3. methodology assume a business firm decides to borrow money from bank or creditor. this often involves receiving cash now in exchange for a promise to make payments in the future periods. the loan repayment amounts usually are set to include interest and a portion of the principal. based on this notion, each of the next period’s repayment amount is calculated by considering the preceding repayment, the project performance, and the interest rate. at initial repayment period, the repayment amount is assumed to be small and fixed without performance consideration. however, the second period repayment amount grows tessema: uneven cash flows 70 according to the business performance. a business performance measuring tool, eva, can be calculated as follows: eva= [adjusted net operating profit after taxes (anopat)] – [capital], or eva = (roce – wacc) k where, roce is a return on capital employed, wacc is the weighted average cost of capital, and k is capital employed. from the above, a business firm can increase its eva by increasing the nopat generated by existing capital, reducing the wacc, and investing in new projects where the return on capital exceeds the wacc. assume that the first small repayment amount to be 1a the following repayments can be calculated as: 12211 )1)....(1)(1(   nn pipipipa (1) such that repayment period repayment amount 1 1a 2 )1( 11 ipa  3 211 )1( pipa  .. . . n 12211 )1)....(1)(1(   nn pipipipa where n = loan period, pn= performance rate, any arbitrary figure greater than zero, which can be measured as the percentage by which the (n+1) th period eva grows from the n th period eva. that is: international journal of banking and finance, vol. 8, iss. 2 [2011], art. 4 71 n nn n eva evaeva p   1 (2) the reason for the choice of the performance rate as is expressed in relation (2) is to prevent the cash flows from exaggerated changes. evan is projected net income for the nperiod loan; and i is the bank interest rate which is added to the performance rate since the n th period repayment should contain a portion of the principal amount and the interest amount for the period. it is in points, i.e., between 0 and 1. the present value of the future cash flows of the loan repayments can be expressed as:   n nn n n r n r i pipipipipa i pipipa i pipa i ipa i a i a iip )1( )1)...(1)(1)(1( .................... ............. )1( )1)(1( )1( )1( )1( )1( 1)1( )1()1( 123211 4 3211 3 211 2 1111 1 1 1                                    (3) another aspect of a firm is a decision to invest cash now in order to receive cash, goods or services in the future. let the periodic cash investment be as: period investing cash 1a 1 2 a1 (w1-t1) 3 a1 )1( 1 p ) (w2-t2) . . . n a1 )1)(1)...(1)(1( 2321   nn pppp ( w n-1 t n-1) where 1 and 0between 1)(i)(pfor ))(1(1t , 1 nn  ippw nnn (i.e., 0< (pn+1)(i)<1). tn = is the firm performance excluding bank’s deposit interest. the deposit interest rate tessema: uneven cash flows 72 should not be included into the project performance since it is always less than the bank lending interest rate. i = bank interest rate which is between 0 and 1. n = investing period .   11 nn tw a portion of performance rate by which the excess above consumption will be saved in a bank account. therefore the future value of investment cash flows is:           n nn n r n r itwppa itwpaitwaiaiipia 1)1...(p)1)(1(.............. ...........1 )1(11)1)(1()1( 112-n211 3 2211 2 1111 1 1 1             (4) where: )....(a........., ),(a ),(a , , 1211n32134212311211   nn aaapaaapaapapaaa such that; )1)...(1)(1)((a...., ..... ),1)(1)((a ),1)((a ),( , 22111n 21134112311211    nn pppap ppappapapaaa (5) let the th n period cash flow )( n a , performance rate )( n p , and the discount rate )(x belong to the domain of any numbers that can mathematically be expressed as elements of the real numbers of set of r, i.e., .,, rpxa n  elements).for stands ( where then, the sum of cash flow for n periods can be explained as follow: 121 1 1 ............     n n m m aaaa this can be expressed as     1 1 1 n m m x xa a 0 for x , .......... 1 1 1 2 2 2     n n n x xa x xa splitting the terms as: )).....().....a (( 1 ................. ......))())(( 1 ))()(( 1 )0( 1 1 1 11 1 1n1 1 2 21 2 2 3 3311 2 221 xaxaxaxaxax x xaxaxaxaxa x xaxaxa x xa x n n n n n n n n       (6) international journal of banking and finance, vol. 8, iss. 2 [2011], art. 4 73 collecting the terms in the brackets yields:      xaxaxa xx xaxaxa xx xax xx xa n n n n nn n n n n 1 1 11n 11 1 1321 2 221 ..... x 1 11 ................... 11 a 11                                 xaxaxa xx xa xxxx x xxxxxx xa n n n n nn n nnnnn 1 1 11n 1132 2 213221 ..... x 1 11 .............. 11 .... 11 a 11 ... 1111                         xaxaxa xxx xa xx xa xx xa n n n nnnn n nnn 1 1 11112 2 211 .... 111 .......... 1111                       xaxaxa xx x xa x x xa x x xa n n n nnn n nn n n n 1 1 1111 1 2 211 .... 11 ................... 11                               (7) multiplying both sides by :gives 1, for x , 1 1    x x n        xaxax xx x xa x x xa x x xa x a n n nn n nnn m m 1 1 1n 1 2 21 1n1 1 .....a 1 1 1 1 ....... 1 1 1 1 1 x                                     since the formula of geometric progression at ratio = x can be expressed as: 1 ....1 1 1     n n xx x x ( sobel and lerner,1995:515-535) , it follows that:        1 1 2 21 32 2 2 1 .... 1 1 ..................           n n n n nn xaxaxa x xaxxxaxxxa equivalently this can be expressed as:                     1 1 ............... 1 1 2 2121 3 321 2 211 x xaxaxaxaaaxaaaxaaxa n n n n (8) tessema: uneven cash flows 74 let 0 1  n a , then it follows that:            9..... x 1 ..........x 1 x 1 x 11 21 3 321 2 211 1 n n n n z z x x aaa x x aaa x x aa x x a x x xa                                            from this expression let us assume that the cash flow of a project for each period can be defined as )....( 13211   nnn aaaapa which can be written as )....(a........., ),(a ),(a , , 1211n32134212311211   nn aaapaaapaapapaaa or: )1)...(1)(1)((a...., ..... ),1)(1)((a ),1)((a ),( , 22111n 21134112311211    nn pppap ppappapapaaa such that pn + 1  or pn and p0 = 1. substituting this, we have: n nnn x x x paaaaaaax x x paaaaaa x x x paaaax x x paax x x a                                       1 )......().....(.......... 1 )()( 1 )()( 11 11211321 4 3321321 3 22121 2 1111          1 1)1().....(.............. .... 1 1)1()( 1 1)1( 1 1)1()( 1 11321 4 3321 3 221 2 11 2 1             x x xpaaaa x x xpaaa x x xpaa x x xpa x x a n nn multiplying both sides of the equation by x-1, we obtain:            )11)x((p.... ...................... ......)1)1((1)1(p x1)1( 1-n121 4 3321 3 221 2 11 2 1 1 1 n n n n m m xaaa xxpaaaxxaaxpaxaxa            international journal of banking and finance, vol. 8, iss. 2 [2011], art. 4 75            )10( .................................................... )1)x((p.... ...................... ......))1(()1(p )1( 1 1-n121 45 3321 34 221 23 11 2 1 1 1 nn n n n m m xaaa xxpaaaxxaaxxpaxaxa             if the bank interest rate (i) is expressed as 1 1   i x , by substituting it in relation (2) and multiplying both sides of the equation by (i+1), we have the following:                     n nn nn m m i ipaa i ipaaa i ipaa i ipa i a i a 1 .......a .................... ............. 11111 1 1121 4 3321 3 221 2 111 1                               …… …………(11) suppose that in the above equation ipp nn  , and adding both sides of the equation )1( 1 i a we obtain the following present value formula:   n nn n n r n r i pipipipipa i pipipa i pipa i ipa i a i a iip )1( )1)...(1)(1)(1( .................... ............. )1( )1)(1( )1( )1( )1( )1( 1)1( )1()1( 123211 4 3211 3 211 2 1111 1 1 1                                    (12) the initial cash flow (a1) in the above formula grows progressively from one period to another by multiplying progressive performance rates of the periods. each of the periodic cash flows contains an interest rate that embodies inflation rate and risk which might happen in the future. furthermore each of the periodic cash flow contains the period performance rate that can move along with the strength of the business. tessema: uneven cash flows 76 now, let us consider equation (10). setting x = (i+1), deducting 2 1 )1( ia from both sides, and then adding both side of the equation )1(1 ia , we have the following future value formula: it is correct                 n nnn n n m m itwaa itwaaitwaiaiiaia 1......a.............. ...........11111 11121 3 2221 2 11111 1 1            this can be written as:           n nn n r n r itwppa itwpaitwaiaiipia 1)1...(p)1)(1(.............. ...........1 )1(11)1)(1()1( 112-n211 3 2211 2 1111 1 1 1             (13) where 1 and 0between 1)(i)(pfor ))(1(1t , 1 nn  ippw nnn (i.e., 0< (pn+1)(i)<1) ; pn = performance rate such that 1np  or np , and tn = is the firm performance excluding the bank’s deposit interest. the rest of the notations are as defined before (see the segment above relation (4)). in sum, we have focused on how business firms determine their cash inflow or outflow based on their economic profit (or economic value added). as shown by the formulae above, the first payment, which is excess above consumption, for payment of debt or for saving into bank account is relatively small. the next after the first amount progresses or grows along with the business firm’s performance rate. since this performance rate shows the relative level of growth of one’s firm, current to last economic profit, it embraces all activities of the firm. as eva fluctuates from period to period, the net cash left to the firm also fluctuates from period to period. these fluctuations are reflected by the rate ( n p ). international journal of banking and finance, vol. 8, iss. 2 [2011], art. 4 77 financing organs, such as banks, use ordinary annuity formula to determine the loan capacity as well as the loan repayment of the borrower. the formula does not contain any measure of the performance of the borrower. some loans become nonperforming. to avoid this situation, banks resort to their own rules and regulations to minimize nonperforming loans. the present value formula stated in this paper calculates the projected fluctuating repayment amounts along with the performance of the borrowing organ based on the real cash on hand, which is the excess above the consumption. the present value formula enable the borrower, who has no excess cash on hand for investment, to plan to save a portion of income according to the earnings growth. this also may prompt plans for saving by those who have low income and those who are salaried. 4. application in ethiopia, most business units use eva as a measure of both value and performance. notable examples are construction firms and banks (colin et al., 2007). the following applications are based upon real past data of an ethiopian bank for illustration. table 1 contains present value and table 2 and the rest future value calculations. tessema: uneven cash flows 67 table 1: computation using the formula developed in this paper, millions as we have discussed above, performance rate of each years is calculated on the basis of the following formula: n nn n eva evaeva p   1 yielding: 18.0 11 )1113( 1   p 08.0 13 )1314( 2   p 14.0 14 )1416( 3   p 75.2 16 )1660( 4   p 10.0 60 )6066( 5   p 12.0 66 )6674( 6   p 11.0 74 )7466( 7   p assume that 10 aeva  is the base year economic value added which can be assumed as an initial amount. interest rate (i) =10%. present value of each year’s eva is calculated as: first year pv of eva     10 110.0 11 1 11 1    i a second year pv of eva years base (year=0) 2003 (year=1) 2004 (year=3) 2005 (year= 4) 2006 (year=5) 2007 (year=6) 2008 (year=7 ) 2009 (year=8) net operating profit after tax 12 15 17 19 63 71 81 74 less: cost of capital 1 2 3 3 3 5 7 8 economic value added 11 13 14 16 60 66 74 66 performance rate 0.18 0.08 0.14 2.75 0.10 0.12 -0.11 pv of eva 10 11.6364 0.8463 1.5887 35.1784 4.4772 5.8611 -5.9588 international journal of banking and finance, vol. 8, iss. 2 [2011], art. 4 68         6364.11 1.1 08.14 110.0 10.0118.011 1 )1( 222 11       i ipa third year pv of eva         8463.0 1.1 1264.1 110.0 )08.0(10.0118.011 1 )1( 333 211       i pipa fourth year pv of eva         5887.1 1.1 3260.2 110.0 )14.0)(10.0108.0(10.0118.011 1 )1)(1( 444 3211       i pipipa fifth year pv of eva         1784.35 1.1 6551.56 110.0 75.2)10.0114.0)(10.0108.0(10.0118.011 1 )1)(1)(1( 5 5 5 43211       i pipipipa sixth year pv of eva seventh year pv of eva         4772.4 1.1 9317.7 110.0 )10.0)(10.0175.2)(10.0114.0)(10.0108.0(10.0118.011 1 )1)(1)(1)(1( 6 6 6 543211      i pipipipipa         8611.5 1.1 4217.11 110.0 12.0)10.0110.0)(10.0175.2)(10.0114.0)(10.0108.0(10.0118.011 1 )1)(1)(1)(1)(1( 7 7 7 6543211       i pipipipipipa tessema: uneven cash flows 69 eighth year pv of eva using present value formula for uneven cash flow, the present value of each period’s eva is easily summed by the formula as follows: present value of eva =10+11.6364+0.8463+1.5887+35.1784+4.4772+5.8611+(5.9588) = 63.6293 the accuracy of this result can be checked by the following present value formula )1( )1()1( 1 1 1 1                  n n r n r i a iip =[(0.18+1+0.10)(0.08+1+0.10)(0.14+1+0.10)(2.75+1+0.10)(0.10+1+0.10)(0.12+1+0.10)(0.11+1+0.10)+   )18(110.0  ][  8110.0 11  ] = 63.6293 (which is the surplus dollar amount on a project). if the bank had projected a cash inflow or outflow standing on a base year, the value of those cash amounts within the given life time would have been the above calculated result. this helps the bank to decide, by comparing with the initial investment, whether the investment is feasible or not. each of the above calculated pv of eva is positive except the last term for 2009, when the bank shows the lowest performance. positive eva increases the value of the firm whereas negative eva reduces the value. as per this paper’s recommendation, the bank         9588.5 1.1 7732.12 110.0 )11.0)(10.0112.0)(10.0110.0)(10.0175.2)(10.0114.0)(10.0108.0(10.0118.011 1 )1)(1)(1)(1)(1)(1( 8 8 8 76543211         i pipipipipipipa international journal of banking and finance, vol. 8, iss. 2 [2011], art. 4 70 should have shown progressive performances, otherwise the formulae detect as the firm has not been performed well in that given period. it is very strong part of the formulae. take now the case of the bank deposits. the main objective of a bank is accepting deposits and lending them to those with liquidity shortage. under the annuity formulas, the loan repayments disregard inflation and risk in interest fluctuations. these may create hardships for the borrowers, including defaults. the future value formula of this paper helps the bank in setting its reserve requirements while considering these factors. starting with the relatively small initial amount that will progress according to the performance of the bank, eva is calculated. using again the above data and based on table 1 above, if the bank had projected its cash reserve deposits by the end of 2009, it would have had the following: initial amount (eva0) = 11 deposit interest rate (i)=10% table 2: computational data using the formula developed in this paper, millions year base (year=0) 2003 (year=1) 2004 (year=2) 2005 (year=3) 2006 (year=4) 2007 (year=5) 2008 (year=6) 2009 (year=7) performance rate   n p 0.18 0.08 0.14 2.75 0.10 0.12 -0.11 1 nn pw 1.18 1.08 1.14 3.75 1.10 1.12 0.89  ipt nn 11  0.882 0.892 0.886 0.625 0.89 0.888 0.911 nn tw  0.298 0.188 0.254 3.125 0.21 0.232 -0.021 tessema: uneven cash flows 71 accordingly, the calculation of the current cash flow to get the future value up to year 2008 is shown below: year cash flow 1 10.12)10.1(11)1(1 ia 2 9664.3)10.1)(298.0(11)1)(( 22 111  itwa 3 2480.3)10.1)(188.0)(18.1(11)1)()(1( 33 2211  itwpa 4 2132.5)10.1)(254.0)(08.1)(18.1(11)1)()(1)(1( 44 33211  itwppa 5 4297.80)10.1)(125.3)(14.1)(08.1)(18.1(11 )1)()(1)(1)(1( 5 5 443211   itwpppa 6 2951.22)10.1)(21.0)(75.3)(14.1)(08.1)(18.1(11 )1)()(1)(1)(1)(1( 6 6 5543211   itwppppa 7 8033.29)10.1)(232.0)(10.1)(75.3)(14.1)(08.1)(18.1(11 )1)()(1)(1)(1)(1)(1( 7 7 66543211   itwpppppa summing the cash investment flows, we have =12.10 +3.9664 +3.2480 +5.2132 +80.4297 +22.2951 +29.8033 = 157.0557. all the current cash investments of each period are positive in magnitude except the last year. the negative sign indicates that the bank performance in this period is less than the prior year. as can be seen above, the steps to arrive at the final result is very tedious and time consuming. however, the future value formula of this paper calculates the above cash flow as easily as follow =11(1.10)[(1.18)(1.08)(1.14)(3.75)(1.10)(1.12)( 7 10.1 )-(0.10)] =157.0557           1 1 1 )1)(1()1( n r n r iipia           1 1 1 )1)(1()1( n r n r iipia international journal of banking and finance, vol. 8, iss. 2 [2011], art. 4 72 5. conclusion the performance of a firm can be projected from its income statement. however, in order to project a firm net income, there should be a record of past trend regarding the firm’s performance. using periodic economic profit within an economic value added (eva) approach, performance rate of each period is measured as the ratio of (n+1) th period’s projected net eva increment to the n th period’s eva. this will result in uneven cash flows. further, it implies that each projected net economic profit and loan repayment occur in the same direction, which in turn implies that projected net income, loan repayment and projected net cash flow will have the same direction. as such, the borrower will be expected not to incur cash shortages since net income, loan repayment, and net cash flow are in the same direction. thus, the borrower is guided to meet his/her loan repayment obligation properly. in most instances the firm’s cash flows are uneven and as such one cannot project the future cash receipts or payments. a good experienced manager may forecast flexible cash flows based on the past data such as income statement and balance sheet of the business firm. as the performance rate of the business moves up or down, the projected cash flow moves in the same direction. in this paper, the performance of a business firm is assumed to be progressive, i.e., accelerated growth. the firm can achieve its growth objective either by expanding its existing market or entering into new market. as the performance rate moves from one period to another, the loan repayments move along with the direction of this fluctuation. hence, we have uneven cash flows. we have shown how the present and future values of such uneven cash flows may be calculated. the method that we have presented may also help investors in projecting their cash inflow or outflow according to their business progressive performance rate. the formulas may also facilitate the creditors or investors decision making process in arriving at realistic measures of firm’s performance. tessema: uneven cash flows 73 author information: ameha tefera tessema is a staff at the construction and business bank s.c, addis ababa, ethiopia. he may be reached via e-mail ambet22002@yahoo.com or 251-912-36-55-49 references alexander hamilton institute, (1998), financial management handbook, global books and subscription services, new delhi, chandra, prasanna,(1993). fundamentals of financial management, third edition, tata mcgraw-hill publishing company ltd, new delhi, india. damodaran, aswath, (2001). corporate finance, second edition, john wiley & sons, inc., usa. brealey, r.a., and myers, s.c., (2003). principles of corporate finance, mcgraw-hill inc., new york, usa. construction and business bank sc: www.cbb.com.et firer, colin, stephen a. ross, randolph w. westerfield and brandford d.jordan, (2007). fundamentals of corporate finance, third edition, mcgraw hill companies, london, united kingdom. pandey, i m., (1999). financial management, eighth edition, vikas publishing house ltd, new delhi, india. mosich, a.n., (1963). intermediate accounting, sixth edition, mcgraw-hill inc., usa. middelton, d.r., (1995). the essence of financial management, prentice hall international ltd, united kingdom. proctor, ray, (2006). managerial accounting for business decisions, second edition, pearson education ltd, england. reilly, r.f., and schweihs, r.p., (2000). handbook of advanced business valuation, mcgraw-hill companies, inc., new york, usa. . sobel, m.a., and lerner, n., (1995). college algebra, fourth edition, prentice-hall, inc., new jersey, usa. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 4 mailto:ambet22002@yahoo.com a review of ipo under pricing phenomenon in asia : 1-26 1 a review of ipo under pricing phenomenon in asia ali albada othman yong* graduate school of business universiti kebangsaan malaysia, malaysia *corresponding author: othmanyo@ukm.edu.my ________________________________________________________________ a r t i c l e i n f o _____________________________ article history: received 30 august 2018 revised 14 november 2018 accepted 18 november 2018 published 6 january 2020 ____________________________ keywords: ipos, initial public offerings, asian ipos, ipo under pricing, regulatory environment. jel codes: g11, g12, g14, g24, g32. a b s t r a c t _________________________________ this paper reviews the current status of ipo research in asia, especially when it comes to the issue of the under pricing phenomenon. this interest stems from the increasing attention that some of the asia pacific region countries have, due to their higher average initial returns in comparison to other developed and developing countries. in this review, we identify the determinants for the high levels of under pricing as reported in the literature. we find that the regulatory environment of these asian countries is the most reasonable source for such under-pricing, as it sets it apart from other developed and developing countries. the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 1 1. introduction this paper is a sequel to yong’s (2007) work, where he reviewed the ipo research in asia before 2007. furthermore, he investigated some of the unique features that are related to the asian ipo markets such as pricing mechanisms, family business ipos, privatization ipos, legal systems, listing requirements, small investors, and price limits. in his review paper, yong also suggested that the ipo research in asia was still lacking in exploring unresolved issues such as the right time for issuing an ipo, the effect of the volume or size of ipos, the difference in asian laws and their enforcement, cultural differences, governmental interference, pricing mechanism, marketing of ipos, and auditor reputation and how they affect ipo under pricing. the purpose of the present paper is, therefore, to review the asian ipo literature since then till the most 2 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 recent one, in order to see how much of the previous gaps have been addressed in the literature. even though this paper is supposed to be a sequel to yong (2007), we still include some of the previous works (before 2007) due to the following reasons: (a) some of these works are either ignored or overlooked by yong (2007); (b) these works are the basis for future related research and thus their inclusion is important in terms of continuity of discussion; and (c) those earlier works, especially in the developed markets, are the basis for some of the future works (or result in future works) in asia. 2. short-term under pricing of asian ipos the ipo under pricing (or positive initial returns) puzzle was documented in many different countries, where the literature showed that under pricing was not exclusive to the united states (us) market. according to heerden and alagidede (2012) and yong (2007), most of the work regarding ipo initial returns had been carried out using data from the us. among the first empirical studies that tested under pricing and reported a positive mean ‎initial return in the us ipo market were ibbotson (1975), logue (1973), and stoll and curley (1970). a study conducted by ritter and welch (2002) documented an average initial return of 18.8% in the us. table 1. empirical evidence of under pricing in the asia-pacific region country avg. initial return* size time period classification sample source asian-pacific region china 118.4% 2,512 1990-2013 developing chen et al. (2008); jia et al. (2014) hong kong 15.8% 1,486 1980-2013 developed zhao and wu*; mcguinness (1992); ljungqvist and yu (2003); fung et al. (2007) india 88.5% 2,964 1990-2011 developing marisetty and subrahmanyam (2008); ritter(2003) indonesia 24.9% 464 1990-2014 developing suherman* japan 41.7% 3,236 1970-2013 developed dawson and hiraki (1985); hebner and hiraki (1993); fukuda (1984); pettway and kaneko; hamao et al. (2000); kaneko and pettway (2003) korea 59.3% 1,720 1980-2013 developed ihm*; dhatt et al. (1993); choi and heo*; mosharian and ng*; cho*; joh*; dealogic* malaysia 56.2% 474 1980-2013 developing isa (1993); isa and yong*; yong*; ma*; dealogic* pakistan 22.1% 80 2000-2013 developing mumtaz et al. (2016) philippines 18.1% 155 1987-2013 developing unite and sullivan (2003); dealogic* (continued) a review of ipo under pricing phenomenon in asia : 1-26 3 country avg. initial return* size time period classification sample source asian-pacific region singapore 25.8% 609 1973-2013 developed lee, taylor and walter; dawson; dealogic sri lanka 33.5% 105 1987-2008 developing samarakoon (2010) taiwan 38.1% 1,620 1980-2013 developed chen (1992); chiang* thailand 35.1% 500 1987-2012 developing lonkani and tirapat*; wethyavivorn and koo-smith (1991); ekkayokkaya and pengniti (2012); vithessonthi (2014) australia 21.8% 1,562 1976-2011 developed lee et al. (1993); woo*; pham*; ritter (2003) new zealand 18.6% 242 1979-2013 developed vos and cheung (1993); camp and munro (2000); alqahtani*; dealogic* * the source of data, from individuals mentioned by loughran et al. (1994), is based on the latest updated version (february 16, 2015) of the website, https://site.warrington.ufl. edu/ritter/files/2015/05/initial-public-offerings-international-insights-2015-05-21.pdf, as visited on june 14, 2018. researchers in different countries had tested the ipo under pricing phenomenon and many of these studies had confirmed a positive initial return in various markets around the world. this overview regarding the under pricing literature was made possible through the latest updated version (february 16, 2015) of loughran’s et al. (1994) table, which presents the average under pricing history for 52 countries. loughran et al. (1994) constructed the table by collecting figures from various studies at that time for the purpose of understanding the differences in under pricing between countries. the original table covered 25 countries from 1960 to 1992. the present review paper is focused on the asian region, and for that reason in this review, we use the table by loughran et al. (1994) to cover only the asian countries (see table 1). china comes in first with the highest initial return in the asian region, reporting an average initial return of 118.4% and covering the period ranging from 1990 to 2013 with a sample size of 2,512 firms. according to moshirian et al. (2010), the high initial return in china was due to the distinctive corporate governance system, security laws, and lack of law enforcement. the asia pacific region has gained particular attention among researchers because some of these countries had documented a higher average initial return in comparison to other developed and developing countries. in an early study, jenkinson (1990) covered the period from 1985 to 1988 to test the ipo performance, by comparing the direct and indirect costs of raising the initial equity finance in the united kingdom (uk), the us, and japan. the study reported that the prices of the new issues relative to the market, after one week of trading, increased by over 12% for the uk, between 9% and 11.4% for the us, and nearly 74% for japan. jenkinson (1990) concluded that the under pricing 4 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 was due to the new equity issues being typically priced at a discount relative to their subsequent trading prices in the after-market. in the case of the uk, the ‎discount was around 7.0%, while in the us, this value was around 10%. in contrast, the japanese ipo prices soared, on average, by nearly 55% after one week. furthermore, ritter (2003) who reported that the average initial return in the us ipo market was significantly lower than the average initial return in the asian ipo markets and substantiated this finding. his study tested the under pricing of ipos in 38 countries, for which among the sample were 11 asian countries, namely china, malaysia, korea, thailand, india, singapore, taiwan, japan, philippines, hong kong, and indonesia with an average initial return of 256.9%, 104.1%, 74.3%, 46.7%, 35.3%, 31.4%, 31.1%, 28.4%, 22.7%, 15.9%, 15.1%, respectively. moshirian et al. (2010) also examined the initial return, where their sample covered the asia pacific region by focusing on the six major markets (china, hong kong, japan, korea, malaysia, and singapore). they employed a cross-sectional analysis to test the initial returns, with a sample consisting of 4,439 ipos and covering the period from 1991 to 2004. they documented a gradual increase in the level of excess returns for ipo under pricing in the asian region. furthermore, moshirian et al. (2010) reported that the abnormal excess returns had been one of the distinguishing features of the asian ipos before the financial crisis of 1997. however, the ipo markets of the asian region had shown significant declines in the abnormal excess returns since the beginning of the 1997 crisis. after the brief decline, the asian ipo markets started to resurrect its previous glory during the so-called “technology boom” period of 1999–2000. moreover, the ipo markets were accompanied by a greater degree of under pricing. the ipo literature also consists of studies that investigated the under pricing of individual asian countries. for example, in an early study in the philippines, unite and sullivan (2003) tested the ipo performance of 104 ipos from 1987 to 1997. they reported an average initial return of 22.7%. in another example of an early study, chi and padgett (2005) studied the chinese ipo market by covering 668 ipos from january 1996 to december 2000. they reported an average 129.2% under pricing in the chinese ipos. other early studies that examined different asian countries include: mcguinness (1992) on hong kong; kim et al. (1995) on korea; mok and hui (1998) and su and fleisher (1999) on china; hwang and jayaraman (1993), pettway and kaneko (1996), and cai and wei (1997) on japan; and koh and walter (1989) and lee et al. (1996) on singapore. in more recent studies such as darmadi and gunawan (2013) reported an average under pricing of 22.20% in indonesia by covering 101 ipos from 2003 to 2011. boonchuaymetta and chuanrommanee (2013) reported an average 18.03% under pricing for thailand covering 153 ipos for the period 2001–2011. moreover, covering a sample of 948 chinese ipos, song et al. (2014) reported an average initial return of 66%. komenkul and siriwattanakul (2016) documented an average initial return of 25.36% for 245 ipos that were issued in thailand between 2001 and 2012. pandya (2016) reported a cumulated average abnormal return of 15.73% for the study period between 2003 and 2013, using a sample of a review of ipo under pricing phenomenon in asia : 1-26 5 183 firms listed on the indian stock exchange. sochi and islam (2018) reported an average under pricing of 198.8%, for 50 ipos listed on the dhaka stock exchange from june 2011 to june 2016. moreover, sochi and islam (2018) found that oversubscription rate and offer size had a substantial influence on the ipo under pricing. finally, cheunga et al. (2018) reported the short-term performance of ipos in hong kong from 1994 to 2014, and their results showed that under pricing magnitude varied annually and ranged from a negative 32.5% to a positive 93.5% for the sample of 938 hong kong ipos. overall, 601 ipos (64.9%) obtained positive initial returns, with an average initial return of 9.6%. 3. regulatory environment and under pricing in an early study related to the regulatory environment, loughran et al. (1994) concluded that the degree of under pricing varied enormously across countries due to the contractual mechanisms used and the characteristics of the firms going public. furthermore, they argued that countries with the highest average initial returns tended to be countries where institutional constraints were binding (the average initial returns at the time of the study for malaysia, korea, and brazil were 80.3%, 78.1, and 78.5%, respectively). meanwhile, countries with the lowest average initial returns i.e. below 10.0% (e.g. canada, france, and the netherlands) tended to be countries in which most of the firms going public were relatively large firms with a long operating history and where the contractual mechanism used had auction-like features. finally, they reported that the move in recent years by most east asian countries to reduce regulatory interference during the 1990s in the setting of offering prices resulted in less under pricing in comparison to the 1980s. moshirian et al. (2010) documented an interesting fact regarding the listing requirements. they concluded that the initial return in the emerging markets, where the listing requirements were lenient, was higher in comparison to the developed markets, where the listing requirements were more stringent. they reported an initial return of 202.63%, 70.30%, and 61.81% for the emerging markets of china, korea, and malaysia, respectively. they also reported an initial return of 21.43%, 34.04%, and 33.10% for the following developed markets in asia, which are hong kong, japan, and singapore, respectively. hence, they concluded that investors could earn abnormal returns by subscribing to the newly issued shares during the first day of listing and sell them at the end of the listing day. in an early study by jelic et al. (2001), which covered a sample of 182 malaysian ipos from 1980 to 1995, they concluded that, in immature ipo markets, initial returns were high. they explained that this was due to a weak regulatory environment and the failure of investment bankers to adequately manage the process of listing new issues. liu et al. (2014) used 963 chinese a-share ipos that went public on the shanghai stock exchange (shse) and shenzhen stock exchange (szse) ipos from 1997 to 2009 to investigate the relationship between under pricing and legal protection, by controlling for time-invariant characteristics of regions. 6 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 they concluded that firms from a province with more developed legal framework experienced less under pricing after controlling for time-invariant, province-fixed effect. furthermore, the extent of under pricing was decreased by the strength of legal protection of property rights. importantly, these tendencies were evident only after the introduction of a book-building system, which provides issuers and underwriters with discretion on offering price determination. zeng and zhang (2015) investigated how family involvement and political connection affect ipo under pricing of private firms. they used handcollected information on private companies listed on the small and medium enterprises (smes) of shenzhen stock exchange and covered the period from 2004 to 2012. they concluded that private firms with top-level political connections and family involvement had lower ipo under pricing. furthermore, they reported that family involvement lowered ipo under pricing in the mideast region of china where the institutional environment was poor. however, the effect of political connection was not significant in the mid-east region of china. moreover, they documented that both political connections and family involvement significantly lowered the ipo under pricing in the eastern region of china, where the institutional environment was well developed. finally, they argued that under pricing for large companies could be reduced by political connections, while for small companies the under pricing could be reduced by family involvement. using a sample of 7,627 ipos issued during 2000–2008 from 32 countries (among these countries were australia, china, hong kong, india, japan, malaysia, korea south, singapore, taiwan, thailand), espenlaub et al. (2016) explored the impact of the legal system on ipo survival. they argued that the legal systems effectively protect investors’ interests, enforce contracts and control information and agency costs of external financing, and not only encourage companies to list their stocks (as shown in previous studies) but also ensured that ipo companies are able and willing to stay listed. they documented that a one unit increase in the legality index constructed by berkowitz et al. (2003), i.e., an increase in the legality index from its value in the us (20.85) to that in switzerland (21.91), increases survival times by over 17%. moreover, they reported that ipos in countries with legal systems of above-median quality [i.e., with values of the legality index above that of hong kong (19.11)] have between two to 2.6 times longer survival times than ipos in countries with below-median legality (such as china, india, korea south, malaysia, taiwan, and thailand). finally, they reported that ipos in countries with legal institutions originating in the uk common law system have nearly 40% longer survival times than those with civil law systems. wadhwa et al. (2016) included all ipos and seasoned equity offerings (seos) listed on the bombay stock exchange (bse) and the national stock exchange from 1991 (the year of financial deregulation) to 2012. they examined whether equity issues through ipos or seos were due to mispricing or growth opportunities. to address this question, they followed the method of rhodes-kropf et al. (2005) to decompose market-to-book ratio (m/b) into a review of ipo under pricing phenomenon in asia : 1-26 7 market-to-value ratio (m/v), to account for mispricing, and value-to-book ratio (v/b), to control for growth opportunities. they documented that the proceeds from equity offerings were used to finance investment opportunities, such as real assets, inventory, and capital expenses. moreover, they argued that the tight regulations and high transparency limited issuing firms from the tendency to exploit investors and motivated them to focus on value-creating activities. finally, they documented that the market penalized firms that issued overvalued stocks to increase their cash holdings, as exhibited by poor long-term stock performances. chen et al. (2015) investigated two inconclusive issues regarding ipo under pricing in the chinese ipo market, due to china being the largest socialist country in the world. the first issue is on whether private firms or state-owned enterprises (soes) under-price their ipos more than private-owned firms. the second issue is on the effect of institutional environment on ipo under pricing. their final sample contained 675a-share ipos and covered the period 1999– 2007. they concluded that soes under-priced their ipos more than private firms. specifically, soes controlled by the central government (csoes) underpriced their ipos 27 percentage points more than private firms, whereas soes controlled by local governments (lsoes) under-priced their ipos 7 percentage points more than private firms. moreover, they used the national economic research institute index marketization (neriim) to proxy for the institutional environment. they found that one index score improvement in institutional environment was associated with a two-percentage-point reduction in ipo under pricing. finally, they reported that a better institutional environment reduces ipo under pricing most effectively for private firms, followed by lsoes, and the least for csoes. 4. significant variables in determining under pricing in asia 4.1 issue-specific factors prospective investors diligently evaluate information disclosed by the issuer through draft offered documents in the prospectus. the investors conduct their assessment of firm valuation and hence have a perception regarding its offer price. in addition to the disclosed information, various issue-specific attributes are evaluated and thus the price perception of investors determines the success factor of the ipo (katti & phani, 2016). 4.1.1 underwriter and auditor reputation yon and park (2009) examined the role of underwriters’ reputations on the ipo pricing process and its effect on subsequent initial returns by analysing 275 ipos between july 2002 and december 2006 listed on the korean securities dealers automated quotations (kosdaq). they documented that underwriters with higher reputations exercised more bargaining power than issuing firms or 8 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 institutional investors in the offer price decision process. on the other hand, the underwriters’ certification role was not sufficiently carried out to build a reputation on price discovery. in a recent study, tong and ahmad (2015) reported a negative relationship between underwriter reputation (big 5 underwriters) and under pricing in the malaysian ipo market, which means that the higher the reputation of an underwriter, the lower the level of under pricing. their study covered 322 ipos, from january 2002 to december 2008. however, in a more recent study, ammer and ahmad-zaluki (2016) found that ipos underwritten by big 5 underwriters, which had a high market share and charged low underwriting spread experienced higher levels of under pricing. furthermore, they reported that under pricing increased in ipos with biased earnings forecasts. finally, they revealed that the more accurate the earnings forecasts were, the more minimised would the asymmetric information be and hence, the less would be the ipo under pricing. their study sample covered 265 ipos listed on the main market of bursa malaysia, for the period between 2002 and 2012. espenlaub et al. (2016) using data on 7627 ipos issued during 2000– 2008 from 32 countries, conducted a study on the indirect impact of the legal systems on the survival of ipos, through enhancing the effectiveness of ipo certification by venture capital (vc) investors and reputable underwriters and auditors. they documented that a well-functioning (above-median) legal system helped in amplifying the effect of ipo certification on ipo survival, through doubling the certification effect in the case of vc certification and tripling it in the case of underwriter certification. furthermore, they concluded that the effectiveness of the auditor certification depended almost entirely on a wellfunctioning (above-median) legal system. sundarasen et al. (2017) examined the influence of underwriters and auditors’ reputation on ipo initial returns in malaysia, using a sample of 228 ipos for the period 2005–2012. the results showed that auditors’ reputation had a significantly positive relationship with ipo initial returns. reputable auditors increased the demand for the ipos in the secondary market because they were perceived by investors as providers of true and fair value of the audited reports. this subsequently caused the closing price on the first day to increase, thus influencing the initial returns. nevertheless, after the global financial crisis of 2008, a negative relationship was documented. as for the relationship between underwriters’ reputation and ipo initial returns, a negative relationship was documented. reputable underwriters were cautious in their reputation capital and thus avoided overly under pricing the ipos and leaving issuers money on the table. the negative relationship could also be due to investors’ perception that reputable underwriters would have placed the offer price of ipos as close as possible to the market/fair value of the shares. these were most evident during the crisis. rumokoy et al. (2017) utilized the social network analysis to investigate the effect of underwriter network centrality within the investment banking industry on ipos outcomes in the china’s primary market between 2006 and 2012. their study sample covered 1,157 chinese firms issuing a-share ipos and a review of ipo under pricing phenomenon in asia : 1-26 9 the sample was retrieved from the china and stock market accounting research (csmar) database. they documented that the network centrality improved the underwriter capability of drawing a high number of institutional investors through effectively tapping into peers’ institutional client base. this would lead to higher under pricing, because the underwriter needed to compensate these institutions for their private information. furthermore, they documented that ipos underwritten by central underwriters outperformed other ipos that were underwritten by less central underwriters. finally, they concluded that hiring a more central underwriter could be used as a signal to indicate the quality of the issuing firm. khurana et al. (2017) investigated the effect of auditors’ reputation on ipo under pricing using a comprehensive sample of 14,029 ipos from 37 countries over a period from 1995 to 2014. their study sample covered the following asian countries: australia, china, hong kong, india, indonesia, japan, malaysia, new zealand, philippines, singapore, taiwan, and thailand. they documented, after controlling for country, industry, and year-fixed effects, that issuing firm with big 4 auditors were associated negatively with under pricing, especially in ipo markets where investor protection laws were weak. furthermore, they argued that issuing firms in countries plagued with weak legal regimes could use auditor reputation as a signalling tool to signify their quality and reducing the cost of issuing new equity. jiao et al. (2017) examined the benefits of overallotment option (oao) to issuers in the japanese ipo market by covering 1,008 ipos listed on all japanese stock exchanges and markets from january 2002 to december 2011. they documented that oao benefited the issuers through allowing them to have lower total issuing cost, more accurate pricing, larger issue size relative to shares outstanding, and better post-ipo performance. in other words, oao option helped in reducing under pricing cost on the issuing firms. mohamed and saadouni (2018) examined the impact of incentive fees in mitigating conflicts of interest between the ipo firms and their underwriters, using a sample of 285 ipos listed on the hong kong stock exchange from january 2003 to december 2014. they documented that a high-quality firm which was characterized to be large in size, capable of raising more proceeds, had lower growth opportunities, went public during high market volatility, and were underwritten by reputable underwriters, offered higher incentive fees. furthermore, they categorized a firm with the previous characteristics to be a less risky ipo and tended to use incentive fees at the time of listing to minimize under pricing and underwriters’ compensation. 4.1.2 board structure darmadi and gunawan (2013), using 101 indonesian ipos from 2003 to 2011, documented that board independence was positively related to the level of under pricing. moreover, they reported that there was a negative relationship between board size and institutional ownership and the ipo under pricing. they 10 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 argued that these two governance mechanisms play important roles in mitigating information asymmetry between the issuer and the potential investors. they also found that ownership concentration was insignificant in explaining the first-day returns. finally, the type of corporate control revealed that government controlled companies tended to experience higher under pricing. in the case of malaysia, yatim (2011) examined the relationship between board structure and initial return using 385 malaysian ipos from 1999 to 2008. in her study, she looked at four board structure variables, which were board independence, dual leadership structure, board reputation, and board size. she concluded that due to the risky nature of ipo firms, malaysian investors did not favour the separation of board chair and ceo positions, instead preferred stability and continuity in executive leadership. this resulted in a positive and significant relationship of dual leadership structure with the initial return. however, yatim (2011) reported a negative correlation between board reputation and the initial return. finally, both board size and board independence were not significantly correlated with the initial returns. in a more recent study, badru et al. (2017) examined the impact of board characteristics on the amount of capital raised through an ipo for a sample of 220 malaysian ipos over the period from 2005 to 2015. using ols regression, they reported that board with ethnic malay directors had a significant and positive association with the amount of capital raised, while a weak significance was found for board size. however, using the robust regression techniques, they managed to document that other than board ethnicity, other board characteristics namely board size, board independence, and ceo duality were significantly associated with the amount of capital raised. mcguinness (2016) examined the economic benefits of board gender diversity for the state and privately controlled firms in the hong kong ipo market. mcguinness (2016) concluded that the pursuit of such board gender diversity was value enhancing in relation to the longer-run performance of initial public offerings (ipo) stocks, especially where female board members were unencumbered by family-connection with other directors. furthermore, mcguinness (2016) reported that there was little evidence of a link between female board representation and ipo under pricing. on the issue of board structure, a recent study by handa and singh (2017) investigated the signalling effect of board structures and ownership attributes on ipo under pricing using 404 indian ipos. they concluded that board size and board committees exhibited a significantly positive relationship with ipo returns on listing day. in indian markets characterized by concentrated familyowned firms, promoter ownership did work as an effective signal for investors who took cues of firm potential from ownership patterns. finally, they reported that corporate governance measures had a minuscule contribution in explaining the under pricing of indian ipos and indicated that investors did not incorporate these as a major consideration in their investment decision. xu et al. (2017) investigated the effect of boardroom heterogeneity on ipo under pricing. boardroom heterogeneity is characterized by the functional background, educational background, age, and length of tenure. they covered 355 firms listed on china’s growth enterprise market from its beginning in 2009 a review of ipo under pricing phenomenon in asia : 1-26 11 until 2012. in their study, they concluded that functional heterogeneity and age heterogeneity had a significantly negative relationship with ipo under pricing. however, educational heterogeneity had a positive relationship with ipo under pricing, while heterogeneity in tenure did not have a significant relationship with ipo under pricing. furthermore, they argued that functional, educational and age heterogeneity conveyed signals to potential investors regarding a firm’s quality. hanafi and setiawan (2018) studied 182 ipos listed on the indonesia stock exchange from 2006 to 2015, to investigate the effect of ownership concentration and institutional ownership on ipo under pricing. they documented that ownership concentration did not have an effect on ipo under pricing, while institutional ownership had a negative relationship with ipo under pricing. furthermore, they investigated whether the effect of institutional ownership was not constant across the level of institutional ownership. they found that the negative effect of institutional ownership on ipo under pricing was stronger below the 0.8 level, and the negative effect disappeared above that level. they suggested that the monitoring effect of institutional investors disappeared in a high level of institutional ownership due to the increase in the “principal-principal” conflicts. finally, they argued the importance of ownership in companies’ affairs from the policy perspective, and that this might affect ipo under pricing. cheunga et al. (2018) examined the effect of corporate governance factors on ipo under pricing, by focusing on the multiple roles of the founder in hong kong. they categorized ipos into four groups based on the role of the founder, which are (1) no-founder firms (companies with no specific founder); (2) purefounder firms (companies whose founder is neither the company’s chairman of the board nor its ceo); (3) founder-chairman/ceo firms (companies whose founder is either the company’s chairman of the board or its ceo); and (4) founder-chairman-ceo firms (companies whose founder is the chairman and ceo). they found that the level of under pricing differed among the four categories, where the under pricing levels of no-founder, pure-founder, founderchairman/ceo and founder-chairman/ceo firms were 14.9%, 9.5%, 8.4% and 7.7%, respectively. 4.1.3 lock-up period in their study, boonchuaymetta and chuanrommanee (2013) were interested in investigating the relationship between underwriter reputation, ownership concentration, book-building, ipo allocation, the length of the lock-up period, and investor interest and ipo under pricing in the thailand ipo market. they covered 153 ipos listed between 2001 and 2011. they found that ipo allocation was the strongest factor in influencing under pricing in a negative relationship. they also found that the length of the lock-up period, issue size, industry, and hot issue market had significant positive relationships with under pricing. however, underwriter reputation was not associated with under pricing as the choice of underwriter was restricted by thai regulator’s requirements. moreover, they 12 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 reported that institutional investors played very limited roles in explaining under pricing in thai ipos. finally, they argued that provisions that could restrain insider dealings such as a longer lock-up period could yield a higher initial return and changing the ownership concentration by a small percentage did not affect under pricing. mohd rashid et al. (2014) used a sample of 384 ipos listed on bursa malaysia between 2000 and 2012 to investigate if the lock-up ratio and lock-up period had a positive relationship with the initial return in the malaysian ipos market. in their study, they showed that the effect of the lock-up period was more pronounced than the effect of the lock-up ratio, which they considered to be preliminary. they documented that both the lock-up period and the lock-up ratio were more appropriate for signalling the quality of the issuing firm than the risk of the issuing firm. furthermore, they concluded that the relationship with the initial return was more pronounced in the case of lock-up periods rather than lock-up ratios, and lock-up periods were more appropriate for signalling the quality of the firm. huang et al. (2015) used a sample of 218 taiwanese ipos from 2005 to 2011 to investigate the behaviour of institutional investors during announcement day and expiration day of lock-up. they found that the trading of institutional investors produced insignificant abnormal returns on announcement lock-up day, but the cumulative return on announcement day was significantly different from zero. moreover, they found that the relationship between lock-up ratio and institutional investor on announcement day was insignificant. finally, they examined the relationship between the lock-up ratio and the institutional investors’ influence on the initial return of ipos. they found that the abnormal trading of foreign institutional investors and dealers produced an insignificant initial return. recent studies regarding the lock-period are more interested in investigating the effect of the lock-up provisions on trading volume changes around lock-up expiry dates (zameni & yong, 2016) and on trading volume changes around lock-up expiry dates (zameni & yong, 2017). zameni and yong (2016) sample comprised of 379 malaysian ipos, issued from january 2001 to december 2011. they reported a positive abnormal trading volume at lock-up expiry date for ipo market, except for the ace market, which is the acronym for “access, certainty, efficiency”, the construction sector, and the technology sector were negative. they argued that the high trading volume at and around the lock-up expiration was compatible with shareholders’ selling due to diversification reasons and wealth recognition and which was also an indication of insiders’ lack of confidence about a company’s future prospect. furthermore, the significant negative trading volume can be interpreted in a way that insiders of those related boards and sectors do not sell their shares significantly but would rather watch what would happen to the market and are optimistic about the market’s future. zameni and yong (2017) documented that a review of ipo under pricing phenomenon in asia : 1-26 13 the share price drop around the lock-up expiry was due to the increase in the adverse selection element of the bid-ask spread, which resulted from trading volume increase around the lock-up expiry. in australia, both mandatory lock-ups (ml) and non-mandatory lock-ups (nml) firms co-exist. this is because, in the australian regulatory setting, ml is applied to insider shares of ipo firms that do not satisfy profit or asset tests. taking advantage of this unique ipo policy, haman et al. (2017) investigated the effect of different lock-up types on long-run returns (buy-hold abnormal return at the fifth year post listing). their study sample consisted of 571 ipo firms listed on the australian securities exchange (asx) and covered the period from 2003 to 2007. their sample contained 460 firms with ml only and 111 nml. they indicated that the average market-adjusted under pricing of ml firms was significantly lower than for nml firms, at the one percent level. furthermore, they reported that an increase in market-adjusted returns by one percent decreased long-run returns by 0.47 percent. moreover, they showed that insider equity ownership subject to lock-ups, the weighted average number of days of the lock-ups, the natural logarithm of the number of days from the registration of an ipo prospectus to the ipo listing date, and reputable underwriter had a positive relationship with long-run returns. on the same subject, their study concluded that long-run returns for ml firms were significantly lower (higher) than for nml firms, and good corporate governance was positively associated with long-run returns for both lock-up type firms. finally, they found that the listing survival rate of ml firms was higher than for nml. taking the advantage of regulatory change on the ipo lock-up for institutional investors and a unique disclosure of institutional investor bidding information in china, gao et al. (2017) investigated the impact of ipo lock-up removal on ipo pricing. their study sample consisted of 474 chinese ipo from november 1, 2010, to december 31, 2012. they documented that institutional investor bid price increased due to the ipo lock-up removal. furthermore, they found that the effect was stronger for sub-samples with less reputable underwriter and ipos with higher value uncertainty. they concluded that the impact was channelled to a higher ipo offer price and a lower first-day return. their results suggest that ipo lock-up removal (ipo lock-up) increases (decreases) ipo share valuation. 4.1.4 other factors yong (2013) used 357 ipos and covering the period from january 2001 to december 2008 to investigate the issues of winner’s curse, size effect, and bandwagon effect in explaining the under pricing phenomenon in the malaysian ipo market. he concluded that the average initial return for the malaysian private placement ipos (a proxy for informed investors) was significantly lower than that of the non-private placement ipos (a proxy for uninformed investors), which supported the winner’s curse hypothesis, where uninformed investors demanded a higher initial return in the absence of informed investors. this finding is further 14 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 supported by mohd rashid et al. (2014), who concluded that institutional investor participation had a significantly negative coefficient, indicating that uninformed ‎investors (non-private placement subscribers) tended to demand higher initial returns compared to informed investors (private placement subscribers). finally, on the issue of ipo flipping (the immediate selling of an ipo by the successful subscribers upon ipo listing in the open market), che-yahya et al. (2014) examined the influence of institutional investors’ participation on the flipping activity of malaysian ipos. they concluded that the greater participation of institutional investors during an ipo was expected to be an effective strategy to control aggressive flipping activity; this means that the greater participation of institutional investors during an ipo is expected to be an effective strategy to control aggressive flipping activity. deng and zhou (2015) employed 355 ipos listed on shenzhen stock exchange (szse), covering the period from october 30, 2009, to december 31, 2012, to re-examine the determinants of initial returns for chinext (a nasdaqstyle board of the szse) ipos that were initiated under the new chinese-style book-building process in a more market-oriented environment. they managed to identify three factors that drove the initial under pricing of chinext ipos: the pre-issue share allocation multiplier from institutional investors (offline oversubscription), issue size (size effect), and the listing day stock market conditions (market momentum). kang et al. (2015) examined the effects of insiders’ ownership levels and ownership changes in pre-ipo periods on ipo under pricing and insiders’ shareselling behaviours after lock-up expirations, and paying special attention to the form of ownership: direct ownership and indirect ownership through affiliated firms. their study sample consisted of 669 ipo listed on the korea exchange (krx) and covering a period from january 2000 to november 2011. they found that insiders’ ownership level was positively related to under pricing. furthermore, they documented that ipo under pricing was influenced by direct ownership and not by indirect ownership. moreover, they reported tha-t both insiders’ direct ownership level and the increase in insiders’direct ownership in pre-ipo periods helped in increasing the probability of insiders’ share selling after the lock-up. this implies that insiders can strategically use their direct ownership in share-selling activities during the ipo process in their effort to improve their wealth. banerjee (2016) investigated the effect of company-specific and market sentiment related factors on the under pricing of indian ipos. company-specific factors include the pre-issue financial position, corporate governance, and post issue promoter holding, firm age at the time of the issue, reputation of the lead manager of the issue, reputation of the credit rating agency, ipo grade, and the appetite of the retail and institutional investors. market sentiment related factors were the change in the money supply, foreign institutional investors’ (fii) inflows, price to earnings (pe) ratio of the market, and market return. out of the 16 factors they examined, only nine factors had statistically significant effects on under pricing of ipos. firm age, debt-to-equity(de) ratio, and investment a review of ipo under pricing phenomenon in asia : 1-26 15 bank’s (ib) reputation had a negative relationship with under pricing, while a number of independent directors, pe ratio, money supply, market return, returnon-net worth (ronw), and ipo grade had a positive relationship with ipo under pricing. choie (2016) investigated the causes that led under pricing to be higher in the korean ipo market in comparison to the us ipo market. choie (2016) first used the optimal offer price from the perspectives of both the issuing firm and its underwriting investment bank. choie (2016) concluded that the ipo price was likely to be lower than the actual initial market price because both issuing firms and its underwriters would tend to agree upon ipo price below the expected initial market price. furthermore, he concluded that the magnitude of ipo under pricing in the korean market had a positive relationship with the variance of the probability distribution of the initial market price, and the marginal cost of making the after-market. moreover, the underwriter’s spread had a negative relationship with the ipo under pricing. gumanti et al. (2017) used a sample of 290 indonesian ipos that went public between 1989 and 2005 to examine the number of risk factors, use of proceeds for investment, size of the issue, and the level of ownership retained and their effect on ipo under pricing. the number of risk factors was found to be positively related to the level of average positive initial returns (under pricing). the level of ownership retention had a negative but insignificant relationship with the level of under pricing. firms that used the proceeds from the offering for investment or expansion purposes were less under-priced than their counterparts that used the funds for operating purposes. furthermore, size of the issue was negatively associated with the level of under pricing. chang et al. (2017) took advantage of a unique pre-ipo market in taiwan that had an organized trading platform and was mandatory for firms aiming for an ipo, and in january 2005 firms had to be traded on the emerging stock market (esm) in taiwan for at least six months before they applied for an ipo. their sample consisted of 218 firms that went public between october 25, 2005, and march 1, 2011. they concluded that ipo offer price was largely dependent on the pre-market price. the pre-market price-multiple alone explained about 90% of the variations in the offer price-multiple. furthermore, they found that peers’ prices were no longer important in determining the offer price after taking into account the issuer’s own pre-market price. moreover, the more volatile or riskier the stock, the less informative the pre-market price and the greater the discount taken in setting the offer price relative to the pre-market price (and similarly, a higher first day return). however, despite the informative pre-market price, they documented that the level of ipo under pricing was still at a high average level, 55.3%, in contrast to the prediction of asymmetric informationbased book building theories. they argued that the high level of ipo under pricing was due to the underwriters’ monetary incentives to under-priced shares, where the stronger the underwriter’s incentives to under-price and the stronger their bargaining powers, the higher the under pricing. finally, building on the previous argument they concluded that agency problems could lead to high 16 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 levels of ipo under pricing, even with little information asymmetry or valuation uncertainty about the stock. 4.2 country-level factors banerjee et al. (2011) investigated the influence of four country-level characteristics on ipo under pricing. these characteristics are differences in level of information asymmetry (analyst following and stock price synchronicity are the proxies), investors› home-country bias (home-country bias is the proxy), effectiveness of contract enforcement mechanisms (country-level anti-selfdealing index is the proxy), and accessibility of legal recourse (country-level prospectus liability index is the proxy). their sample included 36 countries, among them, are china, hong kong, india, indonesia, japan, singapore, malaysia, taiwan, and thailand. the sample consisted of 13,386 international ipos from january 1, 2000 to december 31, 2006. they concluded that there was a positive and significant effect of country-level information asymmetry on ipo under pricing. secondly, they found that lower cost to entice the block holders, measured by domestic investors› home-country bias, reduced ipo under pricing. thirdly, effective contract enforcement mechanisms helped in reducing ipo under pricing. finally, they documented a positive relationship between the accessibility of legal recourse and ipo under pricing. with regard to country-level factors, vithessonthi (2014) used a sample of 187 ipos in thailand between january 2000 and december 2010 to investigate whether the financial development, domestic interest rates, and interest-‎rate differentials simultaneously explains the variations in the short-run ipo under pricing. the study documented that stock market development drove the initial return of the ipos by: (1) lowering the level of information asymmetry; and (2) attracting foreign investors into the ipo market. furthermore, the stock market’s p/bv (price-to-book value) ratio was positively related to the initial return. vithessonthi (2014) argued that the development of stock markets generally resulted in lower initial returns, while favourable stock market conditions led to higher initial returns as they shifted the demand curve for the ipos outwards. moreover, they reported that the t-bills market capitalization, government bond market capitalization, and private bond market capitalization, which were used as an indicator for the development of the bond markets, had a negative effect on the initial return. in addition, the amounts of new t-bills and government bond issuances had a negative effect on the initial return, suggesting that all else equal, the development of bond markets might cause an inward shift of the demand curve for the ipos, thereby lowering the initial returns. finally their results showed that interest-rate differentials had a positive effect on the initial return when the interest-rate differential was positive, suggesting that ipos listed in the high interest rate differential environment were likely to experience higher initial returns. boulton et al. (2017) covered 13,285 ipos from 36 countries to examine the influencing effect of accounting conservatism on the under pricing. they a review of ipo under pricing phenomenon in asia : 1-26 17 reported that ipos were under-priced less in countries in which existing public firms practiced more accounting conservatism. according to a study by boulton et al. (2017), the following asian countries were ranked from the most to the least conservative in financial reporting practices: new zealand, australia, thailand, hong kong, philippines, taiwan, india, singapore, malaysia, japan, and indonesia. new zealand had the lowest ipo under pricing and indonesia had the highest ipo under pricing. furthermore, they examined the relationship between conservatism and under pricing alternative measures of conservatism, such as country mean regressions, sample country exclusions, and endogenous treatment models. they concluded that higher country-level conservatism was associated with lower country-level information asymmetry, which explains the lower ipo under pricing. they also found that common country-level factors, such as legal origin, influenced the relation between accounting conservatism and ipo under pricing. 4.3 ipo pricing method reddy (2015) covered 133 ipos, issued through book-building route, during january 2007 through december 2009 and which were subsequently listed on the national stock exchange of india ltd (nse). the study reported negative returns in the long run and concluded that under pricing had occurred in the indian ipos market. the results inferred that post-market offers assured positive returns in the short run but tended to plunge and became negative in the long run. however, boonchuaymetta and chuanrommanee (2013) concluded that bookbuilding pricing method did not explain under pricing in thai ipos. chen and wu (2015) developed a measure of the expected net wealth gains of pre-ipo shareholders to investigate who controls the decision for choosing the issuing method (auction, or fixed-price public offer): is it the underwriter or the issuer? their study covered 712 ipos listed either on the taiwan stock exchange (twse) or on taiwan’s over-the-counter market (otc) during the period of 1996–2003. their sample consisted of 623 ipos that used the fixed price public offer. the remaining 89 ipo firms employed the hybrid auction procedure. they concluded that the choice for the issuing method was mostly dependent on the issuer, because issuers took into consideration their firm’s characteristics and the principle of wealth maximization. furthermore, they documented that under pricing under the fixed-priced method was higher than the auction method. however, they argued that most issuers were more concerned with goals that maximized the dynamic objective function rather than minimized under pricing. dhamija and arora (2017) reported that the level of under pricing for the small and medium enterprise (sme) ipos listed on the bombay stock exchange (bse) and mumbai and the national stock exchange (nse) platforms, were found to be lower than that of ipos listed on the main board stock exchanges 18 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 in india. this may be partially due to the fact that the sme platform was at its infancy stage and had failed to attract investors’ interest. this was reflected in a low level of oversubscription of sme ipos at 1.35 times on average. furthermore, they reported that the type of offer, size of the issue, promoter holding, the extent of oversubscription, lead manager prestige and the stock exchange of listing as the key determinants in the under pricing of sme ipos. post listing, these ipos had significantly outperformed the benchmark index. their study covered 100 ipos starting from february 23, 2012 (when the first such ipo was made) to march 31, 2015. with regard to ipo mechanism, huang et al. (2017) investigated the effects of three initial public offering (ipo) methods (namely fixed-price, auction, and book-building, all of which have been practised in taiwan), on post-ipo performance and the trading behaviour of institutional investors. of these three methods, the empirical results showed that book-building ipos exhibited significantly higher initial returns and institutional ownership of stock offerings, and that auction ipos had higher long-run returns. the price effect of institutional herding and flipping behaviour with respect to fixed-price ipos significantly eroded long-run ipo performance. ipo data were from the 1996– 2012 period and the final ipo sample included 1,198 firms comprising 663 fixed-price ipos, 91 hybrid-auction ipos, and 444 hybrid-book-building ipos. in malaysia, the fixed-priced mechanism is the most popular practice when it comes to pricing an offer price (low & yong 2013; yong, 2015). according to low and yong (2013) and yong (2015), the fixed-priced mechanism helped in increasing the divergence of prospective investors’ opinion regarding the true value of the ipo because the settled price does not reflect their beliefs and expectations regarding the true value of the issue. however, with the bookbuilding and auction offerings methods, potential investors are solicited by the issuers regarding the value of the ipo before the offer price is set up (low & yong, 2013; yong, 2015). not factoring the prospective investors in the offer price will lead to heterogeneity of opinion among investors (yong, 2015; chowdhry & sherman, 1996). the first-day price spread is the most common way to proxy the heterogeneity of investors regarding the true value of an ipo (yong, 2015). low and yong (2013) and yong (2015) documented a significant relationship between under pricing and first-day price spread. they argued that the higher the under pricing the higher the heterogeneous beliefs among investors.yong (2015) investigated the relationship between investor heterogeneity and variables such as initial return, the ratio of the first-day volume over total unit offered, and listing board. he concluded that ipos that were characterised with high initial return, high first-day trading volume, and listed on the ace market suffered from a very high divergence of opinion regarding their true values among the investors due to their speculative nature. moreover, low and yong (2013) documented ipos that were highly under-priced, small in offering size and were listed on the mesdaq market tended to have a high level of heterogeneous beliefs among investors. a review of ipo under pricing phenomenon in asia : 1-26 19 5. conclusion this paper reviews the recent status of ipo research in asia, especially when it comes to the issue of the under pricing phenomenon. this interest stems from the increasing attention that some of the asia pacific region countries have, due to their higher average initial returns in comparison to other developed and developing countries. the review starts with showing the levels of under pricing in different asian countries that is made possible with the help of the latest updated version (february 16, 2015) of the table by loughran et al. (1994) that contains the average under pricing history for 52 countries. furthermore, based on the updated table by loughran et al. (1994), we are able to identify the top five asian countries with the highest levels of under pricing in the region, in the following order: china, india, korea, malaysia, and japan, with a reported average initial return of 118.4%, 88.5%, 59.3%, 56.2%, and 41.7%, respectively. this paper concludes that the regulatory environment is an important influencing factor in determining the level of under pricing. the second part of the review deals with the literature which investigated such issues, where most of the literature regarding such issues was related to the chinese ipos. from the literature (moshirian et al., 2010; liu et al., 2014; espenlaub et al., 2016; wadhwa et al., 2016; chen et al., 2015), we find that the level of under pricing is higher in countries with lenient regulatory environments and lower in countries with tighter regulations and higher transparency. the next part of the paper reviews the various determinants (i.e. issuespecific factors, country-level factor, and ipo pricing method) of under pricing in the asian region. we find that issue-specific factors become more relevant in influencing the initial returns of ipos, especially in countries characterised with lenient regulatory environments and high levels of information asymmetry among the main participants of the ipo process (i.e. investment banks, issuers, and prospective investors). this is due to prospective investors relying on such preavailable information (in the ipo prospectus) to build their own perception about the success factors of an ipo (katti & phani, 2016). for example, the reputation of the main investment bank that undertakes the responsibility of underwriting the issues of the listing firm helps prospective investors in identifying good investment opportunities. this is due to the perception of the investors that the reputation of the investment bank has the ability to signal the quality of the listing firm and to reduce the level of information asymmetry around its issues (rumokoy et al., 2017). furthermore, the reputation of the auditing firm has the ability to help prospective investors identify good investment opportunities through signalling on the quality of the issuing firm (khurana et al., 2017). however, the effect of such signals on under pricing is still quite contradictory, which varies from negative to positive and sometimes to non-significant. this conclusion is based on beatty and welch’s (1996) explanation in justifying such results, where they argued that the results would differ depending on the economic conditions during different periods. the same can be said about board structure. we also find that the lock-up period has a positive relationship with 20 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 under pricing, specifically, the longer the lock-up period the higher the level of under pricing (boonchuaymetta & chuanrommanee, 2013; mohd rashid et al., 2014). moreover, the new trend in the lock-up literature is steering towards investigating the effect of the lock-up provisions on trading volume changes around the lock-up expiry date. literature on country-level factors shows that any microeconomic factors that helped in reducing the level of information asymmetry, tightening country regulations and increasing the country’s financial transparency will have a reducing effect on under pricing. finally, the pricing method of an ipo plays an important role in determining the level of under pricing, where countries that used auction and book-building pricing methods have lower under pricing levels than countries that used the fixed-priced method. this is because auction and bookbuilding methods allow prospective investors to be part of the pricing process by providing their opinions and beliefs regarding the true value of an ipo, which helps in reducing the level of information asymmetry between the participating parties in the ipo process and thus will lead to lower under pricing. on the other hand, the fixed-priced method does not allow prospective investors to participate in the process of setting up the offer price of an ipo. this in turn will lead to increase in the level of information asymmetry among the participating parties in the ipo process and thus will result in higher under pricing. references ammer, m. a., & ahmad-zaluki, n. a. (2016). the effect of underwriter’s market share, spread and management earnings forecasts bias and accuracy on under pricing of malaysian ipos. international journal of managerial finance, 12(3), 351-371. badru, b. o., ahmad-zaluki, n. a., & wan-hussin, w. n. (2017). board characteristics and the amount of capital raised in the malaysian ipo market. journal of multinational financial management, 42-43, 37-55.‎ banerjee, s. (2016). determinants of under pricing of graded ipos in the indian capital market. singapore management journal, 5(1), 45-66. banerjee, s., dai, l., & shrestha, k. (2011). cross-country ipos: what explains differences in under pricing? journal of corporate finance, 17, 1289– 1305. beatty, r. p., & welch, i. (1996). issuer expenses and legal liability in initial public offerings. journal of law and economics, 39(2), 545–602. berkowitz, d., pistor, k., & richard, j. f. (2003). economic development, legality, and the transplant effect. european economic review, 4, 165– 195. boonchuaymetta, e., & chuanrommanee, w. (2013). management of the ipo performance in thailand. journal of multinational financial management, 23(4), 272-284. a review of ipo under pricing phenomenon in asia : 1-26 21 boulton, t. j., smart, s. b., & zutter, c. j. (2017). conservatism and international ipo under pricing. journal of international business studies, 48(6), 763– 785. cai, j., & wei, k. c. j. (1997). the investment and operating performance of japanese initial public offerings. pacific-basin finance journal, 5(4), 389–417. camp, g., munro, r. (2000). under pricing of initial public offerings in new zealand: a comparison of the fixed price and book-build methods. unpublished working paper (university of melbourne). chang, c., chiang, y. m., qian, y., & ritter, j. r. (2017). pre-market trading and ipo pricing. review of financial studies, 30(3), 835–865. chen, h. c., & wu, s. c. (2015). who makes the choice on ipo underwriting methods? issuers vs. underwriters. financial management, 44(4), 753783. chen, h. l. (1992). the price behaviour of ipos in taiwan. unpublished working paper (university of illinois). chen, z., choi, j. j., & jiang, c. (2008). private benefits in ipos: evidence from state-owned firms. available at ssrn:https://ssrn.com/abstract=1089563. chen, y., wang, s. s., li, w., sun, q., & tong, w. h. s. (2015). institutional environment, firm ownership, and ipo first-day returns: evidence from china. journal of corporate finance, 32, 150-168. cheunga, y. l., dai, y., ouyang, z., & tan, w. (2018). who leaves money on the table? the role of founder identity in hong kong. applied economics, 50(7), 774-788. che-yahya, n., abdul-rahim, r., & yong, o. (2014). moderating effect of investor demand on the relationship between institutional investors’ participation and flipping activity. labuan bulletin of international business & finance, 12, 1-12. chi, j., & padgett, c. (2005). short-run under pricing and its characteristics in chinese initial public offering (ipo) markets. research in international business and finance, 19(1), 71–93. choie k. s. (2016). factors of ipo under pricing. international journal of economics and finance, 8(2), 1-8. chowdhry, b., & sherman, a. (1996). international differences in oversubscription and under pricing of ipos. journal of corporate finance, 2, 359–381.‎ darmadi, s., & gunawan, r. (2013). under pricing, board structure, and ownership: an empirical examination of indonesian ipo firms. managerial finance, 39(2), 181-200. dawson, s. m., & hiraki, t. (1985). selling unseasoned new shares in hong kong and japan, a test of primary market efficiency and under pricing. hong kong journal of business management, 3, 125-134. deng, q., & zhou, z. g. (2015). offline oversubscription, issue size, and market momentum: the driving forces for chinext ipos’ initial under pricing. chinese economy, 48(2), 114–129. 22 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 dhamija, s., & arora, r. k. (2017). initial and after-market performance of sme ipos in india. global business review, 18(6), 1536–1551.dhatt, ms., kim, y. h., & lim, u. (1993). the short-run and long-run performance of korean ipos: 1980-90. unpublished working paper (university of cincinnati and yonsei university). ekkayokkaya, m., & pengniti, t. (2012). governance reform and ipo under pricing. journal of corporate finance, 18, 238–253. espenlaub, s., goyal, a., & mohamed, a. (2016). impact of legal institutions on ipo survival: a global perspective. journal of financial stability, 25, 98–112. fukuda, s. (1984). sinki-jojo-kabushiki no kakaku performance, shogakukenkyu. kansei gakuin university, 32, 83-105. fung, s., gul, f. a., & radhakrishnan, s. (2007). investment banks’ repeated ipo business opportunities and ipo under pricing. available at ssrn: https://sci-hub.tw/https://ssrn.com/abstract=998394. gao, s., liu, j., & chan, k. c. (2017). does the removal of the ipo lockup matter in ipo pricing? finance research letters, 23, 246-252. gumanti, t. a., lestari, a. r., & abdul-manan, s. s. (2017). under pricing and number of risk factors of initial public offerings in indonesia. business: theory and practice, 18, 178-185. haman, j., chalmers, k., & fang, v. (2017). ipo lock-ups, long-run returns, and growth opportunities. journal of international financial markets, institutions and money, 49, 184-199. hamao, y., packer, f., & ritter, j. r. (2000). institutional affiliation and the role of venture capital: evidence from initial public offerings in japan. pacific-basin finance journal, 8, 529-558. hanafi, m. m., & setiawan, a. (2018). ownership concentration, institutional ownership, and ipo under pricing: evidence from indonesia stock exchange. international journal of governance and financial intermediation, 1(1), 3-17. handa, r., & singh, b. (2017). performance of indian ipos: an empirical analysis. global business review, 18(3), 734-749. hebner, k. j. & hiraki, t. (1993). japanese initial public offerings, in: walter, i., & hiraki, t. eds., restructuring japan’s financial markets (business one/ irwin, homewood. il). heerden, g., & alagidede, p. (2012). short run under pricing of initial public offerings (ipos) in the johannesburg stock exchange (jse). review of development finance, 2(3-4), 130–138. huang, h., chiang, m., lin, j., & lin, y. (2017). fixed-price, auction, and bookbuilding ipos: empirical evidence in taiwan. finance research letters, 22, 11-19. huang, p. k., chang, p., & chiang, y. m. (2015). lock-up provision and institution investors trading. available at ssrn: https://ssrn.com/abstract=2565293. hwang, c. y., & jayaraman, n. (1993). the post-listing puzzle: evidence from tokyo stock exchange listing. pacific-basin finance journal, 1(2), 111– 126. a review of ipo under pricing phenomenon in asia : 1-26 23 ibbotson, r. g., & jaffe, j. f. (1975). “hot issue” markets. journal of finance, 30(4), 1027–1042. isa, m. m. (1993). list of malaysian ipos. unpublished working paper (university of malaya). jelic, r., saadouni, b., & briston, r. (2001). performance of malaysian ipos: underwriters reputation and management earnings forecasts. pacificbasin finance journal, 9(5), 457–486. jenkinson, t. (1990). initial public offerings in the united kingdom, the united states, and japan. journal of the japanese and international economies, 4(4), 428–449. jia, c., xie, z., & zhang, d. (2014). analyst coverage in the remarket of ipos. available at ssrn: http://ssrn.com/abstract=2450012. jiao, y., kutsuna, k., & smith, r. (2017). why do ipo issuers grant overallotment options to underwriters? journal of corporate finance, 44, 34-47. kanekoa, t., & pettway, r. h. (2003). auctions versus book building of japanese ipos. pacific-basin finance journal, 11, 439-462. kang, s. k., kang, h. c., kim, j., & kim, n. (2015). insiders’ pre-ipo ownership, under pricing, and share-selling behaviour: evidence from korean ipos. emerging markets finance and trade, 51(3), 66-84. katti, s., & phani, b. v. (2016). under pricing of initial public offerings: a literature review. universal journal of accounting and finance, 4(2), 35-52. kavita, w., reddy, v. n., goyal, a., & abdulkadir, m. (2016). ipos and seos, real investments, and ‎market timing: emerging market evidence. journal of international financial markets, institutions and money, 45, 21-41. khurana, i., ni, c., & shi, c. (2017). the role of big 4 auditors in the global primary market: does audit quality matter most when investors are protected least? presented at asian bureau of finance and economic research annual conference, gardenia, azalea, 2017. kim, j. b., krinsky, i., & lee, j. (1995). the after-market performance of initial public offerings in korea. pacific-basin finance journal, 3(4), 429–448. koh, f., & walter, t. (1989). a direct test of rock’s model of the pricing of unseasoned issues. journal of financial economics, 23(2), 251–272. komenkul, k., & siriwattanakul, d. (2016). how the unremunerated reserve requirement by the bank of thailand affects ipo under pricing and the long-run performance of ipos. journal of financial regulation and compliance, 24(3), 317-342. lee, p. j., taylor, s. l., & waiter, t. s. (1993). australian ipo pricing in the short and long run. unpublished working paper (university of sydney). lee, p. j., taylor, s. l., & walter, t. s. (1996). expected and realised returns for singaporean ipos: initial and long-run analysis. pacific-basin finance journal, 4(2&3), 153–180. liu, j., uchida, k., & gao, r. (2014). legal protection and under pricing of ipos: evidence from china. pacific-basin finance journal, 27, 163-187. 24 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 ljungqvist, a., & yu, x. (2003). stock market development, liquidity and corporate governance. unpublished working paper (new york university). logue, d. e. (1973). on the pricing of unseasoned equity issues: 1965-1969. journal of financial and quantitative analysis, 8(1), 91–103. loughran, t., ritter, j. r., & rydqvist, k. (1994). initial public offerings: international insights. pacific-basin finance journal, 2(2&3), 165-199. low, s. w., & yong, o. (2011). explaining over-subscription in fixed-price ipos: evidence from the malaysian stock market. emerging markets review, 12(3), 205–216. low, s. w., & yong, o. (2013). initial public offerings and investor heterogeneity: evidence from malaysia. american journal of finance and accounting, 3(1), 41-56. marisetty, v., & subrahmanyam, m. (2008). group affiliations and the performance of initial public offerings in the indian stock market. unpublished working paper (stern school of business, new york). mcguinness, p. (1992). an examination of the under pricing of initial public offerings in hong kong: 1980-90. journal of business finance and accounting, 19(2), 165–187. mcguinness, p. (2016). ipo firm performance and its link with board officer gender, family-ties and other demographics. journal of business ethics, 1, 1-23. mohamed, a., & saadouni, b. (2018). incentive fees: do they bond underwriters and ipo issuers? available at ssrn: https://ssrn.com/abstract=3009871. mohd rashid, r., abdul-rahim, r., & yong, o. (2014). the influence of lockup provisions on ipo initial returns: evidence from an emerging market. economic systems, 38(4), 487–501. mok, h. m., & hui, y. (1998). under pricing and after-market performance of ipos in shanghai, china. pacific-basin finance journal, 6(5), 453–474. moshirian, f., ng, d., & wu, e. (2010). model specification and ipo performance: new insights from asia. research in international business and finance, 24(1), 62–74. mumtaz, m. z., smith, z. a., & maqsood, a. (2016). an examination of shortrun performance of ipos using extreme bounds analysis. estudios de economia, 43(1), 71-95. pandya, f. (2016). after market pricing performance of initial public offerings (ipos). jindal journal of business research, 5(1), 1–16. pettway, r. h., & kaneko, t. (1996). the effects of removing price limits and introducing auctions upon short-term ipo returns: the case of japanese ipos. pacific-basin finance journal, 4(2-3), 241–258. reddy, k. s. (2015). the aftermarket pricing performance of initial public offers: insights from india. international journal of commerce and management, 25(1), 84-107. rhodes–kropf, m., robinson, d. t., & viswanathan, s. (2005). valuation waves and merger activity: the empirical evidence. journal of financial economics, 77, 561–603. a review of ipo under pricing phenomenon in asia : 1-26 25 ritter, j. r. (2003). differences between european and american ipo markets. european financial management, 9(4), 421-434. ritter, j. r., & welch, i. (2002). a review of ipo activity, pricing, and allocations. journal of finance, 57(4), 1795–1828. rumokoy, l. j., neupane, s., chung, r. y., & vithanage, k. (2017). underwriter network structure and political connections in the chinese ipo market. pacific-basin finance journal, available online 7 november 2017. samarakoon, l. p. (2010). the short-run under pricing of initial public offerings in the sri lankan stock market. journal of multinational financial management, 20, 197–213. sochi, m. h., & islam, r. (2018). what impacts ipo under pricing? evidence from dhaka stock exchange. international journal of business and management, 13(3), 1-10. song, s., tan, j., & yi, y. (2014). ipo initial returns in china: under pricing or overvaluation? china journal of accounting research, 7, 31–49. stoll, h. r., & curley, a. j. (1970). small business and the new issues market for equities. journal of financial and quantitative analysis, 5(3), 309–322. su, d., & fleisher, b. m. (1999). an empirical investigation of under pricing in chinese ipos. pacific-basin finance journal, 7(2), 173–202. sullivan, m. j., & unite, a. a. (2001). the influence of group affiliation and the underwriting process on emerging market ipos: the case of the philippines. pacific-basin finance journal, 9, 487–512. sundarasen, s. d., khan, a., & rajangam, n. (2017). signalling roles of prestigious auditors and underwriters in an emerging ipo market. global business review, 19(1), 69–84. tong, l. y., & ahmad, r. (2015). the signalling power of the investment banks’ reputation on the performance of ipos on bursa malaysia. jurnal pengurusan,45(1), 1-16. unite, a. a., & sullivan, m. j. (2003). the effect of foreign entry and ownership structure on the philippine domestic banking market. journal of banking and finance, 27(12), 2323–2345. vithessonthi, c. (2014). what explains the initial return of initial public offerings after the 1997 asian financial crisis? evidence from thailand. journal of multinational financial management, 27, 89-113. vos, e. a., & cheung, j. (1993), new zealand ipo under pricing: a reputation based model. small enterprise research, 1, 13-22. wadhwa, k., reddy, v. n., goyal, a., & mohamed, a. (2016). ipos and seos, real investments, and market timing: emerging market evidence. journal of international financial markets, institutions and money, 45, 21-41. wethyavivorn, k., & koo-smith, y. (1991). initial public offerings in thailand, 1988-89: price and return patterns. in: s. g. rhee & r. p. chang, (eds.) pacific-basin capital markets research, vol. 2 (north-holland, amsterdam). xu, z., wang, l., & long, j. (2017). the impact of directors’ heterogeneity on ipo under pricing. chinese management studies, 11(2), 230-247. 26 the international journal of banking and finance, vol. 14, 2018-2019 : 1-26 yatim, p. (2011). under pricing and board structures: an investigation of malaysian initial public offerings (ipos). asian academy of management journal of accounting and finance, 7(1), 73–93. yon, k. h., & park, s. h. (2009). the effect of underwriters’ reputations on post-deregulation ipo pricing: price discovery ability versus bargaining power. asia-pacific journal of financial studies, 38(4), 611-656. yong, o. (2007). a review of ipo research in asia: what’s next? pacific-basin finance journal, 15(3), 253–275. yong, o. (2013). when do after-market ipo prices stabilize? evidence from malaysian fixed-price ipos. international review of business research papers, 9(4), 77–90. yong, o. (2015). what is the “true” value of an initial public offering? journal of scientific research and development, 2(10), 78–85. zameni, a., & yong, o. (2016). lock-up expiry and trading volume behaviour of malaysian initial public offering’s. international journal of economics and financial issues, 6(s3), 12-21. zameni, a., & yong, o. (2017). effects of lock-up expiry on bid-ask spread of malaysian ipos. american journal of finance and accounting, 5(1), 64-84. zeng, j., & zhang, q. (2015). family involvement, political connection and ipo under pricing: evidence from the companies listed on the sme board of china. available at ssrn: https://ssrn.com/abstract=2648147. financial liberalization or financial development? tests using delphi-based index of liberalization the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 95 financial liberalization or financial development? tests using delphi-based index ofliberalization nicolaas groenewold, jiangang peng, guanzheng li and xiangmei fan university of western australia, hunan university and hunan normal university abstract most empirical analysis of the finance-growth nexus has used measures of financial development such as the ratio of monetary or financial assets to gdp to measure financial development. we argue that from a policy perspective measures of financial liberalisation or reform are of greater interest and, besides, are less likely to be beset by endogeneity problems which have dogged the empirical growth literature. we develop such a measure by combining the ‘delphi’ method and principal components analysis to construct an index of financial liberalisation for china. much of china’s financial development has been policy-driven and we could expect to find a distinct difference, at least in timing, between measures of financial reform and financial development. we compare our financial liberalisation index to a number of standard measures of financial development and find that there is pervasive evidence that financial liberalisation granger-causes financial development but not vice versa. keywords: financial liberalisation, financial development, economic growth jel classification: g18, o16, o43, o53 1. introduction empirical research on the determinants of economic growth has shown a resurgence over the past decade or so. a recent strand of this literature focuses on the relationship between finance and growth. it, too, has expanded rapidly over the past decade – a recent survey by levine (2005) lists about 300 references, most of which have appeared since the late 1990s. the continuing lively research activity in this area of investigation attests to the importance of the questions: ijbf 96 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 why is it that some countries have grown quickly over long periods while others have stagnated, and can policy be designed to stimulate the growth process? the policy question has been particularly important for the finance and growth strand of the literature since there has been a rapid development of the financial systems of a number of developing countries fuelled, partly at least, by government policy which has freed up financial systems previously subject to pervasive regulations. we would argue, however, that much of the literature has been only indirectly relevant to the policy questions which have often motivated it since the explanatory variables in growth regressions have typically been measures of financial development which, in contrast to measures of financial liberalisation, are one step removed from policy. thus levine (2005) in the theoretical section of his survey paper sets out to “describe models where market frictions motivate the emergence of distinct financial arrangements [which] … may influence economic growth” (p.869).1 the emphasis is therefore clearly on (endogenous) developments in financial markets and their effects on economic growth. the empirical literature generally follows this line and uses measures of financial development. the pioneering work of goldsmith (1969), for example, used measures of the size of the financial sector such as the ratio of assets of financial intermediaries to gdp to proxy the state of financial development. subsequent research has generally followed this lead although with variations. thus an influential paper in the 1990s, king and levine (1993a), use not only the ratio of financial institutions’ assets to gdp as a measure of the extent of financial intermediation but also attempt to include a measure of the importance of the private sector both in providing and in using funds. other work included measures of the importance of the stock market such as the ratio of stock market turnover or capitalisation to gdp; see, e.g., levine and zervos (1998a). the experience of the chinese economy has also provided data for the analysis of the relationship between finance and growth and, although the literature using these data is rather limited, similar variables have been used. two early (and similar papers) by li and liu (2001) and liu and li (2001) examine the relationship between “financial liberalisation and growth in china’s economic reform” by regressing aggregate gdp growth on investment growth disaggregated into four components according to the source of funding, the argument being that the source of funding has changed with the development of the financial system so that it provides (indirect) evidence about the effect of such development on growth. they conclude that a shift from state appropriation to self-raised funds has stimulated economic growth, providing evidence of the positive effects of development on growth. more recent papers include those by allen et al. (2005), hao (2006) and liang and teng (2006). of these, the last two both directly address the financegrowth question, the former using a panel of provincial data and the latter 1 a recent exception is the paper by bekaert, harvey and lundblad (2005) which focuses on liberalisation but restricts its attention to the equity market. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 97 aggregate time-series data. liang and teng use traditional financial development measures representing bank credit and the deposit/liability ratio while hao argues that such measures tell only part of the story and finds that the financing method (the switch from state budget appropriations to loans as in liu and li above) and the volume of saving are more powerful determinants. hao concludes that “the development of financial intermediation exerts a positive, causal and economically large impact on china’s economic growth” (361) while the results in the paper by liang and teng are more ambiguous: there is no causation from financial development to growth but some evidence that causation runs in the opposite direction. thus, on the whole, the finance-growth literature has used measures of financial development rather than of financial liberalisation. we argue that this has two disadvantages. both are related to the fact that financial development results not only from policy changes designed to enhance the functioning of the financial system but also from the interaction of policy (and indeed the financial system itself) with the rest of the economy. it is therefore difficult to link the results of empirical work directly to policy. moreover, it is likely that financial development measures are generally endogenous in an econometric sense making it difficult to interpret regression results in cause-and-effect terms, a problem that has dogged the empirical growth literature in general. neither of these disadvantages are likely to be as severe for measures of financial liberalisation which more closely reflect the policy changes driving the development of the financial system. on the other hand, it is, of course, true that financial development will be broader than financial liberalisation and will capture events which are not policy-driven but will nevertheless have an important influence on the ability of the financial system to support economic growth. thus, while we think there is a case for redressing the balance in favour of financial liberalisation, we do not wish to make exclusivist claims. our aim in this paper is quite modest – we do not set out to re-examine all questions in terms of financial liberalisation which have previously been analysed using measures of financial development. rather, we provide a detailed explanation of the construction of a financial liberalisation index for china and briefly compare it to some common measures of financial development. we choose china as our case study since it is relatively under-researched and, more importantly, because the pace of financial deregulation in china since the onset of reforms in 1978 has been rapid and sustained so that, here of all places, we may to observe a clear difference between measures of financial development and liberalisation which might be difficult to distinguish for countries where the pace of reform has been slower. the contribution of this paper is, therefore, that we (a) construct an index of financial liberalisation, (b) compare it to more common measures of financial development and, in so doing, (c) provide a reasonably detailed account of financial reforms in china since the beginning of economic reforms and ‘opening-up’ under deng xiaoping in 1978. we conclude that for china during this period, at least, there is a clear distinction between the liberalisation 98 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 index and measures of financial development and, in particular, that financial liberalisation granger causes the three measures of financial development we use but that the causation does not run in the opposite direction. the structure of the paper is as follows. in section 2 we provide background information on the recent reforms of the chinese financial system; in section 3 we explain the construction of our financial liberalisation index and in section 4 we compare it to some common measures of financial development using granger causality analysis. conclusions are presented in the final section. 2. a brief history of financial system reform in china since 1978 economic reform has proceeded apace since reform and ‘opening-up’ began in 1978 and financial deregulation has been a central part of this reform. financial liberalisation has been continual and widespread since 1978 and in this section we present a brief survey of the main aspects under seven headings: the diversification of financial institutions, the reform of credit quotas, interest rate deregulation, the establishment of a financial regulatory system, the development of a stock market, increased openness to the rest of the world, and the development of the financial legal framework.2 2.1 institutional diversification the development of greater diversity in financial institutions can be characterised as having taken place in four stages: (i) the development of a two-tier banking system (1979-93), (ii) the establishment of the three policy banks and the commercialisation of the four state-owned banks (1994-97), (iii) the management of non-performing loans and financial restructuring (1997-2001) and, (iv) the opening of the financial system to the rest of the world and increased privatisation (2001 to the present). we discuss each of these stages in turn. the first stage saw the development of the two-tier banking system, beginning in 1979 when the agriculture bank of china, the bank of china and the construction bank were established as state-owned specialised banks and the china international trusted investment companies (citic) emerged to target international financial investment. in the same year, the first urban credit cooperative (ucc) was set up to meet the financial needs of the urban non-state sector that was not well served by the state-owned special banks. subsequently, the growth of the uccs surged with official acceptance and the rapid development of the non-state sector of the economy. in 1984 the state council designated the people’s bank of china (pbc) as the central bank, and also established the industrial and commercial bank, the fourth state-owned specialised bank, to take over the deposit-taking and lending functions of the pbc. between 1985 and 1992, the chinese government 2 these correspond to the first seven of eight aspects covered by our financial liberalisation index, the last aspect of which is a general policy variable not covered as a specific heading below. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 99 fostered increased competition by permitting the establishment of new smalland medium-sized commercial shareholding banks, which initially offered banking services to households and firms, mainly in the regions and cities. this group included the bank of communication (1986), the shenzhen development bank (1987), the citic industrial bank (1987), the china merchants bank (1987), the guangdong development bank (1988), the fujian industrial bank (1988), china everbright bank (1992), the hua xia bank (1992) and the pudong development bank (1993). the second stage of institutional diversification covered the period 1994 to 1997. it centred around the establishment of the three policy banks and the accelerated commercialisation of the four state-owned special banks. three policy banks were set up in 1994 to separate policy lending from commercial banking activities, with the policy lending allocated to the policy banks. in addition, to create a more competitive commercial banking sector, the shareholding banks, the city commercial banks (ccbs) and the rural credit cooperatives (rccs) were allowed to develop quickly during this period. the first shareholding bank funded completely by private capital, the minsheng bank, was also set up during this stage. most of the uccs set up during the previous stage were reorganised into a smaller number of ccbs. the rccs have, in turn, undergone consolidation since 1996. during the third stage of institutional diversification, attention was devoted to the growing problem of non-performing loans (npls) in the chinese banking system, which had increased to alarming proportions by 1997. in response to increased concern, the government set up four asset management companies in 1998 to deal with the bad assets of the four state-owned commercial banks and recapitalised them with an injection of 270 billion yuan. in addition, in november 2001 the pbc approved the consolidation of 1,658 rccs into 81 joint stock city or rural commercial banks depending on the assessment of assets and capital, writing off some bad debts, and encouraging new shareholders. this reorganisation was part of a pilot reform of the rural financial system. other reform measures during this period included the closure and bankruptcy of several insolvent financial institutions including the hainan development bank and three trust and investment companies which had become highly insolvent and were closed in 1997-1998. the fourth stage consists of the continuing privatisation of the financial sector and the opening of the financial industry to the rest of the world. since china’s accession to the wto, its banking sector has been gradually opened to foreign influence. in december 2003 the chinese government announced the us$45 billion recapitalisation of the bank of china and china construction bank so that they could be listed on the stock exchange and on august 26 and september 21, 2004, the two banks were respectively transformed into jointstock companies. currently the china industrial and commercial bank, the agricultural bank of china and the bank of communications are also preparing for public listing. 100 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 in 2004, the pbc sped up the reform of rural credit cooperatives. initial pilot programmes were carried out in 8 provinces, the success of which saw the state council extend the reform to include another 21 provinces in the reform program, including beijing, tianjin and hebei. by the end of 2004, the pbc, together with the china banking regulatory commission (cbrc), had approved rural credit cooperatives in 620 counties to purchase in four groups central bank special bills totalling 35.55 billion yuan. rural credit cooperatives included in this reform have experienced significant improvement in their financial situation with the mitigation of historical problems. 2.2 the evolution of the credit allocation system prior to 1978 the allocation of credit was arranged through the pbc and its branches according to the credit plan, which was prepared in the form of a source-and-use-of-funds statement to match the estimated demand for physical resources. after establishing the two-tier banking system (central bank and specialised banks), the authorities revised the regulations in 1984 so that specialised banks were allowed a certain freedom in the use of funds and were responsible for profits and losses although they were still obliged to submit projections on loans and deposits to the pbc for approval. the credit quotas for the specialised banks were strictly enforced with the banks forbidden to extend credit that exceeded the ceiling. during the 1980s the overall trend was away from direct controls and by 1990 only the sobs remained subject to any credit quotas. it is interesting that, as the sobs had become gradually more prudent in terms of credit allocation in the face of the deteriorating performance of the state owned enterprises (soes), only 80 percent of the credit quota was fulfilled by 1997. the pbc finally removed credit quotas for the sobs in 1998, moving instead to the implementation of management principles based on asset-liability ratios. since february 2004, the pbc has begun to manage capital sufficiency to deal with operating and market risk in order to increase the capital base of the sobs to the point of the capital adequacy requirements of the basel capital accord. 2.3 interest rate deregulation china has taken a cautious approach to the deregulation of centrally controlled interest rates. during the pre-reform period, interest rates were controlled by the pbc and fixed at very low levels and rarely varied. lin et al. (1998) have argued that such repression created several problems such as a severe shortage of capital and inefficient capital allocation. in 1983 the pbc first allowed some interest-rate flexibility for the sobs when floating interest rates were introduced for certain types of workingcapital loans. in 1986 banks were allowed to adjust lending interest rates within a certain margin about the administered rate, but such flexibility on deposit interest rates was not granted. in october 1993, following a decision by the 14th national congress of the chinese communist party that “the central bank shall promptly adjust benchmark interest rates according to changes in market the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 101 supply and demand, and allow the commercial banks to flexibly set their own rate on loans and deposit within a specified range”, the pbc has taken steps to widen the floating band on lending rates for financial institutions and to adjust the reference rates more frequently, cutting the deposit and lending rates on nine occasions between may 1996 and october 2004. the pbc also began to set a reference rate for the inter-bank market in january 1996; the inter-bank bond market was launched in june 1997; and a market-based auction method began to be used in october 1999 for all government bonds. more recently important steps have been taken to further advance the market-based interest rate reform with the band for lending rates of the financial institutions being further widened from the beginning of 2004. in october 2004, the upper limit on the loan rates charged by commercial banks was removed and that for the rural and urban credit cooperatives was raised to 2.3 times the benchmark rate. a lower limit was no longer applied to the interest rate of rmb deposits. therefore, by late 2004 interest rates on both sides of bank balance sheets were largely market-determined. 2.4 the financial regulatory system given the history of central planning in china since 1949, it is not surprising that the development of the financial regulatory system experienced a proliferation of agencies and rules interrupted by several attempts to streamline and centralise the regulatory infrastructure. the pbc as central bank was first designated as the key agent for financial market supervision in 1984 and officially retained this function until 1992. during this period the bank exercised comprehensive regulatory as well as administrative jurisdiction over the financial sector. with the two major stock exchanges emerging in 1990, the state council established the state council securities commission (scsc) in 1992 and it became an important body for developing policies for financial markets, although it did not become a fullblown regulator. this task was assigned to the chinese securities regulatory commission (csrc), created by the state council in 1993. the csrc was given some lawmaking power and issued listing requirements as early as 1993. however, until the adoption of the securities law in 1999, and the strengthening of the csrc through its merger with the scsc in 1998, the state council issued most of the path-setting rules and policies that govern financial markets, including the 1993 regulations on the management of stock exchanges and securities fraud, and the 1995 adoption of b share regulations. although the reallocation of control rights over securities markets from the pbc to the scsc and eventually to the csrc has not come without opposition, the csrc eventually became the only regulatory and supervisory institution for securities markets in 1999. in 1998 the china insurance regulatory commission (circ) was entrusted with the supervision of insurance companies, including preparations for the opening of the insurance sector to foreign players. in april 2003 the china banking regulatory commission (cbrc) was established to take over 102 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 the regulatory function of the banking sector from the pbc in order to leave the pbc free to concentrate on monetary policy matters. the reserve requirements system in china was established in 1984. the statutory reserve ratio was very high by the standards of developed financial markets. initially, reserve ratios were set at different levels for different deposits: 20 per cent for enterprise deposits, 40 per cent for savings deposits and 25 per cent for rural deposits. the reserve requirements were unified and reduced to 10 per cent for all deposits in 1985 but increased to 13 per cent in 1987, a level which was maintained until 1998. in addition, the sobs were required to hold excess reserves of around 5-7 per cent. the sum of these two ratios amounted to approximately 20 per cent. this high reserve ratio policy was reversed in 1998 when the pbc consolidated the statutory reserve account and excess reserve account to a single legal reserve account, and the legal reserve ratio was reduced from 13 per cent to 8 per cent. in 1999, the pbc further lowered the reserve ratio from 8 per cent to 6 per cent but it was raised by 1 percentage point to 7 per cent on september 21, 2003. in addition, the pbc also lowered the interest rate on excess reserves of financial institutions from 1.89 per cent to 1.62 per cent on december 21, 2003 in order to encourage financial institutions to more efficiently use their resources. the interest rate on statutory reserves was kept at 1.89 per cent. thus by 2004 bank liquidity management by the central bank was proceeding along lines similar to those used in developed financial markets. 2.5 stock market development the development of china’s stock market is one of the most important elements of china’s reform of the financial system and china’s stock market has experienced amazing growth since its beginnings in the early 1990s although its capitalisation still accounts for only a small part of total financial assets. in july 1984 beijing tianqiao department store was the first company limited by shares in post-1949 china and in november of the same year the shanghai feile acoustics company issued irredeemable shares to the public. by the second half of the 1980s the issuing of equity had become increasingly frequent and shares began to change hands through non-market channels. shenyang became the first city to initiate formal over-the-counter (otc) trading in july 1986, followed by shanghai two months later. by the end of 1988 there were nine financial institutions conducting otc trading. in december 1990 and july 1991, two stock exchanges, the shanghai stock exchange and the shenzhen stock exchange were established. initially, there were only eight companies listed on the shanghai stock exchange and six on the shenzhen stock exchange. a unique feature of the chinese stock market is the two types of shares, a shares and b shares traded on each exchange. a shares are exclusively sold to chinese nationals and trade is carried out in local currency. b shares, the first of which was listed on the shanghai exchange in february 1992, are traded in foreign currencies (hong kong dollars in shenzhen and us dollars in shanghai) by foreign investors. since february 2001, domestic investors are also allowed to trade in b shares, although trading is still restricted to be in the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 103 terms of foreign currency. in addition to the a and b shares, chinese companies can issue h shares on the hong kong stock exchange, n shares on the new york stock exchange and s shares on the singapore stock exchange but these account for relatively little of their capitalisation. in general, china’s stock market is a policy-driven market. at the beginning of the 1990s, the stock markets grew largely because of promotion by local government bodies in shanghai and shenzhen while the central government paid little attention to stock market activity. this changed, however, from 1996 onwards; in 1997, the government endorsed a plan to transform more than 10,000 soes into publicly listed companies. later, many soes were listed but with a large proportion of their shares being non-tradable state-owned shares. trading shares in companies which were still effectively government-owned proved predictably problematical and was addressed in june of 2001 when the state council announced plans to reduce state share ownership. this, however, caused a significant decline in stock prices as investors feared the unloading of large parcels of government-owned shares and the state council had to suspend the state-owned share reduction scheme a year later. in order to increase the supply of funds to the domestic share market, the csrc instituted the qualified foreign institutional investor (qfii) system in december of 2002 by which limited access was given to foreign investors in a shares. on june 25, 2004 china launched a new small and medium enterprises board on the shenzhen securities exchange. on april 29, 2005, the csrc announced the resumption of reforms to address the imbalance between tradable and non-tradable shares and by the end of 2006, 95% of the listed companies issuing a shares had at least begun participation in the reform process. state-owned share holdings are still substantial, however. 2.6 external liberalisation and openness until the beginning of reforms in 1978, china under mao zedong was largely closed to the outside world (with occasional exceptions of other communist countries) but since reform began it has gradually widened its economic relationships with the rest of the world although decision-making concerning external liberalisation has been highly centralised. all major policies in this area have been decided by the state council and has involved four departments which, in the current government institutional setup are: the national development and reform commission, the state administration of foreign exchange (safe), the ministry of commerce, and the pbc. the liberalisation process can be broken down into five phases: (i) early liberalisation from 1978 to 1986; (ii) a period of increased regulation from 1987 to 1991; (iii) re-accelerated liberalisation from 1992 to 1996; (iv) the strengthening of capital controls following the asian financial crisis from 1997 to 2000; and (v) the post-wto era from 2001 to the present. in the first, early liberalisation, phase from 1978 to 1986 china maintained a dual exchange rate system until the beginning of 1985 when the rates were 104 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 unified. strict foreign exchange controls were in force during this period although they were relaxed for four coastal cities (shenzhen, zhuhai and shantou in guangdong province and xiamen in fujian province), initially selected by the authorities as foci to attract foreign direct investment (fdi) into the manufacturing sector; this group of cities was later extended to 14 coastal cities in 1984 and to the deltas of the yangtze, pearl and minnan rivers in 1985. in all cases the authorities allowed domestic firms and local governments to retain part of their foreign exchange earnings. the second phase which ran from 1987 to 1991 was a period during which regulations were increased. an initial foreign exchange control and monitoring system was set up during this period. in 1988, the fdi policies were expanded to hainan island and another 140 coastal cities and counties, including nanjin, hangzhou and shenyang. the official exchange rate in this period was adjusted several times in line with changes in purchasing power parity although adjustments of the official exchange rate generally lagged price changes. effective 9 april 1991, the official exchange rate regime was changed from a periodical adjustment to a managed float, allowing the authorities to adjust the rate more frequently. in the face of high inflation in 1989, some rules and regulations were introduced to strengthen supervision and monitoring of external transactions including regulation of foreign-funded enterprises’ foreign investment. following the strengthening of regulation in the late 1980s, the third phase saw a re-acceleration of liberalisation and of the opening-up process from 1992 to 1996. in 1992 the b-share market was launched on the shenzhen and shanghai stock exchanges, allowing foreign access to chinese shares and on july 15, 1993 the first chinese firm, qingdao beer, listed its shares on a foreign exchange – the hong kong stock exchange. in 1994, the foreign exchange retention and quota system was repealed and replaced by a foreign exchange surrender and purchase system which realised conditional current account convertibility. in late 1996 china moved to full current account convertibility. the fourth phase began with the asian financial crisis in the second half of 1997 which prompted the strengthening of capital controls and monitoring until 2000. capital outflows became an increasing problem in late 1997 and early 1998. to prevent illegal capital outflows, the restrictions on financial institutions and enterprises permitted to provide external guarantees were significantly tightened in 1997. safe launched a targeted examination of import payments in 1998. in contrast to this regulatory tightening, 13 cities were opened to foreign banks in 1997 to conduct business in rmb and in 1999 the geographic restriction on foreign banks was repealed, although the geographic restrictions on foreign banks’ conducting business in rmb were retained. the fifth and final phase, the liberalisation of external economic relations, began with china’s accession to the wto and continues to the present. it has been a period of all-round external liberalisation in china. in the banking sector, foreign banks were allowed to provide foreign exchange services to all firms and individuals and provide rmb services to domestic firms in open cities from the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 105 2003. in capital markets, domestic investors including individual residents are allowed to invest in b shares with self-owned foreign exchange earnings since 2001. from 2002 qualified foreign institutional investors (qfii) have been allowed to invest in the domestic capital (bond and stock) markets and since 2004 insurance companies have been allowed to use their own foreign exchange to invest in foreign capital markets. in 2005, the first foreign company was listed on the shanghai stock exchange. in the same year, domestic firms were allowed to set up corporations abroad with the special purpose of facilitating overseas listing, mergers and acquisitions. in the foreign exchange market, the pbc continued rmb exchange system reform when, on april 21, the rmb was pegged to a basket of currencies and allowed to fluctuate within an extended range of 3 per cent. 2.7 the improvement of the legal framework the legal framework is a tool to regulate the relationship among financial supervisors, financial institutions and investors in financial products in order to provide an environment in which laws are transparent and consistently enforced. anecdotal evidence of widespread violation of these principles of good governance in chinese financial markets has been persistent despite considerable advances made by the authorities in establishing and enforcing laws and regulations to control the financial system. a large number of rules and regulations for the financial sector have been formulated since reform began in 1978. laws such as the law of the people’s bank of china (1995), the commercial banking law (1995), the insurance law (1995), the securities law (1998), the law on trust and investment companies (2001), and the law on banking regulation and supervision (2003) which regulate financial supervision, financial institutions and their behaviour, have formed the fundamental legal framework for the financial sector. the law on administrative penalty (1999), the law on administrative appeal (2003), and the law on administrative license (2003) are designed to regulate the administrative behaviour of financial supervisors. in addition, the law on negotiable instruments (1995), the guarantee law (1995), and the company law (1993, revised in 1999, 2004 and 2005) are designed to regulate civil and commercial relations. the pbc is maintaining its efforts to improve the legal framework for the financial sector with the current amendment of the criminal law, the securities law and the enterprise insolvency law which were finished in 2006 and the law on anti-money laundering which came into effect in november 2006. in addition, the pbc is facilitating the formulation of the regulations on credit information management, the new enterprise insolvency law, the regulations on the bankruptcy of financial institutions, the regulations on deposit insurance and the rules on administration of the securities investor protection fund amongst other initiatives. despite this plethora of legislation and regulatory activity, it must be remembered that this is occurring with the context of a developing and emerging 106 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 financial system during a period of rapid economic, financial and political transition. with the deepening of china's financial system, the accelerated opening-up to the rest of the world and the rapid innovations in the financial sector, the legal framework is barely able to keep pace with the changing requirements of the financial sector and great efforts will be required on the part of the authorities to bring china’s financial system framework into the 21st century. in conclusion, there has been rapid, substantial and continual change in the chinese financial system from 1978 onwards. we have described a system which less than 30 years ago consisted of a single government-owned bank and which has developed into a highly diversified, sophisticated system serving a dynamic economy with over 1.3 billion inhabitants. it may well be argued that there have been significant hiccups in the reform process and that there is still substantial development necessary before china has a smoothly functioning, transparent and modern financial system. 3. the financial liberalisation index the measurement of financial liberalisation in a form suitable for econometric investigation is problematical for at least two reasons: how can the process of financial liberalisation be measured numerically and how can the multidimensional nature of liberalisation be accommodated in a single or relatively few variables? the most common response to the measurement problem has been to use readily observable financial variables such as the ratio of bank loans to gdp to represent liberalisation. examples of this abound in the literature on the relationship between finance and growth reviewed briefly in section 1. while this may be an acceptable approach for the measurement of financial development (although see levine, 2005, for an expression of concern about the gap between theory and measurement even here), for the purposes of measuring the extent of financial liberalisation it seems unsuitable. at best, it measures the outcome of the liberalisation process and not the process itself and, moreover, it is likely to be the outcome of the interaction between liberalisation and other economic factors. an alternative resolution of the measurement problem is to examine the history of liberalisation itself and construct an artificial index of liberalisation by assigning numerical values to particular deregulatory events. while this is more likely to capture liberalisation as such, it will be plagued by a certain arbitrariness in the assignment of numbers to events. this difficulty notwithstanding, a similar approach has been widely used in the literature on institutions and growth to capture the evolution of the political environment (see glaeser et al., 2004, for a recent example). the second problem associated with the construction of a liberalisation variable is that of the multi-dimensional nature of liberalisation. again, two alternative methods have been used in the face of this problem. the most the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 107 common has been to use a single representative measure or a limited number of measures entered as regressors simultaneously. however, the results of using a single measure may be sensitive to choice of representative measure (although this weakness may be ameliorated by assessing the sensitivity of the results to the particular measure chosen) while the simultaneous use of multiple measures has been plagued with problems of multicollinearity given the inter-relatedness of different aspects of liberalisation. a second approach has been to construct a variable which combines in a single measure the multiplicity of components. an early cross-country variable of this type measuring financial development (of the stock market) is one by demirguc-kunt and levine (1996) in which they compute an index as a simple average of standardised measures of financial development such as market capitalisation/gdp, value traded/gdp and so on. an application of this approach is in a recent paper by love (2003) who assesses the effects of financial development on individual firms’ investment decisions. in a similar approach but from a political-science perspective, quinn (1997) constructed a financial liberalisation index from three components, focussing on openness of the balance of payments – one component covered the capital account, one the current account and one international legal agreements. each component is captured by a sub-index which starts from 0 with 1 being added each time a significant incident of de-regulation is observed. the sub-indexes are combined by simply adding them. this index was used in subsequent work by adsera et al. (2001), also focussing on political questions, in this case the determinants of government effectiveness and corruption. a later cross-country study of the determinants of liberalisation by abiad and mody (2005) used a similar but somewhat more sophisticated index – they combine 6 aspects and assign a value of 0, 1, 2 or 3 to each country for each aspect depending on the magnitude of the deregulatory change. they aggregated the resulting subindexes by simply adding them together although they report that alternative aggregation schemes were tried but without substantial effects on the outcome. applications to questions of the economic effects of liberalisation have also used various ways to address the multiplicity problem ranging from a recent paper by koo and shin (2004) which simply uses a 0-1 dummy variable to more complex treatments by laeven (2003) who used a simple addition of six 0-1 dummy variables, to bandiera et al. (2000) who assign a 0 or 1 in each of eight categories but combine the resulting dummy variables using principalcomponents analysis to construct a single index of liberalisation. a similar approach is used by koo and maeng (2005) although they depart from the 0-1 scheme and allocate an additional 1 for each period in which there was a significant liberalisation event. there has, therefore, been general recognition that financial liberalisation is difficult to measure in a form suitable for econometric analysis and any approach will be less than ideal. we overcome the measurement problem by constructing sub-indexes for a variety of aspects of the liberalisation process. we attempt to overcome the subjectivity of the assignment of numbers to deregulatory events 108 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 by using the delphi method. this involved choosing 15 experts (a mixture of academics in the areas of finance and economics, commercial bankers, central bankers and government officials) and asking them to rate each of a set of eight aspects of liberalisation over the sample period, assigning a value between zero and one for each aspect for each year of the sample. the resulting 15 individual responses were averaged to derive a sub-index for each aspect ranging between zero at the beginning of the sample to one at the end for each aspect. we analysed the individual responses to detect outliers and it was interesting (and reassuring) that the individual responses were highly correlated, indicating a high degree of common assessment of the regulatory changes over the period. the correlation matrix for the individual assignments for the first aspect, “policy”, is reported in panel a of table 1 while panel b reports column averages for the first aspect in the first row and for the other seven aspects in the remaining rows. for the first aspect, it is clear that, with the exception of respondent 11, all responses are very highly correlated with correlation coefficients usually well over 0.9 and, from the remainder of panel b, it is clear that respondent 11 is not a consistent outlier and that outliers even of this magnitude are rare. we are reasonably confident, therefore, that the responses are reflective of generally perceived changes in the regulatory structure of chinese financial markets over the sample period. the second problem we had to solve was the multi-dimensional nature of deregulation. it is clear from our discussion so far that we have chosen not to focus on one or a few easily measured proxies for liberalisation but have instead measured eight different aspects of the liberalisation process itself. they are: • “policy”: major central government policy shifts • “institutions”: the diversification of financial institutions, • “allocation”: the reform of credit allocation, • “interest”: interest rate deregulation, • “regulations”: the establishment of a financial regulatory system, • “stock”: the development of a stock market, • “open”: increased openness to the rest of the world, and • “legal”: the development of the financial legal framework. using all of these indexes in a regression model would doubtlessly lead to serious multicollinearity due to the likely inter-relationships between aspects of the liberalisation process. to use only one or a few would risk the problem of omitted variables. we therefore use an approach which has been used in some recent papers reviewed above, viz. principal-components analysis, to combine the eight indexes into one index which we denote fli. principal-components analysis is a method of long standing to assess and summarise the common contents of a set of variables. consider a set of k variables x k (k=1,2,…,k), each with t observations, combined in the matrix x. the cross-products matrix x’x has k eigenvectors a k and associated eigenvalues λ k . if we arrange the eigenvectors and eigenvalues in decreasing order of magnitude of the eigenvalues we can define the kth principal component of the x matrix as t h e in tern a tio n a l jo u rn a l o f b a n kin g a n d f in a n ce, 2008/09 v ol. 6. n um ber 1: 2008: 95-115 109 table 1: correlation of individual responses panel a: correlations for “policy” 1.000 0.989 1.000 0.980 0.985 1.000 0.983 0.989 0.980 1.000 0.979 0.988 0.953 0.981 1.000 0.952 0.950 0.932 0.971 0.966 1.000 0.946 0.950 0.913 0.967 0.977 0.993 1.000 0.940 0.974 0.935 0.957 0.913 0.923 0.944 1.000 0.943 0.942 0.936 0.978 0.957 0.989 0.979 0.929 1.000 0.986 0.996 0.993 0.994 0.980 0.946 0.940 0.970 0.951 1.000 0.879 0.838 0.840 0.823 0.857 0.774 0.766 0.763 0.770 0.838 1.000 0.950 0.976 0.965 0.934 0.957 0.941 0.930 0.812 0.906 0.961 0.862 1.000 0.977 0.998 0.972 0.977 0.996 0.953 0.960 0.888 0.943 0.990 0.871 0.974 1.000 0.985 0.994 0.977 0.998 0.994 0.974 0.974 0.977 0.973 0.993 0.824 0.962 0.994 1.000 0.979 0.987 0.963 0.994 0.996 0.983 0.986 0.976 0.978 0.983 0.812 0.958 0.991 0.998 1.000 panel b: average correlation coefficients 0.965 0.970 0.955 0.968 0.966 0.950 0.948 0.927 0.945 0.968 0.835 0.939 0.965 0.975 0.972 0.967 0.975 0.969 0.975 0.982 0.947 0.969 0.907 0.967 0.971 0.972 0.979 0.980 0.980 0.978 0.984 0.984 0.972 0.977 0.983 0.984 0.971 0.969 0.987 0.941 0.982 0.978 0.982 0.988 0.985 0.971 0.985 0.974 0.974 0.981 0.971 0.982 0.972 0.974 0.972 0.981 0.976 0.976 0.983 0.971 0.962 0.936 0.945 0.965 0.924 0.949 0.960 0.871 0.944 0.898 0.965 0.966 0.961 0.960 0.949 0.973 0.973 0.965 0.976 0.888 0.963 0.962 0.958 0.970 0.936 0.970 0.962 0.963 0.955 0.967 0.984 0.988 0.987 0.987 0.975 0.972 0.961 0.978 0.987 0.987 0.989 0.956 0.986 0.979 0.989 110 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 z k = x.a k and the ratio (λ k /∑λ k ) measures the proportional contribution of the kth principal component to the total variation in the x variables. thus if, as we shall find, the first principal component captures most of the variation in the x k s, we can use it to summarise the information in the data set. in cases where the first principal component does not include most of the variation in the data set, a single index may be constructed as a linear combination of the most important principal components. in table 2 we report the eigenvalues of the eight possible components as well as the proportion and the cumulative proportion of the variation in the variables explained by each. it shows that the first principal component explains by far the greatest proportion of the variation in the individual sub-indexes and that the first two principal components explain over 97% of total variation. we use a weighted average of the first two principal components as the fli, with the weights being the proportion which each explains of total variation. in the last two columns of the table we report the loadings on the eight variables for the first two principal components (the elements of a 1 and a 2 ); they show that the first principal component is roughly a simple average of the eight variables while this is not true of the second.3 a graph of the first two principal components and the fli is given in figure 1 below. the graph makes it clear that the fli is approximately equal to the first principal component and that the index increases steadily over the sample period. table 2: eigenvalues and scores of principal components component /variable eigenvalue % of variance explained cumulative % of variance explained score principal component 1 score principal component 2 1 7.379 92.232 92.232 0.132 0.044 2 0.411 5.134 97.366 0.134 -0.052 3 0.084 1.053 98.420 0.124 0.926 4 0.057 0.714 99.134 0.129 -0.691 5 0.031 0.386 99.520 0.131 -0.521 6 0.016 0.204 99.724 0.134 -0.125 7 0.012 0.155 99.879 0.126 0.833 8 0.010 0.121 100.000 0.132 -0.341 3 note that, for the purposes of the second, third and fourth columns, the numbers in the first column refer to the components, the z k s (of which there are a maximum of eight), whereas for the purposes of the last two columns the numbers in the first refer to the x k s. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 111 figure 1: the financial liberalisation index the fli clearly increased monotonically over the sample period reflecting the general impression that reform is cumulative on the whole – while there may be particular reforms which are wound back (temporarily), the general thrust of reform of the financial system in china has been a continuing one. there have, however, been several episodes when reform appears to have accelerated (or, more precisely, has been perceived by the experts surveyed to have accelerated). in the early years following the demise of mao zedong and the consolidation of power by deng xiaoping in 1978, reform seems to have proceeded slowly and cautiously. there was a spurt, however, starting in 1984 which reflects the establishment of the people’s bank of china as the central bank in april 1984. this event affected the fli in various ways – through the credit control, institution building and prudential regulation components of the index. the establishment of the people’s bank also coincided with the institution of a system of statutory reserve requirements for the banks – the ratio was initially set very high but substantially reduced in 1985. reform seems to have faltered somewhat between 1985 and the early 1990s. the first half of the 1990s saw steady institution building and diversification, relaxation and modernisation of the system of credit controls as well as the opening up of the financial system to the world economy with the move to a managed floating exchange rate. the brief reform spurt in 1998 coincided with the removal of direct credit controls, the consolidation of the system of reserve requirements and the greater flexibility in the way in which banks could set their interest rates. finally, an acceleration of reform in the two last years of the sample reflects continued institution building, the further relaxation of interest rate controls and, in 2005, substantial changes to the foreign exchange arrangements. 112 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 the overall behaviour of the index therefore looks plausible in the light of events in the chinese financial system over the period. 4. the causal relationship between financial liberalisation and development in this section we briefly consider the relationship between fli and more common measures of financial development to assess our conjecture that liberalisation is likely to precede financial development. we use three measures of financial development commonly used in the literature: fir (financial assets to gdp ratio), depth (the ratio of m2 to gdp) and credit (total credit to gdp ratio). the sample period for these is 1978 to 2005 as for the fli and the data are taken from “china statistical yearbook” and “almanac of china’s finance and banking” (china statistical bureau, various years). we analysed the relationship between our fli and the three measures of financial development using a test for granger causality. the test is straightforward and widely used. thus, for example, to test whether fli granger-causes fir we estimate an equation with fir as the dependent variable and lagged values of each of fli and fir (and a constant) as regressors: fir t = α + β 1 fir t-1 + β 2 fir t-2 + … +β p fir t-p + δ 1 fli t-1 + δ 2 fli t-2 +…+ δ p fli t-p + ε t and test the null hypothesis h0:δ 1 =δ 2 =....=δ π =0, i.e. that fli does not granger-cause fir. if the null hypothesis is rejected (i.e., δ 1 to δ p are jointly significant) we conclude that fli granger-causes fir. we chose lag length using standard criteria: the akaike information criterion, the schwarz bayesian criterion and the hannan-quinn information criterion. generally three lags were indicated but in one case two seemed sufficient. we used three lags throughout but checked the robustness of the results by re-running all tests with one, two and four lags – our conclusions were unaffected. table 3 reports the results of tests of granger causality between fli and each of our three financial development variables in turn. in each case we can clearly reject the hypothesis that fli does not granger-cause the other variable but we cannot reject the hypothesis that the measure of financial development does not granger-cause fli. we therefore have strong evidence that our measure of financial liberalisation is causally prior to each of the common measures of financial development. this confirms our argument set out in the beginning of the paper that financial liberalisation precedes development and suggests that if we want to analyse the effects of finance on growth or other economic aggregates such as saving or investment we should start from financial liberalisation, especially if our interest in these questions is motivated by policy concerns. moreover, the use of financial liberalisation measures is likely to be less plagued by endogeneity both because it is prior to development and because it is closer to policy changes which themselves are likely to be exogenous. while we would not argue for the complete replacement of financial development variables by financial liberalisation ones (for reasons also set out earlier in the paper), the use the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 113 of the latter would clearly be an informative supplement to the current practice which concentrates on measures of financial development. table 3: tests of granger causality between fli and financial development variables f statistic p-value fli → credit 4.0285 0.0247 credit → fli 0.5593 0.6491 fli → depth 3.2032 0.0497 depth → fli 2.2616 0.1182 fli → fir 13.6643 0.0001 fir → fli 1.0779 0.3849 notes: fli is the financial liberalisation index described in the text, credit is the ratio of total credit to gdp, depth is the ratio of m2 to gdp and fir is the ratio of total financial assets to gdp. the f-statistics are for a test that the first variable does not granger-cause the second and the p-value is the corresponding marginal probability. 5. conclusions in this paper we have argued for the importance of the distinction between financial development and financial liberalisation. financial liberalisation measures the process of government policy to deregulate the financial markets while financial development may be partly the effect of earlier financial liberalisation. in the literature which examines the links between finance and macroeconomic performance (mainly growth) the general practice is to use measures of financial development such as the ratio of bank deposits to gdp rather than a measure of financial liberalisation (which is typically much more difficult to construct). nevertheless, financial liberalisation may be a preferable measure both because it is closer to the effects of deregulatory policy which often motivates the analysis and because it is less likely to be plagued by endogeneity problems. to test our conjectures about the relationship between measure of liberalisation and development, we constructed an index of financial liberalisation for china for the period 1978 to 2005. our method of construction was based on an innovative combination of the delphi method (by which we obtained information about liberalisation for a range of aspects of the financial system) and principal components analysis which we used to combine the various subindexes obtained from the delphi procedure. the resulting index was compared to three common measures of financial development by the use of granger causality. our results provided strong evidence that the index granger-causes all three measures of financial development. there is no evidence of reverse causality. we conclude that while empirical measures of financial liberalisation are more difficult to construct than most measures of financial development, they are 114 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 likely to have a useful place in the analysis of the relationship between finance and economic aggregates such as growth because they are more closely related to policy which often motivates our interest in these questions and because, being causally prior to measures of financial development, they are less likely to be contaminated by endogeneity which has plagued the empirical literature on finance and growth (and, indeed, the empirical growth literature in general). author statement: submitting author: nicolaas groenewold, university of western australia.e-mail: nic.groenewold@uwa.edu.au. he holds a visiting appointment at the hunan normal university. the authors are jointly responsible for any errors, and wish to thank the editorial office for the review and copy editing of the submission. references abiad, a. and mody, a. (2005), ‘financial reform: what shakes it? what shapes it?’, american economic review, 95, 66-88. adsera, a., boix, c., payne, m. (2001), ‘are you being served? political accountability and the quality of government’, inter-american development bank research department, working paper 438. allen, f., qian, j. and qian m. (2005), ‘law, finance and economic growth in china’, journal of financial economics, 77, 57-116. bandiera, o., caprio, g.., honohan, p. and schiantarelli, f., (2000), ‘does financial reform raise or reduce saving?’, review of economics and statistics, 82, 239-263. bekaert, g., harvey, c. and lundblad, c. (2005), “does financial liberalization spur growth?”, journal of financial economics, 77, 3-55. demirguc-kunt, a. and r. levine (1996), “stock market development and financial intermediaries: stylised facts”, world bank economic review, 10, 291-321. glaeser, e., la porta, r., lopez-de-silanes, f., shleifer, a., (2004), “do institutions cause growth,” journal of economic growth, 9, 271-303. goldsmith, r. w. (1969), financial structure and development, yale university press, new haven, usa. hao, c. (2006), ‘development of financial intermediation and economic growth: the chinese experience’, china economic review, 17, 347–362. king, r. g.. and levine, r. (1993), ‘finance and growth: schumpeter might be right’, quarterly journal of economics, 108, 717-738. koo, j. and shin, s. (2004), ‘financial liberalization and corporate investments: evidence from korean firm data’, asian economic journal, 18, 277292. koo, j. and maeng, k. (2005), ‘the effect of financial liberalization on firms’ investments in korea’, journal of asian economics, 16, 281-297. laeven, l. (2003), ‘does financial liberalization reduce financing constraints?’ financial management, 32, 5-34. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 95-115 115 levine, r, (2005), ‘finance and growth: theory and evidence’, chapter 12 in p. aghion and s. n. durlauf (eds.), handbook of economic growth, volume 1a, elsevier, 865-934. levine, r. and s. zervos (1998), ‘stock markets, banks and economic growth’, american economic review, 88, 537-558. li, k. w. and liu, t. (2001), ‘financial liberalization and growth in china’s economic reform’, the world economy, 24, 673 – 687. liang, q. and teng, j. z. (2006), ‘financial development and economic growth: evidence from china’, china economic review, 17, 395–411. lin, j. y., cai, f. and li, z. (1998), the china miracle: development strategy and economic reform, the chinese university press, pp.198-201. liu, t. and li, k. w. (2001), ‘impact of financial resources liberalization in china’s economic growth: provincial evidence’, journal of asian economics, 12, 245−262. quinn, d. (1997), ‘the correlates of change in international financial regulation’, american political science review, 91, 531-551. international journal of banking and finance 9-1-2008 financial liberalization or financial development? tests using delphi-based index of liberalization nicolaas groenewold jiangang peng guanzheng li xiangmei fan recommended citation cetak a4 ijbf6(2)09.indd the international journal of banking and finance, 2009/10 vol. 6. number 2: 2009: 139-154 139 ijbf separate legal entity under syariah law and its application on islamic banking in malaysia: a note1 zainal a. zuryati, mohamed yusoff and ahmad n. azrae university utara malaysia abstract the principle of separate legal entity – that is, after the incorporation of a company, it is regarded as an artifi cial person or juridical person who has the rights and responsibilities similar to a living person has been widely accepted and applied in the world of business, trade and industry. in malaysia, an islamic banking institution is incorporated under the companies act 1965 where after its incorporation, it becomes a legal entity separated from its members and shareholders. in the case of bank islam malaysia berhad v. adnan bin omar (1994), the court held that bank islam malaysia berhad is a corporate institution created by statute. this case has been decided based on civil law system, not under the islamic legal system. since the products offered by an islamic banking institution is solely in harmony with islamic principles, one fundamental legal question yet to be resolved is whether the principle of separate legal entity is recognised under the syariah law. this paper aims to discuss the issue of separate legal entity in islam and its application to the islamic banking institutions. keywords: separate legal entity, islamic banking institutions, syariah, juridical person, company jel classifi cation: k29 1 editors’ note: this article is not on the operational aspects of banking as all our accepted articles have been over the last six years, but on the lacuna in law as to what is an islamic bank. it is important enough to be included in this journal as this article points to a fundamental legal problem of recognition of islamic bank as a legal persona. it appears that the term ‘islamic bank’ that can be traced to the fi rst islamic bank founded in 1963 has had no historical precedent in islamic laws, and remains yet to be formally recognized in syariah laws some 45 years after this entity has been invented and nurtured over the years. ht tp :// ijb f.u um .e du .m y 140 separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 1. introduction one of the important elements in the common law concept of separate legal entity is that once a company has been incorporated, it will be considered as ‘a person’. the corporation inter alia has the right to sue and to be sued, own property in its name and it is separate from the owner or shareholders of the corporation. the liability of the owner or the shareholder is limited to only the value of shares invested in that particular company. in islam, there are views concerning this legal matter. the traditional islamic jurists opinion is that the principle of separate legal entity does not exist in islam. it is for a very simple fact that a company which is regarded as ‘a person’ is only a fi ctitious person and only an imaginary person, as such, has no capacity to engage in any commercial transaction. this is one view. besides, if a company or the fi ctitious entity is declared insolvent, the creditors have only the rights over the assets of the company which is only registered under the imaginary person whereby if the assets are not suffi cient to settle the amount, it is simply written off and cannot be claimed from anyone. it is therefore pertinent to determine the existence of the principle of separate legal entity in syariah and its application to islamic banking institutions in malaysia. in sections 2 and 3, we examine the grounds for this legal entity within the english system of laws and its growth in malaysia. in sections 4 and 5, the reader will fi nd a discussion of the historical development of bodies outside of a real person with legal recognition given under the syariah laws. there it is clear that there is a clear difference of opinion for and against the recognition of a legal person outside that of a real person although there are such bodies – not banks – that had been given such recognition in historical time by jurists. we discuss the need for clarity by way of some an initiative to arrive at a ruling to give recognition of legal persona to the banking institutions in malaysia. 2. the principle of separate legal entity under common law a company is a corporation. a corporation is a succession or collection of persons having at law an existence, rights and duties, separate and distinct from those of the persons who are from time to time its members (keenan and bisacre, 2005, p. 2). a corporation carries with it several distinguishing features, namely it is a persona at law, i.e. an artifi cial person; and it has perpetual succession, i.e. the company carries on and is not dissolved despite the changes in its membership since its existence is maintained by the constant succession of new persons who replace those who die or being removed. the doctrine of separate legal entity was originated from the case salomon v. salomon & co. ltd. [1897]. the facts in this case disclosed that a company had been incorporated by mr. salomon in which he and members of his family were the only shareholders. the issue arises when the company’s business turn to be a failure. the value of the assets was insuffi cient to pay out both mr. ht tp :// ijb f.u um .e du .m y separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 141 salomon and the company’s other creditors. consequently, the creditors raised an issue whereby they argued that mr. salomon should not receive the payment from the company because the degree of control he exercised over the company. it was held by the house of lords that despite mr. salomon having the control over the company, it was neither his agent nor trustee. this is because, a company was treated as operating the business in its own right, and as being separate from its controller, i.e. in this case of mr. salomon. therefore, the charge given by the company to mr. salomon was valid and he was entitled to be paid his debt even though other creditors of the company would not be paid because the company had insuffi cient assets to pay all its creditors. thus, it shows that a company is a legal person separate and distinct from its individual members or directors as in the words of cave j in re sheffi eld & south sheffi eld yorkshire permanent building society, in liquidation [1889]: ‘a company is a legal persona just as much as an individual’. in macaura v. northern assurance co. ltd. [1925], mr macaura, had formed an estate company. thereupon, he sold his owned timber estate for 42,000 to the estate company, whereby the purchase money was paid by the company in the form of issuance of 42,000 fully paid shares of 1 each. macaura had effected an insurance policy on the timber in his own name, and not in the company’s name. eventually, on 23 february 1922, most of the timber was destroyed by fi re. thus, macaura claimed under his insurance policies. however, it was ruled by the court that macaura have no insurable interest. the insurance policy effected by him could only be on the basis of a creditor or a shareholder of the company, which neither two has an insurable interest in the assets of the company based on the principle that a company is an independent entity. a company also has perpetual succession despite of the death of its shareholders. in lee (catherine) v. lee’s air farming ltd. [1960], the facts disclosed that in 1954, mr. lee had formed the respondent company carrying on the business of crop spraying from the air. mr. lee owned 2,999 of the company’s 3,000 1 shares. apart from that, he also was the company’s governing director whereby he had appointed himself as the only pilot of the company at a salary arranged by himself. in march 1956, mr. lee was killed while piloting an aircraft during the course of top-soil dressing. the appellant want to claim compensation from the company as the employer of her husband under the new zealand workers’ compensation act 1922. the issue for determination was whether there existed the relationship of employer and employee between the company and mr. lee. the court held in favour of appellant and she was entitled to compensation. following the grounds of the decision in salomon’s case, mr. lee was employed by the company in the sense required by the act 1922. according to keenan & bisacre (2005, p. 27), the principle set out in the above case has also led to the use of the term ‘the veil of incorporation, which is said to hang between the company and its members and in law at least, act as a screen between them. under certain circumstances, this doctrine of separate legal entity produces what appear to be unjust and purely technical results. as a result, judges tend ht tp :// ijb f.u um .e du .m y 142 separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 to come out under a moral or intellectual pressure to sidestep the principle of separate legal entity and produce a result which seems more ‘just’ (pettet, 2005, p. 24). there are many reported cases that show the attitude of the judges to by-pass the precedent of salomon’s case in order to achieve justice by ‘lifting’ or ‘piercing’ the corporate veil (re a company (1985) at p. 241; creasey v. breachwood motors ltd. [1992] at p. 647). lord denning mr suggested that the corporate veil could be lifted, that the companies were in reality a group, and should be treated as one (dhn ltd. v. tower hamlets [1976]). on the other note, the house of lords criticized lord denning’s approach and held that the corporate veil could only be lifted in this way in circumstances where the company is a ‘façade’ (woolfson v. strathclyde dc (1978)). in adams v. cape industries plc [1990], the court gave strong support to the idea that there is really only one well-recognised exception to the rule prohibiting the piercing of the corporate veil. that is ‘it is appropriate to pierce the corporate veil only where special circumstances exist indicating that it is a mere façade concealing the true facts. in trustor ab v. smallbone & ors. (no. 3) [2002], morritt v-c held that the court would be ‘entitled to pierce the corporate veil and recognize the receipt by a company as that of the individual in control of it if the company was used as a device or façade to conceal the true facts thereby avoiding or concealing any liability of those individual’, and that on the facts this was satisfi ed. 3. separate legal entity under malaysian law this doctrine established in english law has received its application in malaysia via the companies act 1965 (act 125). section 16(5) laid down the effect of its incorporation, namely: • a company shall be regarded as a body corporate, capable of exercising all the functions of an incorporated company. the term ‘body corporate’ is not defi ned under the companies act 1965. however, generally it covers both the ‘companies’ and ‘corporation’. both of these terms are defi ned under section 4 of companies act 1965 which will be discussed later on. ‘corporation’ is one of an artifi cial legal person (zuhairah ariff, 2003). according to salleh abas fj, it is ‘artifi cial’ the company’s legal persona is the result of statutory acts of the registrar of companies under s. 16 of companies act 1965 (tan lai v. mohamed bin mahmud [1982]). in addition to that, zakaria yatim, j in people’s insurance co (m) sdn bhd v. people’s insurance co ltd & ors [1986] held that under the ordinary rules of law, a parent company and its subsidiary company, even a wholly owned subsidiary company, are distinct legal entities. ht tp :// ijb f.u um .e du .m y separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 143 in one of the court of appeal’s decision, it was held inter alia that even though an individual is the director of two companies, both companies possesses two different entities (nanyang union sdn. bhd. v. gloveline industries (m) sdn. bhd. [2006]). the fact in the present case disclosed that mr. radwan alami was the director of two companies; namely gloveline industries (m) sdn. bhd. and safeline industries (m) sdn. bhd. nevertheless, both these companies are not the same as each company has their own separate entity. • a company will have the right of suing and being sued. in lee eng eow (as director of lee guat cheow & co sdn bhd) v mary lee (as executrix of the estate of low ai lian) & ors [1999], the court of appeal had laid down the statutory effects of an incorporation, in that an incorporated association has a legal personality of its own apart from the persons who comprise it; even though it is not specifi cally provided in the companies act 1965. if a director breaches his duty to the company, it is the company who has the right to sue him. a member cannot sue the director on the company’s behalf. similarly, if a contracting party breaches his contract with the company, it is the company who has the right to sue the contractor. a member of the company cannot sue on the company’s behalf to enforce a company’s rights. this rule is known as the ‘proper plaintiff rule’ or the rule in foss v. harbottle (1843). • a company will have perpetual succession. to illustrate on this point, in the case of abdul aziz bin atan v. ladang rengo malay estate sdn. bhd. [1985], despite changes in the membership, the corporate entity continues unchanged. in re noel tedman holdings pty ltd [1967], the company may even continue to exist despite the death of all its shareholders and directors. • a company will have the power to hold land and other property. article 9 of the third schedule to the companies act 1965 provides that a company possesses the power to purchase, take on lease or in exchange, hire and otherwise acquire any movable or immovable property. besides, such rights are also conferred onto states by the national land code, where section 43(b) conferred on the state authority with the power to dispose the land to the corporations. even though section 16(5) of companies act 1956 only mentions the right to own land, a company also possesses the right to own other sort of property (tan cheng han, 2005). the property will be treated as the company’s own and not the shareholder. therefore, even if a person owns all the shares in the company, he does not own the company’s property nor does he have any legal or equitable interest therein (macaura v. northern assurance co. ltd. [1925]). ht tp :// ijb f.u um .e du .m y 144 separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 • the liability on the part of the members to contribute to the assets of the company in the event of its being wound up are provided by the companies act 1965. for example, according to s. 214(1)(d) of companies act 1965, in the case of a company limited by shares, the liability of its members is limited to the amount unpaid on his or her shares in the company. this was noted as one of the benefi t enjoyed by the members of the limited company (rachagan et al., 2005). in other words, section 16(5) of companies act 1965 adopted the principle of separate legal entity as embodied in salomon’s case; i.e. the creation of a person in the eyes of the law upon incorporation of a company. similarly, under common law system, the malaysian cases also accepts the principle of lifting the corporate veil. peh swee chin, scj in yap sing hock & anor v. public prosecutor [1992] said that: “such veil has been lifted by statutes, e.g. the companies act 1965 itself or by the courts for certain specifi c purposes. the lifting of the veil clearly constitutes a violation of the primary principle but this has come to be treated correctly as an exception to the primary principle with such exception being subject to the parameters and perimeters indicated by the statutes or by the specifi c purposes so far indicated by the courts when lifting the veil”. to highlight on this point, we could refer to the case of aspatra sdn bhd & 21 ors v. bank bumiputra malaysia bhd & anor [1988] where the majority of supreme court held that the lifting of the corporate veil in a group of companies in order to do justice, particularly when there is an element of fraud involved, was proper to expose the true owner of the company’s assets (i.e. lorrain osman) in granting a mareva injunction. in this case, the respondents had brought an action against lorrain osman, who was a director of the fi rst respondent and the chairman of the second respondent, for the total sum of $27,625,853.06 which they claimed to be secret profi ts made by lorrain without their knowledge and approval. the respondents also made an ex parte application for a mareva injunction to restrain lorrain from transferring his assets out of jurisdiction and also for an order of discovery for lorrain to disclose the value, nature and whereabouts of all his assets. the fact disclosed that only 32 out of 21,796,395 shares in the appellant companies did not belong to lorrain. he also exercised the effective or sole control of the companies by holding more than 99 percent of the total paid up capital of the 22 appellant companies. in addition, lorrain also was a director in 15 of them. based on this, the learned judge lifted the corporate veil to see whether the companies and their shareholders were in fact the same entity. the secret profi ts received by lorrain were not denied on affi davit evidence; only the legal capacity under which lorrain had received them was being contested. no doubt that it was admittedly an element of fraud in the receipt of the secret profi ts whatever might be the capacity in which lorrain had received them. thus, it is suffi cient for the court to lift the corporate veil for the purpose of determining whether the assets of the companies are really owned by them as envisaged in salomon’s case [1897] and section 16(5) of the companies act 1965. ht tp :// ijb f.u um .e du .m y separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 145 although the learned judge did not say so expressly in his judgment, he in fact found the existence of such an abuse from the very structure of the companies after lifting the corporate veil. the learned judge found that lorrain was the alter ego of the companies, and the assets of the appellant companies are in fact and in law lorrain’s assets. therefore, we may conclude that there are two types of person recognized by the law. the fi rst one being the natural person or human beings; and the second would be the artifi cial person (tan lai v. mohamed bin mahmud [1982]), which includes any being other than human being which the law recognized as having duties and rights. one of the most recognized artifi cial person is the corporation. thus, we can see that the doctrine of separate legal entity is a fundamental legal principle which draws a distinction between an incorporated company and those people who have a control over it. a company will continue unchanged even if the identity of the participants in it changes (aiman nariman et al., 2002). 4. the principle of separate legal entity and syariah law the principle of separate legal entity as introduced under the common law system fi nds no express provisions from any syariah law resources. zuhairah ariff (2003) noted that some of the modern jurists of fi qh like mustafa ahmad al-zarqa, muhammad abu zuhrah and etc. justify the existence of entity other than human being as a legal person based on the theory of fi qh known as aldhimmah. ‘al-dhimmah’ basically means ‘guarantee’ or ‘accountable’ (nasri naiimi, 2007). according to al-kabashi (1989), some jurists view that the term aldhimmah and al-ahliyyah (capacity) as synonymous, both of which represent the entity that have rights, bear responsibilities and obligation. thus, if an entity (al-dhimmah) is recognized to be in existence, either artifi cial or real, it will have a certain capacity (al-ahliyyah) and therefore will be subjected to certain obligations (iltizamat). nevertheless, this opinion was heavily criticised by the majority of the jurists because al-dhimmah refers to anything which has attributes of human being that denotes to it rights and responsibilities (al-izibn, 660h, al-bahuti 1041h, ibn ‘abidin, 1252h.). besides, al-bazdawi (483h) and al-nawawi, (676h) opined that al-dhimmah is a ‘zat’ which is real and not fi ctitious, since the syariah only imposed obligations and rights on real person. moreover, the term al-dhimmah is not applicable at all and it has no relevancy in relation to interpretation of liability and obligation (al-tahanawi, 1996). however, the modern jurists like taqi usmani propagate the existence of the principle of separate legal entity under syariah law based on precedents of other islamic institutions and practice like waqf, masjid (mosque), baitul mal, joint stock, inheritance under debt and al abd al-ma’thoon (the slave who is permitted by his master to trade). briefl y, his argument as the basis for the existence of separate legal entity will be discussed. according to taqi usmani, waqf is a legal and religious institution wherein a person dedicates some of his properties for a religious or a charitable purpose, ht tp :// ijb f.u um .e du .m y 146 separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 creating separate legal entity. the properties, after being declared as waqf, no longer remain in the ownership of the donor. the benefi ciaries of a waqf can benefi t from the corpus or the proceeds of the dedicated property, but they are not its owners (taqi usmani, 2006 at p. 105). muslim jurists (fuqaha’) have treated waqf institution as separate legal entity based on two rulings; fi rst, if a property is bought by using the waqf’s money, the property does not become part of waqf, but become the property owned by the waqf institution. in other words, it demonstrates that the waqf institution may own a property. another example given by him is that any donation given to the masjid, the donation belongs to the masjid and not become part of waqf. furthermore, according to ahmad aldardir, a distinguished maliki jurist, a bequest made in favour of a mosque is accepted in islam. another precedent that formed the basis for taqi usmani’s (2006, pp. 105106) argument for the existence of separate legal entity in syariah law is the baitul mal. according to him, imam al-sarakhsi, the well-known hanafi jurist, in his work “al-mabsut” stated that baitul mal has some rights and obligations which may possibly be undetermined. al-sarakhsi also pointed out that “if the head of an islamic state needs money to give salaries to his army, but he fi nds no money in the kharaj department of the baitul-mal (wherefrom the salaries are generally given) he can give salaries from the sadaqah (zakat) department, but the amount so taken from the sadaqah department shall be deemed to be a debt on the kharaj department”. based on the views of al-sarakhsi (1986) in almabsut, it clearly shows that a department may borrow and indebted to another department. furthermore, the principle in a joint stock company which is found in the fiqh of imam shafi ‘i, also formed the basis of taqi usmani’s argument on the separate legal entity in islam. it is a settled principle of shafi ‘i school that if more than one person run their business in partnership, where their assets are mixed with each other, the zakat will be levied on each of them individually. however, the zakat will be payable on their joint-stock as a whole, so much so that even if one of them does not own the amount of the nisab, but the combined value of the total assets exceeds the prescribed limit of the nisab, zakat will be payable on the whole joint-stock including the share of the former, and thus the person whose share is less than the nisab shall also contribute to the levy in proportion to his ownership in the total assets. he was not subject to the levy of zakat, had it been levied on each person in his individual capacity. the same principle, which is called the principle of ‘khultah-al-shuyu‘’ is more forcefully applied to the levy of zakat on the livestock. consequently, a person sometimes has to pay more zakat than he was liable to in his individual capacity, and sometimes he has to pay less than that. this principle of ‘khultah-al-shuyu‘’ which is also accepted to some extent by the maliki and hanbali schools with some variance in details, has a basic concept of a juridical person underlying it. it is not the individual, according to this principle, who is liable to zakat. again, it is the ‘joint-stock’ which has been made subject to the levy. it means that the ‘joint-stock’ has been treated a separate entity which is very close to the concept of a ‘juridical person’ (taqi usmani, 2006, pp. 106-107). ht tp :// ijb f.u um .e du .m y separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 147 lastly, when a deceased leaves liabilities exceeding his property, this property is neither owned by the deceased, nor is it owned by his heirs, for the debts on the deceased have a preferential right over the property as compared to the rights of the heirs. it is not even owned by the creditors, because the settlement has not yet taken place. they have their claims over it, but it is not their property unless it is actually divided between them. being property of nobody, it has its own existence and it can be termed a legal entity. the heirs of the deceased or his nominated executor will look after the property as managers, but they are not the owners. if the process of the settlement of debt requires some expenses, the same will be met by the property itself (taqi usmani, 2006, p. 107). the basis of the opinion for the existence of the principle of separate entity by taqi usmani however is heavily criticised in a book entitled “the concept of limited liability – untenable in shariah” by mujlisul ulama of south africa (http://books.themajlis.net/node/251). on waqf, it is refuted and argued that the assets purchased with the income of a waqf do not become part of the original waqf property for the simple reason that for anything to become waqf there has to be a ‘waaqif’ (a human being who dedicates the asset in the path of allah (god) as waqf). the waaqif of the waqf property specifi cally and expressly made the property waqf so that it could be employed to generate income for distribution to whatever charitable cause he/she (the waaqif) had designated. if the income of the waqf too has to become waqf automatically, the very aim and purpose of the waqf will be defeated and it will be devoid of utility. this is a dissenting view. as for masjid, it is concurred that the ownership vests in allah s.w.t. and allah is not fi ctitious. ownership of the masjid in this context means ownership of the owner of the masjid. allah s.w.t. who is the true owner of the masjid becomes likewise the owner of all assets made waqf for the expenditure of the masjid, and of all income generated by these waqf assets. as such, it negates the idea of the waqf being a fi ctitious entity like the common law principle of separate legal entity when the syariah law emphasises the real ownership of the waqf to allah s.w.t.? is it not against the syariah principles to negate allah’s ownership for relegating waqf to the limbo of fi ction merely for substantiating the concept of separate legal entity or ‘juridical person’? the argument applying baitul mal as the basis for the recognition of separate legal entity is also been criticised by mujlisul ulama of south africa. the different classes of wealth such as zakat and kharaj have to be expended in different avenues or for different purposes. if, for example, zakat money is used to construct a masjid, the obligation of zakat will not be discharged. if zakat money is thus spent in a category of expenditure which does not result in the discharge of the zakat obligation, it (zakat) will have to be made good by person who had used the funds for another purpose. this is not exclusive with the baitul mal. this law applies to everyone. the khalifah is empowered by the syariah law to take money from one category of funds and utilize it for a different purpose other than for what the money has to be used for according to the syariah law. but in relation to the khalifah such use is not misappropriation because the shariah law permits this. however, when later funds of the particular kind are received, ht tp :// ijb f.u um .e du .m y 148 separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 the khalifah has to replace it to ensure that correct distribution is achieved. it is simply an issue of replacing the funds which the khalifah had borrowed by virtue of the right the syariah law has given him. the inanimate vaults in which are kept the funds do not lend. the khalifah does not represent the baitul mal. he does not contract on behalf of the baitul mal which has no such capacity. he represents allah s.w.t. and acts in terms of the mandate assigned to him by allah s.w.t. the khalifah is not in need of a concept such as ‘separate legal entity’ to carry out his mandate. the obligation of zakat only applies to muslims. zakat will not be obligatory, according to the shafi ’e mazhab on the joint stock owned by a muslim and a non-muslim. if the stock of a non-muslim and the stock of a muslim are combined, khultah has taken place. if the obligation of zakat devolves on the joint inanimate rice and barley which were mixed or on any other combined stock, it would logically follow that zakat will have to be paid regardless of one partner being a muslim and one a non-muslim. only the muslim will pay zakat on his share of the mixture. the non-muslim will not pay zakat on his share of the admixture regardless of the incidence of khultah. if the joint stock was truly a separate entity or a juridical person in the western conception of the term, then zakat should have been obligatory on the stock by virtue of the principle of khultah regardless of the faiths of the owners of the joint stock. faith does not apply to the inanimate ‘juridical person’ (themajlis.net). imran (2003, p. 117) expressly stated that artifi cial person or juridical person cannot perform religious duties including zakat. it is therefore meaningless to portray the joint stock as a juridical person and divert the obligation of zakat from the joint stock to only the muslim. imran (2003) further opposed the existence of separate legal entity in syariah law, though termed it as fi ctitious person or artifi cial person. the concept of separate legal entity that a company is a fi ctitous person relied on the instances of waqf, baitul mal and the estate of deceased has been regarded as misplaced assertions by modern muslim jurists. according to him, the concept of syarikah will lose its signifi cance if the separate legal entity concept acknowledged in syariah law. in consequence, the acceptance of the principle will shatter the whole structure and violates the fundamental principles of syariah law particularly law of contract. hence, there is a strong opposition to the concept of legal persona for a corporation. 5. islamic banking institution in malaysia the establishment of islamic banking institutions refl ects the desire of muslims to live fi nancial aspects of their lives in accordance with the teaching of islam, as interpreted by a majority of scholars. practically, the emergence of islamic banking in malaysia can be traced back to 1960s. nevertheless, much of their principles are derived from the qur’an and the sunnah of the prophet muhammad p.b.u.h. more than 1400 years ago, so it is claimed by the majority opinion now prevalent on this institutional practices. ht tp :// ijb f.u um .e du .m y separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 149 islamic banking act 1983 (act 276) (hereinafter referred to as iba) served as the legal basis for the establishment of the islamic banks in malaysia. it is one of the few islamic legislations passed by the parliament to deal with the banking system. iba came into effect on 7 april 1983. iba had underlined on numerous occasions the control and supervision of the central bank of malaysia (“central bank”) over islamic banks. apart from that, the government also played a part to ensure the smooth running of the islamic banks. pursuant to this, the government investment act 1983 was enacted at the same time of iba to empower the government to issue government investment issue, which is a government security similar to a treasury instrument issued based on syariah principles. section 2 of iba 1983 defi nes ‘islamic bank’ as any company which carries on islamic banking business and holds a valid licence; and all the offi ces and branches in malaysia of such a bank shall be deemed to be one bank. further, iba defi nes ‘islamic banking business’ as to mean banking business whose aims and operations do not involve any element which is not approved by the religion of islam. thus, the establishment of an islamic banking institution is very much focused on the transactions that are syariah’s compliance. apart from that, iba also stressed on the requirement of licensed islamic bank. this has been highlighted under section 3 of iba where it states that an islamic banking business shall not be transacted in malaysia except by a company which is in the possession of a licence in writing from the minister authorizing it to do so. therefore, if a company desires an authority to carry on islamic banking business, it shall apply for such licence through the central bank to the minister. here, the central bank’s role is to consider the application and to make recommendation to the minister stating whether a licence should be granted or not and the conditions, if any, to be attached to the licence. then, the minister shall decide whether to grant a licence with or without conditions, or refuse a licence. as discussed above, particularly in reference to iba, the word ‘company’ has been used in numerous occasions to describe islamic bank, for example under section 2 and 3 of iba. section 2 of iba 1983 defi nes ‘islamic bank’ as any company which carries on islamic banking business and holds a valid licence; and all the offi ces and branches in malaysia of such a bank shall be deemed to be one bank. this shows that an islamic bank in malaysia must be a company (halsbury’s laws of malaysia, vol. 14, 2002 at p. 255). however, iba did not defi ne the term ‘company’ or ‘corporation’ itself (section 2, iba 1983). nevertheless, the section 4 of the companies act 1965 defi ne the term ‘company’ as to mean a company incorporated pursuant to the companies act 1965 or pursuant to any corresponding previous enactment. the same section also defi nes the term ‘corporation’ means any body corporate formed or incorporated or existing within malaysia or outside malaysia and includes any foreign company, but excluding four specifi c categories of corporation as enumerated in that section. apart from that, section 2 of the companies act ht tp :// ijb f.u um .e du .m y 150 separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 1965 also defi nes ‘banking corporation’ to mean a licensed bank, a licensed merchant bank and an islamic bank. all of this indicates that a banking institution, particularly islamic bank, is a company incorporated under s. 16(5) of companies act 1965, and consequently the rule of separate legal entity which embodied in a company is also applicable to the islamic banking institution. in addition to that, section 55 of the companies act 1965 states that since islamic bank is incorporated under the companies act 1965, than it shall be subjected to both of the provisions under companies act 1965 and iba 1983. but, in case of confl ict or inconsistency between the two acts, the provisions of iba shall prevail. this has been explained by norhashimah mohd yassin (1997) while commenting on the case of bank islam malaysia bhd. v adnan bin omar (unreported), where it deals with the issue of jurisdiction of syariah court over islamic banking cases. the court in that case had decided that bimb is a corporate body which has no religion, and consequently it is not within the jurisdiction of syariah courts. list ii (state list) of the ninth schedule to the federal constitution stated that the syariah courts shall have jurisdiction only over persons professing the religion of islam and in respect only of any matters included in the state list. nh chan j held that the matter was rightly brought before the civil court because islamic institutions such as bimb and syarikat takaful are corporate institutions created by statute and do not have a religion (norhashimah mohd yassin, 1997 at p. cxcv-cxcvi). another interesting case which attracted the application of doctrine of separate legal entity in islamic banking is the case of dato’ hj nik mahmud bin daud v. bank islam malaysia bhd. [1996]. the fact disclosed that a bay’ bithaman ajil fi nancing agreement was entered into by plaintiff and bimb to develop his piece of land in kelantan. the land is held under the kelantan malay reservation enactment 1930; whereby section 7(i) prohibits any transfer or transmission or vesting of any right or interest of a malay in reservation land to or in any person not being a malay. the issue here is whether there is a transfer or a vesting of right or interest involving a non-malay. this is due to the fact that bimb is a bank with neither a malay nor a native of kelantan. however, to reconcile this matter and to enable the bank to register the charge, schedule d of the 1930 enactment allows rulers in council to grant malay status to a bank for the purpose of registering a charge document (norhashimah mohd yassin, 1997 at p. cc-cci). 6. suggestion and conclusion looking at the discussion above, it is clear that the issue on the existence of the principle of separate legal entity in syariah law draws differences of opinions among the jurists. however, to draw back from the principle will defi nitely causing a huge impact on the islamic banking system and institutions which already are so advanced and well accepted throughout the world. even the ht tp :// ijb f.u um .e du .m y separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 151 dissenting jurists on the acceptance of separate legal entity in syariah law (based on dhimmah, baitul mal, waqf etc) acknowledged the importance of justifying and accommodating the legal validity of a fi ctitious legal personality or separate legal entity under syariah law whenever it is clashed (imran, 2003, p. 17). hence, to reconcile the matter, it is indeed an urgent need to have a specifi c fatwa or islamic injunction by the muslim jurists on this matter, what more if it can be a collective fatwa of them to become an ijma’ (considered consensus). as advocated by taqi usmani (2002), this ijtihad should preferably be undertaken by the syariah jurists at a collective level, yet, as a pre-requisite, there should be some individual efforts which may serve as a basis for the collective exercise. thus, it is worthy to note an opinion by wahbah al-zuhayli on “syarikah ‘inan” because the concept of syarikah ‘inan has become the basis for the establishment of islamic banking institution throughout the world. on the concept syarikah ‘inan, there are several categories of companies which are akin to company under common law, namely, syarikah al-tawsiyyah albasitah (company limited by guarantee), syarikah al-musahamah, al-tawsiyyah bil-asham and zatul mas’uliyyah al-mahdudah (company limited by share). the various names of the syarikah/company limited by share are due to fl exibility and compatibility of islam based on the principle of ‘urf (custom) to suit the changing nature of modern world of trade and business. the shareholders of these companies have limited liabilities according to the portion of shares held (wahbah zuhayli, 1995 at p 882-887) just as in english laws. however, wahbah zuhayli does not specifi cally state the principle of neither separate legal entity nor juristic person (syakhsiyah maknawiyah) on the nature of these companies, but it is inferential from the nature of the companies, there are similar to the principle of separate legal entity under common law. this is due to the fact that the basis for sharikah ‘inan as propounded by wahbah zuhayli (2003, pp. 451452) is in fact a form of limited partnership. according to wahbah zuhaily while referring to ibn al-mundhir that ‘inan’ is a form of partnership, where the partners share the capital, as well as profi ts and losses, is approved by consensus. these partners in a sharikah ‘inan need not be equal in their contributions to capital, nor equal in their legal rights for using the property. thus, one party may contribute more than another to the partnership, and one of the partners may have the exclusive right to run the affairs of the partnership. given this potential for great variation in legal rights of dealing in the joint property, each party is only responsible for dealings that he himself performed. thus, while they share the profi ts according to any rule they agreed upon in the contract, the only share losses in proportion to their contributions to the partnership’s capital. the general rule is summarized thus: “profi ts are shared according to the parties’ conditions, but losses are shared according to their shares in the capital” (2003, p. 452). henceforth, as an example, the establishment of bank islam malaysia berhad is founded on the principle of syarikah ‘inan. according to dato’ tajuddin abdul rahman, managing director of bank islam malaysia berhad, the islamic bank is essentially an ‘inan company, a separate legal entity based ht tp :// ijb f.u um .e du .m y 152 separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 on the concept of ai-musyarakah. the ‘inan company is made up of a minimum two shareholders with no limit to the maximum number. it is a business entity, formed by having a group of shareholders, electing representatives to manage the company (board of directors) on the basis of wakalah. at the same time the shareholders retain their management control (voting rights) in the company in proportion to their shareholding. even though taqi usmani is the champion of separate legal entity, he himself has some reservation on its application. the concern is that the company, while the liability of the shareholders are limited, being exploited as a vehicle of fraud. this is also the basis for the adversaries that the principle is incompatible with syariah law. on the other hand, on the apprehension of the company being utilised for avoidance of law or fraudulent purposes, though understandable, may be dealt with the application of exception to principle of separate legal entity that is, lifting the corporate veil. though the exception only exposes the ‘real’ person behind a company, the liability is still limited to the portion of his shares without extending to his personal properties. in other words, the exception to the principle of separate legal entity does not fully solve the problem of companies being used for fraudulent purposes. to sum up, the islamic banking institutions in malaysia incorporated under section 16(5) of companies act 1965 is a separate legal entity, enjoying and having the rights and liabilities alike to the living person. the application of separate legal entity principle upon islamic banking institutions in malaysia is founded on the common law principle as propounded in the case of salomon v salomon co. ltd. and is applied in toto. the questions of whether the principle is recognisable under syariah law, whether in an unchanged or modifi ed application of the principle should be practiced, are pertinent to be answered and analysed collectively by muslim jurists (fuqaha’) for further advancement of islamic banking system. author statement: zainal amin ayub zuryati, mohamed yusoff and ahmad nasyran azrae are teaching staff in the faculty of management and law, university utara malaysia. e-mails: z.amin@uum.edu.my, zuryati@uum.edu. my and nasyran@uum.edu.my. references ahmad tajudin abdul rahman. (2000). workings of an islamic bank: case study of bank islam malaysia berhad. at the fourth international conference on islamic economics, banking and finance loughborough, united kingdom august 13 15, 2000. aiman nariman mohd sulaiman. (2005). commercial applications of company law in malaysia (2nd ed.). singapore: cch asia pte. limited. al-kabashi, al-makashifi taha. (1989). al-dzimmah wa al-haq wa ta’thiruha bi al-mawt fi al-fi qh al-islami: dirasah muqaranah. riyadh: maktabah al-haramayn. ht tp :// ijb f.u um .e du .m y separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 153 al-tahanawi, muhammad ’ali. (1996). kashshaf mustalahat al-funun wa alulum. beirut: maktabah lubnan. al-zuhayli, wahbah. (2003). financial transaction is islamic jurisprudence. vol. 1. beirut, lubnan: dar al-fikr al-mouaser. al-zuhayli, wahbah. (1995). fiqh dan perundangan islam jilid iv. (terjemahan oleh md. akhir hj yaacob et al.). kuala lumpur : dewan bahasa dan pustaka. imran ahsan khan nyazee. (2003). islamic jurisprudence. petaling jaya: the other press, international institute of islamic thought. keenan, d., and bisacre, j. (2005). smith and keenan’s company law (13th ed.). harlow: pearson. mohd. ma’sum billah (2006). shariah standard of business contract. kuala lumpur: a.s. noordeen. muhammad taqi usmani. (2002). an introduction to islamic fi nance. the hague: kluwer law international. mujlisul ulama of south africa. (2007). the concept of limited liability: untenable in the shariah. south africa: young men’s muslims association. retrieved november 15, 2007 from http://books.themajlis. net/node/251 nasri naiimi. (2007, april). peranan perundangan dalam merealisasikan konsep syirkah dalam pengurusan harta pusaka. paper presented at the persidangan undang-undang 2007 ‘undang-undang dan etika: menangani cabaran globalisasi, edc-uum, sintok. norhashimah mohd. yasin. (1996). islamisation/malaynisation: a study on the role of islamic law in the economic development of malaysia: 1969-1993. kuala lumpur: a.s. noordeen. norhashimah mohd. yasin. (1997). islamic banking: case commentaries involving al-bay’ bithaman ajil [1997] 3 mlj cxcii. padmanabha rao, k.v. (2003). company law of malaysia (incorporation and management). petaling jaya: international law book service. pettet, b. (2005). company law (2nd ed.). harlow: pearson. rachagan, s., pascoe, j., and joshi, a. (2005). concise principles of company law in malaysia. malayan law journal. shanthy rachagan. (2002). principles of company law in malaysia. malayan law journal. shams al-din al-sarakhsi. (1406h/1986). al-mabsut. beirut: dar al-ma’rifah. tan cheng han. (2005). walter woon on company law (3rd ed.). singapore: sweet & maxwell asia. taqi usmani. (2006). the principles of limited liability. retrieved from http:// www.meezanbank.com/knowledgecenter/knowledge-islamic-section -7-3. asp. taqi usmani. (2007). musharakah & mudharabah: the principle of limited. retrieved november 15, 2007, from http://www.darululoomkhi.edu.pk/ fi qh/ islamicfi nance/limitedliability.html zuhairah ariff abd ghadas. (2003, october). the myth of corporate personality: an overview from the common law and the islamic law perspective. paper ht tp :// ijb f.u um .e du .m y 154 separate legal entity under syariah law and its application on islamic banking in malaysia: a note: 139-154 presented at the international conference on harmonisation of shariah and civil law, international islamic university malaysia. zuhairah ariff abd ghadas. (1998). islamic law of business organization: corporations. islamabad: international institute of islamic thought and the islamic research institute. zuhairah ariff abd ghadas. (2007). bank negara malaysia. retrieved november 25, 2007, from http://www.bnm.gov.my/index. php?ch=174&pg=467&ac=369. zuhairah ariff abd ghadas. halsbury laws of malaysia. malayan law journal, 14, (2002). remedies: syariah law. statutes: companies act 1965 1965 (act 125). federal constitution of malaysia. islamic banking act 1983 (act 276). national land code (act 56). cases: abdul aziz bin atan v. ladang rengo malay estate sdn. bhd. [1985] 2 mlj 165. adams v. cape industries plc [1990] bcc 786 at p. 822. aspatra sdn bhd & 21 ors v. bank bumiputra malaysia bhd & anor [1988] 1 mlj 97. bank islam malaysia berhad v. adnan bin omar (unreported). creasey v. breachwood motors ltd. [1992] bcc 638. dato’ hj nik mahmud bin daud v. bank islam malaysia bhd. [1996] 4 mlj 295. dhn ltd. v. tower hamlets [1976] 1 wlr 852. foss v. harbottle (1843) 2 hare 461. lee (catherine) v. lee’s air farming ltd. [1960] 3 all er 420. lee eng eow (as director of lee guat cheow & co sdn bhd) v mary lee (as executrix of the estate of low ai lian) & ors [1999] 3 mlj 481. macaura v. northern assurance co. ltd. [1925] ac 619. nanyang union sdn. bhd. v. gloveline industries (m) sdn. bhd. [2006] 1 clj 873. people’s insurance co (m) sdn bhd v. people’s insurance co ltd & ors [1986] 1 mlj 68. re a company (1985) 1 bcc 99,. re noel tedman holdings pty ltd [1967] qd r 561. re sheffi eld & south sheffi eld yorkshire permanent building society, in liquidation [1889] 22 ch 470, 476, 5 tlr 192. salomon v. salomon & co. ltd. [1897] ac 22. tan lai v. mohamed bin mahmud [1982] 1 mlj 338. trustor ab v. smallbone & ors. (no. 3) [2002] bcc 795 at p. 801. woolfson v. strathclyde dc (1978) 38 p & cr 521. yap sing hock & anor v. public prosecutor [1992] 2 mlj 714. ht tp :// ijb f.u um .e du .m y financial intermediaries and economic development: evidence on transaction costs of borrowing by the poor the international journal of banking and finance, volume 8 (number 3) 2011: pages 54-72 financial intermediaries and economic development: evidence on transaction costs of borrowing by the poor vighneswara swamy and b.k. tulasimala indian business school-hyderabad and university of mysore, india ___________________________________________________________ abstract this study, while validating the increasing role for financial intermediaries in economic development, analyzes the importance of reducing the transaction costs for financial deepening and, consequently, economic growth. it shows that higher borrowing transaction costs for the poor in particular will retard the long-term growth of rural financial markets. further, the empirical analysis based on our primary (survey) data indicates that the microfinance models of lending offer considerably lower costs of borrowing than those in regular models of direct lending by banks. the study suggests that microfinance model of lending can provide cost-efficient avenue for speedy financial development and, subsequently, economic growth. keywords: transaction costs; banks; microfinance; nonprofit institutions; ngos; economic development; financial markets, savings institutions, growth jel classification: d23, g21, l31, o16, o43 _____________________________________________ 1. introduction for a rationalist, a cleric is an unwanted intermediary in the path towards the divine. he finds the cleric as superfluous and prefers to remove him in a move towards perfect state of disintermediation where the people can directly deal with the divine. however, on the contrary, a rationalist would choose to accept an intermediary towards the achievement of economic growth in any economy. financial intermediation wields positive influences on financial development as efficient finance contributes to economic development. the intermediation theory has endeavoured to explain why these financial intermediaries exist. savings/investment process in a capitalist economy is structured around financial intermediation, thus making the financial intermediaries as cardinal to economic development. swamy and tulasimala: transaction costs of borrowing by the poor 54 55 conventional intermediation theories are based on the concepts of transaction costs and asymmetric information. these are constructed to suggest reason for the existence of financial institutions, which take deposits or issue insurance policies and channel the funds to the needy enterprises, apart from serving to reduce transaction costs and informational asymmetries in an economy. on the other hand, in recent past there have been noteworthy vicissitudes, which deserve the attention of a researcher. transaction costs and asymmetric information have declined, and intermediation has increased. as deregulation and deepening of financial markets tend to reduce transaction costs and informational asymmetries, the propositions of financial intermediation theory need a review and questioning. this diverges from the emerging point of view that financial intermediation is a value-creating economic process. the given paradox necessitates more justifications from the traditional theory of intermediation. conventional theory is wanting in explaining the persisting and expanding economic importance of financial intermediaries. in view of the enormous socio-economic importance of financial intermediaries particularly in rural finance in the context of a developing economy, this study points to an increasing role of financial intermediaries much beyond the traditional viewpoint of information asymmetries and transaction costs. given these new evolving economic scenarios, this study intends to analyse and discuss the role of intermediation in the contemporary context. obviously, this study argues that many current theories of intermediation, to a great extent, focus on the functions of institutions which are not that crucial in evolving financial systems and inadequate to provide an acceptable answer to the question as to why and how the real life financial intermediaries continue to exist. empirical observations indicate at an increasing role for financial intermediaries in economies that experience vastly decreasing information and transaction costs. this paper goes into this paradox and comes up with a new dimension towards understanding the concept of financial intermediation. this article is structured as follows. in section 2, the concept of financial intermediation and consequent financial development is critically examined with relevant literature wherein it is attempted to infer key predictions with regard to the role of financial intermediaries in modern economic development. however, since not all the developments in this field could be detailed due to various limitations, the focus is on the basic rationale for financial intermediaries i.e. transaction costs. in section 3, the concept of transaction costs and their economic significance in financial intermediation are analysed. we provide the details of objectives of the study, methodology, and research design including area of study, data collection methods employed in international journal of banking and finance, vol. 8, iss. 3 [2011], art. 3 56 section 4. empirical results are presented in section 5 while the limitations, policy choices, and conclusion are presented in section 6. 2. financial intermediation and financial development the finance-growth nexus is an established source of debate among economists. from the time when the seminal work of patrick (1966), in which it was first hypothesized as a bi-directional relationship between financial development and economic growth, subsequent enormous empirical literature has developed testing this hypothesis (levine, 1997). financial intermediation can induce economic growth by influencing the saving rate as well as social marginal productivity of investment. obviously, financial development will be positive for economic growth. then, what is the role of financial intermediaries? the answer lies in the modest view of financial intermediation, which suggests that financial intermediaries serve to transfer financial resources from net savers in an economy to net investors. they play a key role in pooling the savings for economic growth. further, it is observed, as the gross national savings of the economy increases, there is a positive impact on gdp of the economy (see figure 1). the movement of gross national savings (gns) as against the gdp in the organisation for economic co-operation and development (oecd) countries in the period from 2000 to 2008 is presented in figure 1. it is apparent from the figure that gross national savings has a direct relationship on the growth rates of gdp of the economies. the significance of financial intermediaries in the transmission of economic surpluses to the needy areas of investment in the economies is well illustrated for these selected countries. as stated by the modern theory of financial intermediation, financial intermediaries are active because market imperfections prevent savers and investors from trading directly with each other in an optimal way. the most important market imperfections are the informational asymmetries concerning the savers and investors. financial intermediaries (as agents and as delegated monitors) fill information gaps between ultimate savers and investors. this is because they have a comparative informational advantage over ultimate savers and investors. they watch over and scrutinise investors on behalf of savers. swamy and tulasimala: transaction costs of borrowing by the poor 57 figure 1: selected oecd gross national saving (gns) as percentage to gdp and gdp growth rate figure 1-a figure 1-b figure 1-c figure 1-d figure 1-e figure 1-f figure 1-g source: all the above figures has been constructed for the study based on the data collected from the world bank website. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 3 58 however, in essence we can make out three reasons for financial intermediaries: (i) information problems, (ii) transaction costs and (iii) regulatory factors. informational asymmetries problem is an ex ante and by engendering adverse selection, they can generate moral hazard. they can also be of an ex post nature resulting in auditing or expensive verification and enforcement by the state. as a result, these information asymmetries by generating market imperfections as in the neo-classical framework give scope to specific forms of transaction costs. regulatory factors affect money circulation, savings, consumption, and financing in an economy by the intermediaries and agents navigating the twin aspects, solvency, and liquidity. it is evidenced that bank capital affects bank safety, the ability of bank to refinance, and the bank‘s ability to extract repayment from borrowers or its willingness to liquidate them. hence, the legal-based view sees regulation as a crucial factor that affects the efficiency of financial intermediaries. however, it is significant for an economy to have efficient financial intermediaries in order to overcome three crucial issues such as information asymmetry in the market, reducing the transaction costs and enabling financial regulation to work smoothly. intermediation costs in financial sector of an economy can be considered to be one of the indicators of efficiency of financial development. the deregulation of the banking systems and the entry of several private and foreign banks in some of the emerging economies have gradually induced greater competition, prompting banks to alter their business strategy and management practices which, along with technological developments, have led to overall improvement in efficiency. this was reflected in the intermediation cost (operating expenses as percentage of total assets) decelerating from 2.77 per cent in 1996 to 2.11 in 2006 in india. notwithstanding this improvement, intermediation cost of banks in india is still high in comparison with other countries. there is, thus, scope to bring it down to the level of intermediation cost of banks in other asian countries (see figure 2). financial deepening has been shown to ‗cause‘ growth (demirgüç-kunt and maksimovic 1998; rajan and zingales 1998; beck, levine, and loayza 2000; levine, 2005). a doubling of private sector credits to gdp ratio is associated with a 2 percentage points increase in the rate of gdp growth. swamy and tulasimala: transaction costs of borrowing by the poor 59 * intermediation cost represents operating expenses as percentage to total assets. source: developed by the author based on the data obtained from the following sources. 1. mohan (2006c). 2. report on trend and progress of banking in india, various issues, reserve bank of india. 3. report on currency and finance in india, (credit market in india), reserve bank of india. economies with higher levels of financial development experience faster reduction of poverty. levine (1998, 1999) and beck, demirguc-kunt and levine (2007) have noticed a positive effect of finance on poverty reduction. this has been further explained by an extensive body of literature that includes deininger and squire (1998), dollar and kraay (2002), white and anderson (2001), ravallion (2001) and bourguignon (2003). financial depth, defined as the ratio of financial aggregates to gdp, appear to promote economic growth in only low-income developing countries while it has no effect in high-income countries (see figure 3). figure-3a: domestic credit provided by figure-3b: domestic credit provided by banking sector (% of gdp) in oecd countries banking sector (% of gdp) in india source: world bank data source: world bank data in order to achieve financial deepening there is a dire need to strengthen rural financial markets. in the desired process, undoubtedly, there is the necessity to sustain efficient financial intermediaries. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 3 60 3. economic significance of transaction cost in economics and related disciplines, a transaction cost is a cost incurred in making an economic exchange. when rationally evaluating a potential transaction, it is important not to neglect transaction costs that might prove to be significant. transaction cost theorists assert that the total cost incurred by a firm could be grouped largely into two components—transaction costs and production costs. transaction costs, often known as coordination costs, are well defined as the costs of all the information processing necessary to coordinate the work of people and machines that perform the primary processes, whereas production costs include the costs incurred from the physical or other primary processes necessary to create and distribute the goods or services being produced. arguably, transaction cost reasoning became most widely known through oliver e. williamson's 'transaction cost economics'. today, transaction cost economics explain a number of different behaviors. often, this involves considering as transactions not only as the obvious cases of buying and selling, but also of day-to-day emotional interactions, informal gift exchanges, etc. transaction costs have been broadly defined by steven (1992) as ―any costs that are not conceivable in a robinson crusoe economy‖. given the critical role of credit markets in economic development, microfinance is being seen as one of the effective approaches for achieving financial development that eradicates poverty. microfinance helps in three dimensions in attaining economic growth promoting economic opportunity, facilitating empowerment, and enhancing security. transaction costs derive a lot of significance in the field of microfinance in view of the very nature of its provision of finance covering the space and distance of the targeted resource poor. it is an important cost factor involving the time and expenses for the beneficiaries. high transaction cost in rural credit system is one of the core problems that set back the viability of rural banking system. microfinance is affected by transaction cost. many of the credit instruments introduced in microfinance via the banks have not met with success because of the high transaction costs in rural credit delivery via the banks. self help groups (shgs) that have emerged as suppliers of microfinance reduce transaction costs of financial institutions that do business with the poor and that of the shgs themselves. they reduce the cost to financial institutions by acting as intermediary organizations swamy and tulasimala: transaction costs of borrowing by the poor http://en.wikipedia.org/wiki/oliver_e._williamson http://en.wikipedia.org/wiki/oliver_e._williamson http://en.wikipedia.org/wiki/robinson_crusoe 61 or by providing social collateral that substitutes for costly loan appraisals and supervisions (ajay, 2005). llanto and chua (1996) studied the transaction costs of two philippines-based nongovernment organisations (ngos). they concluded that there is an inverse relationship between an organisation‘s transaction costs and its number of years in existence. motivation and retention of ngo staff were critical for transaction costs. gonzalez-vega et al., (1997) studied the transformation of bancosol from an ngo mfi (microfinance institution) to a licensed commercial bank. the ratio of total costs to average number of loans outstanding increased from us$ 149 (1992) to us$ 242 (1994). most of this increase came from higher cost of funds, but the ratio of operational costs to the average number of loans also increased from us$ 103 to us$ 135. one reason was the transformation was accompanied by an increase in the number of branches from 4 to 32. the increased investment in infrastructure, monitoring and communication systems, and additional staff did not immediately generate sufficient number of loans. bancosol compensated by increasing the revenue generating capacity of each loan by increasing loan sizes and increasing maturities. these studies have thrown much of the light on the transaction costs from the point of view of lending non-government organisations. in their study of the microcredit of the nationalized commercial banks in india, puhazhendi (1995), girija, and satish (2000) concluded that non-governmental organizations (ngos) and self-help groups (shgs) in the credit delivery system reduced transaction costs of both banks and borrowers. ajay (2002) identified the factors that influenced group formation costs. the most important factors identified were: the number of groups handled by a field worker and his/her conveyance expenses; group training costs; and average staff salaries in the region. karduck and siebel (2004) studied transaction costs of borrowers and concluded that weekly as against monthly meeting schedules increase transaction costs by 34 per cent in all places in their indian study. these studies have underscored the importance of reduction in transaction costs for the lending organisations. the shgs are projected as potential micro-banks, either on their own or through higher levels of association, capable of using their own resources, grants, and borrowed funds for financial intermediation. within this role for shgs, models and approaches have emerged, representing different methods of ensuring effectiveness and sustainability. these models envisage solely financial intermediation or include other non-financial elements as well. given the diversity of approaches being practiced in financing the poor, it is desirable to examine the borrowing costs for the poor. much of the literature available on transaction costs in international journal of banking and finance, vol. 8, iss. 3 [2011], art. 3 62 microfinance deals with lender‘s perspective mostly ignoring the borrower‘s perspective. the author is of the opinion that it is the transaction costs of borrowing for the poor that is important which needs to be reduced in order to motivate and induce the poor to avail these services instead of focusing only on the transaction costs for the lender even though it too is equally important. this study brings the required focus on the transaction costs of borrowing by the poor, and makes an empirical analysis of the same apart from providing required insights for policy support that is needed to lessen the burden on the microfinance beneficiaries. the principal objective of this study is to bring more focus on the aspects of transaction costs in borrowing for the poor while the poor avail finance form the institutional and informal sources. transaction costs in borrowing for the poor should be studied from the borrower‘s point of view. accordingly, the objectives of this study are:  to estimate the transaction costs of borrowing for the poor in both the prominent models of lending to the poor that are prevailing in india; o under direct lending method from the institutional sources; and o under the self help group linkage approach;  to compare the transaction costs of borrowing for the poor in the above said two methods of lending for the poor; and  to suggest measures to bring down the transaction costs of borrowing for the poor in order to help the borrowing poor and to expand the reach of rural finance. 4. methodology and research design the study sets out to understand the importance of transaction costs of borrowing for the poor and estimate them in different models lending to the poor. in addition, this study attempts answer the question like: (1) what are these costs for a borrowing poor. (2) how significant are these for the poor in making a decision for borrowing from institutional source or informal source? the study has drawn extensively on recent literature on shgs and microfinance in ascertaining the different dimensions of transaction costs shg bank linkage program and other methods of lending in the shg movement. reserve bank of india in 1991, issued a directive to all scheduled commercial banks encouraging them to establish linkages directly with ngos and shgs in swamy and tulasimala: transaction costs of borrowing by the poor 63 india, using the latter as financial intermediaries of the banks to reach the poor. subsequently guidelines were issued by national bank for agriculture and rural development (nabard) for a pilot project to link banks with shgs. the shg promotion movement has gathered momentum since 1998 and there has been phenomenal rise in the number of shgs due to the active support of banks, government organizations and non-government organizations. for the purpose of study, we consider a tenyear period upto 2008 as a suitable period for the study. accordingly, march 2008 has been chosen as the reference period as the study was commenced during this period. in india, southern region has dominated the shg-bank linkage programs since the launch of the pilot project to link shgs with banks. in terms of cumulative number of shgs linked with banks, karnataka has been among the top three, the other are tamil nadu and andhra pradesh. in karnataka state, shimoga has led the way in the formation and linkage of shgs with banks. the district provides an ideal region to undertake the study in view of the diverse culture, climate encompassing the maidan region (temperate plain region) and malnad region (hilly forest region) consisting of thirthahalli, sagara and hosanagara blocks endowed with majestic sahyadri hill ranges and thick forest cover. some of the salient features of the district which prompted us to select this district as study area are; shimoga district has the distinction of having 2755 shgs linked to the banks under linkage programme with cumulative bank loan disbursed upto inr 839.1 million [us$18.24 million] by march 2008 (reference period). the district provides an ideal region to undertake the study in view of the diverse culture, climate, and people. the district has a good mix of quite old and new shgs ranging from 12 years to 1 year old. about 15 to 20 nongovernment organizations (ngos) are actively engaged in shg formation and linkage programs apart from the active participation by local bank branches and the state government‘s women empowerment departments. therefore, shimoga district was selected as a region since all the indicators are very well stabilized. the authors have brought a field perspective to the study. it is based on primary data which are collected by participating in several meetings of the shgs, interviewing the shg members and microfinance practitioners in the study area. field interviews with shg members, shg federations office bearers, other villagers and microfinance practitioners were jotted down and were subsequently detailed out using ms office excel sheet in order to make the data international journal of banking and finance, vol. 8, iss. 3 [2011], art. 3 64 unambiguous to arrive at correct estimates of the costs of the different components involved in the borrowing by the poor. informal interviews to allow others to interact freely and share information – including shg federation office bearers, group members, and neighbors was encouraged in order to gain a broader perspective on the topic. elite interviews with the branch manager of lending banks in the area were conducted to gain the lenders perspectives. data related to ‗opportunity cost of one man day lost in availing finance‘ and the ‗cost of documentation‘ in connection with the availment of loan by the poor are collected using the stratified random sampling method covering all the seven blocks in the district. transaction costs of borrowing for the poor per loan account were estimated based using data collected from 700 survey questionnaires/interview data involving 100 each for all the seven blocks of shimoga district in the state of karnataka in india. transaction costs of borrowing for the poor were estimated using the cost-allocation method. estimates of time spent by borrower for the identified functions or tasks were used to calculate the cost of borrowing for an average loan amount of inr 10,000.00 (us$215.00) per account. 5. empirical analysis when the borrower is required to avail credit from any source of finance, he/she incurs some costs that are other than the interest costs, service charges, and other loan related charges levied by the finance provider. such costs are normally incurred by the borrower for his own needs during the process of securing a loan. these costs normally include, cost of visits to bank branch, cost of document collection, cost of loan applying, cost of loan availment, cost of visit to bank branch by office bearers of the shg and other such activities. we compute and analyse the transaction costs incurred by the poor in securing loans in two broad models: (i) direct loans to individuals by banks model and (ii) shg-bank linkage model. the quantum of loan amount involved in the assessment is presumed for the poor as about (inr) rs10,000.00. 1 direct lending by banks model: this is a traditional model of loan availment by the poor since decades of directed lending by the formal financial institutions in india. under this model, the poor approach the bank branch for loans for their various purposes. the borrower incurs 1 transaction costs are estimated in indian rupee (inr) represented as inr or rupee symbol. swamy and tulasimala: transaction costs of borrowing by the poor 65 transaction costs for loan availment apart from other costs such as interest costs, service charges and other charges levied by the banks. table 1: transaction costs of borrowing by the poor, direct lending by banks sl. no. type of activity money cost in inr total cost in inr 1. cost of initial visit to bank branch 100 1 opportunity cost of one day wages 60 2 transportation cost of the visit 20 3 incidental cost of the visit 20 2. 2. cost of second visit to bank branch 225 1 opportunity cost of one day wages 60 2 transportation cost of the visit 20 3 incidental cost of the visit 20 4 cost of account opening: cost of photos = rs. 15 cost of duplication of address proof = rs.5 transportation cost involved = rs.25 opportunity cost of time spent[one day] = rs.60 incidental costs[refreshments] = rs. 20 125 3. cost of document collection 150 1 opportunity cost of one day wages 60 2 transportation cost of the visit 20 3 incidental cost of the visit 20 4 cost of document collection 50 4. cost of loan applying 120 1 opportunity cost of one day wages 60 2 transportation cost of the visit 20 3 incidental costs of the visit 20 4 related expenses 20 5. cost of loan availment 325 1 opportunity cost of one day wages 60 2 transportation cost of the visit 20 3 incidental cost of the visit 20 4 procurement cost of stamp paper and revenue stamps: cost of stamp paper = rs. 50 cost of revenue stamps = rs.5, cost of photos = rs.10 opportunity cost of time spent [one day] = rs. 60 transportation cost involved = rs.25 incidental costs[refreshments] = rs.20 175 5. service charges for loan[ for an average loan of rs.10000 i.e @ rs.5 per thousand 50 total 920 the various components of transaction costs of borrowing for the poor under the traditional model of direct loans to individuals are fragmented and arranged here according to their order of occurrence. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 3 66 costs involved are: (1) cost of initial visit to bank branch, (2) cost of second visit to bank branch, (3) cost of document collection, (4) cost of loan applying and (5) cost of loan availment. after having estimated the various components of transaction costs involved in borrowing for the poor from the bank under model: direct loans to individuals, the total transaction cost involved is derived as mentioned below. further, we present here below (table 2) the proportion of above detailed activities in the transaction costs of borrowing by the poor under the direct lending by banks model. table 2: transaction costs of borrowing by the poor under direct lending by banks – activity wise distribution of costs sl. no. type of activity amount in inr percentage to total cost 1 cost of initial visit to bank branch 100 11% 2 cost of second visit to bank branch 225 24% 3 cost of document collection 150 16% 4 cost of loan applying 120 13% 5 cost of loan availment 325 36% total 920 100% it is noticed that a significant proportion of transaction costs is involved in visiting the bank branch (i.e 35%) and document collection (16%) and loan applying (13%). 5.1 transaction cost of borrowing for the poor under shg-bank linkage model after the emergence of shgs as the financial intermediaries for the poor under microfinance, the poor are grouped into shgs and in turn linked with banks for availing loans as explained earlier in this study. under this model, the poor members of the group approach the group for loans for their various purposes during their regular meetings. the group‘s borrowing member incurs transaction costs for loan availment in addition to other costs such as interest costs, and other charges levied by the banks. here in this assessment, the service charges of 1 per cent of the loan amount levied by the group on the borrowing members are considered as a component of transaction cost. it is believed that the member of the group does not incur any additional transaction cost apart from what the group has incurred during the linkage process. the various components of transaction costs of borrowing by the poor under the traditional model of direct loans to individuals are fragmented and arranged here below according to their order of occurrence. costs involved are: (1) cost of initial visit to bank branch swamy and tulasimala: transaction costs of borrowing by the poor 67 (by 3 office bearers of the shg), (2) cost of second visit to bank branch for submission of loan application and (3) cost of loan availment. table 3: costs of borrowing by the poor from the bank under shgs sl. no. type of activity money cost in inr total cost in inr 1. cost of initial visit to bank branch 270 1 opportunity cost of one day wages[3x rs.60] 180 2 transportation cost of the visit[3x rs.20] 60 3 incidental cost of the visit 30 2. cost of second visit to bank branch for submission of loan application 275 1 cost of preparation of resolution and loan project report 5 2 opportunity cost of one day wages[3x rs.60] 180 3 transportation cost of the visit[3x rs.20] 60 4 incidental cost of the visit 30 3. cost of loan availment 1690 1 opportunity cost of one day wages[3x rs.60] 180 2 transportation cost of the visit[3x rs.60] 60 3 incidental cost of the visit 30 4 procurement cost of stamp paper and revenue stamps: cost of stamp paper = rs.100 cost of revenue stamps = rs.5, cost of photos = rs.45 opportunity cost of time spent [one day x 3] = rs.60x3= rs.180 transportation cost involved[3x rs.20] = rs.60 incidental costs = rs.30 420 5. service charges for loan[ for an average loan of rs.100000] i.e @ rs.1 per rs.100 1000 total 2235 then, per member transaction cost== rs.2235/10 224 further, we present here below (table 4) the proportion of above detailed activities in the transaction costs of borrowing by the poor under shg-bank linkage model. figure 4 presents the comparison of activity-wise costs of transaction under both models as a comparison. table 4: costs of borrowing for the poor from the bank under shg-bank linkage model activity wise distribution of costs sl. no. type of activity amount in inr percentage to total cost 1 cost of initial visit to bank branch 270 12% 2 cost of second visit to bank branch for submission of loan application 275 12% 3 cost of loan availment 1690 76% total 2235 100% international journal of banking and finance, vol. 8, iss. 3 [2011], art. 3 68 figure 4: comparison of transaction costs of borrowing note: percentage of different sub costs to total transaction costs in models of comparison figure 5: comparison of transaction cost for borrower under direct loan model and shg-bank linkage model note: transaction costs in models of comparison represented in indian rupee (inr) swamy and tulasimala: transaction costs of borrowing by the poor 69 as is evident from these statistics of the survey results, direct lending by banks incurs the most transaction costs. obviously, linkage model attracts lower number of activities, hence lower the costs. figure 5 presents the comparison of transaction costs of borrowing along with activity wise costs under the models studied for comparison. the total cost of linkage model is inr 2,235 against the inr 1,840 for the direct lending model. obviously direct lending attracts less transaction costs. on a comparison, we notice that microfinance through shgs offers significant disadvantages in terms of transaction costs for borrowers when compared to the traditional direct loan model followed by banks (see figures 4 and 5). 6. findings, limitations, policy choices and conclusion shg-bank linkage is advantageous for the poor as the transaction cost of borrowing amounts to inr 224 per individual (table-2) as against that of inr 920 (table-1) under the direct lending by banks model. shg loan model‘s transaction cost of borrowing for the poor is about one fourth the transaction cost under direct lending by banks model. in view of this, it is beneficial to encourage financial provisions for the poor under the shg model in order to cut down the transaction costs for the poor. it is established by the study that transaction costs for the borrowing poor significantly decrease when they use the shg–bank linkage models of microfinance. it is important to ascribe due importance towards reduction of these costs in order to help the borrowing poor thereby inducing them to embrace microfinance to achieve economic development. the significant reduction in transaction costs of borrowing offers stimuli to all the prospective borrowers to show interest in the shg programs that is emerging as a microfinance tool for the benefit of the resource poor. shgs offer a significant reduction in such costs of transaction that makes another selling point for the shgs. generalisations based on the case studies should be done with caution. in view of the difficulty in covering all the areas of the country in sampling, the study area keeping in view of the prominent visibility of the microfinance activity in the area and also the awareness level of the beneficiaries. further, specific studies on transaction costs of borrowing by the poor at the national and global levels involving larger sample would be useful. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 3 70 the policy of supporting shg linkages with banks has merit in a country with a large bank network and as vast in space as india. however, additional efforts are needed to create and nurture competitive mfis willing to penetrate into the remote areas of country. banks should encourage shg-bank linkage since transaction costs of borrowing are significantly lower compared to transaction costs under banks‘ direct lending to individuals. some of the measures that can be taken to reduce the transaction costs are (a) simplifying the documentation process and (b) arranging provision of credit at doorsteps of the poor as far as possible. the local governments can consider utilizing the services of the line department personnel who are in close contacts with the poor to form shgs in order to bring down the transaction costs both for the lenders as well as for the borrowers. banks need to simplify their systems and procedures without compromising on the legalities involved to bring down the costs of transaction for borrowers too. the governments both at the centre and at the state can consider the possibilities of implementing the entire subsidy/margin oriented schemes through the shgs so that recovery performance and end utilization can be ensured along with reduction in the transaction costs of borrowing for the poor. this study has established that effective financial intermediation leads to financial development. to achieve financial development on a sustainable basis, it is important to reduce the transaction costs of not only the lenders but also that of the borrowers. this could be ensured by designing appropriate cost efficient financial intermediaries, which may significantly reduce the transaction costs. innovation in financial intermediation such as the microfinance models can surely spur the rate of financial development, which in turn can lead to economic growth. author information: the submitting author is a professor at the indian school of businesshyderabad, india. he may be contacted at vighneswar@ibsindia.org. his co-author b.k. tulasimala is from university of mysore, india. references: beck, t., levine, r., and loayza, n, (2000). finance and the sources of growth. journal of financial economics, 58: 261-300. beck, t., demirgüç-kunt, a., and levine, r., (2007). finance, inequality and the poor. journal of economic growth, 12(1): 27-49 bourguignon, f., (2003). the growth elasticity of poverty reduction; explaining heterogeneity across countries and time periods, in inequality and growth: theory and policy implications, ed. t. eicher and s. turnovsky, cambridge, ma: mit press. swamy and tulasimala: transaction costs of borrowing by the poor mailto:vighneswar@ibsindia.org 71 steven, cheung, n. s., (1992). on the new institutional economics in l.werin and h. wijkander (eds.), contract economics, basil blackwell publishers 1992: 48-65 deininger, k., and squire, l., (1998). new perspectives on old issues: inequality and growth. journal of development economics, 57(2): 259-288. demirgüç-kunt, a., and maksimovic, v., (1998). law, finance, and firm growth. journal of finance, 53: 2107–2137 dollar, d., and kraay, a., (2002). growth is good for the poor. journal of economic growth, 7: 195–25 gonzalez-vega claudio, schreiner mark, meyer richard l., rodriguez jorge and navajas sergio, (1997). bancosol: the challenge of growth for micro finance organisations, oecd publication karduck stefan and siebel hans dieter, (2004). transaction costs of self-help groups in nabard‘s shg banking program: a study in karnataka state. nabard, india levine, r, (1997). financial development and economic growth: views and agenda. journal of economic literature, 35: 688-726. levine, r, (1998). the legal environment, banks, and long-run economic growth. journal of money, credit and banking, 30: 596-613. levine, r, (1999). law, finance, and economic growth. journal of financial intermediation, 8: 36-67 llanto gilberto, m, and chua ronald t, (1996). transaction costs of lending to the poor: a case study of two philippine non-governmental organizations. www.bwtp.com nair, ajai, (2005). sustainability of microfinance self-help groups in india: would federating help? policy research working paper series, world bank patrick, h., (1966). financial development and economic growth in underdeveloped countries. economic development cultural change, 14: 174-189. puhazhendi, v., (1995). transaction costs of lending to the rural poor: non-governmental organizations and self-help groups of the poor as intermediaries for banks in india" (www.bwtp.com) rajan, raghuram, zingales g. luigi, (1998). financial dependence and growth. american economic review, 88 (3): 559-586 ravallion, m., (2001). growth, inequality and poverty: looking beyond averages, world development, 29(11): 1803-1815 sen, amartya, (2000). development as freedom, anchor books, new york international journal of banking and finance, vol. 8, iss. 3 [2011], art. 3 http://www.bwtp.com/ 72 srinivasan girija and satish, p., (2000). transaction costs of shg lending impact on branch viability. banker‘s institute of rural development, lucknow, india tankha, ajay, (2002). self-help groups as financial intermediaries in india: cost of promotion, sustainability and impact, sa-dhan, india white, h., and anderson, e., (2001). growth vs. redistribution: does the pattern of growth matter? development policy review, 19(3): 167-289. swamy and tulasimala: transaction costs of borrowing by the poor cetak a4 ijbf6(2)09.indd the international journal of banking and finance, 2008/09 vol. 6. number 2: 2009: 111-127 111 ijbf does noise signal affect flipping activities? chong fen nee, ruhani ali and zamri ahmad university technology mara sarawak and university sains malaysia abstract in this paper, we report the explanatory power of noise signal and fundamentals on fl ipping activities of share trading. flipping is defi ned as the percentage of opening day trading volume divided by the number of shares offered on the fi rst trading day (miller and reily, 1987, and aggarwal, 2003) in an offer for sale. it is affected by investors’ opinion about, for example, the new issue’s future prospect on the fi rst listing day. the initial premium which is defi ned as the difference between the opening price and the offer price divided by the offer price is used as a proxy for noise signal. using initial public offers listed on the main board of bursa malaysia during the period of 1991 to 2003, we fi nd support for the relationship between noise signal and fl ipping activity in the immediate aftermarket as evident in several models tested as well as the bullish and bearish market models. among the fundamental factors included in this study, bigger size of offer was found to discourage fl ipping activities. keywords: initial public offers, flipping, traded volume, issue size jel classifi cation: g12 1. introduction there has been an increasing interest in studying investors’ aftermarket trading dynamics in the initial public offerings or ipo market particularly in the u.s. this emerging trend was partly attributed to the resurgence of the behavioral fi nance paradigm. ellis, michaely and o’hara (2000) documented a positive infl uence of initial aftermarket behaviour on initial return of new issues. there has been no study of a smaller market as of yet, and it is worthwhile to test this idea in such a market with an active new issues. flipping refers to the immediate sale of the new issues when the issue begins trading. practitioners and academicians condemn fl ipping as a detrimental activity that depresses the early stage aftermarket performance of the new listings ht tp :// ijb f.u um .e du .m y 112 does noise signal affect fl ipping activities?: 111-127 (fishe, 2002; aggarwal, 2003). it was believed that investors with new listings will fl ip either based on superior information or opening trade performance. few have studied the implications of noise signal on fl ipping activities in new issues market. from the perspective of behavioral fi nance theories, every person is unique and therefore, given the uncertainty on the returns from subscribing to a new issue, homogeneity of opinion is impossible. under such a scenario, investors tend to make different estimates of expected returns from their investments. this difference in expected returns for the new issues is especially high due to the lack of track records at the time of the issue (miller, 1977). since behaviour is driven by expectations, and expectations are formed based on the information and noise trading information available to the investors, this study intends to study the effect of noise signal on fl ipping activity with ex-ante factors representing fundamental information about the issues. an average of about 91.35 percent of the investors in bursa malaysian are individual investors (bursa malaysia data and research centre, 2007), who are normally not informed investors and ipo trades are known to be based on noise (chen, hung and wu, 2002). add to this a different slate of disclosures as befi tting a developing market, this study of fl ipping activity in this market place may yield surprising results compared with the results reported for the more liquid, informed investors in that market. the malaysian market can be classifi ed as relatively illiquid (although not at the time of new issues) and with limited disclosures (campos, newell and wilson, 2002). indeed, the ineffi ciency of this market is documented by local researchers (mat nor, lai and hussin, 2002; lai, guru and mat nor, 2003; and husni, 2005) against the effi ciency evidence in annuar et al. (1994). hence, a study on the aftermarket trading dynamics and particularly on the effects of noise signal on fl ipping activity may provide modest set of results to add to the growing literature on this increasingly popular market for researchers. the rest of this paper is organized as follows. related literature on this topic is reviewed in the next section. section 3 contains a description of the data and methodology commonly used for studying fl ipping. in section 4, we present the empirical results to be discussed. lastly, in section 5, a summary of the results is found as a conclusion of this study. 2. literature review 2.1 flipping among the various immediate aftermarket activities, the investment activity of fl ipping new listings in the immediate aftermarket for a quick gain has become popular in studies related to aftermarket trading. most of these studies focus on the aftermarket stabilization activities of underwriters. underwriters in the american market often play the role as the market maker to provide liquidity and to offset downward price pressure due to fl ipping particularly in if the new issue is a weak offering. ht tp :// ijb f.u um .e du .m y does noise signal affect fl ipping activities?: 111-127 113 making quick decision under uncertainty based on noise signal has been studied by numerous researchers. the notion of ‘noise’ trading was introduced by black (1986) as he asserted that noise trading is trading based on noise as if it were information. two related studies by delong et al. (1987) and ritter (1991) considered noise traders as irrational traders who are extremely uncertain about future outcomes and fi nd it hard to predict the direction of the investment returns. due to this unpredictability of the noise trader, it is very risky for rational arbitrageurs to inhibit noise trader. consequently, prices can diverge signifi cantly from fundamental values: this phenomenon is well known when a crisis occurs, and noise trading increases. they contend that at times, it would be better off for the rational investors to follow and predict the guesses of the noise traders than choosing an appropriate portfolio based on fundamentals in the short-run. later studies by lux and marchesi (1999) and brown (1999) described noise traders as individual investors who do not believe in fundamental but rely on following the behaviour of others or on noise signals about trade as the primary source of information in assigning a value to an asset. additionally, jones’ (2004) description of noise trader as small speculator with no special information also fi ts well with the malaysian investors profi le which are made up of uninformed individuals forming the majority who move the market. no doubt that modern fi nancial market depends on trading volume for their very existence. it is the trading commissions and spreads that remunerate the brokers, remisiers and other dealers. flipping has provided liquidity to new issues in the period after the listing. however, despite this positive aspect, excessive fl ipping is discouraged as it is detrimental to the market performance of the new listings. in fact, the securities commission has imposed a moratorium (similar to “lock-up” in the u.s.) to prohibit substantial share holders who are normally the promoters of the new listings from disposing their share holdings right after the listing of the stocks as part of the effort to protect minority share holders’ interest. nevertheless, due to the different institutional background, on top of the lock-up period imposed, underwriters in the u.s. also trade to ‘stabilize’ ipo initial price through various activities aimed at reducing selling pressure if severe fl ipping is observed. the u.s. securities commission allows underwriters to buy back up to an extra of 35 percent and retire the shares, as if they have never been offered before. this over-allotment option is also known as the green shoe option. ritter and welch (2002) contended that almost all ipos contain an overallotment option for up to 35 percent of the shares offered. this claim has been supported by aggarwal (2000). the fear of losing market share due to excessive overpricing or underpricing of an ipo (beatty and ritter, 1986) could be the main motive which drives underwriters to be actively involved in stabilizing activities. in addition to price stabilization during heavy fl ipping, another precautionary activity that underwriters undertake to restrict heavy fl ipping is by allocating more ipos in ‘strong hands’ who are committed to retain the allocations rather than ‘weak hands’ that are likely to fl ip for profi t shortly after the issue (aggarwal ht tp :// ijb f.u um .e du .m y 114 does noise signal affect fl ipping activities?: 111-127 and rivoli, 1990). normally, the ‘strong hands’ are allocated with ipos via private placements. if strong hands fl ip, they are likely not to be included in the future initial offerings. using a sample of 137 ipos take place during the period of may to july 1997, aggarwal (2000) found that more than half the issues have short positions averaging 10.75 percent of the shares offered. a total of 22 transactions were performed over 16.6 days in the aftermarket to cover the short positions. this aftermarket price stabilization has resulted in a loss of 3.61 percent underwriting fees. she highlighted that the aftermarket activities is not transparent and underwriters manage price support activities by using a combination of aftermarket short covering to stimulate demand, penalty bids to restrict supply to fl ippers and the selective use of the over-allotment options. ellis, michaely and o’hara (2000) confi rmed that lead underwriter is the dominant market maker who takes substantial inventories to support the immediate aftermarket trading. fishe (2002) investigated the effect of stock fl ipping activity and stabilization activity. he contended that in the context of american market evidence, underwriters determine the level of fl ipping activities and subsequent demand comes from demand with or without the fl ippers at the book-building stage. underwriters then incorporate this information in pricing the issue, aiming to increase underwriting profi t at the expense of fl ippers. flipping activity generally lowers the level of underpricing, that is it increases the open-day price. moreover, with a green shoe option, underwriters can oversell the issue and then buy back at a much lower aftermarket price particularly if it is a weak issue. hence, in contrast to aggarwal (2000), fishe (2002) posits that price stabilization could actually allow the underwriter to make a profi t and penalize fl ippers rather than merely reducing the investors’ loss. krigman, shaw and womack (1999) on the other hand, examined underwriters’ pricing errors and the predictive power of fl ipping by studying 1,232 large-cap new issues over 1988 to 1995. they argued that fl ipping is not a cause of poor aftermarket performance as stated earlier, rather, it is a rational and appropriate response to unfavourable pricing. this is because new listings recorded with higher fl ipping activities performed worse in the long-run while those with low fl ipping generate abnormal return of 1.5 percent per month over the fi rst six month after listings. furthermore, it appears that information about the quality of issuing fi rms is available because large institutional investors, who are assumed to be more informed, sell issues that produce worse future performance. therefore, underwriter’s pricing errors are intentional. they also point that larger offerings recorded a higher fl ipping activity while initial return is negatively associated with institutional fl ipping activities. flipping is measured as the percentage of sell-signed dollar volume executed in trades of 10,000 shares or more on the fi rst trading day over the total dollar volume traded on the fi rst day in their study. a later study related to fl ipping by aggrawal (2003), however, documented anomalous result after analyzing 617 cases during may to june 1998. she reported that hot ipo are fl ipped more than the cold ipo. in other words, her ht tp :// ijb f.u um .e du .m y does noise signal affect fl ipping activities?: 111-127 115 fi nding implies that there is a positive relationship between fl ipping and initial return. a recent study on fl ipping activities in australia by bayley, lee and walter (2006) postulates that underpriced ipos are fl ipped more than overpriced ipos. additionally, fl ipping behaviour of informed investors is unrelated to longrun returns whilst uninformed investors consistently fl ipped more of the ipos with better long-run returns than their informed counterparts, the institutional investors. this claim is contradictory to that of krigman, shaw and womack (1999) which showed that institutional investors fl ip more of the ipos which performed worse in the long-run. however, the result of bayley, lee and walter (2006) is consistent with that of aggarwal (2003). additionally, bayley, lee and walter (2006) also report a negative relationship between fi rm size and fl ipping. nonetheless, no relationship was found between underwriters’ reputation and fl ipping activities. 2.2 noise trading noise trading models in fi nance suggest that some investors, especially the less sophisticated individual traders often do not make investment decisions based on fundamentals and that this group is capable of infl uencing stock prices by way of unpredictable changes in their sentiments. black (1986) contended that the impact of noise on the views of the world is profound. noise in the form of expectation need not follow rational rules. noise do not only cause markets to be ineffi cient but also result in infl ation. nevertheless, noise trading plays an important role contributing to enhancing the liquid fi nancial markets. besides, noise trading also provides a solution to the “no-trade” theorems suggested by milgrom and stokey (1982). the creation of noise trading model by delong et al. (1987; 1991) suggests that investor sentiments impacts stock returns and volatility. these two ideas led numerous further studies in this area. while investigating the economic consequences of noise trading, they considered noise traders as irrational traders who are extremely uncertain and it is hard to predict their actions. moreover, with the unpredictability of noise traders’ belief, it is very risky for rational arbitrageurs to inhibit noise trader. because of this, prices can diverge signifi cantly from fundamental. consequently, noise traders with erroneous and stochastic beliefs could affect prices signifi cantly and manage to earn higher returns than the rational investors. these fi ndings have falsifi ed the claim of emh believers – as far as new issues market is concerned who argue that fi nancial markets are effi cient because misinformed noise traders can have little infl uence on asset prices in equilibrium. in addition, they also documented that since stock prices are more volatile than can be justifi ed based on the news of the underlying fundamentals, therefore, it would also be better off for a rational investor to predict the guesses of others than choosing the appropriate portfolio in the short run. lux and marchesi (1999) divided the pool of investors into the fundamentalist and the noise traders. they defi ned fundamentalist as those ht tp :// ijb f.u um .e du .m y 116 does noise signal affect fl ipping activities?: 111-127 who follow the premise of emh whereby, the value of assets are equal to the discounted sum of the future earnings. noise traders, however are investors who rely on the behaviour of others in the market as the primary source of information in assigning the value of the asset. at times, they found that the interactions between these two groups in the market drive the fundamentalists to trade more like noise traders to get short-term profi t. consistent with lux and marchesi (1999), menkhoff (1998) confi rmed that rational market participants use nonfundamental analyses to exploit less-rational noise traders in the german foreign exchange market. brown (1999) extended noise trading study to the closed-end investment funds in the u.s. market. he postulated that noise-trader theory implies a scenario whereby irrational investors acting coherently on a noisy signal as constituting the systematic risk. if noise traders affect prices then, the noisy signal is sentiment and the risk caused by noise traders is manifested as volatility. brown (1999) found that unusual levels of individual investor sentiment are related to greater volatility of closed-ended investment funds as well. moreover, this volatility occurs only when the market is open. jaggia and thosar (2007) shed some light on the infl uence of noise trading in the new issues market. their investigation on the medium-term aftermarket of high-tech u.s. issues in the late 1990s suggests that, the interplay between noise and informed trading has substantial effect on prices. their model also indicated that behavioral variables are more vital in infl uencing pricing behaviour than fundamental variables. berkman and koch (2007) proposed the dispersion in daily net initiated order fl ow across brokers as proxy for the level of noise trading in a stock. using data from the australian stock exchange, they found daily movements of the noise measure are positively associated with trading volume and market depth. additionally, they also reported a signifi cantly greater sensitivity of stock price to net initial order fl ow on mondays. bloomfi eld, o’hara and saar (2007) used a laboratory market to investigate the behaviour of noise traders and their impact on the stock market. they categorized participants in their experiment into three groups: (i) informed traders who possess fundamental information (ii) liquidity traders who have to trade for exogenous reasons and (iii) noise traders who do not possess fundamental information and have no exogenous reasons to trade. they found behavioral differences between liquidity and noise traders. noise traders were found to exert positive impacts on both the volume and depth of the market. furthermore, noise traders only adversely affect the information effi ciency of the market when informed traders have very valuable private information. finally, their fi nding showed that the imposition of tobin tax to curb speculators produce no differential results between the behaviour of different groups of traders. in fact, tax reduces volume across groups. 2.3 fundamentals (ex-ante factors) market performance is related to investors’ behaviour while the knowledge about an asset in turn affects the investors’ behaviour. numerous researches ht tp :// ijb f.u um .e du .m y does noise signal affect fl ipping activities?: 111-127 117 have developed theories and models based on information asymmetry to explain performance (ibbotson, 1975; miller 1977; baron, 1982; rock 1986; beatty and ritter, 1986; grinblatt and hwang, 1989; ritter, 1991; welch, 1992). pre-listing information proxing quality or ex-ante factors are proven to have predictive power over ipo performance. since behaviour such as fl ipping is affected by the opinions or expectations formed after processing the information available, therefore, ex-ante factors such as operating history, market conditions, subscription ratio, underwriters’ reputation and size of offer are included as control variables. these ex-ante variables also represent fundamental information as they are either published in the prospectuses or are refl ections of current market situation. 3. data and methodology the criteria adopted for sample selection are that the new listings were listed on the main board between 1991 and 2003 and that all information is available. issues are only included in the sample from the main board as the issues therein are more or less similar in quality. consequently, a total of 132 new issues formed the sample out of a population of 232 for this study. the sample and required data are drawn from: securities commission, bursa malaysia research and data centre, bank negara malaysia and their relevant publications. the aftermarket behaviour examined in this study is fl ipping ratio. it is defi ned as the percentage of opening day trading volume divided by the number of shares offered on the fi rst trading day (miller and reily, 1987 and aggarwal, 2003). this defi nition is different from that of krigman, shaw and womack (1999) and houge et al. (2001) studies, where only sell-signed block volume is used. model 1 is developed to explore the predictive power of noise signal and ex-ante factors on fl ipping behaviour. it is a regression using the primary explanatory variable, namely, initial premium representing noise signal and the control variables (ex-ante factors) against the dependent variable, fl ipping ratio on the listing day. since the asian financial crisis which started on june, 1997 could have a signifi cant impact on the results of model 1, we further partitioned the sample into bullish market samples which consist of cases listed before 1 june, 1997 and bearish market sample made up of cases listed after 1 june, 1997. the new issues declined substantially due to the crisis. ex-ante risk factors on fundamentals in this study are well-known for their roles in affecting investor expectations and valuations in the prelisting stage. ritter (1984) as well as beatty and ritter (1986) demonstrated a causal relationship between the ex-ante factors and the performance of the ipos. since behaviour is a function of information gauged, therefore, common proxies used for ex-ante uncertainty are: operating history, fi rm size, market conditions, subscription ratio and underwriter reputations. explanations on the expected signs of the independent variables used in model 1 are given below. ht tp :// ijb f.u um .e du .m y 118 does noise signal affect fl ipping activities?: 111-127 model 1: (fr) = α β 1 oh +β 2 mktcon + β 3 sr β 4 logsizeoff β 5 ur + β 6 ip +ε where, fr : flipping ratio, oh : operating history, defi ned as the length of existence of the company prior to ipo, mktcon : market condition, the one-week average market index return prior to listing, sr : subscription ratio, the percentage of new issues over the total of new issues offered, logsizeoff : size of offer, total number of shares fl oated times the subscription price, ur dummy : quality of the lead underwriter; ur dummy equals to ‘1’ for prestigious underwriter and ‘0’ otherwise as per the bank negara malaysia rating, and ip : initial premium, difference between opening price and offer price divided by offer price. from the perspective of divergence of opinion theory, a greater ex-ante uncertainty should lead to a wider divergence of opinion and hence a higher fl ipping activities is predicted (miller, 1977, miller, 2000 and houge et al. 2001). since ex-ante factors are indicators of uncertainty, hence, factors with higher uncertainty will have a positive relationship with fl ipping ratio and vice versa. companies with longer operating history are more well established. these companies have longer track records and therefore, are presumably easier to value and hence, the uncertainties being lower. this claim has been empirically proved in studies (beatty and ritter, 1986; ritter, 1984; 1991; tay, 1993 and klymaz, 2000). since higher the ex-ante uncertainty is predictive of larger disagreement concerning aftermarket prices, so larger disagreement is positively associated with immediate aftermarket behaviour (houge et al., 2001). therefore, operating history is negatively related to fl ipping ratio. following klymaz (2000), market condition is calculated as the average market index return prior to the fi rst trading day. the average klci value one week prior to the listing date are used in this study. some studies proved that issuers are exploiting the ‘window of opportunity’ while other studies suggested that ipos are subject to ‘fad’ (aggarwal and rivoli, 1990; ritter, 1991 and chan, wang and wei, 2004). furthermore, ritter and welch (2002) introduced the market timing theory which suggests that entrepreneurs are more inclined to sell shares in a bull market when investors are over-optimistic. additional evidence on positive relationship between market conditions and level of underpricing is provided by isa and ahmad (1996) and klymaz (2000). generally, a better ht tp :// ijb f.u um .e du .m y does noise signal affect fl ipping activities?: 111-127 119 market condition is associated with a lower expected risk and hence, investors are more optimistic. with that conceptualization, a positive sign is expected from this variable. subscription ratio is defi ned as the demand of a particular new listing. ismail, abidin and nasrudin (1993), yong, yatim and sapian (1999), lee, taylor and walter (1999) and wan-hussin (2002) showed that there is a positive relationship between initial return and oversubscription ratio in malaysia and singapore new issues markets. a higher subscription ratio refl ects the confi dence and the optimism that pre-ipo investors have on new issues. as such, a positive relationship is expected between subscription ratio and fl ipping activity. allen and faulhaber (1989) as well as welch (1989) proposed that signaling can be accomplished through the choice of underwriter. additionally, carter and manaster (1990) stated that high-quality underwriters are typically associated with less underpricing. nonetheless, loughran and ritter (2001) contended that new listings underwritten by high prestige underwriters during the dot-com boom in the late 1990s had higher underpricing than those managed by the less prestigious underwriters. in line with the fi ndings from other asian countries, wan-hussin (2002) found no association between underwriter reputation and ipo valuation. since prestigious underwriters are generally associated with higher quality and less uncertain ipo. thus, this study expects a negative relationship between prestigious underwriter and fl ipping activities. size of offer can be measured using gross proceeds of the new issues. it is a proxy for fi rm size. according to ritter (1984), smaller issues are more likely to be affected by speculative forces and as a result, ex-ante uncertainty is expected to be greater for smaller fi rms. nonetheless, krigman, shaw and womack (1999) discovered that larger offerings recorded a higher institutional fl ipping. in malaysia, consistent with ritter (1984), corhay, teo and rad (2002) discovered a negative relationship between size of offer and market return. with the conceptualization that smaller fi rm is subjected to higher uncertainty, therefore, a negative sign is expected for size of offer. initial premium is defi ned as the difference between the opening price and the offer price divided by the offer price. it is the fi rst pricing indictor available on the fi rst trading day for investors to imitate and form expectations. hence, it also represents noise signal in this study. since initial performance is posited to have a positive impact on fl ipping ratio (krigman, shaw and womack, 1999), thus, a positive relationship is expected. 4. empirical results and analysis this section discusses the empirical results and analysis in three sub-sections. sub-section 4.1 reports the summary statistics of the dependent variables while 4.2 contains a discussion on the correlation coeffi cients. following that, in 4.3, we discuss the empirical results of the cross-sectional analyses. ht tp :// ijb f.u um .e du .m y 120 does noise signal affect fl ipping activities?: 111-127 f li p p in g r at io n figure 1: summary statistics on flipping ratios by years figure 1 shows the average fl ipping ratio of investors by cohort years. flipping ratio is defi ned as the percentage of opening day trading volume divided by the number of shares offered on the fi rst trading day. overall, fl ipping activities account for a rather low trading volume which ranges from 1.5 percent to 11.72 percent with an average of 7.66 percent. the moratorium imposed by the securities commission prohibiting substantial shareholders from disposing their shares right after the listing may have caused the low trading volume. furthermore, the fl ipping ratios reported fl uctuate without specifi c trend by cohort years. nevertheless, the fi nancial crisis which begins in mid-1997 as well as the post crisis till 2001 has recoded an obvious drop in fl ipping activities. a mean of 8.08 percent in fl ipping ratio is reported for the bullish market samples while an average of 6.81 percent is observed in the bearish market. the results of pearson correlation analysis between the variables are presented in table 1. all the independent variables included in the model are found to be signifi cantly related to the dependent variable and the signs are as predicted except for underwriter’s reputation. nonetheless, as discussed in the literature, the relationship between underwriter’s reputation and initial return is inconclusive. referring to table 1, among the independent variables included in the model, some are signifi cantly related but others are not. since all the correlations among the independent variables are below 0.75, multicollinearity problem is not a major concern in this study rendering it to be allowed for further testing through multiple regression procedure. (cavana, delahaye and sekaran, 2001). 4.1 empirical results for cross-sectional analysis our analysis focuses on exploring the explanatory power of the noise signal and ex-ante factors on fl ipping ratio. table 2 reports the cross-sectional regression results of noise signal and ex-ante factors on fl ipping ratio for overall samples. table 3 shows the cross-sectional results for the bullish market samples and lastly, table 4 portrays the fi ndings of the bearish market samples. 0 2 4 6 8 10 12 14 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 listing year f li p p in g r at io f li p p in g r at io ht tp :// ijb f.u um .e du .m y does noise signal affect fl ipping activities?: 111-127 121 t ab le 1 : p ea rs on c or re la ti on c oe ffi c ie nt s a na ly si s fo r p ri m ar y an d c on tr ol v ar ia bl es v a ri a b le s f li p p in g r a ti o o p er a ti n g h is to ry m a rk et c o n d it io n s u b sc ri p ti o n r a ti o l o g s iz e o f o ff er u n d er w ri te r r ep u ta ti o n in it ia l p re m iu m f li pp in g r at io 1 .0 0 o pe ra ti ng h is to ry -0 .1 65 ** 1 .0 0 m ar ke t c on di ti on 0 .1 84 ** -0 .2 04 ** 1 .0 0 s ub sc ri pt io n r at io 0 .2 45 ** -0 .1 92 ** 0 .6 06 ** 1 .0 0 l o g s iz e of o ff er -0 .2 78 ** 0 .1 04 0 .2 47 ** 0. 15 3* * 1 .0 0 u nd er w ri te r r ep ut at io n 0 .0 27 -0 .0 47 0. 06 7 0. 01 1 0. 02 4 1 .0 0 in it ia l p re m iu m 0 .2 65 ** -0 .0 96 0 .6 17 ** 0 .6 09 ** 0 .1 45 * 0. 02 4 1 .0 0 ** c or re la ti on i s si gn ifi c an t at 0 .0 1 le ve l (1 -t ai le d) . * c or re la ti on i s si gn ifi c an t at 0 .0 5 le ve l (1 -t ai le d) . ht tp :// ijb f.u um .e du .m y 122 does noise signal affect fl ipping activities?: 111-127 table 2: cross-sectional regression results of flipping ratio, noise signal and ex-ante factors, overall sample results of the cross sectional regression analysis using fl ipping ratio as dependent variable and noise signal as well as ex-ante factors as independent variables are reported in table 2. overall, the model reports an adjusted r squared of 14.3 percent and this result is signifi cant at the 0.05 signifi cance level. initial premium which is used as a proxy for noise signal is found to be a signifi cant predictor over fl ipping activities. among the fundamental proxies, log size of offer representing fi rm size is found to have signifi cant impact on the fl ipping behaviour. the report of a negative relationship between fi rm size and fl ipping activity is supported as in bayley, lee and walter (2006). the model has a good fi t with a signifi cant f-ratio of 4.647. the explained variation as judged by adjusted r-squared value is small, 14.3 percent. thus, this model appear to provide support for fl ipping activities in the bursa malaysia. the asian fi nancial crisis has affected the performance of the share market. after the commencement of the fi nancial crisis, the share market has been in a bearish mode for a considerable time. hence, to get a better feel for the effect of the crisis on fl ipping activity, we partitioned the sample into the bull and bear market samples using 1 june, 1997 as the cut-off point. the result was 87 cases were listed before the cut-off date under bullish market condition; 45 cases under the bearish market condition. referring to the statistics in table 3, which reports results of the cross-sectional analysis for bullish market samples, it is found that investors fl ipping activity is signifi cantly affected by noise signal. nevertheless, none of the ex-ante factors included in the analysis is having any explanatory power for fl ipping activities. this is consistent with noise trading in a bullish period in that during a bull market, prices are going up, and the behaviour is to fl ip more to get higher return than is possible based on fundamentals alone. the model fi t is signifi cant and the r-squared value is very low, 11 percent. variable dependent variable: flipping ratio independent variables β t -stat operating history -0.001 -1.061 subscription ratio -5.50e-005 -0.182 market condition 4.93e-006 0.733 log size of offer -0.046 -3.558*** underwriter’s reputation dummy 0.003 0.329 initial premium 0.009 2.020** f value 4.647** r2 0.182 adjusted r2 0.143 **signifi cant at 5% level (one-tail test). ht tp :// ijb f.u um .e du .m y does noise signal affect fl ipping activities?: 111-127 123 table 3: cross-sectional regression results of flipping ratio, noise signal and ex-ante factors for bullish market samples table 4: cross-sectional regression results of flipping ratio, noise signal and ex-ante factors for bearish market samples cross-sectional regression results of noise signal and ex-ante factors on fl ipping ratio for the bearish market sample is presented in table 4. findings show that the bearish market model is signifi cant with an f value of 4.649 and variable dependent variable: flipping ratio independent variables β t -stat operating history -0.001 -0.894 subscription ratio 5.88e-006 0.017 market condition 5.62e-006 0.190 log size of offer -0.032 -1.813 underwriter’s reputation dummy -0.003 -0.197 initial premium 0.008 1.981** f value 4.647** r2 0.182 adjusted r2 0.143 **signifi cant at 5% level (one-tail test). variable dependent variable: flipping ratio independent variables β t -stat operating history -0.001 -0.857 subscription ratio 0.002 1.600 market condition 5.04e-005 0.627 log size of offer -0.043 -2.047** underwriter’s reputation dummy 0.006 0.330 initial premium 0.023 1.995** f value 4.649** r2 0.430 adjusted r2 0.337 **signifi cant at 5% level (one-tail test). ht tp :// ijb f.u um .e du .m y 124 does noise signal affect fl ipping activities?: 111-127 an adjusted r-squared of 33.7 percent. consistent with the bullish market model, initial premium proxing noise signal is found to have a signifi cant infl uence on the fl ipping activities in the period after the crisis had hit. the only ex-ante factor which can signifi cantly explain fl ipping activity of the bearish market is log size of offer. consistent with the results on the overall sample, a bigger size of offer is found to discourage fl ipping activities when the market is bearish. these results are also in accord with common sense, and the prior literature on developed markets such as the u.s. and austalia. 5. summary and conclusion cross-sectional regression analysis using fl ipping ratio as dependent variable, initial premium as primary explanatory variable representing noise signal and the ex-ante factors proxying fundamentals found that the model is signifi cant. test results show that independent variables included are able to explain 14.3 percent of the variations in fl ipping activities during the fi rst listing day; 11 percent during bullish market and 33.7 percent during bearish market conditions. initial premium representing noise signal was found to have a signifi cant positive infl uence on the investors’ decisions to fl ip a particular ipo on the listing day. among the proxies for fundamental factors, log size of offer as company size is identifi ed to be a signifi cant predictor over fl ipping activities. similar analyses done on the bullish and bearish market samples uncovered that investors’ fl ipping behaviour is signifi cantly associated to noise signal regardless of the market conditions. none of the ex-ante factors representing fundamentals is found to have an infl uence on the bullish market model. on the other hand, in line with the fi nding provided by the overall sample, investors’ fl ipping behaviour is found to be signifi cantly infl uenced by size of offer that represents. in conclusion, this paper has presented some new fi ndings on how noise signal and ex-ante risks factors affect the investor behaviour described as fl ipping. additionally, it also provides evidence on the relative importance of each ex-ante factors in predicting fl ipping activity on the fi rst listing day. this is to our knowledge a fi rst study to report evidence of fl ipping behaviour in an emerging market. that behaviour appears to be slightly different in bearish and bullish market conditions. to generalize these fi ndings to other markets, more publications on few more emerging markets is necessary. author statement: the submitting author is a lecturer in the faculty of business management, uitm, kampus kota samarahan, jalan meranek, 94300 kota samarahan, sarawak malaysia. e-mail: wuinee@pd.jaring.my. the co-authors are respectively ruhani ali, professor, and zamri ahmad, at the university sains malaysia: the authors express their thanks to the extensive editing of the paper by the editors of the journal following the review process. the authors alone are responsible for any errors. ht tp :// ijb f.u um .e du .m y does noise signal affect fl ipping activities?: 111-127 125 references aggarwal, r., and rivoli, p. (1990). fads in ipo market? financial management, 19, 45 – 57. aggarwal, r., and rivoli, p. (2000). stabilization activities by underwriters after initial public offerings. the journal of finance, lv (3), 1075-1103. aggarwal, r., and rivoli, p. (2003). allocation of initial public offerings and fl ipping activity. journal of financial economics, 68, 111-135. allen, f., and faulhaber, g. (1989). signalling by underpricing in the ipo market. journal of financial economics, 23, 303 – 323. annuar, m.n., ariff, m., and shamsher m. (1994). is kuala lumpur’s emerging share market effi cient? journal of international financial markets, institutions & money, 4(1), 89-100. baron, d. (1982). a model of the demand for investment banking advising and distribution services for new issues. the journal of finance, 37(4), 955976. bayley, l., lee, p., and walter, t.s. (2006) ipo fl ipping in australia: crosssectional explanations. pacifi c-basin finance journal, 14, 327-348. beatty, r.p., and ritter, j.r. (1986). investment banking, reputation and the underpricing of initial public offerings. journal of financial economics, 15, 213 – 232. berkman, h., and koch, p.d. (2007). noise trading and the price formation process. journal of empirical finance. black, f. (1986). noise. the journal of finance, 41(3), 529 -543. bloomfi eld, r., o’hara, m., and saar, g. (2007). how noise trading affects markets: an experimental analysis. retrieved from http://www.ssrn.com. brown, g.w. (1999). volatility, sentiment and noise traders. financial analysts journal, 82-90. bursa malaysia research and data centre. (2007). data on investors’ profi le. kuala lumpur: bursa malaysia. carter, r., and manaster, s. (1990). initial public offerings and underwriter reputations. the journal of finance, 45(4), 1045-1067. cavana, r.y., delahaye, b. i., and sekaran, u. (2001). applied business research: qualitative and quantitative methods. new york: john wiley & sons. inc. chan, k., wang, j., and wei, k.c. (2004). underpricing and long-term performance of ipos in china. journal of corporate finance, 10, 409 430 chen, a., hung, c.c., and wu, c.s. (2002). the underpricing and excess returns of initial public offerings in taiwan based on noisy trading: a stochastic frontier model: review of qualitative finance and accounting, 18(2), 139-158. campos, c.e., newell, r.e., and wilson, g. (2002). corporate governance develops in emerging markets. retrieved from http: www.srrn.com ht tp :// ijb f.u um .e du .m y 126 does noise signal affect fl ipping activities?: 111-127 corhay, a., teo, s., and rad, a. (2002). the long run performance of malaysian initial public offerings: value and growth effects. managerial finance, 28(2), 52-65. delong, b. j., shleifer, a., summers, l. h., and waldmann, r. (1987). the economic consequences of noise traders. working paper (no. w2395). national bureau of economic research (nber), harvard university. delong, b. j., shleifer, a., summers, l. h., and waldmann, r. (1991). the survival of noise traders in fi nancial market. journal of business, 64,1– 19. ellis, k., michaely, r., and o’hara, m. (2000). when the underwriter is the market maker: an examination of trading in the ipo aftermarket. the journal of finance, lv (3), 1039-1074. fishe, r.p.h. (2002). how stock fl ippers affect ipo pricing and stabilization. journal of financial and quantitative analysis, 37(2), 319-340. grinblatt, m., and hwang c.y. (1989). signalling and pricing of new issues. the journal of finance, 44(2), 393-420. houge, t. (2001). divergence of opinion, uncertainty and the quality of initial public offerings. financial management, 30(4), 5-23. husni, t. (2005). price randomness, contrarians and momentum strategies: a study of return predictability in the malaysian stock exchange (doctoral dissertation universiti sains malaysia, 1990). ibbotson, r.g. (1975). price performance of common stock new issues. journal of financial economics, 2, 235-272. isa, m., and ahmad, r. (1996). performance of new issues on the malaysia stock market, malaysia, journal of economic studies, 13(2), 53-66. ismail k., abidin, f., and nasruddin, z. (1993). performance of new stock issues on the klse. capital market review, 1(1), 81-95. jaggia, s., and thosar, s. (2007). the medium-term aftermarket in hightech ipos: patterns and implications. journal of banking and finance (forthcoming). jones, d. (2004). noise traders, a brief review. retrieved from http://www. cisco futures.com klymaz, h. (2000). the initial and aftermarket performance of ipos in an emerging market: evidence from istanbul stock exchange. journal of multinational financial management, 2(2), 213-227. krigman, l., shaw, w., and womack, k. (1999). the persistence of ipo mispricing and the predictive power of fl ipping. the journal of finance, 54(3), 1015-1044. kuala lumpur stock exchange, listing statistics. retrieved from http://www. bursamalaysia.com lai, m.m., guru, k., and mat nor, f. (2003). do malaysian investors overreact? journal of american academy of business, 2(2), 602-609. lee, p.j, taylor, s.l., and walter t, s. (1996). expected and realised returns for singaporean ipos: initial and long-run analysis. pacifi c-basin finance journal, 4(1-2), 153-180. ht tp :// ijb f.u um .e du .m y does noise signal affect fl ipping activities?: 111-127 127 loughran , t., and ritter, j. r. (2001). why has ipo underpricing increased over time? (working paper). university of florida. loughran, t., and ritter, j. r. (2004). why has ipo underpricing changed over time? financial management, 33(3), 5-38. lux, t., and marchesi, m. (1999). scaling and criticality in a stochastic multiagent model of a fi nancial market. nature, 397, 498-500. mat-nor, f., lai, m.m., and hussin, a. m. (2002). price randomness, fundamental factors, and stock market contrarian strategy: further evidence on malaysian stock market. proceedings of the 4th malaysian finance association. malaysia: penang. menkhoff, l. (1998). the noise trading approach – questionnaire evidence from foreign exchange. journal of international money and finance, 17, 547567. milgrom, p., and stokey, n. (1982). information, trade and common knowledge. journal of economy theory, 26, 7-27. miller, e. (1977). risk, uncertainty and divergence of opinion. the journal of finance, 32(4), 1151-1168. miller, e. (2000). equilibrium with divergence of opinion. review of financial economics, 9, 27-41. miller, r. e., and reilly, f. k. (1987). an examination of mispricing, returns, and uncertainty for initial public offerings. financial management, 16, 33-38. ritter, j. f. (1984). the hot issue market of 1980. the journal of business, 57(2), 215-240. ritter, j. r. (1991). the long-run performance of initial public offerings. the journal of finance, xlvi (1), 3-27. ritter j. r., and welch, i. (2002). a review of ipo activity, pricing and allocation. journal of finance, 57(4), 1795-1829. rock, k. (1986). why new issues are underpriced? journal of financial economics, 15, 187-212. tay, s.w. (1993). the long run performance of initial public offerings in malaysia. capital markets review, 1, 61-80. wan-hussin, w.n. (2002). investor protection mechanism and ipo valuation on the kuala lumpur stock exchange. retrieved from http://www.ssrn. com welch i. (1989). seasoned offerings, imitation costs, and the underpricing of initial public offerings. the journal of finance, 44(2), 421-449. welch, i. (1992). sequential sales, learning and cascades. the journal of finance, 47, 695 732 yong, o., yatim, p., and sapian, r. z. (1999). signifi cance of board of listing and type of issue on the under-pricing and after-market performance of malaysian ipos. capital market review, 7(1) & (2), 47-74. ht tp :// ijb f.u um .e du .m y ht tp :// ijb f.u um .e du .m y cetak a4 ijbf6(2)09.indd the international journal of banking and finance, 2008/09 vol. 6. number 2: 2009: 77-93 77 ijbf risk-adjusted returns of american depositary receipts on chinese and indian stocks onur arugaslan and ajay samant western michigan university abstract this study evaluates the risk-adjusted performance of american depositary receipts (adrs) on shares of stock of chinese and indian fi rms. the fi rst part of the study examines the nature of chinese and indian adrs (based on depositary bank, sponsorship status, industry classifi cation and listing). the second part of the study evaluates the performance of these adrs using statistical measures grounded in modern portfolio theory. returns are adjusted for the degree of total risk and systematic risk inherent in each adr, and the securities are then ranked on the basis of risk-adjusted performance. two relatively new evaluation metrics, the modigliani and sortino measures, are used. the objective of the study is to provide documentation to global investors who are contemplating participation in chinese and indian stock markets via depositary receipts. keywords: adrs, portfolio choice, investment decisions, emerging markets jel classifi cations: g15, g11, f30 1. introduction there has been a signifi cant rise in investor comfort over the past decade with global fi nancial securities, aided by the ease and convenience with which transnational corporate information can be accessed via the internet and other means. one of the most convenient vehicles for accessing corporate securities listed outside the investor’s home country is a global depositary receipt. in the united states, these securities are known as american depositary receipts (adrs). as of august 2007, there were 1,703 adrs listed on the new york stock exchange (nyse), american stock exchange (amex), the nasdaq system, and on private trading networks. this study examines the nature and performance of adrs on shares of fi rms incorporated in china and india. as is well known, china and india are the two largest countries in the world in terms of population, and each country has a ht tp :// ijb f.u um .e du .m y 78 risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 diversifi ed industrial base, a rapidly growing consumer market, and an emerging stock market. table 1 summarizes information such as market capitalization, gross domestic product, and labor force for these two countries. as a result of excellent growth opportunities evident in those numbers, each of these countries is under increasing scrutiny by international investors. table 1: summary economic measures for china and india china india market capitalization ($ million)* 780,763 553,074 value traded ($ million)* 586,301 443,175 number of listed domestic companies* 1,387 4,763 average company size ($ million)* 562.9 116.1 gross domestic product (gdp) ($ billion)** 2,518 804 gdp growth (2005-2006) 10.7% 9.2% average gdp growth (1990-2006) 9.8% 6.1% labor force (million)** 798 509.3 * 2005, ** 2006 many of these investors fi nd it inconvenient, for a variety of reasons, to invest directly in stock markets in china and india, and, therefore, prefer to invest in adrs based on chinese and indian stocks. these adrs may be created at the request of investors or corporations whose stock is held in trust as collateral for the adr. these securities serve a dual purpose: they enable fi rms incorporated in china and india to raise funds in developed capital markets without having to meet the stringent listing requirements of u.s. and european stock exchanges, and, at the same time, enable global investors to earn returns on securities listed on emerging markets without the dual inconvenience of having to deal with time difference between countries and with currency conversion. this study examines the nature of chinese and indian adrs, sorted on basis of depositary bank, sponsorship status, industry classifi cation, and stock exchange on which the security is listed. data are obtained from the bank of new york and crsp. the intent of the study is to provide documentation to international investors who would like to hold adrs from china and india in their global portfolios. the study should be of interest to international investors, managers of mutual funds who are exploring opportunities to diversify their global portfolios, managers of corporations who are planning to sponsor the issue of depositary receipts, and to bank managers who provide international fi nancial services. the primary securities that underlie an adr may be corporate stocks or bonds. the earliest adrs (1927) were issued at the request of institutional investors. these adr are “unsponsored.” most of the adrs that are currently listed are “sponsored” programs, issued at the request of the fi rm whose securities underlie the adr. when a sponsored adr is issued, there may or may not be ht tp :// ijb f.u um .e du .m y risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 79 a corresponding creation of new capital. there are four grades of sponsored adrs. level i adrs are trade in the otc market. level ii adrs trade on national stock exchanges (such as the nyse). if new capital is raised during the process of issuing sponsored adrs, then the adrs are categorized as level iii and iv. level iii adrs are listed on national stock exchanges. level iv adrs are privately listed, and are usually issued under rule 144a of the us securities and exchanges commission. this study examines the nature and performance of adrs on chinese and indian fi rms. the rest of the paper is structured as follows. section 2 reviews the literature on adrs and summarizes pertinent studies in the area of modern portfolio theory. section 3 examines the sponsorship status, choice of depositary bank, industrial classifi cation, and market listing. section 4 evaluates the performance of these adrs on a risk-adjusted basis, using the morgan stanley capital international (msci) europe, australasia, and far east (eafe) index as a benchmark for comparison purposes. section 5 concludes the paper. 2. literature review treynor (1965), sharpe (1966), and jensen (1968) pioneered the evaluation of the performance of investment portfolios. they developed statistical techniques that are the most commonly used portfolio performance measures even today. treynor (1965) suggested a way of evaluating the performance of a portfolio by adjusting the mean excess return for the degree of market risk and thus calculating the performance of the portfolio. sharpe (1966) computed mean excess return and adjusted for the degree of total risk involved in the portfolio. jensen (1968) devised a method of determining whether the deviation of portfolio returns from market returns was statistically signifi cant, and, therefore, determining whether the excess return could be attributed to superior management, or purely to chance. the techniques used in these three pioneering studies were further refi ned by kon and jen (1979), henrikkson and merton (1981), and chang and lewellen (1984). later on, modigliani and modigliani (1997) did some pioneering work in the area of fi nancial reward and risk. they proposed a new risk-adjusted performance measure (hereafter referred to as, m squared), which is intuitively quite appealing to investors. the idea that underlies their methodology is to adjust the returns of a portfolio to the level of risk in an unmanaged stock market index and then measure the returns on the risk-matched portfolio. separately, academicians and practitioners in fi nance have shown an interest in downside risk measures for evaluating portfolio performance. the most widely cited performance measure that adjusts for downside risk is the sortino ratio (sortino and price, 1994). in this paper, we use a modifi ed sortino ratio that was introduced by pedersen and satchell (2002), who show that this ratio has a sound theoretical foundation. academics have studied the benefi ts of global diversifi cation of investment portfolios extensively. solnik (1996) presents an excellent summary of these ht tp :// ijb f.u um .e du .m y 80 risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 benefi ts. offi cer and hoffmeister (1987) show that portfolio risk can be reduced signifi cantly by including adrs in a portfolio of purely domestic (u.s.) securities. aggarwal, dahiya, and klapper (2005) analyze the investment allocation decision of mutual fund managers to invest in emerging market fi rms that are listed in their domestic markets and have issued adrs in the u.s. as well. they fi nd that adrs are the preferred mode of holdings if the local market of the issuer has weak investor protection, low liquidity and high transaction costs, and if the fi rm is small and has limited analyst following. the motivation for cross-listing shares on foreign exchanges has also been widely researched (saudagaran, 1988). umutlu, salih, and akdeniz (2007) investigate the consequences of cross listing and fi nd that adr listing does not have any effect on the volatility of the underlying stock. on the other hand, jaiswal-dale and jithendranathan (2001) report that the adrs capture the fl uctuations of both the domestic and u.s. markets. the relation between the price of adrs and the underlying shares has also been studied thoroughly (alexander, eun, and janakiramanan, 1987; alexander, eun, and janakiramanan, 1988). jayaraman, shastri, and tandon (1993) study the impact of international cross-listings using adrs. because adrs can be exchanged for the underlying shares, fi nancial arbitrage usually ensures that the price of an adr is within transactions costs of the price of the underlying share. interestingly, in a recent study eichler and maltritz (2008) develop an options-based approach to model the probability of a currency crisis as a function of the deviation of the adr price from the price of the underlying stock. to the knowledge of the authors, th is is the fi rst study of the nature and performance of adrs on chinese and indian fi rms, particularly, their sponsorship status, industrial classifi cation, names of banks that are active in this business, and exchanges on which these adrs are listed. this is also the fi rst rigorous study of the returns that have accrued to these adrs, from the point of view of us based investors. the results of this study should be of interest to investors and mutual fund managers who are looking for opportunities to diversify their international portfolios, to managers of chinese and indian fi rms who are contemplating sponsoring the issue of these securities in us markets, and to the managers of banks, which provide international fi nancial services. 3. nature of adrs from china and india as of november 2006, there were 73 adr issues on chinese fi rms and 88 adr issues on indian fi rms, listed on u.s. capital markets. all adrs from china and india were sponsored. regarding the fi nancial institutions that have issued the chinese adrs, the bank of new york accounted for 50 of these issues, followed by citibank with 11, j.p. morgan chase with 10 issues, and deutsche bank with two issues. thirty-six adrs from india were issued by the bank of new york, 25 by citibank, 21 by deutsche bank, and six by j.p. morgan chase. ht tp :// ijb f.u um .e du .m y risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 81 regarding the exchanges on which these securities are listed, of the 73 chinese adrs, 20 were listed on nasdaq, 18 were listed on the nyse, 27 were listed on otc (other than nasdaq), and 8 were listed on portal. ten adrs from india were listed on the nyse, three were listed on nasdaq, 73 were listed on portal, and the other two were otc. regarding industrial classifi cation, 8 of the chinese adrs were in the software and computer services industry; 7 in travel and leisure; 5 each in mobile telecommunications and technological hardware and equipment; 4 each in industrial engineering, industrial transportation, and real estate; 3 each in chemicals, electricity, life insurance, industrial metals, and oil and gas producers; 2 each in automobiles and parts, food producers, health care equipment and services, leisure goods, mining, and pharmaceuticals and biotechnology; 1 each in beverages, construction and materials, electronics and electric equipment, fi xed line telecommunications, gas, water, and multiutility, general retailers, media, oil equipment, services, and distribution, and support services. with respect to industrial classifi cation, 12 of the indian adrs were in the software and computer services industry; 9 in chemicals, 7 each in banking and construction and materials; 6 in personal goods; 5 in industrial engineering; 4 each in fi xed line telecommunications, food producers, general fi nance, and pharmaceuticals and biotechnology; 3 each in electricity, electronics and electric equipment, industrial metals, and technological hardware and equipment; 2 each in industrial transportation and travel and leisure; 1 each in automobiles and parts, beverages, food and drug retailers, forestry and paper, gas, water, and multiutility, general industrials, health care equipment and services, household goods, oil equipment, services, and distribution, and support services. all data are obtained from the website of the bank of new york. 4. performance of adrs from china and india a. data and methodology monthly return data for the three-year period january 2003 december 2005 are obtained from crsp. bank of new york adr index includes 35 adrs from china and 12 adrs from india. crsp has full return data for 14 chinese adrs and 9 indian adrs. therefore, the fi nal sample for the performance analysis consists of 23 adrs. the return on u.s. 4-week treasury bills is used as the proxy for the risk-free rate. the msci eafe index is utilized as the market benchmark. monthly returns are averaged over the three-year period to obtain the mean return. risk-free rate of return is subtracted from the mean return to compute the mean excess return. mean excess return of each adr is divided by its standard deviation to compute the sharpe measure: (1)i r f-ri =s i ht tp :// ijb f.u um .e du .m y 82 risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 where r i = mean return on adr i, r f = mean risk-free rate of return, and s i = standard deviation of returns for adr i. mean excess return of each adr is divided by its beta to obtain the treynor measure: (2) where b i is estimated from the market model: where r mt = market return during period t, and e it = error term. expected return of each adr is subtracted from its actual mean return to compute jensen’s alpha: (3) where the expected return for each adr is obtained using the capital asset pricing model: (4) jensen’s alphas are then tested for statistical signifi cance. mean excess return for each adr is divided by the downside deviation of that adr’s return from the risk-free rate of return to compute the sortino ratio: (5) where the downside deviation is estimated as follows: (6) sharpe measure is multiplied by the market standard deviation and then the risk-free rate added to calculate the m squared measure: (7) i r f-ri =t i e+r+=r itmtiiit ][r ieri=i )(][ r f-rmirr ie f ddi r f-ri =soi 2 1 1 ]2}),0(max{ 1 1 [dd n j r ijr fni r f+m i r f-ri =m i 2 ht tp :// ijb f.u um .e du .m y risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 83 finally, the benchmark standard deviation is divided by the adr standard deviation to obtain the leverage factor: (8) leverage factor reports a comparison of the total risk in the adr with the total risk in the market portfolio. for example, a leverage factor less than one implies that the risk of the adr is greater than the risk of the market index, and that the investor should consider unlevering the adr by selling off part of the holding in the adr and investing the proceeds in a risk-free security, such as a treasury bill. on the other hand, a leverage factor greater than one implies that the standard deviation of the adr is less than the standard deviation of the market index, and that the investor should consider levering the adr by borrowing money (if possible, at the risk-free rate of return) and investing in that particular adr. b. results the 23 adrs with full monthly return data are identifi ed in table 2 along with their risk, return, and performance statistics. returns, of course, are reported in us dollars. the adrs are ranked in alphabetical order for each country and chinese adrs are listed fi rst, followed by indian adrs. the adr with the highest mean return is rediff.com india with an average monthly return of 10.63 percent. aluminum corporation of china leads the chinese adrs with an average monthly return of 6.23 percent. in comparison, the monthly mean return of the benchmark msci eafe index is 1.83 percent. the adr with the highest total risk (measured by the standard deviation of returns) is again rediff.com india with a monthly standard deviation of 25.59 percent. the chinese adr with the highest monthly standard deviation (18.74 percent) is netease.com. in comparison, the standard deviation of the benchmark msci eafe index is 3.51 percent. further, table 2 reports the numerical values of the sharpe and sortino measures, which are used to rank the adrs in table 3. the highest sharpe and sortino measures obtained (0.48 and 1.14) are by hdfc bank from india. the highest sharpe measure (0.42) for the chinese adrs is by cnooc (china national offshore oil company) and the highest sortino measure (1.12) is by petrochina. in comparison, the sharpe measure and the sortino measure of the benchmark msci eafe index is 0.48 and 1.05, respectively. table 2 also reports the values of adr betas, m squared measures, jensen’s alphas (and their t-statistics), and treynor measures, all of which are computed using the benchmark msci eafe index. the adr with the highest systematic risk (beta=2.14) is aluminum corporation of china. the indian adr with the highest beta (1.89) is videsh sanchar nigam. in comparison, the beta of the benchmark msci eafe index is, by defi nition, exactly 1.0. the adr with the highest m squared measure (1.83) is hdfc bank from india. among the chinese adrs, the highest m squared measure (1.62) is by cnooc. in comparison, the benchmark msci eafe index has an m squared measure of i m=li ht tp :// ijb f.u um .e du .m y 84 risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 t ab le 2 : 3y ea r p er fo rm an ce o n a m on th ly b as is ( 20 03 -2 00 5) a vg s td a lp ha a d r s c ou nt ry (% ) (% ) s ha rp e s or ti no b et a m s qu ar ed a lp ha tst at t re yn or 1 a lu m in um c or po ra ti on o f c hi na c hi na 6. 23 16 .6 5 0. 37 0. 86 2. 14 1. 43 2. 48 1. 55 2. 85 2 b ri ll ia nc e c hi na a ut om ot iv e h ol di ng s c hi na 0. 18 12 .1 5 0. 00 0. 00 0. 96 0. 16 -1 .5 9 -0 .7 8 0. 04 3 c hi na e as te rn a ir li ne s c hi na 1. 01 9. 69 0. 09 0. 15 0. 68 0. 46 -0 .2 8 -0 .4 8 1. 27 4 c hi na m ob il e c hi na 2. 41 7. 42 0. 31 0. 51 1. 16 1. 22 0. 30 0. 42 1. 95 5 c hi na p et ro le um & c he m ic al c hi na 3. 89 10 .8 4 0. 35 0. 82 1. 39 1. 36 1. 41 1. 08 2. 70 6 c hi na t el ec om c hi na 2. 70 9. 81 0. 26 0. 53 1. 63 1. 06 -0 .1 9 0. 50 1. 57 7 c hi na u ni co m c hi na 1. 10 10 .0 4 0. 10 0. 14 1. 02 0. 48 -0 .7 6 -0 .4 1 0. 94 8 c n o o c c hi na n at io na l o ff sh or e o il c hi na 3. 34 7. 62 0. 42 0. 93 0. 41 1. 62 2. 50 1. 07 7. 82 9 g ua ng sh en r ai lw ay c hi na 2. 37 8. 35 0. 27 0. 61 0. 94 1. 08 0. 63 0. 35 2. 35 10 h ua ne ng p ow er i nt er na ti on al c hi na 1. 95 7. 56 0. 24 0. 46 0. 86 0. 98 0. 35 0. 08 2. 10 11 n et ea se .c om c hi na 6. 10 18 .7 4 0. 32 0. 65 1. 29 1. 26 3. 78 1. 34 4. 61 ht tp :// ijb f.u um .e du .m y risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 85 12 p et ro c hi na c hi na 5. 10 12 .1 5 0. 41 1. 12 1. 55 1. 58 2. 34 1. 55 3. 20 13 s in op ec s ha ng ha i p et ro ch em ic al c hi na 3. 85 13 .4 9 0. 27 0. 54 1. 69 1. 11 0. 85 0. 87 2. 19 14 y an zh ou c oa l m in in g c hi na 3. 64 13 .6 2 0. 26 0. 51 1. 66 1. 05 0. 68 0. 77 2. 10 15 d r. r ed dy ’s l ab or at or ie s in di a 0. 74 8. 89 0. 07 0. 10 1. 07 0. 38 -1 .2 1 -0 .6 9 0. 56 16 h d f c b an k in di a 4. 17 8. 42 0. 48 1. 14 1. 08 1. 83 2. 20 1. 54 3. 72 17 ic ic i b an k in di a 5. 23 13 .0 0 0. 39 0. 76 1. 56 1. 52 2. 45 1. 51 3. 26 18 in fo sy s t ec hn ol og ie s in di a 3. 16 11 .8 7 0. 25 0. 41 0. 60 1. 04 2. 01 0. 64 5. 06 19 m ah an ag ar t el ep ho ne n ig am in di a 2. 40 11 .0 0 0. 21 0. 37 0. 58 0. 87 1. 28 0. 30 3. 91 20 r ed if f. co m i nd ia in di a 10 .6 3 25 .5 9 0. 41 1. 10 0. 06 1. 58 10 .3 8 2. 04 * 16 4. 51 ** 21 s at ya m c om pu te r s er vi ce s in di a 3. 89 13 .7 0 0. 27 0. 53 1. 46 1. 10 1. 28 0. 87 2. 57 22 v id es h s an ch ar n ig am in di a 5. 71 18 .5 2 0. 30 0. 74 1. 89 1. 20 2. 37 1. 23 2. 95 23 w ip ro in di a 2. 78 10 .9 1 0. 24 0. 39 0. 76 0. 99 1. 34 0. 49 3. 45 m s c i e a f e 1. 83 3. 51 0. 48 1. 05 1. 00 1. 83 0. 00 0. 00 1. 69 u s 4 -w ee k t re as ur y b il l 0. 14 0. 08 0. 00 0. 00 0. 00 0. 14 0. 00 -2 .8 8* 0. 00 * s ig ni fi ca nt a t th e 5% l ev el . * * d ue t o th e ve ry l ow b et a, t re yn or m ea su re s ho ul d no t be c on si de re d m ea ni ng fu l. t ab le 2 : 3y ea r p er fo rm an ce o n a m on th ly b as is ( 20 03 -2 00 5 (c o n ti n u ed ) ht tp :// ijb f.u um .e du .m y 86 risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 1.83. the adr with the highest alpha measure is rediff.com india with alpha equal to 10.38, which is signifi cant at the fi ve percent level. the chinese adr with the highest alpha (3.78) is netease.com, but this alpha is not signifi cant. the alpha measure of the benchmark msci eafe index is, by defi nition, zero. none of the other adr alphas is signifi cantly different from zero. finally, the adr with the highest treynor measure (164.51) is rediff.com india. cnooc is the leading chinese adr with a treynor measure of 7.82. in comparison, the treynor measure for the msci eafe index is 1.69). table 3: 3-year ranking (2003-2005) adrs sharpe rank sortino treynor alpha panel a: whole sample country (m squared rank) rank rank rank msci eafe 1 4 19 19 hdfc bank india 2 1 6 8 cnooc china national offshore oil china 3 5 2 3 rediff.com india india 4 3 1 1 petrochina china 5 2 9 7 icici bank india 6 8 8 5 aluminum corporation of china china 7 6 11 4 china petroleum & chemical china 8 7 12 10 netease.com china 9 10 4 2 china mobile china 10 16 18 18 videsh sanchar nigam india 11 9 10 6 sinopec shanghai petrochemical china 12 12 15 14 satyam computer services india 13 13 13 13 guangshen railway china 14 11 14 16 china telecom china 15 14 20 20 yanzhou coal mining china 16 15 16 15 infosys technologies india 17 18 3 9 wipro india 18 19 7 11 huaneng power international china 19 17 17 17 mahanagar telephone nigam india 20 20 5 12 table continues on the next page ht tp :// ijb f.u um .e du .m y risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 87 china unicom china 21 22 22 22 china eastern airlines china 22 21 21 21 dr. reddy’s laboratories india 23 23 23 23 brilliance china automotive holdings china 24 24 24 24 panel b: chinese sub-sample msci eafe 1 2 11 11 cnooc china national offshore oil 2 3 1 2 petrochina 3 1 3 4 aluminum corporation of china 4 4 4 3 china petroleum & chemical 5 5 5 5 netease.com 6 6 2 1 china mobile 7 11 10 10 sinopec shanghai petrochemical 8 8 7 6 guangshen railway 9 7 6 8 china telecom 10 9 12 12 yanzhou coal mining 11 10 8 7 huaneng power international 12 12 9 9 china unicom 13 14 14 14 china eastern airlines 14 13 13 13 brilliance china automotive holdings 15 15 15 15 panel c: indian sub-sample msci eafe 1 3 9 9 hdfc bank 2 1 4 4 rediff.com india 3 2 1 1 icici bank 4 4 6 2 videsh sanchar nigam 5 5 7 3 satyam computer services 6 6 8 8 infosys technologies 7 7 2 5 wipro 8 8 5 6 mahanagar telephone nigam 9 9 3 7 dr. reddy’s laboratories 10 10 10 10 table 3: 3-year ranking (2003-2005) ht tp :// ijb f.u um .e du .m y 88 risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 table 3 panel a reports the rankings of all the adrs. the treynor and alpha ranks in table 3 indicate that, 18 adrs have returns (adjusted for systematic risk) that, exceed the risk-adjusted returns of the msci eafe index. the sharpe ranks indicate that none of the adrs has returns (adjusted for total risk) that exceed the risk-adjusted returns of the msci eafe index. the sortino ranks indicate that only hdfc bank (india), petrochina, and rediff.com india have returns (adjusted for downside risk) that exceed the risk-adjusted returns of the msci eafe index. the ranking based on the m squared measure is identical to the ranking based on the sharpe measure. however, the m squared measure enables us to draw some inferences, which cannot be drawn from the sharpe measure (or, as a matter of fact, from any other measure), and these are detailed at the end of this section. panel b and panel c show the rankings for the chinese and indian sub-samples, respectively. table 4 panel a reports the average returns that accrue to the whole sample of adrs with and without risk-adjustment. the risk-adjustment is performed by using the msci eafe index as the benchmark. the returns are annualized for the convenience of investors. this is done by compounding the monthly mean returns over twelve periods. in that panel, rediff.com india, which ranks fi rst based on unadjusted returns, falls back to rank four on the basis of returns adjusted for risk. aluminum corporation of china ranks second on the basis of unadjusted returns, but falls back to rank seven based on returns adjusted for risk. netease.com, which ranks third on the basis of unadjusted returns, falls back to rank nine on the basis of returns adjusted for risk. on the other hand, msci eafe, which ranks 20th on an unadjusted basis, ranks fi rst when the returns are adjusted for risk. hdfc bank (india), which ranks seventh on an unadjusted basis, ranks second when the returns are adjusted for risk. cnooc, which ranks 12th on an unadjusted basis, ranks third when the returns are adjusted for risk. the leverage factor for this adr is 0.46, which implies that an investor, who is comfortable with bearing the same level of risk as in the benchmark msci eafe index, could unlever the adr (lend 54 percent of her down payment, if possible, at the risk-free rate of interest and invest the rest in the adr) and thereby attain an annual return level of 21.22 percent. the example below details how this return can be obtained. consider an investor who would like to earn superior returns on an adr and, at the same time, bear only an average level of risk. in this context, the average level of risk is measured by the standard deviation of the benchmark msci eafe index, which is 3.51 percent on a monthly basis. now consider the following investment strategy: suppose that the investor has $1,000 to invest. the investor could lend $540 and invest $460 in cnooc. the end of month return from the adr portion of the portfolio will be $460 x 0.0334 = $15.36. suppose that the loaned funds were given at the monthly risk-free rate of 0.14 percent. in that case, the loaned funds will bring $540 x 0.0014 = $0.76. the portfolio return is $15.36 + 0.76 = $16.12, which is a return of 1.61 percent on a ht tp :// ijb f.u um .e du .m y risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 89 t ab le 4 : 3y ea r a nn ua li ze d p er fo rm an ce : u na dj us te d an d a dj us te d fo r r is k a d r s u na dj us te d u na dj us te d a dj us te d a dj us te d l ev er ag e a nn ua li ze d r an k a nn ua li ze d r an k f ac to r p an el a : w h ol e s am p le c ou nt ry r et ur ns ( % ) r et ur ns ( % ) m s c i e a f e 24 .3 5 20 24 .3 5 1 1. 00 h d f c b an k in di a 63 .3 1 7 24 .2 4 2 0. 42 c n o o c c hi na n at io na l o ff sh or e o il c hi na 48 .2 5 12 21 .2 2 3 0. 46 r ed if f. co m i nd ia in di a 23 6. 08 1 20 .7 5 4 0. 14 p et ro c hi na c hi na 81 .7 3 6 20 .6 8 5 0. 29 ic ic i b an k in di a 84 .4 2 5 19 .8 4 6 0. 27 a lu m in um c or po ra ti on o f c hi na c hi na 10 6. 56 2 18 .5 6 7 0. 21 c hi na p et ro le um & c he m ic al c hi na 58 .1 1 8 17 .5 9 8 0. 32 n et ea se .c om c hi na 10 3. 62 3 16 .2 4 9 0. 19 c hi na m ob il e c hi na 33 .0 6 16 15 .6 3 10 0. 47 v id es h s an ch ar n ig am in di a 94 .6 1 4 15 .3 8 11 0. 19 s in op ec s ha ng ha i p et ro ch em ic al c hi na 57 .2 7 10 14 .1 4 12 0. 26 s at ya m c om pu te r s er vi ce s in di a 58 .0 2 9 14 .0 9 13 0. 26 g ua ng sh en r ai lw ay c hi na 32 .4 0 18 13 .7 6 14 0. 42 c hi na t el ec om c hi na 37 .6 5 15 13 .4 8 15 0. 36 y an zh ou c oa l m in in g c hi na 53 .5 1 11 13 .2 9 16 0. 26 in fo sy s t ec hn ol og ie s in di a 45 .2 0 13 13 .1 6 17 0. 30 w ip ro in di a 38 .9 1 14 12 .5 8 18 0. 32 h ua ne ng p ow er i nt er na ti on al c hi na 26 .0 8 19 12 .4 7 19 0. 46 m ah an ag ar t el ep ho ne n ig am in di a 32 .9 5 17 10 .8 9 20 0. 32 c hi na u ni co m c hi na 14 .0 7 21 5. 91 21 0. 35 c hi na e as te rn a ir li ne s c hi na 12 .7 8 22 5. 62 22 0. 36 d r. r ed dy ’s l ab or at or ie s in di a 9. 25 23 4. 65 23 0. 40 b ri ll ia nc e c hi na a ut om ot iv e h ol di ng s c hi na 2. 23 24 1. 88 24 0. 29 t a b le c o n ti n u es o n t h e n ex t p a g e ht tp :// ijb f.u um .e du .m y 90 risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 p an el b : c h in es e s u b -s am p le m s c i e a f e 24 .3 5 12 24 .3 5 1 1. 00 c n o o c c hi na n at io na l o ff sh or e o il 48 .2 5 7 21 .2 2 2 0. 46 p et ro c hi na 81 .7 3 3 20 .6 8 3 0. 29 a lu m in um c or po ra ti on o f c hi na 10 6. 56 1 18 .5 6 4 0. 21 c hi na p et ro le um & c he m ic al 58 .1 1 4 17 .5 9 5 0. 32 n et ea se .c om 10 3. 62 2 16 .2 4 6 0. 19 c hi na m ob il e 33 .0 6 9 15 .6 3 7 0. 47 s in op ec s ha ng ha i p et ro ch em ic al 57 .2 7 5 14 .1 4 8 0. 26 g ua ng sh en r ai lw ay 32 .4 0 10 13 .7 6 9 0. 42 c hi na t el ec om 37 .6 5 8 13 .4 8 10 0. 36 y an zh ou c oa l m in in g 53 .5 1 6 13 .2 9 11 0. 26 h ua ne ng p ow er i nt er na ti on al 26 .0 8 11 12 .4 7 12 0. 46 c hi na u ni co m 14 .0 7 13 5. 91 13 0. 35 c hi na e as te rn a ir li ne s 12 .7 8 14 5. 62 14 0. 36 b ri ll ia nc e c hi na a ut om ot iv e h ol di ng s 2. 23 15 1. 88 15 0. 29 p an el c : in d ia n s u b -s am p le m s c i e a f e 24 .3 5 9 24 .3 5 1 1. 00 h d f c b an k 63 .3 1 4 24 .2 4 2 0. 42 r ed if f. co m i nd ia 23 6. 08 1 20 .7 5 3 0. 14 ic ic i b an k 84 .4 2 3 19 .8 4 4 0. 27 v id es h s an ch ar n ig am 94 .6 1 2 15 .3 8 5 0. 19 s at ya m c om pu te r s er vi ce s 58 .0 2 5 14 .0 9 6 0. 26 in fo sy s t ec hn ol og ie s 45 .2 0 6 13 .1 6 7 0. 30 w ip ro 38 .9 1 7 12 .5 8 8 0. 32 m ah an ag ar t el ep ho ne n ig am 32 .9 5 8 10 .8 9 9 0. 32 d r. r ed dy ’s l ab or at or ie s 9. 25 10 4. 65 10 0. 40 t ab le 4 : 3y ea r a nn ua li ze d p er fo rm an ce : u na dj us te d an d a dj us te d fo r r is k ht tp :// ijb f.u um .e du .m y risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 91 monthly basis or 21.13 percent (slightly off the 21.22 percent in table 4 due to rounding) on an annual basis. note that the monthly risk of the portfolio is 0.46 x 7.62 = 3.51 percent, which is the same as the monthly standard deviation of the benchmark msci eafe index. this investment strategy, therefore, enables the investor to earn superior returns for an average level of risk. it may be noted that the above example assumes that the returns on risk-free us treasury bills are not correlated with the returns on the adr. 5. conclusion adrs represent a convenient vehicle to access emerging markets such as china and india for international investors who are contemplating global portfolio selection. these securities are structured to serve the needs of both investors and issuers. there are 161 adrs from the china-india region that are listed on u.s. markets and they provide a menu of investment opportunity across diverse industry groups. this study examines the nature of these adrs in order to identify the depositary bank, sponsorship status, industry classifi cation, and market listing. prior research has examined the performance of individual chinese and indian stocks in local currencies. however, risk-adjusted returns reported in terms of us dollars are more useful to international investors for security selection and portfolio construction. in addition, from the global investor’s point of view, the instrument of choice for accessing chinese and indian stock markets is the adr, not the underlying stock itself. hence, there is need for rigorous evaluation of the performance of these adrs using metrics grounded in modern portfolio theory. this study contributes to academic and practitioner literature by meeting this need. in order to facilitate comparison with international stock market performance, this study uses the morgan stanley capital international eafe index to evaluate the risk-adjusted returns of chinese and indian adrs. some of these adrs have unadjusted returns which are high, but once risk is factored in, the adjusted returns are less attractive. on the other hand, some adrs with modest returns may be quite attractive to international investors, when their returns are adjusted upward for low risk. global investors may want to examine each of these securities in detail, in order to evaluate them further for possible inclusion in an investment portfolio. of course, the contribution of a security to portfolio return and risk should matter more to the global investor than the return and risk of an individual security. this study provides empirical evidence on the risk and return characteristics of adrs from china and india. it would be benefi cial to update this information on a continuing basis, in order to provide ongoing documentation to international investors who seek to diversify into these markets. future research could focus on decoupling the return to these adrs into its constituent components: riskadjusted return in the domestic currency and fl uctuation in the exchange rate. ht tp :// ijb f.u um .e du .m y 92 risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 author statement: the submitting author is ajay samant, national city endowed chair in finance, western michigan university, kalamazu, michigan, usa. e-mail: ajay.samant@wmich.edu. onur arugaslan is an associate professor at the same university. references aggarwal, r., dahiya, s., and klapper, l. f. (2005). american depositary receipts (adr) holdings of u.s. based emerging market funds. world bank policy research (working paper no. 3538). alexander, g. j., cheol s. e., and janakiramanan s. (1987). asset pricing and dual listing on foreign capital markets: a note. journal of finance, 42, 151-158. alexander, g. j., cheol, s. e., and janakiramanan s. (1988). international listings and stock returns: some empirical evidence. journal of financial and quantitative analysis, 23, 135-151. chang, e. c., and lewellen, w. g. (1984). market timing and mutual fund investment performance. journal of business, 57(1), 57-72. eichler, s., and maltritz, d. (2008). currency crisis prediction using adr market data an options-based approach. efa 2008 athens meetings paper. henrikkson, r.d., and merton, r.c. (1981). on market timing and investment performance ii: statistical procedures for evaluating forecasting skills. journal of business, 54(4), 513-533. jaiswal-dale, ameeta and jithendranathan, thadavillil, (2001). fluctuating returns of dual listings: domestic and adr markets. efma 2001 lugano meetings paper. jayaraman, n., shastri k., and tandon k., (1993). the impact of international cross-listings on risk and return: the evidence from american depositary receipts. journal of banking and finance, 17(1), 91-103. jensen, m.c. (1968). the performance of mutual funds in the period 1945-64. journal of finance, 23(2), 389-416. kon, s. j., and jen, f. c. (1979). the investment performance of mutual funds: an empirical investigation of timing, selectivity, and market effi ciency. journal of business, 52(2), 263-289. modigliani, f., and modigliani, l. (1997). risk-adjusted performance. journal of portfolio management, 23(2), 45-54. offi cer, d., and j. ronald h. (1987). adrs: a substitute for the real thing? journal of portfolio management, 61-65. pedersen, c.s., and satchell, s.e., (2002). on the foundation of performance measures under asymmetric returns. quantitative finance, 2(3), 217-223. umutlu, mehmet, salih, aslihan a., and akdeniz, l. (2007). do adr listings affect the dynamics of volatility in emerging markets? ssrn working paper. ht tp :// ijb f.u um .e du .m y risk-adjusted returns of american depositary receipts on chinese and indian stocks: 77-93 93 saudagaran, shahrokh m. (1988). an investigation of selected factors infl uencing the decision to list on foreign stock exchanges. journal of international business studies, 101-128. sharpe, w. f. (1966). mutual fund performance. journal of business, 39(1), 119-138. solnik, b.h., and mcleavey, d., (2004). international investments (5th ed.). reading, ma: addison-wesley. sortino, f. a., and price, l. n. (1994, fall). performance measurement in a downside risk framework. journal of investing, 59-65. treynor, j.l. (1965). how to rate management of investment funds. harvard business review, 43(1), 63-75. ht tp :// ijb f.u um .e du .m y ht tp :// ijb f.u um .e du .m y remittances and banking sector development in south asia 47 the international journal of banking and finance, volume 8 (number 4) 2011: pages 47-66 remittances and banking sector development in south asia abdullah m noman and gazi s uddin university of new orleans, united states and carleton university, canada ____________________________________________________________ abstract the paper investigates the interaction among foreign remittance, banking sector development and gdp in four south asian nations that export huge pools of labour abroad. multivariate granger causality tests, based on error correction models, are employed with data spanning from 1976 to 2005. a key finding of the paper is that remittances and banking sector development influence per capita income in all four south asian nations. in addition, interactions among the variables are also examined in a panel setting. as in individual country analyses, both remittance and banking sector development have positive and significant influences on the national income of south asian countries. on the other hand, neither domestic products nor advancement in banking sector have significant impact on the remittance flows. this is new findings of the linkage between remittances and economic development, which may also be evident for countries exporting labour pools. key words: remittances; financial development; south asia; panel causality jel classification: c32, f24, o16 _____________________________________________ 1. introduction inflow of remittance, being a prime source of foreign currency and thereby a contribution to the national economy, plays a significant role in the context of the developing nations. since 2000, remittance inflow has been rising by an average rate of 16% per annum in the developing countries: world bank (2006). china, india and mexico, the top three recipients, occupied more than one third of the total remittance inflow, as of 2006. only noman and uddin: remittances and banking sector development in south asia 48 three of the south asian countries, namely: bangladesh, india and pakistan, made a foothold in the list of top 25 recipients. according to the same source 50% of the remittance flow recorded globally passes through the informal funds channel. this particular fact captures the basic idea of this paper that remittance inflow through formal channel may have a potential to contribute more to the south asian economies via developmental impacts of the banking sector (hinojosa-ojeda, 2003; terry and wilson, 2005, and world bank, 2006). remittances enhance the development of banking sector in a number three ways. firstly, remittances supply the households with excess cash that might potentially generate a transaction demand for financial services. secondly, the fees earned through remittance processing, can add to the profitability of a branch. banks with the state of the art infrastructure can explore this new market opportunity. thirdly, banks can target the ‘bottom of the pyramid’ segment of the remittance receiving market, where, a substantial portion of the remittances is likely to remain unbanked. the development of the banking sector promotes an open market competition among the money transfer firms resulting in the reduction of transaction costs and thereby pulling more remittances from the informal to formal channel. remittances may potentially contribute to the long-term growth through higher rates of capital accumulation. in absence of an efficient financial system, as commonly seen in the developing economies, the untapped market with poor credit rating can potentially be approached by the inflow of remittances. therefore, remittances may contribute to alleviate credit constraints to the ‘bottom of the pyramid’ of the market, by improving the allocation of capital, and therefore accelerating economic growth. in addition to promoting banking sector development, foreign remittance has an immense direct impact on economic development of receiving nations as well (adelman and taylor, 1990). remittances inflow can directly work for community development (cordova, 2004; de haas, 2007) which can eventually result in economic development. in addition, remittance inflow, as a source of external capital, can contribute to bridging the gap between domestic savings and investment and thereby promoting economic growth in the long run. in addition, remittances are also a valuable source of foreign international journal of banking and finance, vol. 8, iss. 4 [2011], art. 3 49 currency that could be used for meeting external debt obligation denominated in international currencies keeping the wheel of development active for the receiving nation. this paper presents both country and panel analysis of remittances and finds that remittance and banking sector development have positive and significant influence on the national income of south asian nations. on the other hand, the impact of neither domestic products nor advancement in banking sector on the remittance flow could not be established. remittance seems to flow independent of the macroeconomic aspect of the recipient nation. the findings of this paper clearly suggest that remittance plays a very important role in south asia. in the light of this findings, the government may initiate new programs to maximize the benefits of remittances which could improve the welfare of migrant workers and their families, especially poor rural households by providing institutional support for the promotion of formal and semi formal remittance services and other support services taking advantages at the south asian well established microfinance network. more specifically, governments could encourage remittance inflow through formal and semi-formal channel by providing low cost but reliable services. in addition, government agencies are responsible for creating awareness among the migrant workers and their families about the appropriate uses of the hard-earned remittances. they should also promote better investment opportunities for sustainable and productive use of remittances through support mechanism for development of microenterprises. this paper aims to investigate the causal relationship between foreign remittances, banking sector development and gdp for south asia. unlike some other papers, which include financial development indicators, this paper uses an indicator for development in the banking sector only. the reason for this narrower focus is the direct role of commercial banks in channeling foreign remittances to the recipients of the home country. it is therefore expected that volume of remittances have an immediate influence on banks’ ability for credit expansion. the structure of the paper is as follows. section 2 describes the literature review. section 3 includes the south asian remittances flow section 4 presents the data and noman and uddin: remittances and banking sector development in south asia 50 methodology. section 5 analyzes empirical analysis and results and finally, section 6 presents the concluding remarks. 2. literature review remittances are the major sources of foreign exchange earning and important implications for the remittance-recipient countries because of their increasing volume. these private flows are mostly spent on consumption, housing, and land, and are not used for productive investment discussed in the literature. remittances affect the economy from both the micro and macro perspective. they will influence poverty, inequality, growth, education, infant mortality, welfare, entrepreneurship, local livelihood and etc. the significance of remittances in stimulating the economic growth, specifically at the interaction between remittances and the banking sector development, an aspect ignored in the literature. a well and sound banking system performs a number of key economic functions and their development has been shown to the development of financial system, which will lead to foster economic growth. in this paper, we explore how banking sector development influences a country's capacity to take advantage of remittances. the relationship between remittances and the banking sector is important because some argue that intermediating remittances through the banking sector can magnify the developmental impact of remittance flows ( hinojosa-ojeda, 2003; terry and wilson, 2005, and world bank, 2006). the positive effect of financial development on growth has been extensively documented (beck, demirguc-kunt, & levine, 2004; levine, 1997, studies that relationship between remittances and investment, where remittances either substitute for, or improve financial access, conclude that remittances stimulate growth (giuliano & ruiz-arranz, 2005; toxopeus & lensink, 2006). using cross-country data, aggarwal, demirgüç-kunt, and martinez peria (2006) find evidence that remittances are associated with banking sector development across a broad set of countries such as guatemala, el salvador, ecuador and honduras and mexico,. the differences are large in guatemala, el salvador, ecuador and honduras, but much smaller in mexico, where 19% of remittance-receiving households have accounts international journal of banking and finance, vol. 8, iss. 4 [2011], art. 3 51 compared with 16% of non-recipient households. neither study makes any claim about the causality of the associations they report. giuliano and ruiz-arranz (2006) also finds a positive correlation between the level of remittance flows and measures of bank deposits, but much weaker correlations between remittances flows and bank credit. orozco and fedewa (2007) investigates that households receiving remittances in five latin american countries are more likely than non-recipient households to have bank accounts. freund and spatafora (2005) examine that concentration in the banking sector, financial risk, and exchange rate variability typically increases transaction costs. financial sector reforms that address the structural problems in the receiving, and sending countries are also likely to lower the cost of remittances. cross-border uniformity in regulations related to remittances, and regulatory interventions, where fees are prohibitive have been proposed as other cost-reducing measures (ratha & riedberg, 2005; sander & maimbo, 2005). the link between remittances, banking sector development, and growth is an important issue for remittances dependence countries. development of the banking sector reducing the transaction costs that might stimulate the investor to invest in different investment projects with high rate return and therefore increase the economic growth in the country. remittances might become a substitute for inefficient credit markets by helping local entrepreneurs by pass lack of collateral or high lending costs and start productive activities. remittances affect banking sector for several reasons such as temporary excess cash for banking services, processing fees for transactions and etc. the potential to collect these fees might induce banks to expand their outreach, locate close to remittance recipients, and increase their demands for banking services, since banks offer households a safe place to store this temporary excess cash. entrepreneurs in developing countries confront much less efficient credit markets and available evidence indicates that access to credit is among their biggest concerns (paulson and towsend, 2000). dustmann and kirchamp (2001) find that the savings of returning migrants may be an important source of startup capital for microenterprises based on the micro level data. noman and uddin: remittances and banking sector development in south asia 52 3. data and methodology 3.1 data for the purpose of this paper, real gdp per capita (base year = 2000), remittances flow as percentage of gdp and total bank credit disbursement as percentage of gdp are used. the data span from 1976 to 2005 and are extracted from world development indicator (wdi) database of the world bank 2006 cd–rom. the variable that proxies for the banking sector sector development has also been widely used in the literature. for example, giuliano and ruiz–arranz (2009) uses this measure of financial development to estimate the extent of intermediation provided by the banking sector in an economy. on the other hand, jalil et al. (2010) consider supply of credit to private sector is a better indicator of development in the banking sector than other indicators, for example, amount of liquid liabilities to gdp or m2 to gdp. baltagi et al. (2009) also use private sector credit disbursement as the crucial variable in their study investigating the relationship between financial development and openness. 3.2 methodology the multivariate granger 1 methodology will be applied to identify direction of causality among the variables of interest, i.e. remittance, financial development and gdp. consider the following vector autoregression (var) representation, 11 12 13 21 22 23 31 32 33 ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) t t t t t t t t t l l lgdp gdp rem l l l rem u bsd bsd vl l l                                                     (1) where, gdp, rem and bsd denote three potentially endogenous variables: real gdp per capita, remittance as percentage of gdp, and bsd, (i.e. total bank credit disbursement as percentage of gdp), an indicator for banking sector development, respectively and l is the lag operator.  , and are intercepts and  , u and v white noise error. the null of no joint significance can be tested using f–tests. 1 the method was proposed by granger (1969) and popularized by sims (1972). international journal of banking and finance, vol. 8, iss. 4 [2011], art. 3 53 empirical works based on time series data assume that the underlying time series is stationary. however, many studies have shown that majority of time series variables are nonstationary or integrated of order 1 (engle and granger, 1987). the time series properties of the data at hand are therefore studied in the outset. in addition to applying traditional augmented dickey–fuller (adf) tests, phillips and perron (pp) tests are also applied as the latter tests are more efficient in the presence of a one–time structural break in the data. the above specification of the causality test assumes that the time series at hand are mean reverting process. however, it is highly likely that variables of this study are nonstationary. formal tests will be carried out to find the time series properties of the variables. if the variables are i (1), engle and granger (1987) assert that causality must exist in, at least, one direction. the granger causality test is then augmented with an error correction term (ect) as shown below: 0 1 2 3 4 1 1 1 1 m m m t i t i i t i i t i t t i i i gdp gdp rem bsd z                         (2) 0 1 2 3 4 1 1 1 1 n n n t i t i i t i i t i t t i i i rem rem bsd gdp z u                        (3) 0 1 2 3 4 1 1 1 1 q q q t i t i i t i i t i t t i i i bsd bsd gdp rem z v                        (4) where zt–1 is the ect obtained from the long run cointegrating relationship between gdp, remittance and financial development. the above error correction model (ecm) implies that for each of the model possible sources of causality are two: lagged dynamic regressors and lagged error correction term. if estimated coefficients of either sources of causation turn out to be significant, the null of no causality is rejected. for instance, by equation (2), given the presence of financial development indicators, remittances granger cause gdp, if the null of either 21 0 m ii    or 4 0  is rejected. on the other hand, if we can not reject either of the null hypothesis, it is concluded that no causation exist from independent variables to dependent variable. noman and uddin: remittances and banking sector development in south asia 54 3.3 panel granger causality for the time series analysis of granger causality, the data span needs to be sufficiently long to capture the direction of causality. a motivation of considering the panel granger causality tests is to have more reliable results from a panel than from the individual countries as the former contain more information than the latter (baltagi, 2005). the most common framework for panel causality tests is proposed by holtz–eakin et al. (1988, 1989). examination of causality in a panel framework begins with pooling the data across time and groups and allowing for individual effects. consider the following equations (5) through (7), where, variables are as defined before, subscript i indexes cross section units and t denotes time dimension and the individual effects are represented by the term f . estimation of the equations using pooled ols would, however, be biased as lagged dependant variables are correlated with error terms including individual effects. 0 1 2 3 1 1 1 m m m it i it i i it i i it i gdpi it i i i gdp gdp rem bsd f                   (5) 0 1 2 3 1 1 1 n n n it i it i i it i i it i remi it i i i rem rem bsd gdp f u                  (6) 0 1 2 3 1 1 1 q q q it i it i i it i i it i bsdi it i i i bsd bsd gdp rem f v                  (7) in order to remove correlations between the lagged terms and individual effects with the errors, the variables are differenced as follows 1 2 3 1 1 1 m m m it i it i i it i i it i it i i i gdp gdp rem bsd                    (8) 1 2 3 1 1 1 n n n it i it i i it i i it i it i i i rem rem bsd gdp u                   (9) 1 2 3 1 1 1 q q q it i it i i it i i it i it i i i bsd bsd gdp rem v                   (10) it is important to note that, while differencing has removed the individual effects, a problem of simultaneity has now been introduced as the lagged endogenous variables are international journal of banking and finance, vol. 8, iss. 4 [2011], art. 3 55 correlated with the new differenced error terms. so we check if the differenced errors are ma(1) by way of testing for the absence of second order serial correlation. in addition, we solve this problem by following the methodology adopted in al-iriani (2006). in particular, we have used instrumental variable approach with lagged values of dependent variables being used as instruments. however, empirical literature suggests that presence of too many moment conditions increase biasness as well as efficiency (baltagi, et al., 2009). therefore, the over–identification restrictions are tested using the sargan test. each panel equation from (8) through (10) is then estimated using the generalized methods of moments (gmm) procedure as proposed by arellano and bond (1991). the direction of causality is identified within this multivariate framework by testing for joint hypotheses on the significance of a set of estimated coefficients in the presence of the other variables. for example, in the context of equation (8), we test 21 0 m ii    to identify the influence of remittance on gdp while holding banking sector development as constant. similarly, testing for the joint significance 31 0 m ii    identifies the direction of influence running from banking sector development to gdp. the wald test statistics follow a chi-square distribution with degrees of freedom equaling (k – m) where k is the number of variables. 3.4 the hypotheses based on the review of existing literature, we develop the following hypotheses for testing in this paper. more specifically, the hypotheses would postulate expected relationship among variables under investigation. hypothesis one: in the long run, both remittance and banking sector development will have positive impact on gdp. we expect that past values of both banking sector development and remittance would positively impact gdp of a country. this means that the ecm term would be significant showing the long run impact and cumulative sign on these variables would be positive. noman and uddin: remittances and banking sector development in south asia 56 hypothesis two: increase in gdp will have negative impact on remittance flow but positive impact on banking sector development. as a nation’s gdp level rises, the number of expatriates is expected to fall. workers will find better options at home than to travel abroad. this is especially true the south asian countries as most of the expatriates going abroad fall in the laborer category. on the other hand, the rise in gdp would normally be associated with financial sector development in general and banking sector development in particular. hypothesis three: given the level of gdp, remittance flow and banking sector development will positively influence each other. south asian nations are among the top recipients of foreign remittance. commercial banks in these nations work hard to attract expatriates’ money through their foreign branches. in addition, in order to facilitate disbursement of the remittances to their local beneficiaries, banks have extended a number of services. therefore, it is naturally expected that remittance flow and banking sector development will boost each other. 4. empirical results the estimation begins with an examination of time series behavior of variables at hand. in particular, unit root tests are reported in table 1 where both adf tests and kpss tests results confirm that the variables under consideration are all nonstationary. optimal lag length for adf is selected using general–to–specific down method from a maximum lag equal to 8. the lag truncation parameter for the kpss test is selected using the formula   1 4 4 100t . once it is established that variables are i(1), the next step is to test for existence of any cointegrating relationship between gdp per capita, and remittance and bank credit disbursement. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 3 57 table 1: unit root test results adf kpss country gdp rem bsd gdp rem bsd bangladesh 4.126(4) –0.082(0) 0.450 (3) 1.062* 0.916* 1.068* india 1.950 (0) –0.097 (8) –0.125 (1) 0.886* 1.090* 0.475* pakistan –0.134 (6) –2.478 (7) –2.342 (0) 1.046* 0.672* 0.489* sri lanka –2.304 (7) 1.981 (8) –1.668 (0) 1.096* 0.887* 0.549* note: * denotes significance at 5% level. the johansen (1991) lr test of cointegration is applied and results are shown in table 2. the appropriate var lag length is selected using aic. table 2 shows that the variables for all countries are cointegrated. therefore, the granger causality tests will be modeled using ecm as explained in equations (2) to (4). table 2: cointegration lr test results [ , , ]x gdp rem bsd  ; [max var lag k = 4] bangladesh (4) india (3) pakistan (3) sri lanka (2) null λ–trace [p–value] λ–trace [p–value] λ–trace [p–value] λ–trace [p–value] 0r  81.620 [0.000] 61.023 [0.0002] 66.355 [0.0000] 55.789 [0.0012] 1r  36.165 [0.0013] 26.678 [0.0374] 36.133 [0.0013] 23.245 [0.1021] 2r  7.4633 [0.3077] 8.7052 [0.2048] 9.2167 [0.1714] 5.1351 [0.5848] notes: the r denotes the number of cointegrating vectors. var lag was selected using akaike information criterion (aic) from a maximum lag of 4. presence of both a constant and a restricted trend was assumed in cointegrating regression for all countries. table 3 reports multivariate granger causality tests results. the literature on granger causality documents the fact that the results of such tests are sensitive to the selection of var lag length. the optimal lag length for the var in this paper has been selected by minimizing the aic as liew (2004) proves supremacy of aic and fpe over other selection criteria in small sample. the observations on the results presented in table 3 noman and uddin: remittances and banking sector development in south asia 58 will be made for each country separately. this will help us capture the direction of causality for each case. for bangladesh, unidirectional causality runs from the development of banking sector (long run) and from remittance inflow (short run) to per capita income. this provides some support for the hypothesis one. however, no feedback causality is detected. moreover, it is found that past values of remittances positively affect national income while influences of banking development come from the lagged error correction term only. the development of banking sector is not found to be influenced by either remittances inflow or growth in national income, even in the long run. hypothesis two and three are not supported with empirical findings for bangladesh. a possible reason why banking sector is not significantly influenced by remittance is the overwhelming practice of sending wage earners’ money through unofficial and illegal channels. turning to results for india, we find that there exists of bidirectional causality between gdp growth and remittances in the long run. in the short run, an increase in gdp causes a fall in remittance inflow as seen from the negative sum of coefficients of lagged values of gdp as predicted in hypothesis two. this is not quite unexpected as a rise in recipients’ income may decrease the local need for foreign remittances. in addition, there is only unidirectional long run causality flowing from the banking sector development to remittances volume and gdp per capita. in other words, development in the banking sector affects remittances and gdp, but is not influenced by them providing only partial evidence in favor of the hypothesis three. the results for pakistan reveal that all three variables at hand granger cause each other either in the short run or in the long run. remittances and gdp per capita are influenced by each other, as for the case of india, through bidirectional causality. bidirectional causality is also found between banking sector development and national income supporting hypothesis one. similarly, remittances and banking sector development are also linked to each other through bidirectional causality. thus evidence for the hypothesis three is not available. a surprising feature for pakistan is that all short run causalities produce negative impact when considered in cumulative. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 3 59 table 3: multivariate granger causality results bangladesh gdp rem bsd 1tect  gdp (4) --- 3.166* (+) (0.047) 1.479 (0.309) –2.630* (0.025) rem (4) 2.450 (0.119) --- 0.334 (0.835) –0.374 (0.713) bsd (4) 0.265 (0.911) 0.267 (0.890) --- –1.584 (0.141) india gdp rem bsd 1tect  gdp (1) --- 0.011 (0.918) 0.107 (0.744) –2.887* (0.013) rem (1) 3.037* (–) (0.083) --- 2.124 (0.153) – 2.244* (0.028) bsd (4) 0.317 (0.852) 0.444 (0.777) --- –0.3429 (0.760) pakistan gdp rem bsd 1tect  gdp (4) --- 2.835* (–) (0.066) 2.053 (0.179) –2.762* (0.020) rem (4) 4.429* (–) (0.023) --- 3.519* (–) (0.039) –0.726 (0.501) bsd (4) 0.434 (0.790) 2.157 (0.172) --- –2.255* (0.046) sri lanka gdp rem bsd 1tect  gdp (4) --- 2.621 (0.110) 1.970 (0.165) –0.834* (0.096) rem (4) 0.697 (0.68) --- 1.476 (0.340) –0.861 (0.571) bsd (4) 1.576 (0.268) 1.440 (0.331) --- –2.074* (0.063) notes: bootstrapped p–values are obtained based on 1000 replications, with simulated normal errors 2 and presented within parentheses below the test statistic. the optimal lag length is selected by minimizing the aic from a maximum lag of 4 3 . (*) rejects the null at 0.10 level of significance. (+/–) next to the asterisk shows the cumulative sign of coefficients. as for sri lanka, it is evident that remittances flow granger cause national income per capita in the long run through unidirectional causality. however, there is existence of 2 resampling the residuals produces qualitatively similar results. 3 we also rerun the whole estimation selecting the optimum lag length by maximizing of value of r 2 as suggested by abdalla and murinde (1997). the chosen lag lengths are very much similar and causality test results are qualitatively same. noman and uddin: remittances and banking sector development in south asia 60 bidirectional causality between banking sector development and gdp per capita. between remittances and banking sector development, a unidirectional causality is running from the remittances flow to banking sector development. all the existence of causalities for sri lanka is, however, long run by nature happening through the error correction terms. overall, a key finding of the results is that remittances and banking sector development (one in the presence of the other) influence per capita income in all four south asian nations. the importance of wage earners’ remittances on these economies, possibly also through enhancing banking sector capability, is quite evident. this finding is not surprising given the fact that the four countries being considered in this paper are part of the top 20 remittances receiving country of the world (seddon, 2004). remittances as proportion of total gdp in these south asian nations are also quite significant. for example, for sri lanka, official flow of remittances account for 7% of gdp (stalker 2001, p. 110). this result is indicative support in favor of hypothesis one. a second finding related to the hypothesis two is that remittances flow granger cause banking sector development only for pakistan and sri lanka, but not for bangladesh and india. and a third finding is that development of banking sector granger cause remittances inflow in india and pakistan, but not bangladesh and sri lanka. having discussed the multivariate causality results obtained on individual nations in south asia, we now move to look at the results of the panel causality using method as described in an earlier section. examination of the causality in a panel setting would give us better approximation of the relationship among the variables at hand by exploiting more information embedded in the whole panel. in addition, it would also provide with an opportunity to check robustness of the results obtained from time series analysis as documented above. table 4 presents results of the panel causality tests. it is obvious from table 4 that both remittance and banking sector development have positive and significant influence on the national income of south asian nations. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 3 61 table 4: panel causality test results panel a: wald test statistics dep. var. gdp rem bsd gdp ----51.284 (+) [<0.001] 11.269 (+) [0.003] rem 0.055 (–) [0.972] ----1.488 (+) [0.475] bsd 12.584 (+) [0.002] 8550.73 (+) [<0.001] ---- panel b: diagnostic tests dep. var. gdp rem bsd ar (1) –1.641 [0.101] –1.363 [0.173] –1.318 [0.188] ar(2) 0.698 [0.485] –1.115 [0.265] 1.145 [0.252] sargan tests 104.647 [0.408] 97.350 [0.611] 102.112 [0.478] notes: the wald tests are based on equations (8) through (10). test statistics are reported followed by their p–values within brackets. a +/– sign associated with a test statistic show sign of the cumulative effect of the relevant explanatory variable on the dependent variable. panel b reports related diagnostic test results. this result is in line with expectation under hypothesis one as well as what we found in the individual country analysis. on the other hand, neither domestic products nor advancement in banking sector have significant impact on the remittance flow. remittance seems to flow independent of the macroeconomic aspect of the recipient nation. the negative sign on the cumulative effect of gdp growth on remittance is interesting to note. this means that, as an empirical support for the hypothesis two, an improvement of the national income of the recipient country would cause a slowdown in remittance inflow. this is not surprising as the aggregate need of the remittance recipients would fall as the overall economic development occurs. table 4 also shows that banking sector development, as measured by the private sector credit disbursement by the banking system, is significantly affected by both noman and uddin: remittances and banking sector development in south asia 62 remittance flow and gdp. it indicates to the fact that remittance flow makes additional funds available to banks for loan disbursement. similarly, higher national income spurs enhanced economic activities resulting in increased credit flow to the private sector. the signs of the cumulative effects are also in line with the theoretical expectation. the results obtained in the paper reinforce those found in many other studies that investigate the interplay between remittances, financial sector development and income growth. for example, aggarwal et al. (2006) report findings that support the significant link between remittances and financial development in the developing countries. similarly, gupta et al. (2009) found that remittances have a direct poverty-mitigating effect and a positive impact on financial development. in the same line of inquiry, giuliano and ruiz-arranz (2009) confirms that remittances tend to lower inequality in the developing nations. yet in another paper, mundaca (2009) found growth enhancing effects of remittances through increased availability of financial services aimed for appropriate utilizing remittances. panel b of table 4 presents some relevant diagnostic test results. the null hypothesis of the absence of autocorrelation up to orders 1 and 2 cannot be rejected indicating that the error terms in each equation are free from serial correlation. in addition, we report results of the sragan test of overidentifying restriction for each equation from (8) through (10). as can be seen, in all three cases one cannot reject the null that the instruments used (i.e. lagged values of the variables) are all valid. in other words, the instruments are not correlated with the error terms in the original model. these diagnostic test results lend support to the validity of our estimated models. 5. concluding remarks the paper sought to explore the interrelationship among remittances inflow, banking sector development and national income in south asian countries that are major labor exporters in a multivariate causality framework. time series properties of the data dictated us to use error correction modeling in order to capture both short and long run causality running from one variable to the other. the overall results indicate that international journal of banking and finance, vol. 8, iss. 4 [2011], art. 3 63 remittances and banking sector development influence growth in per capita income in all four south asian nations. however, the causal relationship between remittances inflow and banking sector is not the same for all countries meaning that country specific characteristics of banking sector and remittances inflow are at work within south asian nations that share many economic realities among themselves. the results of the paper suggests that as smaller economies, bangladesh, pakistan and sri lanka stand to benefit to a large extent from banking sector expansion if restriction on remittances flows are eased. the findings have important policy implications as well. government in south asian nations need to design appropriate incentive policies to ensure channeling of remittances through the banking system. remittances, if channeled through commercial banks, can do two jobs at the same time. on the one hand, banks will be able to utilize these hard–earned foreign exchanges to their highest potential as capital goods rather than consumption goods, and thus enhancing the level of national income. on the other hand, commercial banks themselves would find additional resources beyond the national market which they could use to create additional assets for them. this would help the banks grow faster and smarter in the long run. in order to make sure that remittances do go through the banking system, it is imperative for transaction costs to be lower enough to attract wage earners’ attention. author information: abdullah m noman is at the department of economics and finance, university of new orleans la 70148, louisiana, usa. email: anoman@uno.edu. gazi salah uddin is at the department of economics, carleton university, ottawa, on k1s5b6, canada. email: mguddin@connect.carleton.ca noman and uddin: remittances and banking sector development in south asia https://sn2prd0310.outlook.com/owa/redir.aspx?c=vtdg3xywcui_agxrbcbjqnz10bp51c4irnq-c8lrn1f9x9q3oc4h4wif0soqtd6qxabxdlcktju.&url=mailto%3amguddin%40connect.carleton.ca 64 appendix: the data series used in this study (in log) bangladesh india pakistan sri lanka year gdp rem bsd gdp rem bsd gdp rem bsd gdp rem bsd 1976 5.460 0.188 2.965 5.362 0.642 20.259 5.662 3.089 21.932 5.940 0.362 12.975 1977 5.463 0.820 4.990 5.409 0.789 20.610 5.669 5.765 23.407 5.973 0.439 15.694 1978 5.509 0.865 4.534 5.442 0.870 22.635 5.716 7.346 22.260 6.011 1.427 20.311 1979 5.532 1.097 5.339 5.366 0.963 24.284 5.723 7.621 24.813 6.149 1.783 22.626 1980 5.517 1.871 5.771 5.408 1.517 23.993 5.791 8.645 23.407 6.091 3.778 17.165 1981 5.530 1.928 6.961 5.447 1.224 25.048 5.839 7.356 24.040 6.129 5.207 18.136 1982 5.529 2.908 7.344 5.461 1.344 26.920 5.875 8.423 24.703 6.157 6.061 19.624 1983 5.544 3.742 9.258 5.508 1.253 27.348 5.914 10.247 26.376 6.190 5.689 21.076 1984 5.570 2.547 12.125 5.528 1.111 29.125 5.937 8.285 24.218 6.227 4.980 19.303 1985 5.577 2.323 13.441 5.562 1.087 29.615 5.983 8.146 27.782 6.261 4.885 20.684 1986 5.596 2.722 13.157 5.588 0.920 31.261 6.010 7.668 29.786 6.286 5.089 20.356 1987 5.609 3.145 13.678 5.608 0.975 31.173 6.046 6.539 27.644 6.288 5.239 20.192 1988 5.608 2.980 14.944 5.681 0.795 31.174 6.093 4.866 26.369 6.298 5.130 21.800 1989 5.611 2.826 16.567 5.723 0.895 26.978 6.116 5.021 24.913 6.308 5.123 20.181 1990 5.646 2.586 16.656 5.759 0.752 25.242 6.134 5.014 24.157 6.358 4.993 19.616 1991 5.656 2.484 15.920 5.749 1.232 24.155 6.158 3.408 22.322 6.389 4.911 8.821 1992 5.683 2.876 14.546 5.781 1.186 25.122 6.207 3.236 23.617 6.423 5.648 9.056 1993 5.705 3.036 15.294 5.810 1.286 24.297 6.200 2.809 24.552 6.477 6.104 9.833 1994 5.722 3.408 16.271 5.864 1.816 23.993 6.211 3.370 24.006 6.517 6.101 10.957 1995 5.749 3.168 20.882 5.920 1.752 22.844 6.235 2.823 24.207 6.557 6.209 31.067 1996 5.773 3.307 21.597 5.974 2.274 23.858 6.258 2.028 24.694 6.584 6.131 29.896 1997 5.804 3.606 22.785 6.000 2.522 23.907 6.244 2.734 24.646 6.633 6.242 29.442 1998 5.835 3.642 23.236 6.041 2.291 24.104 6.245 1.884 25.114 6.668 6.477 28.722 1999 5.863 3.955 23.549 6.093 2.469 25.895 6.257 1.582 25.474 6.696 6.847 29.258 2000 5.901 4.179 24.666 6.116 2.801 28.845 6.275 1.466 22.528 6.738 7.139 28.835 2001 5.933 4.483 26.709 6.150 2.987 29.048 6.269 2.043 22.023 6.755 7.526 28.107 2002 5.957 6.015 28.930 6.172 3.102 32.760 6.277 4.972 21.925 6.779 7.916 28.580 2003 5.989 6.159 28.755 6.237 3.610 31.935 6.301 4.814 24.867 6.825 7.881 29.930 2004 6.031 6.323 30.141 6.303 3.072 36.883 6.339 4.104 29.297 6.866 7.926 31.547 2005 6.070 7.081 31.466 6.377 2.945 40.797 6.389 3.865 28.436 6.910 8.893 32.919 international journal of banking and finance, vol. 8, iss. 4 [2011], art. 3 65 references abdalla, i. s. a., and murinde, v., (1997). exchange rate and stock price interactions in emerging financial markets: evidence on india, korea, pakistan, and philippines, applied financial economics, vol. 7: 25-35. adelman, i., and taylor, j. e., (1990). is structural adjustment with a human face possible? the case of mexico, journal of development studies, vol. 26: 387– 407. aggarwal, r., demirguc-kunt, a., peria, m., (2006). ”do workers’ remittances promote financial development?” world bank policy research, working paper no. 3957. washington, dc. al-iriani, m., (2006). energy–gdp relationship revisited: an example form gcc countries using panel causality, energy policy, vol. 34: 3342–3350 arellano, m., and bond, s., (1991). some tests of specification for panel data: monte carlo evidence and an application to employment equations review of economic studies vol. 58: 277–297. baltagi, b. h., demetriades, p. o., and law, s. h., (2009) financial development and openness: evidence from panel data, journal of development economics vol. 89: 285–296 baltagi, b.h., (2005). econometric analysis of panel data. chichester: john wiley. beck, t., demirguc-kunt, a., and levine, r., (2004). finance, inequality and poverty: cross-country evidence, nber working paper 10979. cambridge, ma. cordova, e. l., (2004). globalization, migration and development: the role of mexican migrant remittances, intal/itd working paper no. 20, inter-american development bank (idb). de haas, h., (2007). remittances, migration and social development, social policy and development programme paper no. 34, geneva: unrisd. dustmann, c., kirchamp, o., (2001). the optimal migration duration and activity choice after re-migration, iza discussion paper 266. engle, r.f., and granger, c.w.j., (1987). co-integration and error correction: representation, estimation, and testing, econometrica , vol. 55: 251276. freund, c. and spatafora, n., (2005). remittances: transaction costs, determinants, and informal flows. world bank policy research working paper no. 3704. washington, dc. giuliano, p., and ruiz-arranz, m., (2009). remittances, financial development, and growth, journal of development economics 90: 144–152. granger, c. w. j., (1969). investigating causal relations by econometric models and cross-spectral methods, econometrica, vol. 34 no. 3: 424-438. gupta, s., pattillo, c. a., and wagh, s., (2009). impact of remittances on poverty and financial development in sub-saharan africa, world development vol. 37, no. 1: 104–115. hinojosa-ojeda, raul, (2003). transnational migration, remittances and development in north america: globalization lessons from the oaxaca-california transnational village/community modeling project. paper prepared for presentation at a conference on "remittances as a development tool in mexico" organized by the noman and uddin: remittances and banking sector development in south asia 66 multilateral investment fund (mif) of the interamerican development bank (idb), mexico city, mexico, october 28, 2003. holtz-eakin, d., newey, w., rosen, h., (1988). estimating vector autoregressions with panel data, econometrica, vol. 56: 1371–1395. holtz-eakin, d., newey, w., rosen, h.s., (1989). the revenues-expenditure nexus: evidence from local government data. international economic review 30: 415– 429. koechlin, v., and gianmarco, l., (2007). international remittances and income inequality: an empirical investigation, journal of policy reform, vol. 10, no. 2: 123–141 levine, r., (1997). financial development and growth. journal of economic literature vol. 35: 688–726. liew, v. k−s, (2004). which lag length selection criteria should we employ? economics bulletin vol. 3, no. 33: 1−9. orozco, m., and fedewa, r., (2007). leveraging efforts of remittances on financial intermediation, integration & trade journal 27: 211-239. paulson, a. l., and townsend, r., (2000). entrepreneurship and financial constraints in thailand, working paper, northwestern university. ratha, d., and riedberg, j., (2005). on reducing remittance costs. mimeo. washington, dc: world bank development research group. sander, c., and maimbo, s. m., (2005). migrant remittances in africa – a regional perspective, in s. m. maimbo, and d. ratha (eds.) remittances – development effect and future prospects. washington, dc: world bank. seddon, d., (2004). south asian remittances: implications for development, contemporary south asia, vol. 13, no. 4: 403–420. sims, c., (1972). money, income, and causality, american economic review vol. 62 no. 4: 540-552. stalker, p., (2001). a no –nonsense guide to international migration. oxford: new internationalist publications. terry, donald f., and wilson, steven r., (2005). beyond small change: making migrant remittances count. washington, d.c.: inter-american development bank. toxopeus, h. s., and lensink, r., (2006). remittances and financial inclusion in development in t. addison, and g. mavrotas (eds.) development finance in the global economy: the road ahead. new york: palgrave macmillan. world bank (2006). global economic prospects: economic implications of remittances and migration. washington, d.c. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 3 an empirical investigation of new bond issue yield spreads, default risk and split ratings the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 141 an empirical investigation of new bond issue yield spreads, default risk and split ratings timothy s. michael university of houston clarke-lake, united states of america abstract this paper attempts to explain the yield spreads charged to new corporate debt issues by comparing the initial yields of a set of 3,287 securities issued over eleven years in the us. we use the measure of constant maturity treasury rates on the day of issue against the moody’s aaa corporate bond index for the week prior to the issue, and the yield on a daily index of long-term treasury securities on the issue date. the influences of credit ratings and disagreement between rating agencies as reflected in split ratings and the interactions between these characteristics are measured. the contributions of sinking fund provisions, call or refunding status, overseas issue and contractual security arrangements are evaluated separately. the results support the view that the higher yields are observed when ratings of agencies differ and that factors associated with the issues also are significant drivers of the yield difference. key words: yield spreads, split ratings, bond index, credit rating effect, terms of debt issues jel classification: e43, g12 i. introduction the influences of credit ratings and disagreement between rating agencies and the interactions between these characteristics are issues that are addressed in this paper. we include in this paper new findings on the contributions of sinking fund provisions, call status, overseas issues and contractual security arrangements. the yield spreads of debt issues contain important information for investment analysts to judge the level of risk of an issue. therefore, an analysis of the yield spreads to identify the factors in a debt issue is expected to add to our understanding of risk of an issue thus aiding analysts in their investment evaluation. ijbf 142 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 the model tested in this paper was most influenced by a prior work on split bond ratings and security yields by billingsley, lamy, marr, and thompson (1985) (blmt). in their model, blmt measure the yield off-treasury (yot) for new securities, defined as issuing yield less moody’s index of long-term government bonds on the issue date. to explain this spread measure, they used the average daily interest rate (level) on moody’s long-term treasury bond index, an absolute deviation measure of volatility in interest rates for the ten days prior to the issue date, a quality spread of the average yield on long-term corporate bonds over the treasury index rate on the issue date, the natural log of the issue size, the term to maturity of the issue, and the minimum number of years until first call. they also include dummy variables for four moody’s rating classes, the existence of a sinking fund provision, and four separate split rating categories for single interclass splits.1 the rest of the paper is organized as follows. section 2 is a review of the literature relating to the yield spreads at the time of issue relating to split ratings by two rating companies. in section 3 we present the research design of this study, and the results are discussed in section 4. the paper ends with comments in section 5. 2. brief review of related literature for a sample of 258 new industrial bonds issued between 1977 and 1983, blmt find significant relationships between yot and an overall quality spread, the level of interest rates, term to maturity, call term, and all four ratings classes (the omitted class, as represented by the intercept term, was not significant). in addition, the coefficients for the split categories aaa/aa, aa/a and a/ba are also significantly different from zero, and inversely related to the yot. using a standard f-test for a subset of coefficients, blmt show that the coefficients on the class dummies are significantly different in all but the aaa versus aa case. their results indicate that split rating coefficients are significantly related to the higher rating class (with the exception of the aaa/aa split), suggesting that the market may weigh bad information more heavily when assigning initial yields to an issue which has caused a difference in rating agency opinion. in other words, there is a significant difference in yields when (for their sample) the moody’s rating of the issue is higher than that given by standard and poor’s. the potential impact of the blmt (1985) results suffers from a lack of degrees of freedom. as pointed out by perry, liu and evans (1988), of the 258 issues in the blmt study, only 33 have a split rating, making the fine distinctions found by blmt little suspect, or at least in need of confirmation by a larger sample. perry, liu and evans (1988) confirm the relationship between splits, yields, and the higher ratings class for a larger sample, again using each 1 an example of an interclass split would be an issue rated aaa by moody’s and aa by standard and poors. no provision was made to measure the splits across more than just broad classes. revised categories, which distinguish between three different intraclass ratings, were not used in the blmt study. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 143 issue’s moody rating as a base. liu and moore (1987) reaffirm the these prior relationships, but show that the phenomenon is not significantly different for moody’s versus standard and poor’s, suggesting that a large premium is demanded from any issue which is rated lower than the rating on a benchmark portfolio of equally-rated bonds. there is also a stream of literature which explores the relationship between security ratings, default risk and split ratings, and these studies question the ability of credit ratings to capture the market’s full assessment of default risk at different points in a security’s life. this line of research is characterized by the studies of ederington (1986), ederington, yawitz and roberts (1987), and reiter and ziebart (1991).2 evidence from this body of works suggests that ratings may become “stale” for some firms since rating agencies may not revise a firm’s credit ratings at regular or frequent intervals. as a response to concerns about this issue in the current study, it should be pointed out that the inaccuracy of ratings in capturing the default risk of a particular security should be small for new issues relative to seasoned securities. most prior studies designed to evaluate default risk proxies have used samples of outstanding bonds. the default risk literature provides evidence that credit rating categories are significantly related to new issue yields. for example, ederington, yawitz, and roberts (1987) show that ratings are significant and capture most of the variation in yields, even when combined with firm-specific accounting measures. the credit rating assigned a new issue is likely to be the most complete timely measure of default risk available to the majority of bond investors. another question from default risk research is whether yields are determined by credit ratings or credit ratings are determined by issue yields. reiter and ziebart (1991), kao and wu (1994) and liu and thakor (1984) all recognize the simultaneity of bond rating and price determination. this problem is not addressed in the current paper, other than to suggest that the direction of causality for new issues is probably more from ratings to yields. this relationship should be clearest for a new security given the nature and timing of the issuing process, which also impacts new issue liquidity. the direct effect of illiquidity on seasoned corporate bond prices is well documented. zivney, bertin and torabzadeh (1993) review the impact of infrequent trading on the availability and quality of high-yield bond price data. for the present study, confining the sample to yields on new issues is partly an attempt to minimize liquidity differences between a particular new issue, treasury securities, and the bonds included in moody’s aaa corporate bond index. if systematic liquidity effects exist, they should be smallest at issue time. this assertion acknowledges the discretionary nature of most debt issues investment bankers advise firms (or savvy firms know) when to issue based 2 kaplan and urwitz (1979) show that studies attempting to explain bond ratings using standard ols models and a categorical dependent variable for rating class are flawed. as a result of their original study, ordered probit models of ratings have been developed over the past decade. these newer models are used in most of the recent research on default risk. 144 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 on the banker’s ability to sell at some target price. the final data used here are made up of only those securities for which a substantial amount of information is present at the time of issue; these are likely to be the most actively traded securities once they are issued. the models tested in this paper are also guided by the sinking fund model of kidwell, marr, and ogden (1989), which uses a standard yot equation with interaction terms to measure the potential slope differences for sinking fund issues as they relate to the overall level of interest rates and an ordered probit model measure of default risk. the sample included 192 new utility issues during 1977 and 1982 period. both of these interactions are highly significant in that study, but a simple dummy variable for sinking fund presence was only significant at the 0.10 acceptance level. a similar model explaining yield level rather than a quality spread is estimated by kidwell, marr and thompson (1984). both studies included some measure of term to maturity, the overall level of interest rates, and some recognition of the call provision associated with the issue. these variables added explanatory power in each case. other related models are the yield models of sorensen (1981 and 1979). hsueh and chandy (1989) developed a similar model for the net issuing cost of insured and uninsured municipal bonds. 3. research design a basic model of the quoted interest rate on a debt security is given by brigham and gapenski (1994, pg. 103): y i = k* + ip i + lp i + mrp i (1) where, y is the quoted rate on security i, k* is the real risk-free rate of interest, ip is an inflation premium, drp is a default risk premium, lp is a premium for liquidity risk, and mrp is a premium for maturity risk. at any point in a debt security’s life, it should be possible to decompose its yield into subsets resembling those in the quoted rate expression. similarly, the simplest way to measure any set of the right-hand-side premia is to subtract from both sides the yield on a security which is similar to the measured security in all but the targeted set of characteristics. a treasury yield, for example, contains the real rate, inflation expectations, and if it is of the same term as the measured bond, the same maturity risk premium, leaving default and liquidity risk to explain the yield differential.3 liquidity effects can be reduced by using new issue information, so this example would end up being primarily a model of default risk. 3 kidwell, marr and ogden (1989) indicate that new issue yields should not reflect any differences in price due to different coupon rates on the compared securities because most issues are priced close to par at issue. to make sure that this kind of influence was removed from the sample, original-issue-discount bonds were dropped, as described in section 4. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 145 for the set of corporate debt securities which have been rated by moody’s and standard & poor’s, the default risk component of a security’s yield can be broken into two parts. as a base, the spread between an index of aaa corporate bonds and the constant maturity treasury rate for a particular term represents the minimum default spread charged in the market on a given issuing day. the default premium on any new security should be greater than or equal to this aaa quality spread on the day the reissuing price is set, which may be up to a week prior to issue.4 in other words, the difference between the measured new issue yield and a treasury rate will contain or at least be a function of the default premium on an index of aaa bonds on the issue date. the difference between the premium paid by a aaa bond (or index) and the new issue will contain premia to pay for all other sources of default risk. in reality, and as measured here, this difference will contain components of liquidity risk and perhaps maturity risk as well. the base model evaluated in this paper is of the form: y i,d,t = t d,t + [(aaa index) d – t d,t ] + drp i + other premia (2) yot i = (y i,d,t – t d,t ) = [(aaa index) d – t d,t ] + drp i + other premia (3) where, y i,d,t is the observed yield on new issue i on issue data d with a term to maturity of t years. t d,t is the day d observed yield on a treasury security with a term to maturity of t years.5 the aaa index measure is the last weekly yield on moody’s aaa corporate bond index for the week up to and including the issue date, and drp i is a unique default risk premium for the new issue. the general category “other premia” may contain a great deal of pricing information. the analysis in this paper limits the investigation of this category to residual term effects, the effect of gross issue size, the yield premium demanded for issuing abroad (“euro” securities), the level of volatility in an overall quality spread for corporate bond yields for some period prior to issue, and the effect of having a call or refunding provision. differential default risk is measured as outlined above, using a quality spread for corporate yields, defined as moody’s aaa corporate bond index less the constant maturity treasury yield corresponding to the exact term to maturity of the measured issue.6 default 4 to the extent that the aaa bond index is a noisy measure of the current yield paid by high quality credits, this model is flawed. the data series used here -weekly values of moody’s aaa corporate bond index -is the most accurate available from public electronic sources. better precision could be obtained by matching each new issue with an actual aaa corporate bond of the same term on the issue date, but this would introduce other factors into the model. 5 i chose to subtract treasuries from issue yield in order to reduce measurement error and collinearity influences on the right-hand-side of the model; measurement error in the yot affects the size of the error term. 6 moody’s aaa corporate bond index for one and two weeks prior to the issue date were also tested in relation to each of the models developed. the results reported were the most significant. 146 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 risk is also measured by categorical variables for six separate moody’s rating classes: aaa, aa, a, baa, ba, and b. in addition, dummy variables for secured issues, the existence of sinking fund provisions, and disagreement between credit rating firms (split ratings) are included. following an initial specification and estimation, separate categories indicating the magnitude (1 through 3 revised rating subclasses) and direction (positive or negative) of rating splits are introduced and tested. finally, a respecified model adds interactions between the rating classes and the split magnitude and direction terms. the issue’s yot is a spread over some base rate, which may or may not be known or well-proxied by an observable rate. the right side of this equation may contain many items related to the issue that are not found in the base rate or its proxy, yet there is no sound theoretical reason that an interest rate level component should contribute to the yield spread on the left. models which use full interest rates to explain yield spread are assuming a great deal about the timeseries behavior of default risk.7 as another consideration, an interest rate level may be highly collinear with other explanatory variables. at least two existing studies (perry, liu and evans (1988) and liu and moore (1987)) avoid this question and measure the impact of ratings splits on yields spreads in directly. the results of both of these studies are robust to this criticism. this study was originally designed to take advantage of data recently made available by the capital markets division of the federal reserve board of governors.8 the full data base is composed of 10,286 new corporate debt securities of varying characteristics issued between 1982 and 1993, with information originally taken from the wall street journal, s&p credit week, moody’s bond survey, investment dealer’s digest, international financing review, and bloomberg business news (guedes and opler (1995)).9 though incomplete for some items, the data set includes for most issues the gross amount, initial yield, coupon rate, issue date, maturity date, standard & poor’s rating, moody’s rating, date (year) of first call, first refunding date, and the first sinking fund payment date (year). in addition, there are several classification items provided which describe some of the relevant characteristics of each issue, such as whether it is convertible, an original-issue-discount (oid) security, or was sold in relation to a merger or swap. using all of these indicators combined, it was possible to reduce the initial set to a group of 4,647 debt securities with issue dates between january 3, 1983 7 given the obvious popularity of using rates levels to explain rate spreads, there may be extremely valid theoretical reasons for including the absolute magnitude of interest rates in models. it stills seems counterintuitive to the author. 8 the complete data set in raw form is available in both comma-separated-value (csv) and excel formats from dr. tim opler’s world-wide web page at http://www.cob.ohiostate.edu/dept/fin/osudata.htm. 9 guedes and opler (1995) report 10,287 records in the original file. for this paper, the data file was read into dbfast 2.0 as text and number fields and converted to a dbase iii format data matrix. the original number of observations in the data set when read was 10,286. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 147 and june 30, 1993. approximately 1,300 issues were dropped for not having a listed maturity date. issues were kept if they were rated at issue by both moody’s and standard and poor’s (approximately 6,300 issues). securities without an entry for either rating or with a un or nr listing were deleted. issues without yield or gross issue size numbers were also removed. from the remaining securities, the following major categories were deleted: convertibles, warrant bonds, unit issues, foreign bonds, guaranteed or government-backed issues, private placements, oid issues, participation certificates, resets, “poison put” issues, merrill-lynch lyons, variable or adjustable rate securities, merger related securities, and securities with specific event risk covenants. the remaining observations should include only short, medium and longterm straight debt, notes, mortgage bonds (and other secured issues) meeting the above criteria. the original data does not provide any way to divide the individual issuers into their appropriate industries.10 for this reason, the data set contains issues by utility, industrial and financial firms. an anticipated extension of this paper will be to correct the data for obvious industry effects by removing all but industrial firms from the sample. this data also include multiple issues by the same firm if they meet the qualifications listed. each valid observation was subsequently matched to a daily treasury yield by term and issue date. information for comparable treasury securities was obtained from federal reserve statistical release h.15.11 for each issue date, this source provides a daily yield on an index of government bonds with maturities between ten and thirty years, daily constant maturity (cm) treasury yields for 1, 2, 3, 5, 7, 10 and 30 year maturities, and most daily 20-year cm rates between 1982 and 1987. after matching, 3,293 securities had issue dates and terms corresponding to the available treasury yield data. these matching rates were subtracted from issue yield to obtain ysprcma, the actual yield spread on the date of issue for each security. these were then matched to weekly observations (daily averages) of moody’s aaa corporate bond index in order to calculate the explaining spread variable cmqsyld, which is the aaa index yield minus the actual comparable constant maturity rate for each issue on the issue date. table 1 gives the distribution by year and term for the final sample of new securities.12 10 guedes and opler (1995) report that out of 10,287 observations, they were able to match 7,369 to standard and poors compustat (presumably by cusip number, which is included in the data set). this would allow for industry identification. in addition, they report that their models often used fewer than 7,369 observations due to missing data in the compustat files. 11 weekly and daily numbers for key interest rates are available in a variety of files and formats from the federal reserve at either its chicago or st. louis district banks. the electronic resources of the entire federal reserve system can be accessed in the fred data base, located at http://www.stls.frb.org. 12 note that the totals in each table are adjusted for the removal of outliers as discussed in section v., below. 148 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 table 1: annual security issues by term to maturity for the sample of 3,285 new corporate debt securities issued between january 3, 1983 and june 30, 1993, with known maturity dates and yields which could be matched by term to maturity with a constant maturity treasury rate on the date of issue. this sample does not include any convertible, original-issue-discount, floating rate or foreign issues. note: 20-year constant maturity treasury rates are only available prior to 1987 and after june 1993. term was measured as the number of years from the issue date until the listed maturity date (when available) or as the number of years between the maturity year and the year of issue. year term to maturity (in years) avg. 1 2 3 5 7 10 20 30 total 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1 1 2 25 6 7 9 2 1 2 1 2 3 7 9 8 14 4 2 7 6 10 3 8 1 12 23 39 21 4 10 16 38 38 47 12 22 44 93 52 7 11 32 52 41 35 16 22 48 127 78 39 51 111 184 111 111 85 50 85 222 157 13 10 11 21 61 25 73 167 91 59 64 48 108 123 111 127 116 251 472 302 265 210 169 323 627 425 20.34 13.89 15.32 16.64 14.49 12.80 14.21 13.39 14.67 11.83 13.59 totals: 53 51 147 376 467 1,206 55 930 3,287 14.65 grand total: 3,287 issues table 2 presents descriptive statistics for the important measured and calculated variables yield, cmyld, yot, cmqsyld and gross for each year of sample data. these statistics indicate the means, variability and the range of values by year. the means of the four variables indicate that the measures are in line with previously reported figures, although these measures are from a very large sample used in this study. the data set contains issue with maturities of 1 to 30 years (see the columns) across 11 years as rated by the rating agencies. there is a total of 3,287 observations in this study. table 3 shows the breakdown of the four non-rating categories (euro, secdum, crfdum and sfdum) and the split dummy variable for each year. the categories in table 3 are not mutually exclusive. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 149 table 2: annual descriptive statistics for selected data series for the sample of 3,285 new corporate debt securities issued between january 3, 1983 and june 30, 1993, with known maturity dates and yields which could be matched by term to maturity with a constant maturity treasury rate on the date of issue. this sample does not include any convertible, original-issue-discount, floating rate or foreign issues. year is the year of issue; n is the number of observations in that year; yield is the measured yield-atissue of the new security, in percent; cmyld is the constant maturity treasury rate for the same term as the new issue on the issuing date; yot is (yield cmyld); cmqsyld is the average yield on moody’s aaa corporate bond index (weekly) for the issue date minus cmyld; gross is the total size of the security issue in millions of dollars. year n variable name yield (%) cmyld (%) yot (%) cmqsyld (%) gross ($m) 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 125 116 251 472 302 265 210 169 323 627 425 mean st. dev. min. max. mean st. dev. min. max. mean st. dev. min. max. mean st. dev. min. max. mean st. dev. min. max. mean st. dev. min. max. mean st. dev. min. max. mean st. dev. min. max. mean st. dev. min. max. mean st. dev. min. max. mean st. dev. min. max. 12.08 1.2611 7.00 15.50 13.60 1.4326 10.30 17.61 11.53 1.5613 6.72 16.03 9.37 1.5681 6.50 14.31 9.86 1.7698 6.07 16.94 10.41 1.9150 5.14 17.51 10.30 1.8420 5.15 15.50 9.56 0.7068 7.75 14.60 8.84 1.0969 5.05 13.50 8.09 1.8244 3.45 13.33 7.35 1.6289 3.48 13.00 10.84 0.6135 9.26 12.15 12.24 0.7438 10.29 13.99 10.44 0.7228 7.94 12.00 7.55 0.6167 6.42 9.67 8.11 0.7549 6.37 9.96 8.69 0.3744 7.59 9.40 8.51 0.6298 7.53 9.85 8.42 0.3042 7.47 9.08 7.69 0.6119 4.98 8.53 6.68 0.8829 3.36 8.06 6.05 0.7436 3.28 7.48 1.23 0.9697 -3.78 3.88 1.36 1.1695 -0.45 5.14 1.09 1.4404 -3.55 6.54 1.82 1.5554 -0.98 7.21 1.75 1.5629 -2.48 10.57 1.72 1.7982 -3.86 8.43 1.78 1.7619 -3.50 7.09 1.14 0.6143 0.07 6.29 1.14 0.8578 -1.37 6.39 1.41 1.4785 -0.46 6.72 1.29 1.4355 -0.24 7.47 1.036 0.3876 0.42 2.51 0.48 0.4701 -0.57 1.92 0.86 0.3554 0.23 3.17 1.45 0.3192 0.30 2.35 1.09 0.4179 0.34 2.92 1.065 0.3486 0.36 2.17 0.77 0.3175 -0.06 1.38 0.92 0.2784 0.37 1.78 1.07 0.5417 0.44 3.57 1.45 0.8226 0.22 4.58 1.55 0.7071 0.39 4.33 89.84 45.10 15 250 90.85 56.60 8 300 103.61 65.78 3 500 120.02 72.43 8 440 124.07 90.76 2 900 133.07 75.21 3 500 152.47 102.09 3 750 166.91 117.93 15 750 173.95 132.24 4 1,500 151.60 114.65 1 900 158.96 132.13 1 1,250 150 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 table 3: annual security issues by type of issues for the sample of 3,285 new corporate debt securities issued between january 3, 1983 and june 30, 1993, with known maturity dates and yields which could be matched by term to maturity with a constant maturity treasury rate on the date of issue. this sample does not include any convertible, original-issue-discount, floating rate or foreign issues. note: categories here are not mutually exclusive. year is the year of issue; n is the number of observations in that year; euro is a dummy variable indicating whether the security was issued in a foreign market by a domestic firm; sfdum is a dummy variable which is 0 if a sinking fund provision exists for the issue; secdum is a dummy variable indicating that the issue is secured (mortgage, lease or equipment trust related); crfdum is a dummy variable which is 1 if the call or refunding date is included in the original issue announcement; split is a dummy variable which is zero if the issue’s moody’s rating is equivalent to the rating assigned by standard and poors. year n variable name euro sfdum secdum crfdum split 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 totals 127 116 251 472 302 265 210 169 323 627 425 3,287 6 1 33 29 10 17 7 1 14 16 4 138 15 12 23 44 46 28 20 7 5 11 4 215 64 44 61 151 81 56 54 47 61 211 151 981 87 62 159 255 149 135 80 40 49 144 123 1,283 73 74 142 291 176 151 106 76 144 315 230 1,778 4. findings and discussion a) a simple model the second column of table 4 shows the results of an initial regression model (model 1a) fit to the full comparable data set.13 the maturity-matched 13 the techniques and specifications in this section were originally applied to the full 4,676 observations as well as those with directly comparable maturity matches (3,293 observations with outliers). for the full data set, yot was calculated as the issue yield minus the daily yield on a long-term treasury index; the rhs spread variable used was the yield on moody’s aaa corporate bond index less the daily treasury bond index yield. these regressions never produced a significant coefficient for the spread variable, even when all different lagged values of spread and volatility (standard deviation or meanabsolute-deviation) were estimated in turn. a full interest rate level measure on the rhs didn’t improve any of these models over the specifications reported, either. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 151 spread measure, issue term, volatility (measured by mean absolute deviation) and four of five bond class coefficients are significant at the 0.01 acceptance level, with the remaining rating class (aaa) dummy significant at the 0.05 levels. the size factor and the call/refund dummy are significant at the 0.10 levels. the intercept, which corresponds to the a rating class, is not different from zero. an f-test for overall significance rejects the null hypothesis at the 0.01 level.14 initial results (not presented) were sensitive to the choice of the omitted rating class. early specifications of the model using aaa as the omitted class were designed to make interpretation of the difference between lower yields and the quality spread variable clearer, but these models showed high levels of multicollinearity between the size variable and several bond classes.15 when the omitted class was changed to b, these problems disappeared, but several of the other regressors lost significance. finally, the a class was chosen due to its large number of observations relative to the other classes and the balanced results it provided.16 in addition to the original specification used in model 1a, the model was estimated for measured values of the quality spread and volatility variables observed one and two weeks prior to the issue date (these results are not presented). these models were no more significant or intuitive than the base case. the model was also tested using one and two-week lagged values of the prior 26-week standard deviations in place of the mean-absolute-deviation term; there was a slight decrease in significance for several regressors, and the overall explanatory power dropped. as shown by its retention in model 1a, the term variable performs better than its natural log, although gross does not. after adjusting the omitted class, the regressor, variance inflation factors, in model 1a dropped to near one.17 in addition, the model’s condition number, 14 models 1b, 2 and 3 are also significant at the 1% level. 15 aaa was originally made the omitted class to avoid problems with the lack of negative splits associated with this class. a standard and poors security rating can never be higher than aaa, so there would never be any observations of an interaction term between the class aaa and the split categories which represent one subclass rating (e.g., aa+ versus aa) higher than the current class, two classes higher, and three classes higher. 16 this specification problem suggests that the overall relationship between default risk and quality spreads not likely to be linear. even so, the relationship between individual rating classes and yield spreads may still be linear. 17 the vif option in sas proc reg produces variance inflation factors, which are the inverse of the tolerance for each regressor. freund and littell (1991) note that these should be below the reciprocal of one minus the r-squared for the overall regression, since the vif is reciprocal of one minus the r-squared for a regression of each regressor on the others; a high vif indicates that the other independent variables are more closely related to the variable in question than to the dependent variable. for this initial regression, the suggested threshhold value is 3.352. freund and littell also note that other authors consider 10 to be an acceptable threshhold value. 152 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 formerly above forty, dropped to below 25.18 multicollinearity, obviously present in each of these models, was not necessarily harmful to the results. the initial model was tested for the presence of influencial outlier values. eight observations (out of 3,293) were found to be extreme and were dropped from the data set used to estimate the remaining regressions. dropping these data points increased the fit of the overall regression and increased the significance of most of the coefficients. this is model 1b in table iv. in addition, the coefficient on euro was found to be negative and significantly related to the yot at the 0.05 level.19 these eliminations improved the fit of models 2 and 3 as well. the error variances were examined for heteroskedasticity using goldfeldquandt tests as described in kennedy (1992) and pindyk and rubenfeld (1981). tests were performed for cmqsyld, term and lngross as suspected related regressors; only cmqsyld marginally rejected the null. a weighted least squares performed to correct this relationship only increased the goldfeldquandt test statistic, which suggests that there may be a relevant variable left out of the model.20 the most likely omitted variable is one which conveys information about the general level of interest rates. as noted above, most prior models have incorporated some value of contemporaneous interest rates to increase their explanatory power. a measure of the moody’s aaa corporate bond index yield minus the rate on a daily index of long-term government bonds was added as a regressor; the aaa corporate bond index yield and the treasury bond index yield were each added separately in turn. none of these changes lower the level 18 the condition number is reported in sas using the collin option with proc reg. this number is given along with the variance proportion for each independent variable. freund & littell (1991) suggest that a condition number greater than thirty may indicate a problem with multicollinearity. a high condition number for a given eigenvalue combined with variance proportions of over 0.5 for any regressors indicates that these regressors may cause collinearity problems. in the initial screen of model 1a (with outliers), the model was tested without the lngross term to attempt to correct for a multicollinearity problem according to a high variance proportion and a large condition number. no significant changes resulted. 19 diagnostic values of the hat matrix, rstudent, covratio, dffits, and dfbetas were generated using the influence model option available in sas proc reg. these were output to a separate data set and tested for cutoff values as described in sas/stat user’s guide, version 6 (1989) using summary and screening procedures, for fifteen parameters estimated (p) and the number of observations [n]. dfbetas for this model were inspected individually. in all cases, the deleted observations had rstudent values greater than two in absolute value and h values greater than [2*p/n]. none of the observations had extreme values of dfbetas, covratio, or dffits. 20 the reweighting in this case consisted of dividing all observations by cmqsyld, as suggested by pindyk and rubenfeld (1981). the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 153 of heteroskedasticity in the model, nor does the addition of the actual constant maturity treasury yield on an issue-by-issue basis (cmyld).21 22 for model 1b, f-tests are used to evaluate the joint impact of the set of rating class coefficients; these are significantly different from zero as a group (including the intercept) at the 0.01 level. the coefficient for the aaa class is not significantly different from that of the aa class, and the aa class is not different from the a class (intercept). the first of these results confirms the findings of blmt (1985) regarding the explanatory power of different rating class categories. the remaining adjacent class coefficients are different from each other at the 1% level, consistent with prior research. these results suggest that the default risk distinction is not as obvious between higher rating classes than between ratings overall. the coefficients for crfdum (the existence of call or refunding provisions) and euro (overseas issue) are both different from zero at the 5% level in model 1b. the coefficients for secdum (secured issues) and sfdum (sinking fund) are not significantly related to yield spreads in this sample. these four relationships hold for each of the models discussed below as well. b) more complex specifications from table 4, it is clear that the ratings categories explain a great deal of the variance in an issue’s yot, but the split designator has no statistical influence on the model results. in model 2, the split category variable is replaced by six dummy variables which indicate the direction and the magnitude of the split; those securities with equal moody’s and standard and poors ratings make up the omitted class.23 these results are shown in column four of table 4. the six split categories represent one, two or three numerical values away from the moody’s rating in either the positive (p) or negative (n) direction. all of the positive split coeffients are individually different from zero at the 0.05 level and directly related to yot, indicating a direct relationship between a higher moody’s rating and an issue’s yield spread in the subclass categories 21 both breusch-pagan and white tests, as described by kennedy (1992), were used to confirmed the general nature of the unequal error variance in the original model. another weighted least squares procedure, also from kennedy, was estimated by weighting the observations by the square root of the estimated error variance from the original regression to produce gls estimators. a goldfeld-quandt test on this regression also indicated significant heteroskedasticity at the prior levels. for this reason, the remaining analysis and extension was done on the original equation and additions to it; an extension of this paper would be to re-estimate these models using an appropriate specialized panel data approach. 22 durbin-watson statistics were tested to assess the level of autocorrelation in the error terms. these are close to 2 for all models tested, and the first-order autocorrelations calculated for residuals are generally below 0.13. for the four models presented, the first-order rhos are less than 0.10. 23 issues showing differences of +/4 and 5 revised categories were deleted from the sample because there were few in each class. a total of approximately forty observations were lost for this reason. 154 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 higher by standard and poor’s) is different from zero at the 10% level and inversely related to measured yot. f-tests on the subset of coefficients for the split categories indicate the coefficients are significantly different from zero, as a group. in addition, the positive split group is jointly different from the intercept term; the negative splits, indicating higher standard and poors ratings for particular issues, are not. some of this difference may be due to the presence of more positive split observations in the later years of the sample. of the six intra-group boundaries, only the coefficients for a positive three and a positive two split are significantly different from each other. the rating classes are still significantly different from one another in model 2. an f-test confirms that the same class relationships hold from model 1b. evidently, the basic relationship connecting splits, ratings and yields is not completely described by this specification. the model results suggest that split issues generally have higher yield spreads, but that this higher spread may be due to the impact of moody’s versus standard and poor’s as the rater assigning the higher issue. in other words, split ratings may not be symmetric with respect to the rater which assigned the higher rating. table 5 contains the distribution of split magnitudes across the six broad classes of bond rating and each of the seven subclasses for each issue year. as noted above, aaa bonds cannot have any negative split observations. split magnitude and direction refer to the standard and poor’s “revised” rating subtracted from the moody’s revised rating. nineteen sub-categories (such as aa1, aa2 and aa3 for moody’s and aa+, aa and aafor standard and poor’s) were initially developed from the data. model 3 was created to exploit the relationships evident in table 5. this model adds interaction terms for split magnitudes and directions and specific ratings classes for those interactions which had thirty or more observations over the sample period.24 as shown in table vi, the positive split coefficients lose most of their explanatory power compared with model 2, though an f-test indicates that the positive split interaction estimates as a group are still different from zero at the 0.01 level.25 the positive three-split coefficient remains different from zero at the 0.05 level. the p1baa coefficient is positively related to issue yield spread and different from zero, as would be expected from model 2. although only one of the model 2 negative split terms is significant, when these observations are grouped by rating class the coefficients for n2ba, n1ba, and n1b are inversely related to yot and significant at the 0.01 level. in addition, the coefficients of n1b and n1ba are different from the model intercept at the 0.01 level, and the p1baa term is different from the intercept at the 0.10 level. none of the adjacent interactions were significantly different from each other. 24 the same model was estimated using all interactions terms having existing observations; the significance of the model coefficients and the model’s overall explanatory power were no different. 25 coefficients found using all interactions which had observed values were not significantly different from zero as a group. this may be due to contextual outliers not identified before due to the limited information used by the original the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 155 table 4: initial regression results the model explains new security yield off-treasury measured with an exact maturity match constant maturity treasury rate. cmqsyld is the difference between moody’s aaa corporate bond index and the matchedmaturity cm rate; lngross is the natural log of the issue size (in millions); term is the term to maturity of the issue (in years); dmad1wlg is the mean absolute deviation in the spread between moody’s aaa index yield (weekly) and an index of long-term (10+ years) government bonds (weekly) for the 26 weeks prior to the issue date; sfdum, crfdum, secdum, euro and split are 0/1 dummy variables for the existence of a sinking fund, a known call or refund date, a secured issue, a euro issue, and a split between the ratings given the issue by moody’s and s&p. aaacl through bcl are dummy variables for the general ratings class for the issue’s moody’s rating, with a (1,280 obs.) being the omitted class. mdifn3 through mdifp3 are dummy variables for the magnitude and sign of the difference between the issue’s moody’s rating and its standard & poors rating, with 0 (no difference) being the omitted class. model 1a was estimated before the elimination of 8 outlier observations, as indicated by values calculated by the influence option in sas proc reg. model 1b is identical to 1a with this exception. all four models were significant at the 0.01 level. *** indicates individual coefficient significance at the 0.01 level; ** indicates significance at the 0.05 level, and * indicates significance at the 0.10 level. regressor model 1a (w/ outliers) model 1b model 2 intercept cmqsyld lngross term dmad1wlg sfdum crfdum secdum euro split aaacl aacl baacl bacl bcl mdifn3 mdifn2 mdifn1 mdifp1 mdifp2 mdifp3 -0.02817 (0.10940) 0.17227*** (0.02676) 0.03412* (0.01779) 0.02177*** (0.00182) 1.47257*** (0.30805) 0.07705 (0.05920) 0.06449* (0.03307) 0.00203 (0.03614) 0.07547 (0.07250) 0.04567 (0.02889) -0.21133** (0.07574) -0.18943*** (0.04129) 0.49822*** (0.03724) 2.15609*** (0.06510) 3.82686*** (0.05185) -0.00465 (0.10316) 0.15008*** (0.02530) 0.03858** (0.01676) 0.02139*** (0.00171) 1.50049*** (0.28991) 0.07432 (0.05568) 0.06915** (0.03110) -0.00105 (0.03401) -0.14941** (0.06945) 0.03365 (0.2720) -0.40039*** (0.07211) -0.18186*** (0.03887) 0.48291*** (0.03505) 2.14868*** (0.06123) 3.81258*** (0.04877) -0.01685 (0.10309) 0.15290*** (0.02521) 0.04556** (0.01679) 0.02118*** (0.00170) 1.45888*** (0.28889) 0.07330 (0.05544) 0.06636** (0.03098) -0.02190 (0.03417) -0.14740** (0.06917) -0.42812*** (0.07204) -0.19303*** (0.03888) 0.47730*** (0.03496) 2.14332*** (0.06140) 3.78290*** (0.04892) -0.09849 (0.11757) -0.12102* (0.06236) -0.05249 (0.03667) 0.10159** (0.03456) 0.13497** (0.06011) 0.39338** (0.12077) r-square adj. r-square number of obs.: 0.7017 0.7004 3,293 0.7300 0.7288 3,285 0.7327 0.7311 3,285 156 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 table 5: refined split model regression the model explains new security yield off-treasury measured with an exact maturity match constant maturity treasury rate. cmqsyld is the difference between moody’s aaa corporate bond index and the matched-maturity cm rate; lngross is the natural log of the issue size (in millions); term is the term to maturity of the issue (in years); dmad1wlg is the mean absolute deviation in the spread between moody’s aaa index yield (weekly) and an index of long-term (10+ years) government bonds (weekly) for the 26 weeks prior to the issue date; sfdum, crfdum, secdum, and euro are 0/1 dummy variables for the existence of a sinking fund, a known call or refund date, a secured issue, and a euro issue. aaacl through bcl are dummy variables for the general ratings class for the issue’s moody’s rating, with a (1,280 obs.) being the omitted class. mdifn3 through mdifp3 are dummy variables for the magnitude and sign of the difference between the issue’s moody’s rating and its standard & poors rating, with 0 (no difference) being the omitted class. n1aa through p1b are interaction terms for the ratings class and the magnitude and sign of the split, for those categories which had more than 30 observations across the full sample period. the overall model is significant at the 0.01 level. *** indicates individual coefficient significance at the 0.01 level; ** indicates significance at the 0.05 level; * indicates significance at the 0.10 level. regressor model 3 regressor model 3 intercept cmqsyld lngross term dmad1wlg sfdum crfdum secdum euro aaacl aacl baacl bacl bcl mdifn3 mdifn2 mdifn1 mdifp1 mdifp2 mdifp3 -0.01968 (0.10388) 0.15494*** (0.02512) 0.04349** (0.01674) 0.02167*** (0.00170) 1.42844*** (0.28798) 0.06345 (0.05527) 0.06113** (0.03088) -0.01057 (0.03414) -0.14894** (0.06892) -0.40939*** (0.07237) -0.21150*** (0.05043) 0.44811*** (0.04731) 2.24283*** (0.10176) 3.84057*** (0.06259) -0.13238 (0.11740) -0.03094 (0.06765) 0.02295 (0.05357) 0.02279 (0.05409) 0.10451 (0.07256) 0.36138** (0.12038) n1aa p1aa n1baa p1baa p2baa n2ba n1ba p1ba p2ba n1b p1b 0.03493 (0.09896) 0.08257 (0.09362) -0.04814 (0.08976) 0.19301** (0.08787) 0.05189 (0.15058) -0.73332*** (0.17900) -0.65103*** (0.17081) -0.14259 (0.17135) -0.10888 (0.19505) -0.51393*** (0.13227) 0.11469 (0.10225) r-square: 0.7367: adj. r-square: 0.7342: number of obs.: 3,285 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 157 the main conclusion implied by the model 3 results is that additional information provided by the general rating classes in conjunction with the magnitude and direction of a rating split adds to the overall explanatory power of the model, given an adequate number of issues for each interaction tested. in general, the results of model 3 suggest that the relationship between splits, new issue yield spreads and the more highly rating agency may depend on the direction and size of the split as well as the broad rating class assigned to the security. separately, the impact of splits appears to be smaller for lower credit rating classes, consistent with the idea that default risk is not priced in equal intervals across rating classes. 5. conclusion this study presents evidence confirming new findings: (1) the ability of credit ratings to explain differing yield spreads on new issues; (2) the higher yield premiums are paid by issues with split ratings; and (3) the need to recognize the interactions between ratings and split magnitude and direction in any model of new issue yields off-treasury. several of the results also suggest that the relationship between spreads and split ratings may not be symmetric across issues rated higher by standard and poor’s and those rated higher by moody’s. the data set developed for this study is more comprehensive in the number of issues and issue dates than data used in prior research on these questions. a natural extension of this work would involve reducing the potential firm, industry and time-series effects that remain uncorrected in the data. author statement: timothy b. michael is a professor in the school of business, university of houston, clear lake. he may be reached at michael@uhcl.edu. he wishes to record his appreciation to the journal editors for the review of this article and also their editing of the article to incorporate the changes. the responsibility for remaining errors rests with the author. references billingsley, randall s., robert e. lamy, m. wayne marr, and g. rodney thompson. (1985). split ratings and bond reoffering yields. financial management 14(2): 59 65. brigham, eugene f. and louis c. gapenski (1994). financial management: theory and practice. fort worth, texas, dryden press. ederington, louis h., jess b. yawitz and brian e. roberts (1987). the informational content of bond ratings. journal of financial research 10(3): 211 226. freund, rudolf j. and ramon c. littell (1991). sas system for regression, second edition. cary, north carolina, sas institute, inc. greene, william h. (1993). econometric analysis, second edition. englewood cliffs, new jersey, prentice-hall. 158 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 141-160 guedes, jose and tim opler (1995). the determinants of the maturity of corporate debt issues. working paper: ohio state university. hsueh, l. paul and p.r. chandy (1989). an examination of the yield spread between insured and uninsured issues. journal of financial research 12(3): 235 244. kao, chihwa and chunchi wa (1994). a re-examination of the impact of credit ratings and economic factors on state bond yields. review of quantitative finance and accounting 4: 59 78. kaplan, robert s. and gabriel urwitz (1979). statistical models of bond ratings: a methodological inquiry. journal of business 52(2): 231 261. kennedy, peter (1992). a guide to econometrics, third edition. cambridge, massachusetts, mit press. kidwell, david s., m. wayne marr and g. rodney thompson (1984). sec rule 415: the ultimate competitive bid. journal of financial and quantitative analysis 19(2): 183 195. kidwell, david s., m. wayne marr, and joseph p. ogden (1989). the effect of a sinking fund on the reoffering yields of new public utility bonds. journal of financial research 12(1), 1 13. liu, pu and william t. moore (1987). the impact of split bond ratings on risk premia. financial review 22(1): 71 85. liu, pu and anjan v. thakor (1984). interest yields, credit ratings, and economic characteristics of state bonds: an empirical analysis. journal of money, credit, and banking 16(3): 344 350. perry, larry g., pu liu and dorla a. evans (1988). modified bond ratings: further evidence on the effect of split ratings on corporate bond yields. journal of business finance & accounting 15(2): 231 241. pindyck, robert s. and daniel l. rubinfeld (1981). econometric models and econometric forecasts, second edition. u.s.a.: mcgraw-hill. reiter, sara a. and david a. ziebart (1991). bond yields, ratings, and financial information: evidence from public utility issues. the financial review 26(1): 45 73. sas institute, inc. (1989). sas/stat user’s guide, version 6, volume 2. cary, north carolina, sas institute, inc. sorensen, eric h. (1979). the impact of underwriting method and bidder competition upon corporate bond interest cost. the journal of finance 34(4): 863 870. sorensen, eric h. (1982). on the seasoning process of new bonds: some are more seasoned than others. journal of financial and quantitative analysis 17(2): 195 208. zivney, terry l., william j. bertin and khalil m. torabzadeh (1993). a reexamination of the investment performance of junk bonds. quarterly journal of business and economics 32(2), 78 97. t h e in tern a tio n a l jo u rn a l o f b a n kin g a n d f in a n ce, 2008/09 v ol. 6. n um ber 1: 2008: 141-160 159 appendix: moody’s ratings vs. standard & poor’s ratings by year of issue securities are listed under the column corresponding to the rating given by moody’s. for example, the “1” under aaa, column p2 is an issue rated one subclass category higher by moody’s (aaa) than by standard and poors (aa+) panel a: aaa, aa, and a ratings year aaa p1 p2 p3 n3 n2 n1 aa p1 p2 p3 n3 n2 n1 a p1 p2 p3 1983 2 2 1 1 6 15 4 2 8 16 12 3 1 1984 5 10 1 5 12 12 12 4 1985 8 3 1 1 9 24 11 1 2 6 22 49 19 8 3 1986 12 4 3 1 18 55 19 2 3 5 11 48 78 30 17 2 1987 6 1 3 2 9 27 12 1 5 9 21 39 21 14 3 1988 10 4 5 17 6 2 16 14 44 20 7 1989 13 1 2 3 27 6 3 7 13 24 24 2 1 1990 2 1 22 6 1 4 17 41 18 1 1 1991 16 13 32 12 1 4 40 82 34 4 1 1992 23 2 3 22 48 25 7 6 6 11 55 81 38 9 1 1993 10 1 14 29 31 1 14 44 60 30 6 2 column totals 102 17 8 0 0 11 105 306 132 13 11 23 89 294 526 258 75 15 class totals 127 578 1280 19published by epublications@bond, 2008 16 0 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 41 -1 60 panel b: baa, ba, and b ratings year n3 n2 n1 baa p1 p2 p3 n3 n2 n1 ba p1 p2 p3 n3 n2 n1 b p1 p2 p3 annual total 1983 1 1 10 18 11 2 1 1 3 1 2 1 127 1984 2 10 10 7 3 3 1 3 2 1 9 2 1 1 116 1985 1 9 21 8 3 1 6 4 4 1 4 7 8 6 1 251 1986 3 6 25 19 34 5 2 3 4 3 4 4 3 1 2 2 14 24 5 1 472 1987 2 12 34 19 1 4 6 2 1 5 5 4 14 16 2 2 302 1988 1 1 7 25 13 3 3 2 2 1 2 3 3 1 2 4 16 22 8 1 265 1989 3 4 23 5 4 3 3 3 2 2 1 1 2 14 12 1 1 210 1990 7 28 15 3 1 1 169 1991 8 43 14 2 1 2 5 2 1 2 1 3 323 1992 2 6 30 93 26 6 3 15 20 7 4 16 46 15 15 1 627 1993 1 23 66 10 3 1 1 7 5 9 3 4 1 2 12 21 13 1 425 column totals 7 23 145 380 162 35 4 6 31 33 50 33 24 5 7 17 47 142 117 25 7 class totals 752 182 362 3,287 international journal of banking and finance, vol. 6, iss. 1 [2008], art. 8 international journal of banking and finance 9-1-2008 an empirical investigation of new bond issue yield spreads, default risk and split ratings timothy s. michael recommended citation building approvals as a leading indicator of property sector investment 44 the international journal of banking and finance, volume 9 (number 2) 2012: pages 44-58 building approvals as a leading indicator of property sector investment harry m karamujic the university of melbourne, australia ____________________________________________________________ abstract overall, building approvals for new houses (banhs) are viewed by most economic analysts/commentators as a leading indicator of property investment due to the importance of this sector to the whole economy and employment. this study seeks shed some additional light on modelling this seasonal behaviour of banhs by: (i) establishing the presence of seasonality in victorian banhs; (ii) ascertaining it as to whether is deterministic or stochastic; (iii) estimating out-of-sample forecasting capabilities of the modelling specification; and (iv) speculating on possible interpretation of results. the study utilises a structural time series model of harvey. factors corresponding to june, april, december and november are found to be significant at five per cent level. the observed seasonality could be attributed to both the summer holidays and the end of financial year seasonal effects. irrespective of partially incomplete nature of this research, the findings should be appealing to, among others, researchers, all levels of government, construction industry and banking industry. keywords: property investment, structural time series modelling, building approvals for new houses, summer holiday seasonal effect, end of financial year seasonal effect jel classification: r30, r38 _____________________________________________ 1. introduction to property sector this paper examines the impact of seasonal influences on australian housing approvals, represented by the state of victoria banhs. all construction works in most countries have to be approved by the relevant government authority before building commences. in particular, banh statistics are used to inform as to how many residential buildings are in the pipeline, that is, are expected to be constructed in the near future. in australia, from july 1990, banhs include all approved new residential building work valued at a$10,000 or more. because banhs provide timely estimates of future residential building work, banhs statistics are viewed by many economic analysts/commentators 45 as a leading indicator of property investment. due to the importance of property sector to the whole economy, the general level of economic activity and employment, the data are relevance for the residential property sector as well as the property sector as whole. the demand for new housing fluctuates according to a range of determinants such as costs of building materials, finance costs (such as fees and interest rates) home buyer incentive schemes and employment opportunities. while non-residential building approvals can be viewed as an indicator of investment in a region, the local level residential building approvals are commonly viewed as an indicator of the availability of financial resources and commitment to live in an area. in the years before the global financial crisis (gfc), australian house prices rose faster than house prices in the united states of america (usa) and the united kingdom (uk), implying that the then house price is a bubble.1 house price bubble can be defined as situation when the behaviour of house prices satisfies at least one of the following four conditions: (i) house prices are significantly above its long-term average; (ii) house prices are significantly above comparable property prices in other economies; (iii) house price to house rent ratio is significantly above its long-term average, and (iv) house price to household income ratio is significantly above its long-term average. during the gfc, compared to other developed countries, australian economy proved much more resilient. according to the australian bureau of statistics (abs) (2010), australian house prices fell from march 2008 to march 2009, by 5 per cent.2 nevertheless, already in the year to the last quarter of 2009, house prices increased by 15 per cent. the relentless house prices rises continued in the following quarter (the year to the first quarter of 2010), peaking at 20 per cent. the following yearly, quarter to quarter, house price increases retracted to much more moderate levels, e.g. from june 2010 to june 2009, from september 2010 to september 2009 and from december 2010 to december 2009, house prices increased by 16 per cent, 11 per cent and 6 per cent, respectively. rising house prices, together with persistence of stabile financial and banking systems and prevalence of moderate interest rates, have maintained strong mortgage activity, encouraged overall consumer spending and as such supported australian economic performance. the impact of the gfc on financial markets first became apparent in 2007. however, it was not until september 2008 that the reserve 1 in the usa house prices peaked just before the gfc, in february 2006. by late 2006 house prices started to fall. some argue that they are still to hit the bottom. 2 measured by the weighted average of the eight state / territories capitals. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 3 http://www.ausstats.abs.gov.au/ausstats/meisubs.nsf/0/9b864fed34eaa2a1ca2578290011674c/$file/64160_dec%202010.pdf http://www.ausstats.abs.gov.au/ausstats/meisubs.nsf/0/9b864fed34eaa2a1ca2578290011674c/$file/64160_dec%202010.pdf http://www.ausstats.abs.gov.au/ausstats/meisubs.nsf/0/9b864fed34eaa2a1ca2578290011674c/$file/64160_dec%202010.pdf http://www.ausstats.abs.gov.au/ausstats/meisubs.nsf/0/9b864fed34eaa2a1ca2578290011674c/$file/64160_dec%202010.pdf 46 bank of australia (rba) responded with the first in a series of decreases in official interest rate (the cash rate). from september 2008 to april 2009 the rba reduced the cash rate five times.3 subsequently, with the improvement in the economic condition the rba started increasing the cash rate. in fact, a high level of concern due to the emergence of the gfc existed in the australian economy only in early 2009. this was short-lived owing to the effective and efficient fiscal and monetary policy measures undertaken by the australian government. by the end of 2009, the economic momentum began to lift. speaking at an australian school of business (australian school of business 2009) alumni event, australia’s treasury secretary at the time, dr ken henry, assessed the australian government's response to the gfc as follows: “the rudd government’s immediate announcement of the a$10.1 billion fiscal stimulation package, was timely, targeted and temporary”. the aim of the initial stage of the fiscal strategy was to quickly increase household spending. this was achieved by making two sets of payments directly to households. the following stages of the fiscal strategy were characterised with investment in infrastructure and skill development, aiming to ensure ongoing fiscal stimulus once the initial boost from the payments directly to households abates. the monetary policy response was also deemed a big success. in october 2009 the australian government announced that it would guarantee deposits and wholesale funding of australian banks, building societies and credit unions. this was complemented with the monetary policy settings of the rba, which, from the peak interest rate of 7.25 per cent in march 2008, eased cash rate to a low of 3.00 per cent in april 2009. according to the abs (2011), the australian economy grew by 1.8 per cent in final three months of 2000 after the government stimulus helped it shrug off the worst of the gfc. the growth of similar nature (1.9 per cent on average, between december quarter 2009 and march quarter 2011) continued in the following quarters. the awareness of the sizeable impact of the property sector on the soundness of financial institutions and the level of economic activity is not a new observation. it is commonly accepted that the boom-bust nature of property prices plays an important role 3the cash rate can be defined as interest rate paid by banks in the overnight money market in australia and new zealand. in other parts of the world other terms are used to identify the same interest rate, such as overnight libor (the london interbank offered rate) and overnight euribor (euro interbank offered rate). http://en.wikipedia.org/wiki/interest_rate http://en.wikipedia.org/w/index.php?title=overnight_money_market&action=edit&redlink=1 http://en.wikipedia.org/wiki/australia http://en.wikipedia.org/wiki/new_zealand 47 in explaining business cycles.4 typically, reducing housing prices tend to impose additional pressure on the banking sector. this happens not only because of increases in bad debts for mortgage loans, but also because of deterioration in the balance sheets of corporate borrowers that rely on property as collateral. not surprisingly, fluctuation of housing prices and the extent to which they interact with the financial sector and the whole economy are very much of interest, among others, to the government, the reserve bank and other financial regulators. the conventional literature recognises for a long time that investment in housing and consumer durables lead non-residential business fixed investment over the business cycle (e.g. burns and mitchell 1946). among others, this is corroborated by fisher (2006), who observed that in seven of the ten post-war recessions in the usa, household investment achieved its peak and trough before business investment. ball and wood (1999) conducted comparative structural time series analysis of housing investment in advanced world economies. they looked at the impact of housing investment on the economy and concluded that housing investment fluctuations after 1960s become a destabilizing factor. this finding highlighted the significance of this category of investment and further accentuated the relevance of studies focusing on better understanding housing investment volatility. as previously stipulated, the focus of this study is to examine the impact of seasonal influences on australian housing approvals (represented by victorian banhs).5 victoria is geographically the second smallest state in australia. it is also the second most populous state in australia. victoria has been selected as a test case because of its geographical homogeneity and economic relevance. victoria is australia's most urbanized state: nearly 90 per cent of residents living in cities and towns, it is the most densely populated state (22 people on square kilometre), and has a highly centralised population, with almost 75 per cent of victorians living in the state capital and largest city, melbourne. at the same time, the state of victoria is the second largest economy in australia, only after new south wales, accounting for almost a quarter of the nation's gross domestic product (gdp).6 all other australian states are either geographically dispersed (cover a wide geographical region across different time and climatic zones) or 4 by strengthening the upswing and amplify the downswing. 5australia has six states (new south wales, queensland, south australia, tasmania, victoria, and western australia), and two territories (the northern territory and the australian capital territory). 6according to the abs (2011), in 2008/2009 victoria contributed 22.6 per cent of the australian gdp. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 3 http://en.wikipedia.org/wiki/states_and_territories_of_australia http://en.wikipedia.org/wiki/australia http://en.wikipedia.org/wiki/melbourne http://en.wikipedia.org/wiki/gross_domestic_product http://en.wikipedia.org/wiki/federated_state http://en.wikipedia.org/wiki/new_south_wales http://en.wikipedia.org/wiki/queensland http://en.wikipedia.org/wiki/south_australia http://en.wikipedia.org/wiki/tasmania http://en.wikipedia.org/wiki/victoria_(australia) http://en.wikipedia.org/wiki/western_australia http://en.wikipedia.org/wiki/northern_territory http://en.wikipedia.org/wiki/australian_capital_territory 48 economically insignificant.7 this is very relevant to the focus of this research, as seasonal impacts would be much harder to isolate if the data include different time and climatic zones. furthermore, it is specifically important to note that the focus of the study is not on modelling the behaviour of time series in terms of explanatory variables (the conventional modelling approach). the conventional modelling approach assumes that the behaviour of the trend and seasonality can be effectively captured by a conventional regression equation that assumes deterministic trend and seasonality. instead, the aim is to use a univariate structural time series modelling approach (allows modelling both stochastic and deterministic trend and seasonality) and show that conventional assumptions of deterministic trend and seasonality are not always applicable. specifically, the study seeks to cast some additional light on banhs by: (i) establishing the presence, or otherwise, of seasonality in victorian banhs, (ii) if present, ascertaining is it deterministic or stochastic, (iii) determining out of sample forecasting capabilities of the considered modelling specifications and (iv) speculating on possible interpretation of results. to do so the study utilises a basic structural time series model of harwey (1989). compared to the conventional procedure, harvey’s (1989) structural time series model involves an explicit modelling of seasonality as an unobserved component. empirical evidence of seasonal variations in property related variables is relatively limited. studies come from a range of different perspectives and employ a number of modelling techniques. harris (1989) provided empirical evidence of strong second and third quarter seasonality in the usa house prices. ma and goebel (1991) established the presence of january seasonal effect for securitised mortgage markets, while friday and peterson (1997) and colwell and park (1990) established presence of a january seasonal effect in returns of real estate investment trusts (reits) in the usa. rossini (2000) examined seasonal effects in the housing markets of adelaide, south australia, and, with respect to the volume of detached dwelling transactions, determined presence of statistically significant ‘summer’ and ‘autumn’ seasonal effects. similarly, costello (2001) examined the impact of seasonal influences on housing market activity in 7 as shown in the abs (2011), tasmania is australia’s smallest and the most compact state, nevertheless it has the lowest gross state product (gsp) of all states (in 2008/2009 contributed only 1.8 per cent of the australian gdp). gsp is a measure of the economic output of a state (i.e., of a sub-national entity). it is the sum of all value added by industries within the state and serves as a counterpart to the gdp. http://en.wikipedia.org/wiki/state_(administrative_division) http://en.wikipedia.org/wiki/subnational_entity http://en.wikipedia.org/wiki/value_added 49 perth, western australia, and found that the volume of transactions and hence demand is greatest during the first quarter of a year and lowest during the last quarter. karamujic (2009) confirmed presence of both cyclicality and seasonality in australian residential mortgage interest rates in the two major australian banks (national australia bank (nab) and commonwealth australia bank (cba)).8 most studies, to varying degrees, point to the existence of seasonality. a study of the literature on housing variables found no empirical research focusing on seasonal fluctuations in banhs. it also uncovered that most studies are focusing on house prices. the paper is organised in 4 sections. the following section (section 2) of this paper outlines the methodology used. section 3 provides data specification, presents modelling test results and interprets the modelling results. finally, in section 4, the paper concludes. 2. methodology a structural time series framework approach used in this paper is in line with that promulgated by harvey (1989). such models can be interpreted as regressions on functions of time in which the parameters are time-varying. this makes the approach a natural vehicle for handling changing seasonality of a complex form. once a suitable model has been fitted, the seasonal component can be extracted by a smoothing algorithm. following harvey (1989) and harvey, koopman and riani (1997), the basic structural time series model is formulated in terms of a trend, seasonal and irregular components. all are assumed to be stochastic and driven by serially independent gaussian disturbances that are mutually independent. if there are s seasons in the year, the model is ( )2t ,0~ , εσεεγµ nidy tttt ++= , (1) where the trend, seasonal and irregular are denoted by tµ , tγ and tε , respectively. the trend is specified as follows: ( )2t11 ,0~ , ησηηβµµ nidtttt ++= −− , (2) 8 according to the bankersalmanac.com‘s (2011) selection of the top banks worldwide (ranked on the total assets of a bank in usa$), calculated from yearend figures gained from submitted balance sheets, nab and cba were ranked thirty-fourth and forty-seventh respectively. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 3 50 ( )2t1 ,0~ , ςσςςββ nidtt += − where tµ is the level and tβ is the slope. the disturbances tη and tς are assumed to be mutually independent. setting 02 =ησ gives a trend that is relatively smooth. the seasonal component is generally constructed in terms of stochastic trigonometric functions at the 2/s seasonal frequencies, although deterministic and dummy-variable formulations are also possible. the fundamental point is that, although the seasonal component is non-stationary, it has the property that the expected value of the sum over the previous s time periods is zero. this ensures that the seasonal effects are not confounded with the trend. it also means that the forecasts of the seasonal component will sum to zero over any one-year period. the statistical treatment of the model is based on the state-space form, with 1+s elements in the state vector. estimation, forecasting and signal extraction are carried out by means of the kalman filter and associated algorithms. the trigonometric form of stochastic seasonality used in models of the form (1) where s seasons in the year is [ ] tt s j tjt ,,1 , 2/ 1 , == ∑ = γγ , (3) and each tj ,γ is generated by       +            − =      − − * , , * 1, 1, * , , cos sin sin cos tj tj tj tj j j j j tj tj ω ω γ γ λ λ λ λ γ γ , (4) where sjj πλ 2= is frequency, in radians, for [ ]2,,1 sj = and tω *tω are two mutually uncorrelated white-noise disturbances with zero means and common variance 2ωσ . the basic structural model consisting of the stochastic trend in (2) with trigonometric seasonality is easily put in state-space form by defining the ( ) 11 ×+s state vector ( ),,,,,, *22*11 ttttttt γγγγβµα = . the measurement equation is then ( ) tttty εα += ',0,1 z , (5) 51 where ( ),0,1,0,1' =tz . if the kalman filter is initiated with a diffuse prior, as shown by de jong (1991), an estimator of the state with a proper prior is effectively constructed from the first 1+s observations. on the other hand, if we choose to fix the seasonal pattern in (1), thus specifying a deterministic seasonal component, tγ , may be modelled as: 1 ∑ = = s j jtjt zγγ (6) where s is the number of seasons and the dummy variable jtz is one in season j and zero otherwise. in order not to confound trend with seasonality, the coefficients, jγ , ,,,1 sj = are constrained to sum to zero. the seasonal pattern may be allowed to change over time by letting the coefficients evolve as random walks as in harrison and stevens (1976). if tγ denotes the effect of season j at time t, then ( ) .,,1 ,0,nid~ , 21, sjtjttjjt =+= − ωσωωγγ (7) although all s seasonal components are continually evolving, only one affects the observations at any particular point in time, that is jtt γγ = when season j is prevailing at time t. the requirement that the seasonal components evolve in such a way that they always sum to zero is enforced by the restriction that the disturbances sum to zero at each point in time. this restriction is implemented by the correlation structure in ( ) ( )iiivar ′−= −12 st ωσω (8) where ( )′= sttt ωωω ,,1  , coupled with initial conditions requiring that the seasonal sum to zero at 0=t . it can be seen from the equation above that ( ) .0=′ tωivar in the basic structural model, tµ in (1) is the local linear trend of (2), the irregular component, tε , is assumed to be random, and the disturbances in all three components are taken to be mutually uncorrelated. the signal-noise ratio associated with the seasonal, that is ,22 εωω σσ=q determines how rapidly the seasonal changes relative to the irregular. an example of how the basic structural model successfully captures changing seasonality can be found in the study of alcoholic beverages by lenten and moosa (1999). international journal of banking and finance, vol. 9, iss. 2 [2012], art. 3 52 3. modelling results and their interpretation the data are sourced from the abs. for consistency, the sample for each variable is standardised to start with the first available july observation and end with the latest available june observation. the structural time series model represented by (1) is applied to seasonally unadjusted monthly banhs data for victoria, between 2000:06 and 2009:05. the results are presented below. table 1: estimated coefficients of final state vector state variable /test statistic model with stochastic trend and deterministic seasonality tµ 2533.30 (17.28) tβ 1.17 (0.05) 1γ 288.62 (3.53) 2γ 154.24 (1.89) 3γ 266.56 (3.28) 4γ 31.38 (0.39) 5γ 94.81 (1.17) 6γ 100.83 (1.25) 7γ -390.32 (-4.82) 8γ -657.19 (-8.12) 9γ -50.03 (-0.62) 10γ 20.75 (0.26) 11γ -78.88 (-0.97) tε 274.39 2 sr 0.31 2 dr 0.59 dw 2.09 q 15.08 n 6.37 h 0.41 aic 11.51 bic 11.88 53 as shown in the previous section, all three considered modelling specification include trend (composed of the level ( tµ ) and slope ( tβ )), seasonal and irregular components. two modelling specifications include a stochastic trend, while the third considered modelling specification incorporates deterministic trend. on the other hand, one modelling specification incorporates stochastic seasonal component, while other two modelling specifications incorporate deterministic (fixed) seasonal components. the final state vector is obtained when (1) is estimated by utilising all the information contained in the whole sample for each model. all three modelling specifications were analysed and the modelling results of estimating the univariate time series model for the modelling specification with the best results of goodness of fit measures and diagnostics test statistics are presented in table 1. more precisely, table 1 reports the estimated components of the state vector ( tµ , tβ and 111−γ ), their t-statistics, goodness of fit measures and diagnostics test statistics. with respect to the goodness of fit, the modelling results are well defined. overall, the diagnostic tests are also predominately passed. the only exception is the test for serial correlation (q) which is slightly above the statistically acceptable level. figure 1 provides a visual interpretation of the seasonal elements for the selected modelling specification. the seasonal components evidenced in the figure show a constant repetitive pattern over the sample period, providing a visual evidence of the deterministic nature of the seasonal component (fixed seasonal components) in the number of new dwellings approved in victoria. figure 2 shows this even more clearly with individual monthly seasonals represented by horizontal lines, implying an unchanging seasonal effect across the whole sample period. in summary, the analysis points out that the behaviour of banhs exhibits stochastic trend and deterministic seasonality. as a result, any model based on assumptions of deterministic trend and seasonality is bound to be mis-specified. out of the eleven seasonal factors relating to the model two, presented in table 1, factors corresponding to june ( 1γ ), april ( 3γ ), december ( 7γ ) and november ( 8γ ) are found to be significant at five per cent level.9 9 shown as variables with t statistics values above 1.96. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 3 54 -800 -600 -400 -200 0 200 400 figure 1: seasonal component -800 -600 -400 -200 0 200 400 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 june july august september october november december january february march april may figure 2: individual seasonals the factors corresponding to december ( 7γ ) and november ( 8γ ) exhibit the seasonrelated reduction in the number of banhs, while the factors corresponding to june ( 1γ ) and april ( 3γ ) demonstrate season-related increases. a possible explanation for the observed statistically significant reduction in banhs during december ( 7γ ) and november ( 8γ ) is the reduction of the level of activity caused by approaching to the ‘summer holidays’ season. it is well known that there is an unavoidable time leg between getting the building approval and starting of the construction of the dwelling. furthermore, it is also commonly recognised that construction industry in australia almost completely shuts down from late december to the end of january. anecdotally, the most commonly quoted cause of this is the ‘summer 55 holidays’ season. the ‘summer holidays’ season typically covers the period from the second half of december to the end of january. cause of this seasonality predominantly lies in human-induced factors. for a period of six weeks from the second half of december, summer school holidays and several public/religious holidays take place in australia. not surprisingly, it is also the time when most workers take annual leave. consequently, this is a period when the construction and real estate industries almost shut down. another possible explanation for the ‘summer’ seasonal effect is the slowing of activity in the housing construction sector caused by the hot australian summer. it should be noted that australia (being in the bottom part of the southern hemisphere), has completely opposite climatic seasons to countries in upper part of the northern hemisphere, e.g. when it is summer in european (june to august) it is winter in australia. on the other hand season-related increases during june ( 1γ ) and april ( 3γ ) may be explained by a spike in the level of activity during ‘the end of financial year’ season. ‘the end of financial year’ season typically starts by the end of april or the beginning of may, and finishes at the end of the first week in july. this is another interesting example of a human-induced seasonal factor. this time the seasonal effect is related to the seasonal pattern of government tax collection. this season is generally characterised by an excess in spending in the month or two before the end of the financial year. with respect to banhs, this season is particularly characterised by an increase in demand for housing approvals and subsequent home finance that allows prepayment of interest and, as such, enables the prepaid interest to be claimed as expenditure in the current year. 10 the primary cause of this seasonal effect is that the australian taxation office (ato) accepts prepayment of the interest payable on investment properties.11 in order to test the robustness of the models specified as well as to determine forecasting power of the three models considered, out-of-sample forecasting was undertaken. firstly, the three models are estimated over the period 2000:6 2005:5. 10according to karamujic (2009), with respect to the interest rate repayment structure, all home loans can be viewed as having either interst only (io) or principal and interest (p&i) repayment structures. a typical example of these kind of home loans are fixed rate home loans (frhls). from the repayment point of view, a frhl is offered as either io or p&i. most of contemporary lenders offer io frhls (friohls) as one, two, three, four and five years fixed rate io repayment products, where the interest is payable monthly in arrears. they also can come as one, two, three, four and five year fixed rate io repayments, payable in advance, however these are generally only available for investment purposes. it is important to note that the fixed rate term must be less than, or equal to, the io period. at the end of the io period, the loan typically automatically converts to a p&i standard variable rate home loan (svrhl) for the remainder of the loan term. 11 for more see income tax assessment act (1997). international journal of banking and finance, vol. 9, iss. 2 [2012], art. 3 56 1500 2000 2500 3000 3500 4000 2000 2001 2002 2003 2004 2005 2006 2007 2008 original data model 3 model 1 and 2 figure 3: out-of-sample forecasting these estimates are then used to forecast the behaviour of banhs for the period 2005:6 2009:5. even superficial observation of forecasts presented in figure 3, shows that in all cases, the variability in the actual data was difficult to predict with the exception of the specifications including the fixed seasonals. this is corroborated again in table 2, which reports on the following two statistics that measure the forecasting power: the sum of absolute forecasting errors and the sum of squared forecasting errors. table 2: sum of absolute/squared errors of the forecasting values model sum of absolute errors 9,987 sum of squared errors 5,063,177 4. conclusion this study uses harvey’s (1989) univariate structural time series mode to examine the impact of seasonal influences on the australian housing time series, with the generic objective of enhancing the practice of modelling housing variables. specifically, the paper seeks to cast some additional light on the seasonal behaviour of banhs by: (i) establishing the presence, or otherwise, of seasonality in victorian banhs, (ii) if present, ascertaining is it deterministic or stochastic, (iii) determining out of sample forecasting capability of the considered model and (iv) speculating on possible interpretation of the results. the goodness of fit measures and the diagnostic test statistics indicate that the model, which is comprised out of stochastic trend and deterministic seasonality, is 57 unambiguously superior to the other two specifications. the examination of the out-ofsample forecasting power of the selected model clearly shows that the seasonality apparent in the actual data is well picked up by specifications entailing deterministic seasonal factor, corroborating the earlier finding that the seasonal pattern in the number of dwelling units approved in victoria is deterministic and not stochastic. a possible explanation for the observed statistically significant reduction in banhs during december ( 7γ ) and november ( 8γ ) is the reduction of the level of activity caused by approaching to the ‘summer holidays’ season, while the season-related increases during june ( 1γ ) and april ( 3γ ) may be explained by a spike in the level of activity during the ‘end of financial year’ season and preparation for a surge in contraction activity during the ‘spring’ season. to corroborate the modelling results and explanations provided, the scope of the analysis would need to be extended. it is reasonable to expect that these substantial season-related changes in monthly banhs are, to a large extent, correlated with home loan drawdowns and housing starts. thus, extending the research to include home loan drawdowns and housing starts could be a rewarding area for further research. irrespective of the incomplete nature of the research, due to the commonly accepted perception of banh as a leading indicator of property sector investment, the findings of this research should be appealing to, among others, researchers, all levels of government, construction industry and banking industry. author information: dr harry m. karamujic is a staff member of the property and finance faculty and his postal address is room 311, architecture building, the university of melbourne, victoria 3010, australia. his e-mail address is harryk@unimelb.edu.au. references abs (2009) http://www.abs.gov.au/ausstats/abs@.nsf/exnote/8731.0, september 2011. abs (2010) 6416.0 house price indexes: eight capital cities, dec 2010, http://www.abs.gov.au/ausstats/abs@.nsf/mf/6416.0, viewed march 2011. abs (2011) 1350.0 australian economic indicators, apr 2011, http://www.abs.gov.au/ausstats/abs@.nsf/mf/6416.0, viewed march 2011. abs (2011) 5206.0 australian national accounts: national income, expenditure and product, march 2011, http://www.ausstats.abs.gov.au/ausstats/meisubs.nsf /0/6e76c70f27b705bbca2578a10012b52b/$file/5206001_key_aggregates.xls, viewed may 2011. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 3 http://www.ausstats.abs.gov.au/ausstats/meisubs.nsf 58 australian school of business (2009) rapid response to gfc saved australia’s economy – meet the ceo with ken henry,9http://www.asb.unsw.edu.au/newsevents/ mediaroom/media/2009/november/pages/kenhenrymeettheceo.aspx, may 2011. ball, m. and wood, a. (1999) housing investment: long run international trends and volatility, housing studies, 14(2), 185 – 209. bankersalmanac.com (2011) top banks in the world by bankersalmanac.com, http://www.bankersalmanac.com/addcon/infobank/bank-rankings.aspx may 2011. burns, a. and mitchell, w. (1946) measuring business cycles, new york: national bureau of economic research, studies in business cycles, 2. costello, g.j. (2001) further evidence of seasonal influences in australian housing markets a case study of perth, proceedings of 2001 prres conference, adelaide, south australia, january. colwell, p. and park, h. (1990) seasonality and size effects the case of real-estaterelated investment, journal of real estate finance and economics, 3, 251 – 259. de jong, p. (1991) the diffuse kalman filter, the annals of statistics, 19(2), 1073 – 1083. fisher, j. (2006) why does household investment lead business investment over the business cycle, federal reserve bank of chicago, october. friday, s. h. and peterson, r.d. (1997) january return seasonality in real estate investment trusts: information vs. tax-loss selling effects, the journal of financial research, 20(1), 33 – 51. harris, j. (1989) the effect of real rates of interest on housing prices, journal of real estate finance and economics, 2, 47 60. harrison p.j. and stevens c.f (1976) bayesian forecasting, journal of royal statistical society, 205 – 247. harvey, a. c. (1989) forecasting, structural time series models and the kalman filter, cambridge: cambridge university press. harvey, a.c., koopman, s.j. and riani, m., (1997) the modeling and seasonal adjustment of weekly observations, jbes, 15, 354 68. hilmer, s.c. & tiao, g.s. (1982) an arima-model-based approach to seasonal adjustment, journal of american statistical association, 77, 63 – 70. karamujic, m. h. (2009) the determination of home loan interest rates and an empirical assessment of their cyclicality and seasonality in australia, saarbrucken germany: vdm. kitagawa, g. and gercsh, w. (1984) a smoothness priors-state space modeling of time series with trend ad seasonality, journal of the american statistical society, 79. lenten, l.j.a. and moosa, i.a. (1999) modelling the trend and seasonality in the consumption of alcoholic beverages in the united kingdom, applied economics, 31, 795 – 804. ma, c. k. and goebel, p.r. (1991) on the seasonalities of mortgage-backed security prices, journal of real estate research, 6, 19 – 38. rossini, p. (2000) estimating the seasonal effects of residential property markets – a case study of adelaide, proceedings of 2000 prres conference, sydney, australia, january. abs (2006) census of population and housing. retrieved from http://www.censusdata.abs.gov.au/absnavigation/prenav/locationsearch?readf orm&prenavtabname=location%20search&&&navmapdisplayed=true&textvers ion=false&collection=census&period=2006&producttype=&method=&productl abel=&breadcrumb=l&topic=&, viewed september 2011. australian taxation office (1997) income tax assessment act (1997) http://law.ato.gov.au/atolaw/execute_search_java.htm., viewed september 2011. http://www.asb.unsw.edu.au/newsevents/%20mediaroom/media/2009/november/pages/kenhenrymeettheceo.aspx http://www.asb.unsw.edu.au/newsevents/%20mediaroom/media/2009/november/pages/kenhenrymeettheceo.aspx http://www.bankersalmanac.com/addcon/infobank/bank-rankings.aspx http://www.censusdata.abs.gov.au/absnavigation/prenav/locationsearch?readform&prenavtabname=location%20search&&&navmapdisplayed=true&textversion=false&collection=census&period=2006&producttype=&method=&productlabel=&breadcrumb=l&topic=& http://www.censusdata.abs.gov.au/absnavigation/prenav/locationsearch?readform&prenavtabname=location%20search&&&navmapdisplayed=true&textversion=false&collection=census&period=2006&producttype=&method=&productlabel=&breadcrumb=l&topic=& http://www.censusdata.abs.gov.au/absnavigation/prenav/locationsearch?readform&prenavtabname=location%20search&&&navmapdisplayed=true&textversion=false&collection=census&period=2006&producttype=&method=&productlabel=&breadcrumb=l&topic=& http://www.censusdata.abs.gov.au/absnavigation/prenav/locationsearch?readform&prenavtabname=location%20search&&&navmapdisplayed=true&textversion=false&collection=census&period=2006&producttype=&method=&productlabel=&breadcrumb=l&topic=& http://law.ato.gov.au/atolaw/execute_search_java.htm momentum effect all over the world: 75-93 75 momentum effect all over the world zulfiqar ali imran* school of economic, finance & banking universiti utara malaysia, university of lahore, pakistan woei-chyuan wong rusmawati ismail school of economics, finance & banking, universiti utara malaysia, malaysia *corresponding author: zulfiqaraliimran05@gmail.com ________________________________________________________________ a r t i c l e i n f o _____________________________ article history: received 28 august 2019 revised 26 september 2019 accepted 30 september 2019 published 6 january 2020 ____________________________ keywords: momentum effect, reversals, momentum investment strategies, existence and profitability of momentum effect, efficient market hypothesis. jel code: g11, g12 a b s t r a c t _________________________________ this study is intended to reaffirm the existence and profitability of momentum investment strategies in 40 countries around the world during the period 1996–2018. the contradictory findings of previous research on the existence and profitability of momentum strategies have raised a pertinent question on the validity of efficient market hypothesis. we documented the momentum effect in 90% of our sample countries of which 52.5% exhibited positive momentum effect while 37.5% exhibited negative momentum effect. the findings were robust to two distinct sub-period analyses. the clear rejection of efficient market hypotheses is valuable to momentum traders and stock market regulators. the international journal of banking and finance, vol. 13. number 2, 2017: 75-93 75 1. introduction finance literature carries substantial evidence on the existence and profitability of momentum returns since the seminal work of jegadeesh and titman (1993). many studies have documented instances where stocks kept on outperforming (underperforming) for the next three to 12 months if they had outperformed (underperformed) during several previous months. jegadeesh and titman (1993) 76 the international journal of banking and finance, vol. 14, 2018-2019: 75-93 analyzed the american stock exchange (amex) and new york stock exchange (nyse) using post 1940 stock data. they reported a significant monthly average momentum profit of 1.49% when adopting a zero-cost momentum strategy of buying past winners and selling past losers. similar results have also been documented in prominent cross-country studies such as rouwenhorst (1998) who found momentum returns across all european stock markets. rouwenhorst (1999) noted that 85% and 15% of the sample countries exhibited positive and negative momentum returns, respectively. likewise, griffin, ji and martin (2003) reported that 80% (18%) of their sample countries exhibited positive (negative) momentum returns, whereas 2% of their sample countries exhibited no momentum returns. researchers have been critical of the applicability of efficient market hypothesis (emh) in conventional finance ever since the discovery of the momentum effect in finance literature. the mere existence of the momentum effect in stock returns has invalidated the notion of emh and strengthened the view point of opponents of market efficiency. the presence of momentum effects implied that a stock’s own past prices could be utilized to predict its future prices and could also lead investors to earn abnormal profits. this is contrary to emh which articulates that investors cannot use information based on securities’ own past prices to make an abnormal profit because stock future prices are random in nature and are not affected by previous events (malkiel, 2003; malkiel & fama, 1970). the question arises then as to what extent are our financial markets informationally efficient and whether asset pricing models are reliable. this is because conventional asset pricing models are developed based on the assumption of market rationality, where chances of achieving excess returns are remote. if investors can beat the market through acquiring excess returns then the validity of conventional asset pricing model is highly questionable (chen 2017). despite the fact that many studies have investigated momentum effects in stock markets around the world, there is no consensus among researchers on the existence and profitability of the momentum effect especially in developing stock markets. this contradictory evidence makes momentum profitability highly questionable in the literature and has been the subject of many empirical studies. hameed and yuanto (2002) and chui, titman and wei (2000), for instance, did not find any momentum effect in asian stock markets in their samples. on the other hand, griffin, ji and martin (2004) and chui, titman, and wei (2010) documented the existence and profitability of the momentum effect in some of the asian stock markets. kang, liu and ni (2002) established the presence of positive momentum effect in the china stock market. a more recent study in china by li, qiu and wu (2010) however, found no momentum profit. the presence of the momentum effect proved by griffin, ji and martin (2004) in the turkish stock market was also controversial as it was later denied by ornelas and fernandes (2008). momentum effect all over the world: 75-93 77 one may argue that emerging markets are subject to higher volatility and greater uncertainty which caused the momentum effect, but the momentum effect is not confined to emerging stock markets. there are some developed stock markets such as in australia and japan where the momentum effect is also subject to many contradictions. chui, titman and wei (2010) and hurn and pavlov (2003) for instance detected strong momentum effect in australia which was unsupported in a later study by huynh et al. (2010). similarly, findings by hong, lee and swaminathan (2003) of the insignificant momentum effect in japan contradicted findings by griffin, ji and martin (2004) who perceived significant momentum effects in the japanese stock market. we are unaware of any comprehensive study, after chui, titman and wei (2010) who examined and reaffirmed the profitability of the momentum effect around the world. the ongoing debate on market efficiency (shiller 2003) and contradictory findings on the existence of the momentum effect served as motivation in the current study. therefore the main objective of our study is to revisit the profitability of the momentum effect around the world to confirm whether stock markets are informationally efficient or not. this study also has implications for individual and institutional investors which adopted momentum investment strategies. momentum based strategies are risky and are subject to huge losses. according to barroso and santa-clara (2015), momentum strategies crash up to -91.59% in just two months and -73.42% in three months during the 1932 and 2009 financial crises, respectively. such huge losses cannot be compensated through decades of momentum profits. our study is important for policymakers to understand how far momentum effect prevails in stock markets and whether they should take a more proactive role in regulating the stock markets. the remainder of this paper is organised as follows: section 2 presents the literature review, section 3 explains the data, section 4 describes the methodology, section 5 clarifies the empirical findings and section 6 concludes the study. 2. related literature jegadeesh and titman (1993) discovered short term momentum effect in the u.s. stock markets. they examined the u.s. stock market from 1965 to 1989 and documented momentum investment strategies which generated 1.49% average momentum profits per month. in other words, they found that winner stocks were outperforming loser stocks by 1.49% for the next three to 12 months. however, their study only focused on the u.s. stock markets and disregarded the european and asian stock markets. subsequent studies on momentum returns confirmed the existence of momentum profits outside of the u.s. markets. rouwenhorst (1998) for instance, analysed 12 european stock markets from 1978 to 1985 that included united kingdom, switzerland, sweden, spain, norway, netherlands, italy, germany, 78 the international journal of banking and finance, vol. 14, 2018-2019: 75-93 france, denmark, belgium and austria. he documented the past six months of winner stocks which outperformed the past six months of loser stocks by 1% per month in all the 12 european markets. a later study by rouwenhorst (1999) who studied 1,750 firms in 20 emerging stock markets consisting of argentina, brazil, chile, colombia, greece, indonesia, india, jordan, korea, malaysia, mexico, nigeria, pakistan, philippines, portugal, taiwan, thailand, turkey, venezuela and zimbabwe found that 17 out of the 20 emerging stock markets exhibited momentum effect. although there is ample evidence on the existence and profitability of the momentum effect, there are also numerous studies that contradicted the profitability and existence of the momentum effect. chui, titman and wei (2001) examined eight stock markets in asia which included hong kong, malaysia, indonesia, taiwan, korea, thailand, japan and singapore. they found positive momentum effect in hong kong, malaysia, singapore and thailand but no momentum effect in indonesia, japan, taiwan and korea. this study, however, lacked generalizability as they only considered eight asian countries. the existence of the momentum effect was also opposed by hameed and yuanto (2002) who studied 1,000 firms in six asian stock markets comprising of hong kong, malaysia, singapore, south korea, taiwan and thailand. they applied jegadeesh and titman’s (1993) momentum methodology and concluded that all 16 momentum strategies were consistently insignificant in the six asian stock markets. moreover, hameed and yuanto (2002) argued that the momentum effect was the result of data snooping bias.1 ornelas and fernandes (2008) also did not find evidence of momentum effect in brazil, indonesia, australia, pakistan, poland, romania and turkey. they related their findings of insignificant momentum effects with the improvement in information technology and use of the internet that helped channel information to investors with greater speed which in turn wiped out the momentum effect. if higher information technology and use of the internet wiped out the momentum effect, then all else being equal, developed countries, should have a lower momentum effect as compared to developing countries. this contention was not supported by existing studies which found that developed countries exhibited higher momentum effect than their developing counterparts (rouwenhorst 1999, 1998). although a substantial amount of literature on momentum effect favoured the existence and profitability of momentum strategies in many countries around the world, there are also many studies that contradicted these findings specifically in the context of emerging markets. the existence and profitability of the momentum effect in emerging markets is not unanimous in the literature. moreover, there is no study examining momentum returns on a global scale after chui, titman and wei (2010). thus, it is important to reaffirm the existence of momentum effect around the world. table 1 presents the list of countries, along with researchers who investigated the existence of momentum returns. 1 data snooping refers to statistical inference which researchers decide to perform after looking at related data. momentum effect all over the world: 75-93 79 table 1. countries and momentum returns panel a. list of countries with significant momentum returns country author argentina rouwenhorst (1999); chui, titman and wei (2010); griffin, ji and martin (2003) africa griffin, ji and martin (2003) australia griffin, ji and martin (2003) austria rouwenhorst (1998); chui, titman and wei (2010); griffin, ji and martin (2003) bangladesh chui, titman and wei (2010) belgium rouwenhorst (1998); chui, titman and wei (2010); griffin, ji and martin (2003) brazil rouwenhorst (1999); chui, titman and wei (2010); griffin, ji and martin (2003) nigeria rouwenhorst (1999) jordan rouwenhorst (1999) korea rouwenhorst (1999) malaysia rouwenhorst (1999); griffin, ji and martin (2003) mexico rouwenhorst (1999); griffin, ji and martin (2003) netherlands rouwenhorst (1998); chui, titman and wei (2010); griffin, ji and martin (2003) new zealand griffin, ji and martin (2003) norway rouwenhorst (1998); chui, titman and wei (2010); griffin, ji and martin (2003) pakistan rouwenhorst (1999); chui, titman and wei (2010); griffin, ji and martin (2003) peru griffin, ji and martin (2003) philippines rouwenhorst (1999); chui, titman and wei (2010); griffin, ji and martin (2003) poland chui, titman and wei (2010) portugal rouwenhorst (1999); griffin, ji and martin (2003) singapore griffin, ji and martin (2003) south africa griffin, ji and martin (2003) spain rouwenhorst (1998); chui, titman and wei (2010); griffin, ji and martin (2003) (continued) 80 the international journal of banking and finance, vol. 14, 2018-2019: 75-93 panel a. list of countries with significant momentum returns country author sweden rouwenhorst (1998); chui, titman and wei (2010); griffin, ji and martin (2003) taiwan rouwenhorst (1999); griffin, ji and martin (2003) thailand griffin, ji and martin (2003) turkey rouwenhorst (1999); griffin and martin (2003) united kingdom rouwenhorst (1998); chui, titman and wei (2010); griffin, ji and martin (2003) venezuela rouwenhorst (1999) zimbabwe rouwenhorst (1999) panel b. countries with controversial momentum returns hong kong hameed and yuanto (2002) switzerland rouwenhorst (1998); chui, titman and wei (2010); griffin, ji and martin (2003) malaysia hameed and yuanto (2002) singapore hameed and yuanto (2002) south korea hameed and yuanto (2002); chui, titman and wei (2000) taiwan hameed and yuanto (2002) thailand hameed and yuanto (2002) japan chui, titman and wei (2000); teplova and mikova (2015) indonesia fernandes and ornelas (2008); chui, titman and wei (2000) australia henker, henker and huynh (2010) brazil fernandes and ornelas (2008) pakistan fernandes and ornelas (2008) poland fernandes and ornelas (2008) romania fernandes and ornelas (2008) turkey fernandes and ornelas (2008) 3. data the stock price data used to compute momentum returns was obtained from datastream. the final sample consisted of 40 countries with complete stock momentum effect all over the world: 75-93 81 price data from 1996 to 2018. the initial sample consisted of 40,365 firms from the 40 countries but after the screening process, 9,332 firms were left. during the screening process, firms with incomplete monthly return values from 1996 to 2018 were dropped. table 2 provides the list of countries included in our study along with the total number of firms. table 2. total numbers of firms no. country total number of firms number of firms retained 1 bangladesh 348 97 2 brazil 610 156 3 china 3706 366 4 colombia 99 28 5 india 4860 1791 6 indonesia 629 171 7 kenya 69 39 8 malaysia 965 303 9 morocco 75 17 10 pakistan 365 158 11 philippines 312 159 12 south africa 399 112 13 sri lanka 301 163 14 thailand 1242 413 15 turkey 394 142 16 australia 2172 380 17 belgium 154 46 18 canada 3684 760 19 chile 211 110 20 denmark 162 86 21 finland 172 52 22 france 860 236 23 germany 876 231 24 greece 194 74 25 hong kong 3336 453 26 israel 414 190 27 italy 462 72 28 japan 1174 481 29 korea 2378 473 (continued) 82 the international journal of banking and finance, vol. 14, 2018-2019: 75-93 no. country total number of firms number of firms retained 30 netherlands 130 60 31 new zealand 150 38 32 norway 299 42 33 poland 820 21 34 singapore 522 113 35 spain 275 51 36 sweden 917 99 37 switzerland 283 123 38 taiwan 1970 284 39 united kingdom 1507 373 40 united states 2869 369 total number of firms 40,365 9,332 source: author’s own calculations. 4. methodology this study adopted the momentum strategy j6k6 developed by jegadeesh and titman (1993). their strategy stands as a benchmark to calculate momentum returns and has been used by subsequent researchers such as chui, titman, and wei (2010), griffin, ji, and martin (2003), ji, martin, and yao (2017), and rouwenhorst (1998). stock returns are calculated through the following formula. (1) where pt is a closing price in month t and pt-1 is an opening price in month t-1. next, stocks are ranked in ascending order based on their cumulative returns at the end of each month for the past six months. the top 10% of the stocks are classified as winner portfolios while the bottom 10% are loser portfolios. these winner and loser portfolios are held for the next six months. this strategy is known as j6k6 in the literature where j6 refers to the six months formation period while k6 refers to the six months holding period. for example, a j6k6 portfolio on 1st july 2016 will show the performance of a portfolio from 31 december 2015 to 30 june 2016 (j6) and will be held until 31 december 2016 (k6). to calculate winner minus loser portfolio, the average loser portfolio returns are deducted from the average winner portfolio returns. the average value of winner minus loser portfolio is the momentum return where t-statistics is used to measure the significance level of the momentum returns. 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 = ( 𝑃𝑃𝑡𝑡 − 𝑃𝑃𝑡𝑡−1 𝑃𝑃𝑡𝑡−1 ) ∗ 100 momentum effect all over the world: 75-93 83 5. empirical findings this section discusses the momentum profitability in 40 countries around the world from 1996 to 2018. table 3 shows the monthly average momentum returns for each of these countries along with their respective t-statistics. for robustness sake, the momentum results during the sub-periods of 1997 to 2006 and 2007 to 2018 are also reproduced. table 3 groups the sample countries according to their respective momentum returns (full sample) whether they are positive and significant (panel a), negative and significant (panel b) or insignificant (panel c). panel d tabulates the average momentum returns of all countries according to full and sub-periods. as shown in table 3, 21 and 15 countries exhibited positive and significant, and negative and significant momentum returns, respectively, during the full sample period. these findings were robust to the two sub-period analyses. the exceptions were thailand, netherlands, israel, us and italy where the momentum returns either turned insignificant or flipped during the subperiod analysis. netherlands for instance had positive and significant momentum returns in the full sample period but turned insignificant during first sub-period (1996–2006). the u.s. and italy experienced positive and significant momentum returns in the full sample period but turned negative and significant in the first sub sample period. momentum returns were insignificant in taiwan, colombia, spain and greece for the full sample period as shown in panel c of table 3. panel d shows the average momentum return of all countries. the average momentum returns were 0.25%, 0.18% and 0.24% during the full sample, first and second subperiods, respectively. table 3. country wise average monthly momentum returns (%) s. no country average monthly momentum profits (%) t-stat panel a: positive monthly momentum returns 1 bangladesh full sample (1996–2018) 1.81*** 21.75 first sub-sample (1996–2006) 1.99*** 16.65 second sub-sample (2007–2018) 1.65*** 25.99 2 china full sample (1996–2018) 1.72*** 52.09 first sub-sample (1996–2006) 2.25*** 58.10 second sub-sample (2007–2018) 1.28*** 47.08 (continued) 84 the international journal of banking and finance, vol. 14, 2018-2019: 75-93 s. no country average monthly momentum profits (%) t-stat panel a: positive monthly momentum returns 3 kenya full sample (1996–2018) 0.47*** 4.25 first sub-sample (1996–2006) 0.56*** 4.50 second sub-sample (2007–2018) 0.41*** 4.05 4 morocco full sample (1996–2018) 0.98*** 5.42 first sub-sample (1996–2006) 1.29*** 6.61 second sub-sample (2007–2018) 0.67*** 4.22 5 south africa full sample (1996–2018) 1.56*** 22.56 first sub-sample (1996–2006) 2.14*** 27.44 second sub-sample (2007–2018) 0.97*** 17.22 6 thailand full sample (1996–2018) 0.12*** 5.51 first sub-sample (1996–2006) 0.25*** 10.48 second sub-sample (2007–2018) 0.01 1.37 7 belgium full sample (1996–2018) 2.39*** 24.86 first sub-sample (1996–2006) 2.05*** 15.35 second sub-sample (2007–2018) 2.68*** 32.78 8 chile full sample (1996–2018) 0.82*** 24.19 first sub-sample (1996–2006) 0.64*** 14.25 second sub-sample (2007–2018) 0.97*** 32.47 level of significance, * p<0.01, ** p<0.05, *** p<0.1 momentum effect all over the world: 75-93 85 table 3 (continued) s. no country average monthly momentum profits (%) t-stat 9 denmark full sample (1996–2018) 1.24*** 22.29 first sub-sample (1996–2006) 1.2*** 19.31 second sub-sample (2007–2018) 1.28*** 24.79 10 finland full sample (1996–2018) 3.20*** 27.05 first sub-sample (1996–2006) 3.9*** 25.54 second sub-sample (2007–2018) 2.62*** 28.30 11 germany full sample (1996–2018) 2.04*** 72.20 first sub-sample (1996–2006) 2.63*** 73.32 second sub-sample (2007–2018) 1.55*** 71.27 12 italy full sample (1996–2018) 0.24*** 5.45 first sub-sample (1996–2006) -0.9*** -5.62 second sub-sample (2007–2018) 1.18*** 14.66 13 netherlands full sample (1996–2018) 0.41*** 5.20 first sub-sample (1996–2006) -0.35 -0.64 second sub-sample (2007–2018) 1.03*** 10.06 14 new zealand full sample (1996–2018) 1.44*** 11.15 first sub-sample (1996–2006) 1.97*** 10.91 second sub-sample (2007–2018) 1.01*** 11.35 15 norway full sample (1996–2018) 2.76*** 21.24 first sub-sample (1996–2006) 3.49*** 22.99 second sub-sample (2007–2018) 2.14*** 19.78 16 poland full sample (1996–2018) 1.53*** 6.86 first sub-sample (1996–2006) 2.14*** 9.59 second sub-sample (2007–2018) 1.03*** 4.60 86 the international journal of banking and finance, vol. 14, 2018-2019: 75-93 table 3 (continued) 17 singapore full sample (1996–2018) 1.53*** 6.85 first sub-sample (1996–2006) 2.14*** 9.56 second sub-sample (2007–2018) 1.03*** 4.59 18 sweden full sample (1996–2018) 3.69*** 51.16 first sub-sample (1996–2006) 4.66*** 46.57 second sub-sample (2007–2018) 2.87*** 54.98 level of significance, * p<0.01, ** p<0.05, *** p<0.1 s. no country average monthly momentum profits (%) t-stat 19 switzerland full sample (1996–2018) 2.84*** 57.90 first sub-sample (1996–2006) 3.50*** 63.78 second sub-sample (2007–2018) 2.28*** 53.00 20 united kingdom full sample (1996–2018) 1.46*** 96.10 first sub-sample (1996–2006) 1.90*** 102.71 second sub-sample (2007–2018) 1.10*** 90.58 21 united states full sample (1996–2018) 0.17*** 8.04 first sub-sample (1996–2006) -0.14** -2.07 second sub-sample (2007–2018) 0.43*** 16.46 panel b: negative monthly momentum returns 22 brazil full sample (1996–2018) -0.89*** -22.07 first sub-sample (1996–2006) -1.19*** -21.30 second sub-sample (2007–2018) -0.46*** -22.72 23 india full sample (1996–2018) -2.34*** -28.62 first sub-sample (1996–2006) -3.20*** -30.02 second sub-sample (2007–2018) -1.54*** -27.00 24 indonesia full sample (1996–2018) -6.23*** -52.64 momentum effect all over the world: 75-93 87 table 3 (continued) s. no country average monthly momentum profits (%) t-stat first sub-sample (1996–2006) -10.72*** -37.06 second sub-sample (2007–2018) -2.50*** -65.63 25 malaysia full sample (1996–2018) -2.32*** -45.40 first sub-sample (1996–2006) -3.35*** -33.70 second sub-sample (2007–2018) -1.463*** -55.15 26 pakistan full sample (1996–2018) -1.08*** -18.85 first sub-sample (1996–2006) -2.19*** -36.78 second sub-sample (2007–2018) -0.15*** -3.91 level of significance, * p<0.01, ** p<0.05, *** p<0.1 27 philippines full sample (1996–2018) -2.19*** -23.03 first sub-sample (1996–2006) -0.95*** -13.95 second sub-sample (2007–2018) -3.23*** -30.60 28 sri lanka full sample (1996–2018) -1.13*** -20.41 first sub-sample (1996–2006) -2.12*** -36.43 second sub-sample (2007–2018) -0.29*** -7.07 29 turkey full sample (1996–2018) -1.98*** -33.22 first sub-sample (1996–2006) -2.31*** -34.68 second sub-sample (2007–2018) -1.71*** -32.00 30 australia full sample (1996–2018) -0.38*** -10.15 first sub-sample (1996–2006) -0.81*** -19.25 second sub-sample (2007–2018) -0.03*** -2.58 31 canada full sample (1996–2018) -1.62*** -10.68 first sub-sample (1996–2006) -1.54*** -21.58 second sub-sample (2007–2018) -1.68 -1.59 88 the international journal of banking and finance, vol. 14, 2018-2019: 75-93 table 3 (continued) s. no country average monthly momentum profits (%) t-stat 32 france full sample (1996–2018) -0.39*** -11.22 first sub-sample (1996–2006) -0.16*** -5.03 second sub-sample (2007–2018) -0.59*** -16.38 33 hong kong full sample (1996–2018) -0.59*** -29.76 first sub-sample (1996–2006) -0.52*** -22.14 second sub-sample (2007–2018) -0.66*** -34.94 34 israel full sample (1996–2018) -0.33*** -5.69 first sub-sample (1996–2006) -0.73*** -12.76 second sub-sample (2007–2018) 0.01 0.21 level of significance, * p<0.01, ** p<0.05, *** p<0.1 35 japan full sample (1996–2018) -0.34*** -32.85 first sub sample (1996–2006) -0.15*** -18.92 second sub sample (2007–2018) -0.49*** -44.92 36 korea full sample (1996–2018) -0.36*** -12.18 first sub-sample (1996–2006) -0.44*** -9.77 second sub-sample (2007–2018) -0.29*** -14.19 panel c: no momentum returns based on full sample period 37 greece full sample (1996–2018) -0.42 -0.23 first sub-sample (1996–2006) -0.98 -1.99 second sub-sample (2007–2018) 0.06 1.23 38 taiwan full sample (1996–2018) 0.05 -1.35 first sub-sample (1996–2006) 0.74*** 14.03 second sub-sample (2007–2018) -0.52*** -14.17 momentum effect all over the world: 75-93 89 table 3 (continued) s. no country average monthly momentum profits (%) t-stat 39 colombia full sample (1996–2018) 0.08 1.15 first sub-sample (1996–2006) 0.34*** 3.33 second sub-sample (2007–2018) -0.14 -0.67 40 spain full sample (1996–2018) -0.08 -0.48 first sub-sample (1996–2006) 0.21 -0.17 second sub-sample (2007–2018) -0.32 -0.74 panel d: overall average monthly momentum returns in 40 countries all countries full sample (1996–2018) 0.25*** 4.86 first sub-sample (1996–2006) 0.18*** 4.78 second sub-sample (2007–2018) 0.24*** 4.93 level of significance, * p<0.01, ** p<0.05, *** p<0.1 table 4 compares findings of the current study with five main international studies on momentum returns. rouwenhorst (1999) found that the j6k6 momentum strategy produced 0.39% monthly average momentum returns in asian stock markets where 85% and 15% of the sample countries exhibited positive and negative significant momentum returns, respectively. hameed and yuanto (2002) examined 16 momentum strategies in six asian stock markets and observed that none of the sample countries were generating momentum returns. griffin et al. (2003) studied momentum effects in 39 countries around the world. they discovered that the j6k6 investment strategy produced 0.49% monthly average momentum returns across the countries in their sample. a total of 80% and 18% of the countries had positive and negative momentum returns, respectively. whereas, 2% of the sample countries did not exhibit momentum returns. fernandes and ornelas (2008) found positive momentum returns and negative momentum returns in 20% and 80% of their sample countries, respectively. chui et al. (2010) examined momentum returns across 40 countries and perceived that 62.5% of their sample countries showed insignificant momentum returns. meanwhile, momentum returns were positive (negative) in 52.5% (10%) of the sample countries. the overall monthly average momentum returns were 0.73% across 40 countries. the findings of the current study were also in line with the five international studies mentioned. we studied 40 countries across the world and found 0.25% monthly average momentum returns. 90 the international journal of banking and finance, vol. 14, 2018-2019: 75-93 ta bl e 4. p er ce nt ag e of p os iti ve a nd n eg at iv e m om en tu m r et ur ns a cr os s c ou nt ri es a ut ho r n um be r of c ou nt rie s sa m e pe rio d m on th ly a ve ra ge m om en tu m p ro fit s c ou nt rie s w ith m om en tu m p ro fit (% ) c ou nt rie s w ith m om en tu m l os s (% ) c ou nt rie s w ith n o m om en tu m (% ) m om en tu m st ra te gi es r ou w en ho rs t ( 19 99 ) 20 19 82 –1 99 7 0. 39 % 85 % 15 % j6 k 6 h am ee d an d yu an to (2 00 2) 6 19 81 –1 99 4 0. 53 % 10 0% 1 6 st ra te gi es g rif fin e t a l. (2 00 3) 39 19 75 –2 00 0 19 90 –2 00 0 0. 49 % w or ld 1. 63 % a fr ic a 1. 82 % s ou th a fr ic a 80 % 18 % 2% j6 k 6 fe rn an de s a nd o rn el as (2 00 8) 15 19 95 –2 00 4 20 % 80 % j3 k 3, j6 k 6, j1 2k 12 , j 24 k 24 c hu i e t a l. (2 01 0) 40 19 81 –2 00 3 0. 73 % 60 % 10 % 30 % j6 k 6 c ur re nt st ud y 40 19 96 –2 01 8 1. 49 % 57 .5 % 37 .5 % 5% j6 k 6 so ur ce : a ut ho rs ’ o w n ta bu la tio n. momentum effect all over the world: 75-93 91 our findings confirmed the existence of the momentum effect in 36 (90%) countries out of 40 countries where 52.5% of the countries had positive and significant momentum returns and 37.5% of the countries had negative and significant momentum returns. a total of 10% of the sample countries had insignificant momentum returns for the full sample period. one noteworthy difference between our study and studies by rouwenhorst (1998, 1999), hameed and yuanto (2002) and chui et al. (2010) was the existence of a significant number of countries with negative momentum returns which implied the existence of return reversals, i.e. past losers outperformed past winners and vice versa. the significant number of countries with negative momentum returns was however consistent with fernandes and ornelas (2008) who obtained reversals in 10 out of 15 emerging stock markets. as suggested by de bondt and thaler (1985, 1987), contrarian investment strategies of buying past loser stocks and selling past winner stocks could produce abnormal profits due to the overreaction of investors. both return momentum and reversals invalidated the notion of efficient market hypothesis. jegadeesh and titman (1993) also showed that return reversals could take place after 12 months. another possible reason why our results were different from previous studies was because of the sample selection criteria. rouwenhorst (1999) excluded 5% of the stocks based on their extreme past performance during the ranking (formation) period. moreover, rouwenhorst (1999) used the 30% cutoff point in determining winner and loser stocks. whereas, we applied a 10% cut-off and did not apply any data filtering technique except deleting stocks that did not have values to full length. 6. conclusion this study revisited the existence and profitability of momentum returns on a global scale. this is necessary since 16 years have passed since the study by chui et al. (2010) with a sample period which ended in 2003. it is important to identify the magnitude of momentum returns across countries so that policymakers are aware of the validity of emh in their respective stock markets. our findings offered a clear rejection of the notion of emh as 90% of the sample countries exhibited significant momentum returns. our study has implications for practitioners and policymakers. the main findings of the current study offer guidance to individual investors, investment houses and institutional investors as to which country(s) the j6k6 momentum investment strategies tend to be profitable. on the other hand, investors should also be aware of the losses associated with momentum strategies as supported by our findings on return reversals in 37.5% of our sample countries. in addition, investors should implement momentum strategies with caution due to the rapid speed of information diffusion in stock prices in recent years. this could 92 the international journal of banking and finance, vol. 14, 2018-2019: 75-93 eradicate momentum profits a lot sooner than what investors have predicted. policymakers should take into consideration the behavioural aspects of investors when designing regulations to curb excessive volatility in the market. they need to identify investor psychological biases that cause stock prices to deviate from their fundamental values for a prolonged period. further research can be conducted to identify the determinants of momentum profits and losses by analysing risk factors such as liquidity and book-to-market value. moreover, the effect of information technology and speed of information diffusion can also be explored to find out the source(s) of momentum effect. references barroso, p., & santa-clara, p. (2015). momentum has its moments. journal of financial economics, 116(1), 111–120. de bondt, w. f. m., & thaler, r. (1985). does the stock market overreact ? the journal of finance, 40(3), 793–805. fernandes, j. l. b., & ornelas, j. r. h. (2008). momentum and reversal puzzle in emerging markets. journal of behavioral finance, 5(3), 54–71. chui, a. c. w., titman, s., & wei, k. c. j. (2000). momentum, legal systems and ownership structure-an analysis of asian stock markets. ssrn electronic journal. chui, a. c. w., titman, s., & wei, k. c. j. (2010). individualism and momentum around the world. journal of finance, 65(1), 361–392. daniel, k., hirshleifer, d., & subrahmanyam, a. (1998). investor psychology and security market under-and overreactions. the journal of finance, 53(6), 1839–1885. fernandes, j. l. b., & ornelas, j. r. h. (2008). momentum and reversal puzzle in emerging markets. journal of behavioral finance, 5(3), 54–71. griffin, j. m., ji, s., & martin, j. s. (2004). global momentum strategies: a portfolio perspective. the journal of portfolio management, 31(2), 23– 39. griffin, j. m., ji, x., & martin, j. s. (2003). momentum investing and business cycle risk: evidence from pole to pole. the journal of finance, 58(6), 2515–2547. hameed, a., & yuanto, k. (2002). momentum strategies : evidence from the pacific basin stock markets. journal of financial research, xxv(3), 383–397. hong, d., lee, c. m. c., & swaminathan, b. (2003). earnings momentum in international markets. ssrn electronic journal, (february). https://doi. org/10.2139/ssrn.390107 hurn, s., & pavlov, v. (2003). momentum in australian stock returns. australian journal of management, 28(2), 141–155. momentum effect all over the world: 75-93 93 huynh, t. d., henker, t., & henker, j. (2010). survivorship bias and alternative explanations of momentum effect. ssrn electronic journal. jegadeesh, n., & titman, s. (1993). american finance association returns to buying winners and selling losers : implications for stock market efficiency author (s): narasimhan jegadeesh and sheridan titman source : the journal of finance , vol . 48 , no . 1 (mar., 1993), pp . 6591 pub. the journal of finance, 48(1), 65–91. kang, j., liu, m.-h., & ni, s. x. (2002). contrarian and momentum strategies in the china stock market: 1993–2000. pacific-basin finance journal, 10(3), 243–265. li, b., qiu, j., & wu, y. (2010). momentum and seasonality in chinese stock markets. journal of money, investment and banking, 2010(17), 24–36. malkiel, b. g. (2003). the efficient market hypothesis and its critics. journal of economic perspectives, 17(1), 59–82. malkiel, b. g., & fama, e. f. (1970). efficient capital markets: a review of the theory. the journal of finance, 25(2), 383–417. rouwenhorst, k. g. (1998). international momentum strategies. the journal of finance, 53(1), 267–284. rouwenhorst, k. g. (1999). local return factors and turnover in emerging stock markets. journal of finance, 54(4), 1439–1464. zhang, x. f. (2006). information uncertainty and stock returns. journal of finance, 61(1), 105–137. board gender diversity, corporate reputation and market performance 1 the international journal of banking and finance, volume 9 (number 1) 2012: pages 1-26 board gender diversity, corporate reputation and market performance meredith b. larkin, richard a. bernardi, and susan m. bosco roger williams university, united states of america ____________________________________________________________ abstract this study examines the association between corporate transparency, ethical orientation of fortune 500 companies, the number of females represented on the board of directors as reported in the 2010 annual report data and respective stock performance. our basis for this judgment was whether the firm was listed on either (both) ethisphere magazine’s 2010 ‘world’s most ethical companies’ or (and) corporate responsibility magazine’s 2010 ‘100 best corporate citizens list’. our results indicate that, as the number of women directors increased, the probability of a corporation appearing on these lists increases. finally, while being on one of these lists did not increase corporate return data in a statistically significant sense, it did dramatically reduce the degree of negative returns. key words: ethical orientations, corporate reputation, market performance, gender diversity jel classification: g39, m14 _____________________________________________ 1. introduction bernardi and several coauthors (2002 to 2010) examined various issues associated with female board members. bernardi et al. (2002, 2005) found that corporations were more likely to include pictures of the board in their annual reports when the membership of their board included women (2002) and/or minorities (2005). bernardi et al. also found that corporations with higher percentages of women on their boards were more likely to be on ‘100 best companies to work for’ (2006) and ‘most ethical companies’ (2009) lists; have a higher percentage of female executives (2004); and, engage in activities demonstrating corporate social responsibility (2010). however, the ‘so what’ question remains concerning the increase in female representation on larkin et al.: board gender diversity 2 corporate boards and corporate reputation, which equates to the public’s overall perception of a corporation (fombrun and shanley, 1990). while the number of women on corporate boards has increased by approximately 28 women per year between 1977 and 2001 (bernardi et al., 2006), it is still relatively low at 832 female directors of 5,613 directors on fortune 500 boards (bernardi et al, 2009). stakeholders with legitimate interests (donaldson and preston, 1995) have lost billions of dollars due to recent corporate scandals. following the corporate scandals in 2002 and new regulatory actions, it is surprising to find that the percent of women on boards has not increased substantially. while women directors made up 11.9 percent of fortune 500 boards in 2002, they now make up 14.4 percent (bernardi et al., 2009). the 2.5 percent increase represents an additional 156 female directors of approximately 5,600 directors (bernardi et al, 2002 and 2009). 1 gender diversity on corporate boards associates with financial performance (carter et al., 2008), reduction in the inherent risk (ittonen et al., 2007), positive market reactions (defond et al., 2005), and positive cumulative abnormal returns (huang et al., 2011). the need for organizations to become better corporate citizens and improve their levels of corporate social responsibility has become increasingly evident. our sample includes the 2010 fortune 500 corporations of which 92 (408) corporations appear (do not appear) on corporate responsibility magazine’s (hereafter crm) ‘100 best corporate citizens list’. our sample also includes the 46 (454) corporations that appear (do not appear) on ethisphere magazine’s (hereafter em) 2010 ‘world’s most ethical companies’ list. our research indicates that stakeholder advocate organizations (i.e., crm and em) tend to recognize corporations that have higher proportions of women on their boards. the interaction of this recognition and multiple female board members for the corporations in this sample associated with higher overall returns and lower negative returns for stockholders’ wealth as measured by the market prices of the corporations’ common stock. 1 of the 5,514 directors on fortune 500 boards included in this research, there are 863 female directors (15.1 percent). international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 3 2. literature review and hypotheses 2.1 board duties an organization’s board of directors is responsible for ensuring that a corporation is meeting the objectives of stakeholders as well as developing business strategies to prosper in the future (arfken et al., 2004; peterson and philpot, 2007). when the corporation fails to meet these objectives, many question the ability of the board members. campbell and minguez-vera (2008) indicate that the effectiveness of a board depends heavily on each board member’s qualifications and experience. historically, older white males dominated; consequently, as corporate scandals continue, stakeholders push for changes in the corporate structure (mcdaniel et al, 2001; farrell and hersch, 2005). recent scandals indicate that corporations are not meeting these objectives; this suggests that the current homogenous boardroom is unable to perform its duties (campbell and minguez-vera, 2008; burke, 1997; arfken et al., 2004). companies now face an investing public that demands scrutiny of all corporate decisions and expects board members to be accountable for their actions (arfken et al. 2004). consumers and shareholders question the ability of a homogenous boardroom (arfken et al., 2004); consequently, there has been a call for a higher representation of women on corporate boards (burke, 1997). many feel that the presence of women on corporate boards adds a sense of moral obligation to a corporation’s decision-making process (arfken et al., 2004) which can in turn improve boardroom transparency and limit the likelihood of corporate scandal. the underrepresentation of women on boards became public in 1977 (special report, 1977); research continues to depict this trend (burgess and tharenou, 2002). boards should not overlook their female board members and should take initiatives to ensure the consideration of female board members’ viewpoints. burke (1997) indicates that the benefits to both internal and external stakeholders of considering female board members’ viewpoints include a more comprehensive decision making-process that is both creative and innovative. women are able to bring a new perspective to the homogenous boardroom including raising issues that affect a wider range of stakeholders and using interpersonal skills to promote discussion (kramer et al., 2007). 4 2.2 board diversity research has shown that lack of diversity within a boardroom results in a manila mindset to solving corporate problems (burgess and tharenou, 2002) that can lead to group think issues as well as lack of achievement within the company. over the past decade, homogenous boards have been a contributing factor to spectacular failures and overall poor governance (brown et al., 2002). a more diverse board results in an increased representation of moral and ethical viewpoints in the discussions prior to making decisions (arfken et al., 2004). diversity limits the possibility of a myopic decision-making process that can result in “unhealthy and possibly unethical decisions” (arfken et al., 2004 p. 185) when the board has similar demographics. many studies have cited that diversity not only limits the likelihood of myopic decision-making process but also increases the likelihood of positive occurrences such as fresh ideas, better problem solving, improved strategic planning, and additional accountability (arfken et al., 2004). diversity in the boardroom allows members to make better decisions as a more complete picture of the issues at hand are typically discussed (adams and flynn, 2005). adams and ferreira’s (2009) research indicates that diverse boards are more likely to hold ceos responsible for poor stock price performance and that board compensation is typically equity-based, implying that the board is more aligned with shareholder interests. these findings further the idea that having women on boards can add value to a company. overall, gender-diverse boards have increased levels of boardroom involvement and corporate oversight (adams and ferreira, 2009); boards with a greater female presence have higher levels of meeting attendance. the primary way in which boards operate and conduct business is through meetings and thus, attendance is a crucial factor of a successful board (adams and ferreira, 2009). these authors note that women were less likely to have attendance problems and that having females on boards results in better attendance by male directors. clearly, the female influence in this area is quite important; increasing attendance should result in better boardroom discussion and higher levels of effectiveness. an increased membership of female directors positively associated enhanced corporate reputation (bear et al., 2010). bernardi et al. also found that corporations with higher percentages of women on their boards were more likely to be named as one of the ‘100 best international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 5 companies to work for’ (2006), one of the ‘most ethical companies’ (2009), and a higher number of female executive-level managers (2004). 2.3 ethical orientation the individuals an organization attracts, hires, and retains influence the organization’s ethical climate (schneider, 1987). harrison (1992) indicates that factors essential to economic success include a sense of community (i.e. a trusting and caring environment reynolds, 2003) and a robust ethical system. employees are more likely to support a corporation’s values when the corporation demonstrates a commitment to the welfare of its community (barnett and schubert, 2002). young people are attracted to a company’s social record (goodpaster, 1991); for example, bernardi and guptill (2008) found that women from eight countries who were approaching graduation were more concerned about a corporation’s reputation within its community than were their male counterparts. consequently, as arnold et al. (1997) suggest, the foundation of an ethical organization culminates in an environment that nurtures ethical behavior. bernardi and arnold (1997) and akaah (1989) indicate a difference between males’ and females’ moral reasoning and development implying that the way men and women handle ethical decision-making differs. williams (2003) makes clear the correlation between increased levels of female directors and a company’s involvement in corporate social responsibility activities. the more concerned the firm is with issues of corporate responsibility, the less likely the firm will take actions that are considered unethical or do not promote the overall wellbeing of the firm and the surrounding environment. in a corporate landscape where corruption is rampant, it is essential that corporations work to ensure their culture is ethical and women are able to enhance this important aspect (mcdaniel et al., 2001). bernardi et al. found that corporations with higher percentages of women on their boards were more likely to be on em’s ‘most ethical companies’ list (2009) and engage in activities demonstrating corporate social responsibility (2010). 2.4 transparency while ethical orientation is concerned with the internal decisions that an organization makes, transparency focuses on whether stakeholders have access to this information. an organization’s reputation rests on its stakeholders trust (larkin, 2003), which directly relates to 6 the information that is available about the organization (i.e., the transparency of the company). public disclosure of information has increased in an effort to increase trust in organizations because trust at all levels is essential to a corporation’s legitimacy (the economist, 2000). following instances such as enron, the big-four firms have all indicated a commitment to ethics and transparent reporting (lehman, 1992). deloitte and touche initiated a challenge to restore the profession’s public trust (parrett, 2004) and both kpmg (2003) and pricewaterhousecoopers (2003) call for increased transparency and integrity in corporate reporting. organizations that are not forthcoming with information tend to be hiding essential facts from stakeholders; this process leads to the possibility of corporate scandal. the lack of transparency and audit failure contributed to the enron debacle one of the most discussed financial scandals. reinstein and mcmillan (2004) show that enron’s collapse was not a perfect storm (i.e., a happenstance of rare events that had devastating effects). rather, the audit team from andersen ignored or missed red flags that would have indicated problems with enron’s financial heath (reinstein and mcmillan, 2004). in this case, understanding the organization’s operations would have lead stakeholders to question enron’s profits and financial statements. 2.5 women and economic performance businesses operate with the objective to earn a profit and in turn increase shareholder value. corporate managers, and those who are interested in positive governance, believe that there is a correlation between board diversity and shareholder value (carter et al., 2003). furthermore, investors are willing to pay more for firms with effective corporate governance (smalhout, 2003). jackson (2004) found that most individuals consider reputation in their investment decisions; research also suggests that stock performance associates with corporate reputation (miles and covin, 2000; vergin and qoronfleh, 1998; sparks, 1998; sims, 1994). many corporations recognize that increasing shareholder value should occur in an ethical manner, but the implementation of this process can be difficult. corporations are under increasing pressure to act in a socially responsible manner while still attaining high profit levels. corporate social responsibility is the implementation of policies that recognize the relationship among business ethics, community investment, governance and many other aspects of business (tsoutsoura, 2004; bernardi et al., 2006). international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 7 prior research demonstrated the benefits that having women on boards can bring to an organization, but many question the ability of organizations to be socially responsible while still meeting shareholder expectations. some feel that social responsibility results in increased cost pressures on organizations, which can affect the bottom line. however, corporations that engage in socially responsible practices can more easily obtain capital as their reputation of being less risky (tsoursoura, 2004). together, these aspects help improve the public’s view of the firm thus increasing profitability. tsoursoura also found that financial performance and corporate social responsibility were positively associated. most importantly, tsoursoura found that the industries with the lowest ratings for social responsibility include mining and construction, the same sectors with the lowest number of women on boards (governancemetrics international, 2010). these findings show that having women on boards does in fact positively affect the social responsibility behaviors of an organization. bear et al. (2010) found that the number of female directors positively associated with measures of corporate reputation. bernardi et al. (2006) also found that an increased proportion of female representation on boards associated with the corporation’s inclusion on the ‘100 best companies to work for’ list. additionally, corporations that value diversity have proven to be more competitive in the overall business setting (mcdaniel et al, 2001). farrell and hersch (2005) conducted research on the effect that women board members have on a corporation’s common stock performance. they found that, while adding women to the board positively associated with return on assets, the market failed to react to adding women to a board. this information supports the idea that having women on boards has a direct impact on the bottom-line profits of an organization, but at this point fails to influence investor opinion. 2.6 hypothesis development while overconfidence in decision-making occurs in both men and women, men are typically more overconfident than women are especially in areas considered masculine (i.e., financial decisions) (lundeberg et al., 1994). barber and odean (2000) found that men tended to turn over their portfolios more often and have lower returns than women; they suggest that overconfidence leads to high levels of counterproductive trading. huang and kisgen (2008) found that female cfo’s tended to be more risk adverse, used debt less frequently to finance corporate capital demands, made fewer acquisitions, and outperformed corporations with male 8 cfos. consequently, women can also bring different viewpoints/attitudes to an organization through board membership. carter et al. (2003) indicate that gender diversity enhances understanding of the intricacies of a corporation’s market. women are able to bring their insights to the boardroom and match the diversity of the organization’s consumer base. in general, boards that closely match the makeup of the general population provide improved corporate social performance (bernardi et al., 2006). diversity also allows an organization to view problems in a different manner and reevaluate the way in which they do business. prior research shows that improved performance associates with boards that are diverse with respect to gender (brady, 2007; cohen and kornfeld, 2006). adding female board members has proven to increase an organization’s sense of responsibility. carter et al. (2008) noted that the effect of gender diversity on a board’s audit function associated with financial performance. ittonen et al. (2007) found that, when female board members are on the board’s audit committee, there was a reduction in the inherent risk of financial misstatements. these authors also noted that gender diversity associates with lower audit fees. audit committees that include women tend to be more conservative; thiruvadi and huang (2011) report that, when female directors were members of audit committees, corporations tended to report increased negative accruals, which decrease income. when new audit committee members had accounting expertise, the market reacted positively (defond et al., 2005). huang et al. (2011) found that, when compared to the addition of male board members, the addition of female board members to the audit committee resulted in positive cumulative abnormal returns. 2 gender diversity can be beneficial in situations involving complex tasks, which require creative decision-making (kravitz, 2003). consequently, expanding a board’s viewpoint can facilitate increased discussion, better problem solving tactics, and a better understanding of the marketplace as a whole. gul et al. (2011) found that board-gender diversity encouraged corporations to increase their disclosure of corporate data. bernardi et al. found that corporations with higher percentages 2 while nguyen and faff (2006) found that gender diversity associated with higher firm values, wang and clift (2009) found that gender and racial diversity did not influence firm performance – both studies used listed australian corporations. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 9 of women on their boards were more likely to be on ‘100 best companies to work for’ (2006) and ‘most ethical companies’ (2009) lists. however, this research fails to associate female directors and listings with financial performance, which leads to our research hypotheses (stated in their alternate form): h1: the corporations on crm’s (2010) list will have a higher (lower) proportion of multiple female directors (zero or only one director) than for corporations not on this list. h2: the corporations on em’s (2010) list will have a higher (lower) proportion of multiple female directors (zero or only one director) than for corporations not on this list. h3: membership on crm’s (2010) list will associate with higher (lower) increases (decreases) in common stock prices in 2010. h4: membership on em’s (2010) list will associate with higher (lower) increases (decreases) in common stock prices in 2010. 3. data and methodology 3.1 sample the current sample includes the 2010 fortune 500 corporations of which 92 (408) corporations appear (do not appear) on crm’s list (table 1). the sample also includes the 46 (454) corporations that appear (do not appear) on em’s 2010 list (table 2). appendix a provides the methodology for crm’s list. appendix b provides the methodology for em’s list. we determined the size and gender composition of the corporate boards of directors by referring to the companies’ actual 2010 annual reports or from data included in the mergent online database. 3.2 selection processes and corporate return data crm’s list (appendix a) took into consideration both the transparency and the level of social responsibility of an organization. it is important to note that our basis for considering an organization as transparent lies with the fact that the magazine penalized corporations for not disclosing information relating to social responsibility. em’s list (appendix b) acknowledges corporations for being ethical and following compliance measures through positive leadership. we tested the research question relating to the organization’s financial return using a rate of return for the period between january 1, 2010 and december 31, 2010. we used historic stock 10 prices to determine the price per share of each organization at the earliest available stock price in 2010 in relation to the latest available stock price in 2010 at the close of the trading day. in order to determine the percentage change of the stock price for the given year, we subtracted the beginning (january 1) stock price from the ending (december 31) stock price, which we divided by the beginning stock price. table 1: most transparent companies 3m ford motor northeast utilities abbott laboratories fpl group occidental petroleum advanced micro devices freeport-mcmoran copper & gold oracle air products & chemical gap owens corning alcoa general mills pepsi bottling allergan h.j. heinz pepsico applied materials hess pg&e corp. avon products hewlett-packard procter & gamble ball hormel foods quest diagnostics baxter international intel raytheon boeing international business machines sara lee bristol-myers squibb international paper sempra energy campbell soup itt sherwin-williams chevron j.c. penney southern cisco systems j.p. morgan chase & co. staples citigroup johnson & johnson starbucks coca-cola johnson controls state street corp. coca-cola enterprises kellogg stryker colgate-palmolive kimberly-clark texas instruments conagra foods lubrizol tjx consolidated edison mattel union pacific cummins mcdonald's united parcel service cvs caremark mcgraw-hill verizon deere mckesson wal-mart stores dell medtronic walt disney dominion resources merck weyerhaeuser duke energy microsoft wisconsin energy eaton monsanto xcel energy emc mosaic xerox exelon newmont mining yum brands exxon mobil nike international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 11 3.3 board gender data while our initial analysis included all fortune 500 corporations, 51 of these corporations are not publicly listed; of the 51 corporations that were not publicly traded, three of them were on crm’s list. for the 89 publicly traded corporations that appear on crm’s list, there were 195 female directors and 1057 total directors (18.4 percent). the 89 corporations on crm’s list make up 19.8 percent of the 449 publicly traded companies in the fortune 500. our data indicate that of the 89 corporations on crm’s list: 1.1 percent had no female directors; 21.3 percent had one female director; and, 77.5 percent had multiple female directors. we compared these percentages to those for the remaining 406 corporations that have 586 female directors and 3,904 total directors (15.0 percent). the data for these corporations indicate that: 14.4 percent had no female directors; 32.2 percent had one female director; and, 55.3 percent had multiple female directors. table 2: most ethical companies aflac flour pitney bowes american express ford motor principal financial aramark fpl group rockwell automation ashland gap rockwell collins becton dickinson general electric sempra energy best buy general mills starbucks campbell soup google symantec caterpillar harris target ch2m hill hartford financial services texas instruments cisco systems hewlett-packard time warner cummins international paper united parcel service deere johnson controls waste management duke energy mattel weyerhaeuser eaton nike whole foods market ecolab pepsico wisconsin energy xerox three of the 46 corporations that appear on em’s list are not publicly traded. for the 43 publicly traded corporations that appear on em’s list, there are 93 female directors and 491 total directors (18.9 percent). our data indicate that of those: 2.3 percent had no female directors; 23.3 percent had one female director; and, 74.4 percent had multiple female directors. we compared 12 these percentages to those for the remaining 406 corporations that have 688 female directors and 4,470 total directors (15.4 percent). the data for these corporations indicate that: 12.8 percent had no female directors; 30.8 percent had one female director; and, 56.4 percent had multiple female directors. 4. analyses and findings 4.1 overview for this part of the analysis, we used the data from 449 of the fortune 500 companies that had publicly listed performance data the other 51 companies were not publicly listed. in our analysis, we group corporations by whether or not they appear on a specific list and by the number of female board members: no female board members, one female board member, and multiple female board members. in our examination of the data, we use contingency analysis, as we believe it visually demonstrates our findings with respect to listing by either crm or em, board gender diversity and common stock performance. 4.2 corporate reputation and female board members (h1 and h2) this part of the analysis tests for an association between listing by either crm or em and gender. for the 89 corporations included on crm’s list, there was one corporation (1.1 percent) with no female directors, 19 corporations (21.4 percent) with one female director and 69 corporations (77.5 percent) with multiple female directors. for the 360 corporations that were not included on crm’s list, there were 52 corporations (14.5 percent) with no female directors, 126 corporations (32.2 percent) with one female director and 192 corporations (53.3 percent) with multiple female directors. while the corporations not listed by crm had a higher proportion of corporations with no female directors or only one female director (14.5 and 32.2 percent respectively) than the corporations listed by crm (1.1 and 21.4 percent respectively), the reverse is true for corporations with multiple female directors (53.3 versus 77.5 percent respectively). panel a of table 3 shows the actual and expected number of female directors for each of group international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 13 or corporations. 3 our analysis indicates that all treatments are not proportionally represented (χ 2 statistic = 20.72, p < 0.000). the most significant contributors to this difference were the corporations listed by crm. corporations on crm’s list had higher proportion of corporations with multiple female directors, which supports our first research hypothesis. for the 43 corporations included on em’s list, there was one corporation (2.3 percent) with no female directors, 10 corporations (23.3 percent) with one female director and 32 corporations (74.4 percent) with multiple female directors. for the 406 corporations that were not included on em’s list, there were 52 corporations (12.8 percent) with no female directors, 125 corporations (30.8 percent) with one female director and 229 corporations (56.4 percent) with multiple female directors. while the corporations not listed by em had a higher proportion of corporations with no female directors or only one female director (12.8 and 30.8 percent respectively) than the corporations listed by em (2.3 and 23.3 percent respectively), the reverse is true for corporations with multiple female directors (56.4 versus 74.4 percent respectively). panel b of table 3 shows the actual and expected number of female directors for each of group or corporations. our analysis indicates that all treatments are not proportionally represented (χ 2 statistic = 6.52, p = 0.045). the most significant contributors to this difference were the corporations listed by em. corporations on em’s list had higher proportion of corporations with multiple female directors, which supports our second research hypothesis. 4.3 corporate reputation and performance (h3 and h4) this part of the analysis tests for an association among listing by either crm or em and corporate performance. on an overall basis, the data indicate that the 89 (360) corporations (not) included on crm’s list had an increase of 11.9 (6.7) percent average return of 7.7 percent. the 43 (406) corporations (not) included on em’s list had an increase of 16.2 (6.8) percent 3 in our contingency analysis, we computed the expected number of companies for each group by multiplying the total number of corporations in each column (i.e., the number of female directors on the board) by proportion of the sample (i.e., either the number of transparent or remaining companies divided by the total sample). for example, for the 16 transparent companies that have one female director, we would expect to have 26.8 companies ([19+116] x [89/449]) rather than our actual count of 19 companies. similarly, for the remaining companies with one female director, we would expect to have 108.2 companies ([19+116] x [360/449]) rather than our actual count of 116 companies. 14 table 3: listing by crm or em and board gender composition panel a: crm’s listing number of female directors none one multiple total transparent corporations actual 1 19 69 89 expected 10.5 26.8 51.7 89 χ2 stat 8.60 2.25 5.76 16.61 remaining corporations actual 52 116 192 360 expected 42.5 108.2 209.3 360 χ2 stat 2.13 0.56 1.43 4.11 panel b: em’s listing number of female directors none one multiple total ethical corporations actual 1 10 32 43 expected 5.1 12.9 25.0 43 χ2 stat 3.27 0.66 1.96 5.90 remaining corporations actual 52 125 229 406 expected 47.9 122.1 236.0 406 χ2 stat 0.35 0.07 0.21 0.63 international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 15 average return of 7.7 percent. consequently, on an overall basis, our first two hypotheses about being on a listed by either crm or em and higher common stock prices were supported by the data. panel a of table 4 provides the average returns for the six groups of companies. our analysis indicates that all treatments did not have a 7.7 percent increase in common stock value (χ 2 statistic = 8.06, p = 0.02). 4 the most significant contributors to this difference were the corporations listed by crm with one female director and multiple female directors. it is the higher percent of increase for corporations listed by crm that are driving the difference in treatments. we further divided the data in panel a by whether their common stock price increased (panel b) or decreased (panel c) for additional analysis. for the corporations in panel b, the average increase in their stock price was 19.9 percent. the data in panel b indicate no difference in treatments (χ 2 statistic = 1.23, not significant) for the corporations whose common stock prices increased. for the corporations in panel c, the average decrease in their stock price was 38.0 percent. the data in panel c indicate that not all treatments had a 38.0 percent decrease in common stock value (χ 2 statistic = 54.91, p < 0.000). again, our data indicates that the most significant contributors to this difference were the corporations listed by crm with one female director and multiple female directors. the common stock prices for the 19 corporations listed by crm (average = -9.1 percent) did not decrease as much as the 76 corporations that were not listed (average = -45.2 percent). consequently, the data support our third hypothesis. panel a of table 5 provides the average returns for the six groups of companies. our analysis indicates that not all treatments had a 7.7 percent increase in common stock value (χ 2 statistic = 20.62, p < 0.000). the most significant contributors to this difference were the corporations listed by em with one female director and multiple female directors. it is the higher percent of increase for corporations listed by em that are driving the difference in treatments. we further divided the data in panel a by whether their common stock price increased (panel b) or decreased (panel c) for additional analysis. 4 we did not include the data for the first group (i.e., being listed by either crm or em and no female directors) as there was only one firm in this group (i.e., return of 31.5 percent was not an average). 16 table 4: average percent change in share price and crm’s listings panel a: average percent change for all corporations (average = 7.7 percent) number of female directors none one multiple total transparent corporations (%) 31.3 13.9 10.8 11.9 (n) (1) (19) (69) (89) χ2 stat na 5.04 1.27 6.31 remaining corporations (%) 10.5 5.8 6.3 6.7 (n) (52) (116) (192) (360) χ2 stat 1.04 0.46 0.25 1.71 panel b: average percent change for corporations with positive returns (average = 19.9 percent) number of female directors none one multiple total transparent corporations (%) 31.3 17.8 16.9 17.3 (n) (1) (16) (53) (70) χ2 stat na 0.22 0.46 0.68 remaining corporations (%) 23.2 20.3 20.0 20.6 (n) (43) (92) (149) (284) χ2 stat 0.54 0.01 0.00 0.55 panel c: average percent change for corporations with negative returns (average = -38.0 percent) number of female directors none one multiple total transparent corporations (%) na -6.9 -9.5 -9.1 (n) (0) (3) (16) (19) χ2 stat na 25.47 21.39 46.86 remaining corporations (%) -50.4 -49.8 -41.5 -45.2 (n) (9) (24) (43) (76) χ2 stat 4.05 3.67 0.33 8.05 international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 17 for the corporations in panel b, the average increase in their stock price was 19.9 percent. the data in panel b indicate no difference in treatments (χ 2 statistic = 0.90, not significant) for the corporations whose common stock prices increased. for the corporations in panel c, the average decrease in their stock price was 38.0 percent. the data in panel c indicate that not all treatments had a 38.0 percent decrease in common stock value (χ 2 statistic = 50.93, p < 0.000). again, the most significant contributors to this difference were the corporations listed by crm with one female director and multiple female directors. the common stock prices for the six corporations listed by em (average = -10.1 percent) did not decrease as much as the 89 corporations that were not listed (average = -39.8 percent). consequently, the data support our third hypothesis. this section of our analysis found that corporations on both crm and em’s lists had higher average increases in their common stock prices (panel a of tables 4 and 5). when we separated the companies according to whether they had increasing or decreasing stock prices, there were no significant differences in the two groups with increasing stock prices (panel b of tables 4 and 5). however, we found that the decrease in common stock prices for corporations on both crm and em’s lists was not as large as the decrease for corporations not on these lists (panel c of ables 4 and 5). consequently, our data indicate an association between a corporation’s reputation (i.e., being on either crm or em’s list) and common stock performance. 5. conclusions the major contribution of this paper is that board gender diversity can provide an incremental benefit for corporations that already have a good reputation in ethical behavior, social responsibility and transparency. our data indicate that companies on crm or em’s lists have superior returns and that companies on crm or em’s lists are more likely to have multiple female directors on their boards. consequently, our data suggest an interactive effect between corporate reputation and the number of female directors. 18 table 5: average percent change in share price and em’s listings panel a: average percent change for all corporations (average = 7.7 percent) number of female directors none one multiple total ethical corporations (%) 10.7 15.9 16.4 16.2 (n) (1) (10) (32) (43) χ2 stat na 8.80 9.90 18.70 remaining corporations (%) 10.9 6.2 6.2 6.8 (n) (52) (125) (229) (406) χ2 stat 1.35 0.29 0.29 1.93 panel b: average percent change for corporations with positive returns (average = 19.9 percent) number of female directors none one multiple total ethical corporations (%) 10.7 21.5 20.5 20.4 (n) (1) (8) (28) (37) χ2 stat na 0.13 0.02 0.15 remaining corporations (%) 23.7 19.8 19.0 19.9 (n) (43) (100) (174) (317) χ2 stat 0.72 0.00 0.04 0.75 panel c: average percent change for corporations with negative returns (average = -38.0 percent) number of female directors none one multiple total ethical corporations (%) na -6.6 -11.9 -10.1 (n) (0) (2) (4) (6) χ2 stat na 25.92 17.90 43.82 remaining corporations (%) -50.4 -48.1 -34.4 -39.8 (n) (9) (25) (55) (89) χ2 stat 4.07 2.70 0.34 7.11 18 international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 http://epublications.bond.edu.au/ijbf/vol9/iss1/1 19 when considering the effect that women on the board have on the financial returns, it was interesting to find that stock prices varied by the direction of the return. while the corporations on either crm’s or em’s list had significantly greater increases in their stock prices compared to corporations not on these lists, this was not the case when we divided the sample into groups with gains versus losses. when analyzing the gains separately, our results indicated that being on crm’s list had the opposite of what we anticipated. the corporations not listed by crm had a slightly higher (3.3 percent) gain; however, the difference was not significant. there was not a difference with respect to the corporations on (not on) em’s list. the data did not support our research hypothesis dealing with corporate reputation and stock prices for corporations with positive returns. when analyzing the losses separately, our results indicated that being on crm’s or em’s list was significantly associated with a reduction in price declines the corporations listed by crm had a significantly lower loss in value than the corporations not on this list (-9.1 and -45.2 percent respectively). the corporations listed by em also had a significantly lower loss in value than the corporations not on this list (-10.1 and -39.9 percent respectively). consequently, there appears to be an economic benefit to being on either of these lists. when we tested our data for the proportion for female directors, we used the same groupings as we did to test for changes in stock prices (i.e., overall change and increasing-anddecreasing stock prices. the corporations listed by crm consistently had a lower number of corporations with no female directors or only one female director and a higher than expected number of corporations with multiple female directors. the opposite was true for corporations that were not listed by crm; a higher number of these corporations had no female directors or only one female director and a lower than expected number with multiple female directors. the corporations listed by em had a lower number of corporations with no female directors or only one female director and a higher than expected number of corporations with multiple female directors. the opposite was true for corporations that were not listed by em; these corporations consistently had a higher number with no female directors or only one female director and a lower than expected number of corporations with multiple female directors. the combined findings indicate that corporations on either crm’s or em’s list have a higher than expected number of boards that include multiple female directors. additionally, the 20 corporations on these lists are more likely to have higher overall changes in stock prices and lower decreases in their stock prices. the stock performance figures imply that higher numbers of women on boards may be able to alter the internal elements of an organization, thus decreasing the likelihood of a loss in value to shareholders. there are four inherent limitations to our study. first, we included only corporations listed in the 2010 fortune 500. our second limitation is that we used only crm’s ‘100 best corporate citizens list’ and em’s 2010 ‘world’s most ethical companies’ lists. third, we examined only the effect of having female directors on boards of directors. fourth, we used only changes in common stock prices. these limitations provide opportunities for future research in this area that include examining: a more diverse corporate sample; using other measures of corporate social responsibility; including minorities as board members; and, using other measures of financial performance. future research could take the form of a longitudinal study that determines whether the organizations with a lower female boardroom presence experienced larger stock losses for a longer time period. future studies might also consider using return on assets and return on equity as internal corporate performance measures to test whether the number of women on a board has an impact on these figures. finally, future research could also survey investors to determine whether the number of women on the board associates with their valuation of the organization. author information: the authors are staff members at the gabelli school of business, roger williams university, bristol, ri 02809, united states of america. the corresponding author is richar a. bernardi, professor of accounting and ethics: he may be contacted at e-mail: rbernardi@rwu.edu or phone: 1-(401)-254-3672. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 21 appendix a panel a: corporate responsibility magazine’s criteria for identifying “100 best corporate citizens” “100 best corporate citizens” methodology uses publicly available information to determine the world’s top corporate responsibility ranking. cr magazine contracts with a third party research organization to collect data and develop initial rankings. once all the necessary information was collected, the companies were scored relative to their industry peers 324 data elements in 7 categories. the companies included in the analysis were defined as the 2010 russell 1000. the rankings are determined from the ordinal list of companies that results from applying the corporate citizenship criteria detailed above. data category # data elements 2010 weighting percent environment 133 19.5% climate change 60 16.5% human rights 40 16.0% employee relations 65 19.5% philanthropy 9 9.0% financial 8 12.5% governance 9 7.0% panel b: explanation of corporate responsibility magazine’s process cr magazine’s researchers and editors employed a detailed process. the separate and sequential analyses conducted were: step 1 selection of and contracting with a research firm step 2 determination of evaluation criteria step 3 data collection step 4 data sources step 5 undisclosed data step 6 data validation step 7 review and publication where: steps 1-2 determined that way that analysis would be completed and includes getting input and opinions from ngos, academics, investment analysts, etc. step 3-6 focus on data collection using only publicly available information (company websites, 10-ks, government datasets, etc.). undisclosed information negatively influences the company’s ranking. data validation is done by the research team reviewing their work and by providing the opportunity for companies to correct factual inaccuracies. step 7 allows companies two opportunities to review the datasets determined by the research team (not their rankings), after this period, the information and rankings are provided to cr. from corporate responsibility magazine (2011)* *the 2011 methodology details were used as 2010 details were unavailable 22 appendix b ethisphere’s criteria for identifying “the world’s most ethical companies” world’s most ethical companies™ (wme) methodology analyzes companies that go beyond making statements about doing business ‘ethically’, to translate those words into action. wme winners demonstrate real and sustained ethical leadership within their industries, putting the council’s credo of “good, smart, business, profit” into real business practice. the ethics quotient (eq) framework is consists of a series of multiple-choice questions in five core categories. these are used to capture and rate a company’s performance in an objective, consistent, and standard manner. the categories and associated weighting are: 1. ethics and compliance program 2. reputation, leadership and innovation 3. governance 4. corporate citizenship and responsibility 30% 30% 15% 25% the eq score is derived given the relationship to answers provided and formulas based on demographic qualifiers. the top percentile of performers in each of the 35 industries are then independently researched and analyzed to verify ethics performance. from ethisphere (2010) 22 international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 23 references adams, r. b., and ferreira, d., (2009). women in the boardroom and their impact on governance and performance. journal of financial economics, 94 (2): 291-309. adams, s. m., and flynn, p. m., (2005). local knowledge advances women's access to corporate boards. corporate governance: an international review, 13 (6): 836-846. akaah, i. p., (1989). differences in research ethics judgments between male and female. journal of business ethics, 8 (5): 375-381. arfken, d. e., bellar, s. l., and helms, m. m., (2004). the ultimate glass ceiling revisited: the presence of women on corporate boards. journal of business ethics, 50 (2): 177-186. arnold, v., lampe, j. c. and sutton, s. g., (1997). understanding the factors underlying ethical organizations: enabling continuous improvement, journal of applied business research, 15 (3), 1-18. barber, b., and odean, t., (2001). boys will be boys: gender, overconfidence, and common stock investment. the quarterly journal of economics, 116 (1): 261-292. bear, s., rahman, n., and post, c., (2010). the impact of board diversity and gender composition on corporate social responsibility and firm reputation. journal of business ethics, 97 (2): 207-221. barnett, t., and schubert, e., (2002). perceptions of the ethical work climate and convenantal relationships. journal of business ethics, 36 (3): 279-290. bernardi, r. a. and arnold, d. f., (1997). examination of moral development within public accounting by gender, staff level, and firms. contemporary accounting research, 14 (4): 653668. bernardi, r. a., bean, d. f., and weippert, k. m., (2002). signaling gender diversity through annual report pictures: a research note on image management. accounting, auditing and accountability journal, 15 (4): 609-616. bernardi, r. a., bean, d. f., and weippert, k. m., (2005). minority membership on boards of directors: a case for requiring pictures of boards in annual reports, critical perspectives on accounting, 16 (8): 1019-1033. bernardi, r. a., bosco, s. m., and columb, v., (2009). does female representation of boards of directors associate with the ‘most ethical companies’ list? corporate reputation review, 25 (3): 270-280. bernardi, r. a., bosco, s. m., and reis, k. d., (2004). corporate leadership in the fortune 500: women in executive management positions, journal of business & economics research, 2 (12): 1-10. 24 bernardi, r. a., bosco, s. m., and vassill, k. m., (2006). does female representation of boards of directors associate with fortune's “100 best companies to work for” list? business and society, 45 (2): 235-248. bernardi, r. a., and guptill, s. t., (2008). social desirability response bias, gender and factors influencing organizational commitment: an international study, journal of business ethics, 81 (4): 797-809. bernardi, r. a. and threadgill, v. h., (2010). women directors and corporate social responsibility, electronic journal of business ethics and organizational studies, 15 (2): 15-21. brady, k. t., (2007, november 26). study finds fortune 500 companies with more women board members performed better financially. st. charles county business record, p. 1. brown, d. a., brown, d. l., and anastasopoulos, v., (2002). women on boards: not just the right thing but the "bright" thing. the conference board of canada, 1-17. burgess, z., and tharenou, p., (2002). women board directors: characteristics of the few. journal of business ethics, 37 (1): 39-49. burke, r. j., (1997). women on corporate boards of directors: a needed resource. journal of business ethics, 16 (9): 909-915 campbell, k., and minguez-vera, a. m., (2008). gender diversity in the boardroom and firm financial performance. journal of business ethics, 81 (8): 435-451. carter, d. a., d’sousa, f., simkins, b. j., and simpson, w. g., (2008). the diversity of corporate board, committees and financial performance. working paper. available at: http://www.ssn.com/abstract=1106698. carter, d. a., simkins, b. j., and simpson, w. g., (2003). corporate governance, board diversity, and firm value. financial review, 38 (1): 33-53. cohen, r. and kornfeld, l., (2006). women leaders and the bottom line. bloomberg corporate law journal, 1 (1): 1-8. corporate responsibility magazine, (2011). 2010 100 best corporate citizens, available at: http://www.thecro.com /content/100-best-corporate-citizens defond, m., hann, r. n., and hu, x., (2005), does the market value financial expertise on audit committees of boards of directors? journal of accounting research, 43 (2): 153-193. donaldson, t., and preston, l. e., (1995). the stakeholder theory of the corporation: concepts, evidence, and implications. academy of management review, 20 (1): 65-91. ethisphere magazine: (2011). 2010 world’s most ethical companies, available at: http://www. ethisphere.com/wme2010. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 25 farrell, k. a., and hersch, p. l., (2005) additions to corporate boards: the effect of gender. journal of corporate finance, 11 (1-2): 85-106. fombrun, c., and shanley, m., (1990). what’s in a name? reputation building and corporate strategy. academy of management journal, 33 (2): 233-235. fortune, (2010). fortune 500, arnold bernhard and co., new york. goodpaster, k. e., (1991). business ethics and stakeholder analysis. business ethics quarterly, 1 (1): 53-73. governancemetrics international, (2010). women on boards: a statistical review by country, supersector and sector, pp. 1-13. available at: http://www.gmiratings.com/(bnnl3555sqbsvh551 v2jfs55) /hp/women_on_boards_-_a_statistical_review_from_gmi_-_3 _2_09.pdf. gul, f. a., srinidhi, b., and ng, a. c., (2011). does board gender diversity improve the informativeness of stock prices? journal of accounting and economics, 51 (3): 314-338. harrison, l. e., (1992). who prospers? how cultural values shape economic and political success, basic books: new haven. huang, h., y. yan, j.m. fornaro, and a. elshhat. (2011). market reactions to audit committee director’s gender: evidence from us-traded foreign firms. international journal of banking and finance 8 (1), article 4. huang, j., and d. kisgen. (2008). gender and corporate finance. working paper, boston college. available at: http://www.usc.edu/ schools/business/fbe/seminars/papers/f_10-308_kisgen-gender. pdf. ittonen, k., miettinen, j., and vahamaa, s., (2007), does female representation in audit committees affect audit fees? available at: http://www.aaahq.org/audit/midyear/08midyear/ papers/61_ittonen_doesfemalerepresentation.pdf. jackson, k., (2004). building reputational capital: strategies for integrity and fair play that improve the bottom line, new york: oxford university press. kpmg, (2003). audit committee roundtable highlights, available at: http://www.kpmg.com /aci/roundtable_highlights.asp. kramer, v. w., konrad, a. m., erkut, s., and hooper, m. j., (2007). critical mass on corporate boards: why three or more women enhance governance. directors monthly, february, 19-22. kravitz, d. a., (2003). more women in the workplace: is there a payoff in firm performance? the academy of management executive, 17 (3): 148-149. larkin, j., (2003). strategic reputation risk management. new york: palgrave macmillan. lehman, c., (1992). accounting’s changing role in social conflict, paul chapman: london. 26 lundeberg, m. a., fox, p. w., and punćcohaŕ, j., (2000). highly confident but wrong: gender differences and similarities in confidence judgments. journal of educational psychology, 86 (1): 114-121. mcdaniel, c., schoeps, n., and lincourt, g., (2001). organizational ethics: perceptions of employees by gender. journal of business ethics, 33 (3): 245-256. miles, m. and covin, j., (2000). environmental marketing: a source of reputational, competitive, and financial advantage. journal of business ethics, 23 (3): 299-311. nguyen, h., and faff, r., (2006). impact of board size and board diversity on firm value: australian evidence, corporate ownership and control 4 (2): 24–32. parrett, w. g., (2004). ‘globalization’s next frontier: principled codes of conduct bolster the rule of law.’ speech to international center for corporate accountability, united nations, may 17, available at: http://www.iasplus.com/resource/0405wparrett.pdf. peterson, c. a., and philpot, j., (2007). women's roles on u.s. fortune 500 boards: director expertise and committee memberships. journal of business ethics, 72 (2): 177-196. pricewaterhousecoopers, (2003). understanding the independent auditor’s role in building public trust: a white paper, pricewaterhousecoopers international. available at: http://www.uic. edu/classes/actg/actg593/readings/sarbanes-oxley/pwc-sox-act-white-paper.pdf reinstein, a. and mcmillan, j. j., (2004). the enron debacle: more than a perfect storm. critical perspectives on accounting, 15 (5/6): 955-970. reynolds, s. j., (2003). perceptions of organizational ethicality: do inflated perceptions of self lead to inflated perceptions of the organization? journal of business ethics, 42 (3): 253-266. schneider, b., (1987). the people make the place. personnel psychology 40 (fall): 19-39. sims, r., (1994). ethics and organizational decision-making: a call for renewal. westport, ct: quorum books. smalhout, j., (2003). barron’s, january 27 (1): 35. sparks, r., (1998). the challenge of ethical investment, in i. jones and m. pollitt (eds.), the role of business ethics in economic performance: 141-170. new york: st. martin’s press. special report, (1977). out of the typing pool and onto the board: a list of women directors. business and society review, 22: 27-34. the economist, (2000). business ethics: doing well by doing good. the economist, 355 (8167), 65-67. thiruvadi, s. and huang, h. w., (2011). audit committee gender differences and earnings management. gender in management: an international journal, 26 (7): 483-498. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 1 27 tsoutsoura, m., (2004). corporate social responsibility and financial performance. working paper, haas school of business, university of california at berkeley. vergin, r., and qoronfleh, m., (1998). corporate reputation and the stock market. business horizons, 41 (1): 19-26. wang, y., and clift, b., (2009). is there a “business case” for board diversity? pacific accounting review, 21 (2): 88–103. williams, r. j., (2003). women on corporate boards of directors and their influence on corporate philanthropy. journal of business ethics, 42 (1): 1-10. bank mergers and competition in japan the international journal of banking and finance, volume 9 (number 2) 2012: pages 1-25 bank mergers and competition in japan kang h. park southeast missouri state university, united states of america ____________________________________________________________ abstract using h-statistic of the panzar-rosse model, this paper examines commercial bank merger waves in japan and their effect on competition in the japanese banking market during 1983-2006. the h-statistic is estimated separately for three different time periods, the boom, the burst and the recovery. this paper concludes that the bank mergers that took place in japan have not led to a higher level of market power except during the period of financial crisis around the time of bubble burst. recent mergers in the japan’s banking sector do not seem to harm the competition level in the banking market. an increase in individual bank’s market share and an increase in overall market concentration have not materialized in higher net interest margin in japan. key words: bank consolidation, market competition, japanese banks, bubbles, korea, japan, china jel classification: g21, l10 _____________________________________________ 1. introduction the japanese and south korean banking industries share commonality in their merger activities and market concentration trends. that is, in japan there is an increase in bank mergers, resulting in the creation of mega banks resulting in an increase in the overall market concentration. even with a decrease in market concentration, the chinese banking industry too is still highly concentrated and its level of competition is close to one nearing oligopoly although there have been little movements in banking concentration. however, both the japanese and south korean banking industries have a market structure of monopolistic competition, which is more desirable than that of banks in china with oligopolistic trend. the city banks in japan tend to be more competitive than their south 1 there has been a surge of mergers and acquisitions of business firms over the last twenty years in the world, so too in japan.1 although mergers and acquisitions have occurred for a long time, consolidation of firms occurred in the past two decades at a faster phase, in the so-called fifth merger wave, which has been the most remarkable. in a similar way, mergers and acquisitions in the banking industry have been accelerating across countries over the last two decades. the banking industries all over the world have experienced a fundamental change in its market structure via rapid consolidation in the face of both de-regulations and technology infusion. domestic financial deregulation and financial globalization triggered fierce competition among banks within and across countries. these necessitated consolidation of banks to reduce risk through business diversification and also to take advantage of scale economies therefrom. the waves of mergers in the banking industry raise important questions of public policy as to whether mergers improve bank efficiency, enhance profitability or lessen market competition. many studies have tried to address these questions by examining performance of merged banks. however, most of these empirical studies analyzed bank efficiency, profitability or competition in the united states (us) or europe. the purpose of this paper is to study the effect of bank mergers on the degree of market competition in the japanese banking industry to address its efficiency: we apply a current methodology. this paper is organized in the following manner. section 2 describes japan’s banking system while section 3 discusses the financial crisis that occurred during the late 1990s in japan. a description of the bank merger waves in japan following that financial crisis is attempted in section 4. section 5 is a brief detour on the merger experiences of china, south korea and the us to provide a common base. section 6 briefly surveys the related literature on bank mergers and market competition in the banking sector. the 1 the us banking industry has responded to the pointed reforms in the1980s and in the 1990s by engaging in large scale mergers that led to the 15,000 or so banks being reduced to less than 8,000 in the 2010s. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 korean counterparts while the south korean regional banks are more competitive than their japanese counterparts. given these overall understanding, it is time to examine the efficiency changes in japan over a longer period to see what has occurred intertemporally, hence this study. 2 reader will find in section 7 the model used to test the degree of competition. the main results on japan are presented in section 8 along with description of the data and the variables used in this study as well as interpretation of the estimated results on japan. in section 9, the korean and chinese experiences are documented. the paper ends with conclusions in section 10. 2. banking system in japan the banking system in japan consists of various bank types such as city banks, trust banks, two-tier regional banks, shinkin banks (shinyou kinko) and credit cooperatives (shinyou kumiai). among them, city banks are by far the largest and most influential banks, providing comprehensive banking services including both traditional and nontraditional services to large corporate customers. these banks are nationwide banks with a wide geographical scope. most of these banks act as main banks that maintain shareholdings in keiretsu firms of the main customers and vice versa: this is a unique feature of main banks. there were twenty city banks in the 1980s, grown through bank failures and mergers, which decreased to thirteen by the 1990s: by 2007 there were just six city banks. trust banks in japan are authorized to do both banking and trust activities. traditionally they provide funds to corporations and receive deposits from trusts. in recent years their business focus shifted to asset management. under the financial system reform law of 1992, the number of trust banks almost doubled. these too went through mergers and acquisitions resulting in its number being reduced to twenty in 2007 (liu, 2009). the first tier regional banks (chiho ginko) are chartered under the banking act, operating within each prefecture. this is akin to us state chartering of banks. there were 64 of these banks in 2007. their customer bases are local, small and medium size businesses, households and local governments. almost three-fourths of their deposits are time deposits and these banks are the main source of small and medium business loans. the second tier regional banks (sogo ginko) were established initially as mutual banks and then transformed into regional banks under the banking act of 1992. a total of 45 3 banks were in operation in 2007. although their size is somewhat smaller than the first tier regional banks, their services are similar, serving the needs of small businesses, households and local governments. shinkin banks known as credit cooperatives, with 280 banks in 2007, are different from other banks mentioned above in several aspects. first, they are chartered under the shinkin bank act, a different legislation from the banking act. second, they are nonprofit financial institutions with membership requirements like credit unions in the us and other countries. members take part in ownership and profit-sharing to some limited extent. these banks are also allowed to accept deposits from non-member customers. third, they are small in scale compared to ordinary commercial banks, and focus on financing small business customers within their local areas. credit cooperatives (shinyou kumiai) numbering 164 in 2007 are basically similar to shinkin banks in their banking services provided and the customer bases served. the only difference is that these cooperatives are supervised by the prefecture government while shinkin banks are under the supervision of the central government or the ministry of finance. therefore, their operation is limited to their respective prefectures geographically (uchida and udell, 2009). scale difference among the various types of banks can be seen from the different amounts of average outstanding domestic loans in 2000; 2,379 billion yen for city banks, 212 billion yen for tier 1 regional banks, 95 billion yen for tier 2 regional banks, and 18 billion yen for shinkin banks. in addition to the above ordinary banks, there are several special banks such as long-term credit banks and financial institutions for small businesses, agriculture, forestry and fishery. long-term credit banks provide long-term working capital to corporations in the form of loans and guarantees, and contribute to economic growth of the japanese economy by involving in financing shipbuilding, steel and petro-chemical industries in the past. however, as demand for long-term credit by manufacturers diminishes, these banks are at the verge of extinction with only one bank remaining. in recent years, new types of banks such as settlement banks and internet banks cropped up in the japanese banking system. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 4 3. financial crisis in japan the japanese economy was at its peak performance till the late 1980s and japanese banks by size were then counted to be among the top 10 largest banks in the world. however, the very size precipitated the financial crisis, and led to their slow demise over later time period. the easy monetary policy was carried out by the bank of japan in the 1980s. the result was low cost credit that caused excess demand in various sectors of the economy, for example as real estate overinvestment, industrial overcapacity and the stock market boom. at the end of the 1980s, the japanese economy became a bubble economy, with land prices and the stock price index doubling in a short time period. these phenomena were propelled by excessive bank lending in real estate and construction. when the bank of japan responded with credit tightening in the early 1990s to control the over-heated economic activities, the asset bubble started to burst leading to asset deflation, also the subsequent collapse of the real estate market as well as the stock market. the land prices dropped by 20 per cent in 1992: by 1997 real estate lost about 60 per cent value. with the collapse of the bubble economy, many banks ended up with an excessive level of non-performing loans. the ministry of finance estimated non-performing loans of twelve per cent of total loans and credits in 1997. while the collapse of real estate market triggered the financial crisis, bankruptcy of many large banks and financial institutions burdened with high rates of non-performing loans contributed greatly to a magnifying effect of the financial crisis in the late 1990s. a banking crisis occurred in 1997, when three large financial institutions, including a city bank named hokkaido takushoku bank, failed. in the subsequent year, two long-term credit banks named the long-term credit bank of japan and the nippon credit bank failed (hosono, et al. 2007). the japanese government let hokkaido takushoku bank, sanyo securities and yamaichi securities go bankrupt while other banks such as the long-term credit bank of japan and the nippon credit bank were bailed out. during the 1990s, japanese banks in the midst of a deflationary macroeconomic environment, suffered from very low interest margin charged on loans coupled with high 5 default rates of outstanding loans from previous era. to make things worse, they made big losses from their investments in southeast asia, in korea, indonesia and thailand, as precipitated by the asian financial crisis of 1997-1998. the immediate response by the japanese government was to inject capital as bailout by providing additional funds for deposit insurance. then a series of financial deregulation and financial system reforms were designed and implemented following the financial crisis. for example, strict accounting standards in implementing the basel capital standards were applied and bank recapitalization was required, which prompted bank restructuring, which is the stimulus for the mergers that followed. in 1998, financial supervision agency (fsa) was set up to take over bank supervisory function which was previously done by the ministry of finance. the world was at that time moving away from putting prudential oversight in the central banks, and this was a knee-jerk response to increase attention to the already badly-off banks. 4. bank merger waves in japan the banking system in japan went through fundamental changes after the financial crisis, beginning in the mid 1990s (casu, et al., 2006). financial deregulation permitted financial institutional diversification of their business and also deregulation of bond markets, which forced commercial banks to seek new ways of raising funds and utilizing their assets. this change resulted in an increase in loans to small and medium size banks by large commercial banks, which shunned away from such lending in the past. however, the most significant change in the banking system after financial deregulation was a surge of mergers and acquisitions of banks. banks mergers occurred very rarely in japan since the world war ii before the 1990s. the numbers of city banks, regional banks and shinkin banks remained stable until 1990. only few cases of mergers at different levels of commercial banks, that is, at city bank level, regional bank level and shinkin bank level, were reported. a small number of mergers before the financial crisis can be explained by the government policy of convoy system. the ministry of finance and other governmental regulatory international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 6 authorities, with the intention to stabilize the banking sector, restricted competition among banks, even restricting opening new branches and preventing banks from doing security transactions. furthermore, when some banks were at the risk of bankruptcy, the government asked financially sound banks to rescue financially distressed banks with injection of capital and management help. in some cases, the government directly bailed out the troubled banks with government funding. financially sound banks usually accepted government request for two reasons. first, they do not want to have negative repercussions from the powerful regulatory authorities by refusing to comply. second, they could extend their branch network by obtaining branches of failing banks. so, most of mergers and acquisitions occurred before the financial system reforms were in place by government as request for sound banks to take over failed banks. financial liberalization and financial deregulation which started in the 1980s and continued in the 1990s made this kind of convoy system not workable. in particular, when the risk-based capital requirements on the basis of the basel standard were implemented in 1992, financially weak banks sought for survival through consolidation. two mergers among city banks and three mergers among regional banks occurred in the first half of the 1990s. mergers among shinkan banks also occurred more frequently in the 1990s than before (hosono, et al. 2007). in 2002, fsa announced financial rehabilitation plan, which required financial institutions apply strict accounting standards and reduce their non-performing loans to one-half of the then exiting level. the plan kicked in big merger waves of banks in the early 2000s because financially unsound banks had to be consolidated in order to survive. through several mergers among major city banks during the early 2000s, mega banks were established under three financial holding companies. they are mizuho financial holdings, mitsui-sumitomo banking corporation, mitsubishi-tokyo-ufj financial holdings. many mergers among regional banks and among shinkin banks also occurred in the early 2000s with encouragement and financial support from the government. however, no mega banks were created from the regional or shinkin bank groups, and their numbers are still in excess of 100. during these merger waves, fsa refrained from 7 intervening in mergers by maintaining hands off policy away from the previous practice of rescuing financially distressed banks or arranging or mergers among banks. as a result of the merger waves, the market share of total bank deposits held by the top three banks increased from the 25 per cent level in 1990 to 40 per cent level in 2006. (however, this level is much lower than those in south korea (korea hereafter) and china.) the herfindahl-hirschman index (hereafter hhi) for the japanese banking market in 2006 is about 1,000: hhi is calculated by the sum of the squares of each bank’s market share in total bank deposits. there has been an increase in hhi from about 700 in the 1990s to about 1,000 in the 2000s. compared to the banking markets of other countries, the japanese banking market with this hhi value is less concentrated. 5. bank merger waves in other countries in the us, lifting geographical restrictions on acquisitions and branching in the 1980s accelerated mergers and acquisitions of banks. bank consolidation was further stimulated by the passage of the 1994 riegle-neal interstate banking and branching efficiency act. this established the basis for a true nationwide banking system and accelerated bank mergers. the number of us banks fell from 14,404 in 1980 to about 12,000 in 1990 and then to about 7,100 in 2008: in all a 50 per cent decline over three decades. although the number of banks in the us declined and their average asset and deposit sizes became large, there was no noticeable change in the concentration of the us local banking market measure of hhi. this is because much of bank consolidation in the us is typically characterized by market-extension, that is, acquisitions involving two banks in different geographical markets. figure 1 shows that the u.s. banking industry is much less concentrated in the deposit market than those of other countries. the banking sectors in the united kingdom, france, japan, sweden, canada and russia are all more concentrated than the us banking sector. over in the last twenty years, the south korean banking system went through many changes: financial deregulation, financial crisis, and restructuring. in this process, market concentration decreased as the number of banks increased due to financial international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 8 deregulation prior to the asian financial crisis of 1997. after the crisis, the concentration ratio increased because of a decline in the number of banks due to bank closures and the creation of mega banks through bank consolidation. figure 1 this change in concentration in korea is different from the us experience with bank consolidation. while much of bank consolidation in the us is typically characterized by market-extension mergers, which are acquisitions involving two banks in different geographical markets, bank consolidation in korea was the result of horizontal mergers among banks with overlapping geographical markets. http://i.bnet.com/blogs/bankhhi.png http://i.bnet.com/blogs/bankhhi.png http://i.bnet.com/blogs/bankhhi.png� 9 the korean banking sector prior to the asian financial crisis of 1997-98 can be regarded as a non-concentrated market with hhi less than 1,000 in total assets, total loans and total deposits according to either the korean or the us horizontal merger guidelines. after the crisis, particularly after the second-phase restructuring in 2001, the korean commercial banking market became a moderately concentrated market with hhi ranging between 1,300 and 1,500 according to the korean antitrust guideline. hhi is in excess of 1,800 in some specific sub-markets such as loans to households and deposits in foreign currency (park, 2009). today, the three largest banks, kookmin bank, woori bank and shinhan banks hold about 60 per cent of total bank assets. hhi figures of korean banks are higher than those of banks in other oecd countries of a similar population size. for example, hhi of spanish banks was in the 700s and hhi of italian banks was in the 600s according to the 2000 data of the european central bank. on the other hand, chinese banking system went through a different path from the us, japan and korean experiences. even though there have been many bank foreclosures, takeovers and mergers in recent years, the number of new bank entering the chinese banking market far exceeded the number of banks that disappeared. until 1978, there was one single bank, people’s bank of china, and then along with its economic reform, the chinese government in 1979 and 1984 authorized four state-owned commercial banks with limited competition among them. since then the chinese government allowed many joint equity banks and private banks in order to mobilize needed financial resources for economic development. furthermore, it authorized several policy banks and city banks in the 1990s as a measure of financial liberalization in preparation for the entry to the world trade organization. all these have contributed to a continuous decrease in market concentration of the chinese banking industry. contrary to the increasing trend of korean banks’ market concentration, the chinese banking system has experienced continually decreasing market concentration from 2743 in 1994 to 1642 in 2008 (park, 2011). this change is clearly attributable to a change in the chinese government policy on banking, which allowed establishment of more banks and promoted competition among them. in spite of some mergers of banks occurred in recent years, the number of new banks created far exceeded the number of international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 10 banks foreclosed and merged. before deng xiao ping’s 1978 reform, china had a mono bank, the people’s bank of china, playing both roles of central and commercial banking. with reform, four specialized state banks were split from the people’s bank of china between 1979 and 1984, leaving the people’s bank of china solely functioning as china’s central bank. even though restrictions of these specialized banks to do business in only their designated territories were removed in 1985, competition among them was very limited until the mid 1990s. there was a boost to competition when the chinese government authorized establishment of three policy banks. since 1986, 14 joint-equity banks were established, where shares were held by the government, cooperatives and private sector. during the mid 1990s, the central government allowed local governments to establish local (or city) banks. in 2008, the total number of banks exceeds 200 excluding foreign banks, and the number continuously increases year by year. luc and ariff (2008) show concentration reduction as a result of government policy of privatizing state banks. as the market value of the us and western banks were reduced due to the global financial crisis of 2007-08, four chinese banks ranked among the top 10 banks in the world. furthermore, the top three are all chinese banks, industrial and commerce bank of china, bank of china and construction bank of china. 6. survey of the literature this section briefly reviews the theoretical models and empirical findings on bank competition, our main focus for this study. although studies have investigate the effect of bank consolidation on competition, there is little consensus on appropriate theoretical framework, and the existing empirical findings are inconclusive. as long as there are no sunk costs and hit-and-run entry is possible, then market contestability can yield competitive pricing regardless of the number of firms (baumol, et al., 1982). the efficient structure hypothesis advances that efficient banks obtain higher profitability and greater market share because of their efficiency, which will lead to a more concentrated market. therefore, the association between structure and performance might be spurious 11 unless efficiency is controlled in the model (smirlock, 1985). adverse borrower selection may result in spurious empirical structure conduct performance linkages too (shaffer, 2002). a concern about the effect of consolidation on competition arises from the structure-conduct-performance (scp) paradigm, which dates back to mason (1939). the scp model suggests that increasing market concentration leads to less competitive conduct in terms of higher prices and less output, which results in higher profits at the expense of lower consumer welfare. this paradigm is the basis of the so-called collusion hypothesis. although there is a theoretical basis for these linkages, other equilibrium conditions can lead to different relationship between market concentration and conduct. empirical results on the scp paradigm are mixed. according to gilbert (1984), many studies presented a mixed set of results in aggregate, and tended to suffer from various methodological flaws. weiss (1989) reports that only 21 out of 47 studies support the claims of scp model. two empirical methods have been developed to remedy the natural shortcomings of the scp model by testing the conduct directly, without regard to industry structure. one method is the bresnahan (1982, 1989) and lau (1982) model (b-l model) which estimates the markup of price over marginal cost as a measure of market power. thus, this method is also called the markup test. this model is based on two structural equations, an inverse demand equation and a supply equation derived from the first order condition of profit maximization. the other method is the panzar and rosse (1982, 1987) model (p-r model). this model measures the extent to which a change in a vector of input prices is reflected in gross revenue. thus, this method is also called the revenue test. if the market is perfectly competitive, then the change will be fully reflected in revenue. shaffer (2004) contrasts both methods in detail and discusses their advantages and disadvantages. numerous studies apply the p-r model empirically, beginning with shaffer (1982) who finds monopolistic competition behavior with a sample of new york banks in 1979. nathan and neave (1989) reject the hypothesis of monopoly power of canadian banks. country-specific empirical studies include vesala (1995) for finland, molyneux, international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 12 at al. (1996) for japan, coccorese (1998) for italy, hondroyiannis (1999) for greece, and hempell (2002) for germany. molyneux, et al. (1994) and bikker and groeneveld (2000) find monopolistic competition in several european countries. on the other hand, de bandt and davis (2000) find monopolistic competition for large banks and monopoly for small banks in germany and france. bikker and haaf (2002) find that the banking industries in 23 oecd countries for the period 1998-1999 are generally characterized by monopolistic competition with the exception of australia and greece. gelos and roldos (2002) compare eight european and latin american countries and find that the bank consolidation process in its early stage has not lowered competition. for japanese bank competition, there were several studies based on the stochastic cost frontier approach. the p-r model was first applied by molyneux et al. (1996) to data of japanese commercial banks. they find that the japanese banking market was under monopoly or conjectural variations short-run oligopoly in 1986 while improving the level of monopolistic competition in 1988. uchida and tsutsui (2005), using long-term panel data from 1974–2000, find that the market competition had improved, especially in the 1970s and in the first half of the 1980s, but it had deteriorated after 1997. they also find that competition among city banks in metropolitan areas was stronger than that among regional banks in rural areas. lee and nagano (2008) report that market concentration that was brought about by bank mergers has not necessarily resulted in low competition. there are a few studies on korean banking competition. two different results are reported for korea. lee and kim (1995) and lee (2003), applying the p-r model, find that banking competition in korea improved along with financial deregulation. kim (2003), which measures competition using the b-l model based on aggregate monthly data from 1996 to 2002, finds that the pricing behavior of korean banks during this period is consistent with perfect competition and that they behave more competitively even after the increase in concentration ratio. on the other hand, using the p-r model, kim et al. (2004) and lee and lee (2005) find that banking competitiveness weakened significantly along with an increase in market concentration after the 1997-8 korean financial crisis. recently, park (2009) resolved the conflicting results by dividing the sample period of 1992-2004 into three separate periods, and report korean banking 13 market has been monopolistically competitive during the pre-crisis period and post-crisis period with temporary deviation to the level of perfect competition during the crisis period. studies on bank competition in china are scanty and are mostly descriptive rather than analytical, simply reporting the trend of bank concentration without further investigative analysis. park (2011) shows that even though the competitive conditions of the chinese banking market have definitely improved over time, the chinese banking market still has an oligopolistic market structure. 7. panzar-rosse h-model the p-r model is generally regarded to be robust to the extent that market and bank level data are available, even though bikker at al. (2006) stated that inclusion of a scale variable in the model may cause overestimation of the level of competition. furthermore, shaffer (1982) showed that the h-statistic of the p-r model is inversely related to the index of monopoly power [q/q (∂q /∂q)] of the b-l model under certain conditions, where q is a bank’ s output and q is output by all banks. so, we use the p-r model in this study to assess the competitive nature of the japanese banking industry. let a bank’s revenue function be ri = ri (xi, y1i), where xi = a vector of output of bank i and y1i = a vector of exogenous variables which affect the revenue function of bank i. a bank’s cost function is assumed to be ci = ci (xi, pi, y2i), where pi is a vector of k factor input prices of bank i and y2i = a vector of exogenous variables which affect the cost function of bank i. the vectors y1i and y2i may include common variables. profit maximization by the bank requires that marginal revenue equal marginal cost as ri’ (xi, y1i) = ci’ (xi, wi, y2i). (1) market power can be measured by the degree to which a change in factor input prices affects the change in the revenue function at equilibrium. therefore, panzar and rosse international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 14 (1987) calculated the sum of the elasticities of the revenue with respect to factor input prices from the reduced-form revenue equation and define it as the h-statistics. h = σ (∂ri /∂pki) (pki/ ri) (2) where ri is revenue of bank i and pki is kth input price of bank i. panzar and rosse (1987) showed from the profit maximization condition that the h-statistic is equal to unity (h =1) in a perfectly competitive market, and less than or equal to zero (h ≤ 0) under monopoly. if the market is perfectly competitive, then the change will be fully reflected in revenue. although the panzar-rosse article also shows that 0<h<1 could be consistent with oligopolistic behavior, it is common to regard 0<h<1 as the condition of camberlinian monopolistic competition. this interpretation is valid under the assumption that the observations are in the long-run equilibrium (nathan & neave, 1989). for estimation purpose, following park (2009), the reduced-form revenue equation of a bank can be specified as follows. ln(rit) = α + β1 ln(wit) + β2 ln(cit) + β3 ln(fit) + σ γ1h sh + σ γ2k vk + ε it (3) where rit is bank i’s revenue at time t, wit is bank i’s factor input price of labor at time t, c is bank i’s factor input price of capital at time t, f is bank i’s factor input price of funds at time t, sh is a vector of scale variables, and vk is a vector of control variables other than scale variables that affect the bank’s revenue function. the h-statistic is the sum of β1, β2 and β3. in order to eliminate manual calculation of β1 + β2 + β3 and its standard error, equation (3) can be rearranged as follows. ln(rit) = α + β1 [ln(wit) ln(fit)] + β2 [ln(cit) ln(fit)] + (β1+ β2 + β3) ln(fit) + σ γ1h sh + σ γ2k vk + ε it (4) the h-statistics is estimated by the coefficient of ln(fit) which is β1+ β2 + β3, and its statistical significance can be tested by the standard error. 15 competitive conditions are usually reflected in net interest margin. it is generally expected that the more market is dominated by a few mega banks and the more concentrated the banking market becomes, the higher is the net interest margin. so, to further examine the effect of bank mergers on bank competition, the following equation is estimated. mit = α + βk σ xk,it + γk σ yk, it + δk σ zk,t + ε it (5) where mit = the average net interest margin of bank i at time t, xk,it = bank specific variables of bank i at time t, such as equity ratio, the share of non performing loans and number of branches, yk, it = market power variables of bank i at time t, such as the market share and hhi. zk,t = macroeconomic variables such as economic growth rate, inflation rate and a dummy variable for the crisis period. 8. empirical results: did competition reduce? the data used for this analysis are from the nikkei financial quest database. the japanese banking industry has had three distinctive time periods. the first period is from 1983 to 1988 when banks experienced an unprecedented boom along with the bubbles in the real estate market and the stock market until the bubbles burst. the second period is from 1989 to 1998 when banks experienced hard time with increasing non-performing loans, profit losses eventually leading to bankruptcy of major banks and a bank crisis. the third period is form 1999 to 2006 when the banking sector experienced merger waves and improved its performance thanks to the government’s several financial reforms to promote competition. even though there are several different types of commercial banks in japan as explained in section 2, city banks and regional banks (both tier 1 and tier 2) account for over 80 per cent of total bank assets and total bank loans. so, they are selected as the sample for this study. as some banks entered while others exited during the sample period, unbalanced panel data are used for analysis. the chow breakpoint test is used to see whether we can treat the whole period as a homogenous period or not, that is, whether international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 16 there is no significant difference in the estimated equations between the sub-periods. the test of the null hypothesis of no structural change is rejected. the fixed effect model is used to reflect bank-specific characteristics and to control heterogeneity among banks. the revenue of a bank (rit) is typically measured by interest revenue or its ratio to total assets, presuming that the main function of banks is financial intermediation. however, with weakening of financial intermediation in recent years and diversification of bank assets, total revenue or its ratio to total assets is used in some studies. so, both interest revenue and total revenue are used in this study. for factor input prices, the following formulas are used. wit = personnel expenses/the number of employees, cit = depreciation allowance and other maintenance costs/total fixed assets, fit = interest expenses/the sum of total deposits and borrowings. recently there has been an increase in part-time and contractual workers as banks try to reduce costs. however, only full-time workers and their expenses are considered here. two scale variables are included in the model. total assets (asset) are included to see the size effect while the number of branches (branch) is included to account for the effect of bank networks. several bank specific variables are included as control variables. the ratio of risk-adjusted equity to total assets (equity) is used as a control variable for credit market and operational risk while the ratio of non-performing loans to total loans (npl) is included to control for the risk effect. even though financial intermediation is the core business of banks, the japanese banking sector experienced weakening of its financial intermediation function over time, which is a common phenomenon all over the world. to reflect the effect of changing financial intermediation or diversification, the ratio of non-interest revenue to total revenue (nonint) is included in the estimation equation. the estimated results of equation (4), using both the interest revenue and the total revenue as the dependent variable, for city banks in japan are reported in tables 1. the two scale variables, asset and branch, are highly correlated. furthermore, bikker et al. 17 (2006) stated that inclusion of a scale explanatory variable such as asset in the p-r model may cause overestimation of the level of competition and may distort the test results. so, the variable asset is deleted from the estimation equation, and only the variable branch, representing bank network, is used. table 1: estimation results of equation (4), city banks 1983-1988 1989-1998 1999-2006 interest revenue total revenue interest revenue total revenue interest revenue total revenue ln w (labor cost) 0.082 (0.128) 0.096** (0.042) -0.026 (0.294) 0.034 (0.407) 0.061** (0.043) 0.079** (0.026) ln c (capital cost) 0.102* (0.094) 0.114* (0.085) 0.045 (0.331) -0.032 (0.259) 0.103* (0.073) 0.107* (0.061) ln f (fund cost) 0.536*** (0.002) 0.495*** (0.001) 0.327 (0.103) 0.389 (0.170) 0.659*** (0.003) 0.632*** (0.002) ln branch 0.279*** (0.000) 0.256*** (0.000) 0.325*** (0.000) 0.337*** (0.000) 0.364*** (0.000) 0.414*** (0.000) equity 0.001 (0.463) 0.001 (0.328) 0.002 (0.228) 0.014 (0.267) 0.005** (0.031) 0.012** (0.020) npl -0.012 (0.422) -0.012 (0.341) -0.004* (0.098) -0.006* (0.087) -0.007 (0.156) -0.004 (0.279) nonint -0.293* (0.062) 0.534* (0.073) -0.270 (0.145) 0.421 (0.214) -0.226** (0. 043) 0.139** (0.035) n 88 88 132 132 61 61 adj. r2 0.989 0.991 0.992 0.994 0.997 0.997 h statistic 0.721 (0.001) 0.705 (0.001) 0.346 (0.346) 0.389 (0.258) 8.23 (0.002) 8.18 (0.001) wald test: h=0 (ρ) 105.97 (0.000) 101.29 (0.000) 12.25 (0.002) 14.38 (0.001) 154.26 (0.000) 147.83 (0.000) wald test: h=1 (ρ) 57.21 (0.000) 62.33 (0.000) 124.93 (0.000) 116.42 (0.000) 34.75 (0.000) 29.22 (0.000) 1. the fixed effects model is used for estimation. the coefficients of the constant under the fixed effects model are not reported here. 2. h statistic is the sum of ln w, ln c and ln f, and its ρ value is obtained by estimating equation 4. 3. ρ values are shown in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels respectively. with the interest revenue as the dependent variable, the h-statistic dropped significantly from 0.721 for the period of 1983-1988 to 0.346 for the period of 19891999, but increased to 0.823 for the period of 2000-2006. the wald test rejects the international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 18 hypothesis of monopolistic market structure (h=0) and the hypothesis of perfectly competitive market structure (h=1) at the 1% level for all three periods. a similar pattern is found in the model when the total revenue is used as the dependent variable. the hstatistic changed from 0.705 for the period of 1983-1988 to 0.389 for the period of 19891999, but increased to 0.818 for the period of 2000-06. the wald tests render the same conclusion about the market structure of the japanese commercial banking market. the estimation results of the h-statistic with two different dependent variables, the interest revenue and the total revenue, are robust. among the three factor input prices, the unit funding cost (ln f) is the most influential variable affecting a change in revenue in terms of the magnitude of the coefficient and it is also statistically most significant. however, its effect on revenue diminished during period 2. all three factor input prices usually exhibit positive coefficients. but during period 2, the unit labor cost (ln w) as well as the unit capital cost (ln c) are not statistically significant and sometimes they exhibit negative coefficient. the positive sign and significant coefficient of branch indicates the presence of the network effect. equity became significant after the government carries out the financial reform and applied more strict accounting standards. according to the signal theory, banks that are expected to have better performance in terms of profitability will credibly transmit this information through a higher equity ratio according to the signal theory. significance of the coefficient of equity in period 3 vindicates the signal theory. npl had a significant negative effect on revenue during period 2 while not being a major determinant during the earlier boom or during the recovery period. nonint has a significant negative effect on interest revenue while having a significant positive effect on total revenue as evidence by the weakening of the financial intermediary function of banks. this coincides with the consolidation of banks and creation of mega banks. mega banks have expanded their business into non-loan-related activities such as security investments. it can be inferred that the japanese city banks in major metropolitan areas have been monopolistically competitive with the exception for period 2 when competition decreased sharply with economic stagnation and financial crisis. 19 table 2 reports the estimated results of equation (4), using both the interest revenue and the total revenue as the dependent variable, for regional banks in japan. there are no noticeable changes in the h-statistic across three different periods, whether the interest rate or the total revenue are used as the dependent variable. the values of the h-statistic are lower than those for city banks in period 1 and period 3. similar results are obtained in regard to three factor input prices: significant and positive effect of the unit fund cost compared to the unit labor cost and the unit capital cost. the network effect table 2: estimation results of equation (4), regional banks 1983-1988 1989-1998 1999-2006 interest revenue total revenue interest revenue total revenue interest revenue total revenue ln w (labor cost) 0.151** (0.042) 0.106** (0.031) 0.131* (0.093) 0.115 (0.116) 0.143** (0.013) 0.141* (0.066) ln c (capital cost) 0.056 (0.129) 0.112** (0.025) 0.042 (0.244) -0.026 (0.321) 0.012 (0.103) 0.053* (0.066) ln f (fund cost) 0.314*** (0.000) 0.319*** (0.000) 0.225* (0.062) 0.323* (0.073) 0.341*** (0.003) 0.264*** (0.004) ln branch 0.279*** (0.002) 0.256*** (0.001) 0.273*** (0.003) 0.213*** (0.001) 0.304*** (0.004) 0.272** (0.012) equity 0.002* (0.091) 0.001 (0.132) 0.006* (0.083) 0.004 (0.117) 0.011* (0.094) 0.007* (0.080) npl -0.026* (0.082) 0.015** (0.034) -0.091** (0.041) -0.044* (0.069) -0.019 (0.104) -0.014* (0.089) nonint -0.191 (0.126) 0.074 (0.135) -0.170 (0.198) 0.051 (0.143) -0.124* (0. 093) 0.111 (0.142) n 792 792 1291 1291 872 872 adj. r2 0.994 0.995 0.982 0.975 0.998 0.996 h statistic 0.521 (0.001) 0.537 (0.001) 0.398 (0.004) 0.412 (0.005) 0.496 (0.001) 0.454 (0.002) wald test: h=0 (ρ) 178.34 (0.000) 181.81 (0.000) 35.29 (0.000) 37.48 (0.000) 151.67 (0.000) 127.83 (0.000) wald test: h=1 (ρ) 167.97 (0.000) 156.42 (0.000) 114.39 (0.000) 101.73 (0.000) 163.34 (0.000) 140.16 (0.000) 1. the fixed effects model is used for estimation. the coefficients of the constant under the fixed effects model are not reported here. 2. h statistic is the sum of ln w, ln c and ln f, and its ρ value is obtained by estimating equation 4. 3. ρ values are shown in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels respectively. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 20 which is present for city banks is also confirmed for regional banks by branch’s positive and significant coefficient. equity has a positive effect on revenue while npl has a negative effect on revenue as expected. nonint has a negative effect on interest revenue while having a positive effect on total revenue. contrast to significant effect of nonint for city banks, it is not a significant variable for regional banks, implying that regional banks are less diversified than city banks. the estimated results of equation (5) for both city banks and regional banks, covering the entire sample period, are presented in table 3. table 3: estimation results of equation (5), dependent variable: net interest margin category variable city banks regional banks bank specific variables equity 0.073 (0.129) 0.022* (0.045) npl -0.022** (0.000) -0.016** (0.000) branch 0.182* (0.031) 0.093 (0.067) nonint -0.011** (0.003) -0.027** (0.000) market power variables market share (assets) -0.012 (0.247) -0.144** (0.000) hhi (assets) 0.925 (0.110) -1.347* (0.046) macro variables gdp growth rate -0.012* (0.042) -0.003 (0.241) inflation rate 0.013 (0.271) 0.022 (0.473) dummy (crisis) -0.136** (0.002) -0.189** (0.001) n 281 2955 adj. r2 0.951 0.978 1. estimation results of fixed effects model. the coefficients of the constant under the fixed effects model are not reported here. 2. ρ values are shown in parentheses. * and ** indicate significance at the 5% and 1% levels respectively. 21 the average interest margin for regional banks is usually higher than that for city banks because interest rates for small and medium enterprises typically serviced by regional banks include larger costs and risk premiums than interest rates for corporate loans typically serviced by national banks. instead of dividing the sample into three separate periods, the equation is estimated for the entire sample period, using a dummy variable, crisis as a breakpoint. equity has a positive effect on net interest margin and this result is consistent with the signaling theory, which suggests that a higher equity ratio increases confidence among bank customers, leading to higher interest spread. the coefficient of equity is significant for regional banks, but not for city banks. npl has a significant negative effect for both city banks and regional banks as expected. loans are the major incomeearning asset of banks and higher percentage of non-performing loans negatively affects the interest spread. branch shows a positive influence on interest spread through bank network effect, particularly for city banks. the negatively significant relationship between nonint and net interest margin indicates that japanese banks, regardless of city banks or regional banks, have diversification effect. the banks with higher level of diversification exhibit lower net interest margin. two market power variables are used: market share of each bank and hhi indicating the overall degree of market concentration. the most troublesome interpretation of the estimated results is related to these two market power variables. contrary to what a typical theory of industrial organization would predict, the market share has a negative effect on net interest margin while hhi renders a mixed result, positive for city banks, but negative for regional banks. the gdp growth rate and inflation rate have positive, though not significant in some cases, effects on the average interest margin. the dummy variable, crisis, is significant in explaining the interest spread. an increase in individual bank’s market share and an increase in overall market concentration are not linked to higher net interest margin in japan. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 22 9. comparison to the korean and chinese experiences even though the korean commercial banking market has become increasingly concentrated in the process of restructuring since the crisis, park (2009) shows that bank competition in korea has not been negatively affected by bank consolidation. the korean banking system may have remained competitive despite its consolidation due to the entry of foreign banks and increased foreign ownership of domestic banks. actually the competition condition test shows competition in the korean banking sector increased to the level of perfect competition during the korean financial crisis period, which is quite the opposite of what happened during japan’s financial trouble time. in korea, the net interest margin is positively correlated with the market share as expected while in japan they are negatively correlated. in spite of these differences, japanese and korean banking industries have a common trend in their merger activities and market concentration. both countries experienced an increase in bank mergers, creation of a few mega banks and an increase in market concentration ratio measured by hhi. the degree of market concentration is much higher in korea than in japan (1,500 vs. 1000 in hhi). both banking markets have a market structure of monopolistic competition. comparing the values of the h-statistic obtained by lee and nagano (2008), park (2009) and this study, it may be stated that the city banks in japan tend to be more competitive than its korean counterparts while the korean regional banks are more competitive than its japanese counterpart. china has a quite different picture from japan and korea. the bank merger wave began in the us and spread to europe, japan and korea. however, this wave has not yet hit the chinese banking market because of its restricted financial openness and government regulation on banking. bank mergers in china were typically governmentinitiated rather than market-initiated. while there has been an increase in market concentration in japan and korea along with an increase in merger activities among banks, china moved from a more concentrated market to a less concentrated market. for china, the competitive conditions of the chinese banking industry have definitely improved over time. the chinese banking system progressed from one bank 23 system to 4 state banks system in the 1980s to more than 20 banks including joint-equity commercial banks in the 1990s to more than 200 banks at the present time. in spite of a drastic decrease in market concentration of the chinese banking industry, its competition conditions are far from a competitive market, as evidenced by the h-statistic. the sheer number of banks does not guarantee a competitive market. lowering entry barriers for private banks and foreign banks would further facilitate competition. institutional changes and lifting government regulations on banking are also necessary to speed up competitive behaviors in the market. 10. conclusion on public policy worldwide financial liberalization and financial globalization is causing fierce competition among banks all over the world. this necessitates bank mergers and consolidation within a country and across countries to achieve scale efficiency, to take advantage of diversification or just to survive. the collusion hypothesis suggests that mergers might be initiated by banks in order to extract consumer surplus, and the result would be higher prices to consumers and socially inefficient allocation of resources. in this case enforcement of antitrust policy may be justified. however, this study does not find evidences to support the collusion hypothesis in japan. recent mergers in the japanese banking sector do not seem to harm the competition level in the banking market, contrary to a growing concern over creation of mega banks. this paper concludes that the bank mergers that took place so far in japan have not led to a higher level of market power except over a short period of financial crisis that started the bubble bursting. this conclusion is also confirmed from the empirical evidence that an increase in individual bank’s market share and an increase in overall market concentration have not materialized in higher net interest margin in japan. further study is encouraged to test if other models produce congruent results for japan. author information: kang h. park is a teaching staff at the southeast missouri state university, united states. he may be contacted via e-mail: khpark@semo.edu. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 24 references baumol, w. j., panzar, j.c., & willig, r. d. (1982). contestable markets and the theory of industry structure. san diego: harcourt brace jovanovich. bikker, j. a., & groeneveld, j. m. (2000). competition and concentration in the eu banking industry. kredit und kapital, 33(1), 62-98. bikker, j. a. & haaf, k. (2002). competition, concentration and their relationships: an empirical analysis of the banking industry. journal of banking and finance, 26, 2191-2214. bikker, j.a., spierdijk, l., & finnie, p. (2006). misspecification in the panzar-rosse mdel: assessing competition in the banking industry, dnb working paper 114 (www.dnb.nl). bresnahan, t. f. (1982). the oligopoly solution concept is identified. economics letters, 10, 87-92. bresnahan, t. f. (1989). empirical studies of industries with market power. in r. schmalensee, & r. d. willig (eds.), handbook of industrial organization (vol. ii, chapter 17, pp. 1011-1057). new york: north-holland. luc, can and ariff, m., (2008) cost and profit efficiency of banks: a non-parametric analysis, china economic review, vol 19: 260-278. casu, b., girardone, c. and molyneux, p. (2006), introduction to banking. harlow: ft prentice hall. coccorese, p. (1998). assessing the competitive conditions in the italian banking system: some empirical evidence. bnl quarterly review, 205, 171-191. de bandt, o., & davis, e. p. (2000). competition, contestability and market structure in european banking sectors on the eve of emu. journal of banking and finance, 24(6), 1045-1066. gelos, r. g., & roldos, j. (2002). consolidation and market structure in emerging market banking systems. imf working paper, no. 02/186. gilbert, r. a. (1984). bank market structure and competition: a survey. journal of money, credit, and banking 16 (4 part 2), 617-644. hempell, h. s. (2002). testing for competition among german banks. discussion paper, economic research center for the deutsche bundesbank. hondroyiannis, g., lolos, s., & papapetrou, e. (1999). assessing competitive conditions in the greek banking system, journal of international financial markets, institutions and money, 9, 377-391. hosono, k, sakai, k. & tsuru, k. (2007). consolidation of banks in japan: causes and conequences. rieti discussion paper series 07-e-059. japanese bankers association. (2006). the banking system in japan. tokyo: zenginkyo. kim, b., chun, s. & owe,o. (2004). an effect of financial adjustment on the degree of competition in the korean banking industry. industrial economic analysis, 17, 55–80 (in korean). kim, h. (2003). assessing concentration and competition in the korean banking industry (in korean). kdi policy studies, 25, 57-98. lau, l. (1982). on identifying the degree of competitiveness from industry price and output data. economics letters, 10, 93-99. lee, b. (2003). the effect of financial restructuring on the degree of competition of the http://www.dnb.nl/ 25 korean banking sector. economic analysis, 9, 75–113 (in korean). lee, y. and kim, s. (1995). an effect of new entry to banking competition. economic analysis, 43, 165–82 (in korean). lee, m. & nagano, m. (2008). market competition before and after bank merger wave: a comparative study of korea and japan. pacific economic review, 13, 604-619. lee s. & lee, j. (2005). bank consolidation and bank competition: an empirical analysis of the korean banking industry (in korean). economic papers, 8, 102-144. liu, h & wilson, j. (2009). the profitability of banks in japan: the road to recovery? working paper series 06/09. mason, e. s. (1939). price and production policies of large-scale enterprise. american economic review, 29, 61-74. molyneux, p., lloyd-williams, d. m., & thornton, j. (1994). competitive conditions in european banking, journal of banking and finance, 18, 445-459. molyneux, p., thornton, j., & lloyd-williams, d. m. (1996). competition and market contestability in japanese commercial banking. journal of economics and business, 48(1), 33-45. nathan, a., & neave, e. h. (1989). competition and contestability in canada’s financial system: empirical results. canadian journal of economics, 22(3), 576-594. park, k. (2009). has bank consolidation in korea lessened competition? quarterly review of economics and finance, 49, 2009, 651-667. park, k. (2011). bank competition and concentration: a comparative study of south korea and china. working paper, southeast missouri state university. panzar, j. c., & rosse, j. n. (1982). structure, conduct and comparative statics. bell laboratories economic discussion paper. panzar, j. c., & rosse, j. n. (1987). testing for ‘monopoly’ equilibrium. journal of industrial economics, 35, 443-456. shaffer, s. (1982). a nonstructural test for competition in financial markets. proceedings of a conference on bank structure and competition. federal reserve bank of chicago, 225-243. shaffer, s. (2002). competitive bank pricing and adverse selection, with implications for testing the scp hypothesis. quarterly review of economics and finance, 42(3), 633-647. shaffer, s. (2004). patterns of competition in banking. journal of economics & business, 56, 287-313. smirlock, m. (1985). evidence on the (non) relationship between concentration and profitability in banking. journal of money, credit, and banking, 17, 69-83. uchida, h. & tsutsui, y. (2005). has competition in the japanese banking sector improved? journal of banking and finance 29, 419-439. uchida, h. and udell, g. (2009). banking in japan. in berger, a.n., molyneux, p. and wilson, j.o.s. (eds.) oxford handbook of banking. oxford: oxford university press. vesala, j. (1995). testing for competition in banking behavioral evidence from finland. helsinki: bank of finland. weiss, l. w. (1989). a review of concentration-price studies in banking. in l. w. weiss (ed.), concentration and price. cambridge, ma: mit press, 219-254. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 1 cetak a4 ijbf6(2)09.indd the international journal of banking and finance, 2008/09 vol. 6. number 2: 2009: 37-59 37 ijbf the role of bank loans and deposits in the monetary transmission mechanism in malaysia salina kassim and m. shabri a. majid international islamic university of malaysia abstract this study attempts to determine the importance of the banking sector in the monetary transmission process in a developing economy. the study analyzes the malaysian data focusing on three sample periods: the entire sample period (1989:01-2006:12); the pre-crisis period (1989:01-1996:12); and the post-crisis period (1999:01-2006:12). to achieve this objective, the study relies on two tests: fi rst, the auto-regressive distributed lag (ardl) model for the long-run relationship among the variables and second, the impulse response functions and variance decomposition analysis for the short-run relationship among the variables. the fi nding shows that both bank deposits and loans play crucial roles in the monetary transmission process in the economy, suggesting evidence for the money endogeneity theory of post-keynesian economists. in particular, bank deposits and loans are shown to provide an important link from monetary policy to output. this underscores the importance of ensuring the soundness of banking system as a pre-requisite to economic stability in the absence of such market based tools as market-based actions on exchange rate or interest rates as monetary stabilisation tools. keywords: monetary transmission, bank loans, bank deposits, auto-regressive distributed lag model; impulse response functions, variance decompositions jel classifi cations: e42, e51, g21 1. introduction while it is widely accepted that monetary policy affects macroeconomic outcomes in an economy, a central point of a continuing debate among the policy-making circles is on the transmission mechanism of monetary policy to the real economy. the debate arises from the importance of identifying the key link(s) since such link(s) will determine the effective instrument and reliable intermediate targets for monetary policy implementation. in view of this, ht tp :// ijb f.u um .e du .m y 38 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 identifying the relevant channels through which the impact of monetary policy is transmitted to the real economy has been one of the most challenging and crucial tasks in the conduct of monetary policy. a clear understanding of the monetary transmission channel can help in ensuring an effective and successful implementation of monetary policy. besides, the post-keynesian economists have emphasized the role of banking credits as the key transmission mechanism, although this paper is not directly testing that hypothesis. this aspect has yet been seriously studied, and the only study that exists of a developing country is that of turkey. monetary policy affects the real economy via several channels, but the main two that are often mentioned in the literature are the “money channel” and “credit channel”. the money channel, which emphasises the role of central bank’s money in the monetary transmission process, has secured a strong footing in the economic literature (see, for example, brunner and meltzer, 1972, 1988, 1990; mccallum, 1990, 1999, 2001, 2004; mccallum and nelson, 2005; feldstein and stock, 1993; hess and porter, 1993; dotsey and otrok, 1994; meltzer, 1969, 1970, 2001; christiano and rostagno, 2001; and nelson, 2003a, 2003b). according to this view, an expansionary monetary policy causes bank deposit and money supply to expand, which results in real interest rate to decline, thus it is also known as the interest rate channel. studies that are supportive of the existence and importance of the bank lending channel in general fi nd that, bank loan behave in a predictable manner when a policy shock is exerted on the economy. when a central bank conducts open market operations to reduce liquidity in the system, bank’s deposit and liquidity are reduced, thereby restricting the ability of banks to supply new loans. credit is less available in the economy, resulting in fi rms to cut down on investment and production, thus lower aggregate output in the economy. on the other hand, studies that do not support the importance of the bank lending channel generally fi nd that banks offset the decline in liquidity following a tight monetary policy by adjusting their balance sheet items. such ability to do so renders monetary policy ineffective in infl uencing supply of bank loan remaining relatively unchanged even after the implementation of the tight policy stance. for example, suzuki (2001) employs the var analysis to examine the bank lending channel using the australian data and report that bank loans contracted following a tight monetary policy, the contraction. studies by morris and sellon (1995) which used us data for 1974-1990 also show that bank loan is not an important channel for monetary policy since bank loan supply does not decline in response to tight monetary policy. earlier studies such as scholnick (1996) and tan (1997) on malaysian situation adopted the credit-rationing model and estimated the disequilibrium model of lending rate. both studies support the presence of credit rationing as the lending rate adjusted very slowly and did not clear the loan market. more recent studies such as ibrahim (2005), azali (2003), tang (2000, 2002) and vaithilingam et al. (2003) employed more recent investigation techniques such as the var that enable more detailed and conclusive fi ndings. in general, ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 39 these studies are supportive of the important role played by bank loan in the transmission of monetary policy. the need to understand the interaction between bank loan and bank deposit with monetary policy indicator is critical, hence this study is motivated to address this issue. bank lending accounted for more than 100 percent of the country’s total gross domestic product (gdp) and is expected to remain high due to the moderate economic performance. indeed, the important role played by bank lending in achieving the overall economic objective is clearly acknowledged by bank negara malaysia. “as the nation’s monetary authority, the central bank is responsible to the government for promoting monetary stability and a sound fi nancial structure, and for infl uencing the credit situation to help achieve the nation’s overall economic objectives.” (1994, p. 91). an aspect of novelty of this study is that not only that it uses the latest empirical methodology, it also considers two sub-sample periods following the crisis in 1997/1998. the rest of this paper is organized as follows: the next section provides some background information on the monetary transmission process. section 3 discusses the data and empirical framework. section 4 presents the empirical fi ndings and the discussions of the results. lastly, section 5 concludes this paper. 2. an overview of related literature empirical studies analyzing the importance of the banking sector in the monetary transmission process focus on the relationship and impact of disturbances in the banking sector on the aggregate economic activity: see, for example friedman,1984; bernanke and blinder, 1988 and 1992; bernanke and gertler, 1995; bernanke and mihov, 1998; dale and haldane, 1998; domac and ferri, 1998; garretsen and swank, 1998; guender, 1998; gertler and gilchrist, 1993; walsh and wilcox, 1995 and suzuki, 2001. in general, these studies measure the responses of selected macroeconomic variables such as output, unemployment rate and infl ation and the major bank balance sheet variables, which are deposit (representing bank liability) and loan (representing bank asset) to changes in the monetary policy indicator such as interest rates. however, it is often shown that the macroeconomic responses to policyinduced interest rate changes are larger than that implied by the money view. this gives rise to the lending channel, which suggests that bank loans may provide an additional channel for the monetary authority to affect the real economy. an infl uential study by bernanke (1983) fi nds that the output decline in the us during the 1930-1933 great depression cannot be fully explained by monetary disturbances as described by the money view. using the federal funds rate as an indicator for monetary policy, that study analyzed the responses of selected macroeconomic variables such as unemployment rate and infl ation, and the bank balance sheet items to changes in the fed funds rate. it was found that in the short run, banks offset the decline in liquidity following a contractionary monetary policy by selling-off their securities holdings, thereby leaving the supply of ht tp :// ijb f.u um .e du .m y 40 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 loan in the economy virtually unchanged. however, in the longer term, the tight monetary policy causes banks to terminate the existing loans and stop making new loans. the brunt of the tight monetary policy is especially felt by borrowers who are dependent on bank lending. consequently, the tighter monetary policy results in a decline in the aggregate economic output. until now, the reduction in loan supply has often been quoted as one of the reasons for the depth and persistence of the great depression. as lessons learned from that crisis, the central banks the world over are loosening monetary policy since august 2007 in response to the sub-prime loan crisis of 2007/08. in malaysia, the banking system forms an integral part of a larger fi nancial system which is made up of the fi nancial institutions and the fi nancial markets. the fi nancial institutions can be grouped into the banking system and the nonbank fi nancial intermediaries. the banking system is made up of commercial banks, fi nance companies and merchant banks, while the non-bank fi nancial intermediaries comprised the provident and pension funds, insurance companies, development fi nancial institutions, savings institutions and other fi nancial intermediaries (such as cooperative societies, leasing and factoring companies, venture capital companies and housing credit institutions). the operations of malaysia’s banking system fall under the purview of the central bank, the bank negara malaysia (bnm). the fi nancial system experienced a rapid growth in tandem with the robust economic performance during 1976-1996. by the end of 2006, the total assets of the fi nancial system was about equal to the gdp of the country and it was rm2,091.2 or us$ 635 billion compared to rm329.3 billion in 1990 and rm74.2 billion in 1980. the rapid growth fi nancial sector refl ects the process of fi nancial deepening due to increased demands for fi nancial products and services in line with the structural transformation of the economy. more importantly, the banking system remained a major contributor to the total assets of the country’s fi nancial system. of the total assets of the fi nancial system at end-2006, 68 percent were contributed by the banking system while 32 by the non-bank fi nancial intermediaries. in terms of growth, the total assets of the banking system grew by approximately 11.3 percent in the period of 1990-2006, comparable to that of the fi nancial system at 11.5 percent. apart from its signifi cant contribution to the growth of the country’s fi nancial system, the banking system has also been instrumental in fi nancing the country’s economic growth. this is well-refl ected by the high ratio of total loan extended by the banking institutions to total gross domestic product (gdp) at more than 100 percent as at end-2006. one important lesson learned from the 1997/98 asian fi nancial crisis was that fragility of the banking sector could lead to a loss of confi dence which would affect the real economy negatively. in view of this, bnm banking policies in the post-crisis period are directed towards strengthening and increasing the resilience of the banking system as well as to preserve the integrity and stability of fi nancial sector. in this regard, one of the major initiatives taken was to consolidate the banking institutions in order to create a core of strong domestic banking institutions with large capital base. prior to the consolidation program, ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 41 as at end-1997, the banking system comprised 35 commercial banks, 39 fi nance companies and 12 merchant banks. as a result of the consolidation program, as at end-2006, the number of commercial banks has been reduced to 22 operating within ten holding companies, with the operations of the fi nance companies fully being absorbed by the commercial banks and the merchant banks being re-categorized into the 14 investment banks comprising the merchant banks and discount houses. this led to a strong capital base of risk-weighted capital ratio at 13.5 percent as at end-2006. the non-performing loan ratio declined from above 13.9 percent in 1998 to 3.7 percent. 3. data and methodology a. data the overnight rate, henceforth denoted as onr is used as the monetary policy indicator, while the objective variables comprised of the bank balance sheet items, namely, bank loans (bl) and bank deposits (bd) and other variables which are consumer price index (cpi) and industrial production index (ipi). the cpi and the ipi are used as the proxies for infl ation and level of economic activity, respectively. given that the economy is a highly open one, the conduct of monetary policy may be infl uenced by foreign shocks, thus we also include real effective exchange rate (rer) as a control variable. all series are real (adjusted by price index with the year 2000 as base year) and variables are expressed in logarithm, except for the interest rate. the study uses monthly data from january 1989 to december 2006. during this period, the economy experienced a fi nancial crisis. to avoid the disturbance from affecting the results of the analysis, tests are done over separate subperiods and also over the whole period. the pre-crisis period covers the period from january 1989 to december 1996. in this period, the data set excludes the period from 1997 to 1998, which were the heights of the east asian fi nancial crisis. the post-crisis period runs from january 1999 to december 2006. all data are sourced from bnm’s monthly statistical bulletin, except for the rer which is gathered from international financial statistics published by international monetary fund. b. empirical framework ardl bound testing approach the long-run dynamic relationships between monetary policy variables and bank balance sheet items are estimated by using the newly proposed ardl bound testing approach: pesaran et al. (1996). the ardl has numerous advantages. first, unlike the most widely used method for testing cointegration, the ardl approach can be applied regardless of the stationary properties of the variables in the samples and it allows for inferences on long-run estimates, which is not possible under the alternative cointegration procedures. this procedure can be applied irrespective of whether the series are i(0), i(1), or fractionally integrated (pesaran and pesaran 1997; and bahmani-oskooee and wing ng, 2002), thus ht tp :// ijb f.u um .e du .m y 42 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 avoiding problems resulting from non-stationary time series data (laurenceson and chai, 2003). second, the ardl model takes suffi cient numbers of lags to capture the data generating process in a general-to-specifi c modelling framework (laurenceson and chai, 2003). it estimates (p+1)k number of regressions in order to obtain optimal lag-length for each variable, where p is the maximum lag to be used, k is the number of variables in the equation. finally, the ardl approach provides robust results for a smaller sample size of cointegration analysis. since the sample size of our study is small, particularly for preand post-crisis subperiods, this provides a motivation for the study to adopt this model. ideally, for the purpose of this study, we needed to incorporate all the variables in the model, but a var model can be poorly estimated in a fi nite sample, as the addition of a variable will quickly exhaust the degree of freedom. following gertler and gilchrist (1993, 1994), we estimated a series of separate var models including inf, ipi, and rer, policy variable (onr), and the bank balance sheet items. thus, our models contain only fi ve variables. the ardl models for the bank balance sheet items used in this study can be written as follows: (1.1) (1.2) the error correction version of ardl framework pertaining to the variables in the equations (1.1) and (1.2) can be reproduced as follows: (2.1) (2.2) in the above equations, the terms with the summation signs represent the error correction dynamics, while the second part (term with λs) correspond to the long-run relationship. the null of no cointegration in the long-run relationship is defi ned by h 0 : λ 1 ≠ λ 2 ≠ λ 3 ≠λ 4 ≠ λ 5 = 0 which is tested against the alternative of h 0 : λ 1 ≠ λ 2 ≠ λ 3 ≠λ 4 ≠ λ 5 ≠ 0, by the means of familiar f-test. however, the asymptotic distribution of this f-statistic is non-standard irrespective of whether the variables are i(0) or i(1). pesaran et al. (1996) have tabulated two sets of appropriate critical values. one set assumes all variables are i(1) and another assumes that they are all i(0). this provides a bound covering all possible classifi cations of the variables into i(1) and i(0) or even fractionally integrated. if the f-statistic lies above the upper-bound level, the null hypothesis is rejected, which indicates the existence of cointegration. however, if the f-statistic falls below the bound level, the null cannot be rejected, showing that no cointegration bdt = 0 + 1onrt + 2ipit + 3rert + 4inft + €t blt = 0 + 1onrt + 2ipit + 3rert + 4inft + t p 0i p 0i it p 0i iit p 0i iitiitiit p 1i i0t infreripionrbdbd t11t51t41t31t21t1 infreripionrbd p 0i p 0i it p 0i iit p 0i iitiitiit p 1i i0t infreripionrblbl t11t51t41t31t21t1 infreripionrbl ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 43 exists. if, however, it falls within the band, the result is inconclusive. finally, in order to determine the optimal lag-length incorporated into the model and select the ardl model to be estimated, the study employs the akaike information criteria (aic). impulse response functions and variance decompositions we estimate var models and generate impulse-response functions to study the impacts of interest rate shocks on the systems as indicated in equations 1.1 and 1.2. an impulse response function (irf) measures the time profi le of the effect of shocks at a given point in time on the (expected) future values of variables in a dynamical system (pesaran and shin, 1998). the approach is wellsuited because not only that it allows for the relative strength of various shocks to be quantifi ed in terms of their contributions to variations in a particular variable of interest, but it also enables the pattern and direction of the transmission of shocks to be traced. meanwhile, the variance decomposition analysis (vda) shows the fraction of forecast error variance of a variable attributed to shocks in other variable particularly to inference of relative strength of innovations in the variable of interest. it is a method of providing a literal breakdown of the change in the value of a variable in a given period arising from changes in the same variable and in other variables during previous periods. variance decomposition analysis provides an indication of the dynamic properties of the system. vdc, which is termed an out-of-sample causality tests, by partitioning the variance of forecast error of a certain variable into proportions attributable to innovations (or shocks) in each variable in the system including its own, can provide an indication of these relatives. according to sims (1992), a variable optimally forecast from its own lagged values will have all its forecast error variance accounted for by its own disturbances. it is generally observed that in applied research, it is typical for a variance to explain almost all its forecast error variance at short horizons and smaller proportions at longer horizons. 4. empirical results and analysis a. results based on the ardl model in estimating the long-run relationships between the monetary policy variable and the objective variables (consisting of the bank balance sheet items and the selected macroeconomic variables), we needed to determine the lag-length of the fi rst-differenced variables. bahmani-oskooee and bohl (2000) have shown that the results of this fi rst step are usually sensitive to the lag-length. to verify this, we imposed the optimal lag length of 12 on the fi rst difference of each variable to compute the f-statistics for the joint signifi cance of lagged levels of variables for equations (2.1) and (2.2). however, only selected computed f-statistics for lag-lengths for all models are reported as in table 1. as reported in table 1, the test outcome of the signifi cance levels for the ardl models varies with the choice of lag-length. for the entire sample ht tp :// ijb f.u um .e du .m y 44 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 period, the computed f-statistics are signifi cant at least at 0.90 level when the order of lags ranges from 1 to 9 for the deposit and at 0.99 level when the order of lags ranges from 1 to 3 for loan. this suggests that there are cointegrating relationship among the selected variables with bank loans and deposits in the whole sample period, implying that there is a tendency for the variables to move together towards a long-run equilibrium. however, a breakdown of the whole sample period to pre-crisis period and post-crisis period shows somewhat puzzling results. for the pre-crisis period, the computed f-statistics is found to be signifi cant when the order of lags = 1 and 3 at the 90% level for deposits but not for loans. similarly, in the post-crisis period, only the computed f-statistic for deposit is found to be signifi cant at 0.05 probability level at lag-length 3 and 5, but not for loan. this suggests that in the preand post-crisis periods, there are cointegrating relationships among the selected variables and deposit, thus the deposit tend to move together towards a long-run equilibrium with other selected variables, but not loans. these results are considered preliminary and indicate that in estimating equations (1.1) and (1.2) we must retain the lagged level of variables. table 1: f-statistics for testing the existence of a long-run equation lag length entire sample (1989:1 to 2006:12) pre-crisis period (1989:1 to 1996:12) post-crisis period (1999:1 to 2006:12) deposit loan deposit loan deposit loan 1 6.2445*** 11.4565*** 4.2705** 3.2333 2.2765 2.3551 3 4.8321** 7.8294*** 3.5698* 1.0575 4.2278** 1.2158 5 5.0051** 6.0578*** 1.6817 1.9647 4.3039** 1.8131 7 4.4483* 3.6087 2.3935 3.0527 3.3626 1.4383 9 5.3777** 2.9743 3.4045 2.4663 2.9822 2.3469 11 4.0312 2.432 3.001 2.1263 2.324 1.6912 note: the relevant critical value bounds for the preand post-crisis periods are given in table c1.iii (with unrestricted intercept and no trend; the number of regressors = 4). they are 3.74 – 5.06 at the 99% signifi cance level; 2.86 – 4.01 at the 95% signifi cance level; and 2.45 – 3.52 at the 99% signifi cance level (see table c1.iii, pesaran et al., 2001). meanwhile, the relevant critical value bounds for the entire sample period are given in table c1.v (with unrestricted intercept and unrestricted trend; the number of regressors = 4). they are 4.40 – 5.72 at the 99% signifi cance level; 3.47 – 4.57 at the 95% signifi cance level; and 3.03 – 4.06 at the 99% signifi cance level (see table c1.v, pesaran et al., 2001). *, **, and *** denotes that f-statistics falls above the 90%, 95% and 99% upper bound, respectively. ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 45 the fi ndings from this section suggests that there seems to be a long run cointegration among the bank deposit (bd), overnight interest rate (onr), industrial production index (ipi), infl ation (cpi), and real effective exchange rates (rer) during all periods of the study. however, the bank loan are only found to be cointegrated with other selected variables during the entire period of the study, while no cointegration is found during the preand post-crisis periods. for further analysis, the study only focuses on the cointegrated variables. the next step involves estimating equations (2.1) and (2.2) using the appropriate lag-length and model selection criterion based on the akaike information criterion (aic). note that the loan variable is only included in the entire sample period but not in the pre-and post-crisis sample periods based on the earlier fi ndings that there is no long run cointegration relationship between loan and the rest of the variables in the preand post crisis periods. as suggested in table 1, the appropriate lag length for the estimated equations is 1 for precrisis and entire sample periods and 5 for post-crisis period. the study simply selects the order of lags for the estimated models based on the higher value of the computed f-statistics. table 2: the long-run ardl model estimates entire sample (1989:1 to 2006:12) pre-crisis period (1989:1 to 1996:12) post-crisis period (1999:1 to 2006:12) deposit [1,1,0,0,0] loan [1,0,0,1,0] deposit [1,1,0,0,0] deposit [1,2,0,1,0] c -4.4052 (-0.6233) 14.1914 (0.4186) -11.1661 (-0.9931) 2.8885 (0.9479) onr 0.0788** (2.4103) 0.0442 (0.8610) -0.0729 (-0.7338) 0.0978 (1.4564) lnipi 1.9806* (1.8358) 7.9258 (1.0274) -1.7030 (-0.5643) -.10683 (-0.5144) lninf 0.1594 (0.0606) -10.6086 (-.6009) 8.4568 (1.1535) 3.7116*** (4.9072) lnrer 1.3481 (1.4034) 2.0268 (0.7363) -1.2292 (-0.4168) -1.4554*** (-3.4990) adj-r2 = 0.998 d-w = 1.977 adj-r2 = 0.996 d-w = 1.993 adj-r2 = 0.995 d-w = 1.7534 adj-r2 = 0.997 d-w = 2.066 note: *, ** and *** denotes signifi cantly at 10%, 5% and 1% level of signifi cance, respectively. figures in the parentheses and squared parentheses are the t-statistics values and the selected ardl model. d-w denotes durbin-watson test for autocorrelation. dummy variable is included in the entire sample period; d = 0 for the non-crisis period (preand post-crisis periods), while d = 1 is incorporated during the crisis period. ht tp :// ijb f.u um .e du .m y 46 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 the results for the long run estimates based on the ardl model are provided in table 2. for the entire sample period, the results provide evidence that deposits are signifi cantly affected by onr and ipi. in the context of this study, we highlight the signifi cant and positive effect of monetary policy variable, the overnight interest rate, on deposits in the long run. however, none of the variables including the monetary policy variable is signifi cant in affecting loans in entire sample period. in the context of monetary transmission mechanism, this result implies that monetary policy shock is transmitted through bank deposits (the money view) and not through bank loan (the credit view). test results from a breakdown of the entire sample period into the pre and post-crisis periods show that monetary policy variable is not signifi cant in affecting deposits in both the periods. in particular, none of the selected variables are signifi cant in affecting deposits in the pre-crisis period. in contrast, in the post-crisis period, infl ation and real exchange rate are signifi cant in affecting deposit in the long run. b. results based on the impulse response functions the impulse response functions (irf) allow for the analysis of the impacts of interest rate shocks on the bank balance sheet items of both banking groups. the irf shows the magnitude and timing of the responses of the objective variables (the bank balance sheet items) to a shock in the interest rate variable. this enables a comparison of the extent of responses of the bank balance sheet items to the policy shocks. in all cases, impulse response functions are reported for 36 months. to provide some idea of uncertainty surrounding the estimated response, based on sim and zha (1995), one standard deviation of confi dence bands have been obtained by monte carlo integration methods with 1000 replications. figure 1 provides the impulse response functions of the responses of the macroeconomic variables to shocks in bank deposit and bank loan for the entire sample period (1989:1 to 2006:12). in terms of the relevance of the bank balance sheet items to monetary policy shocks, the short run analysis indicates that bank deposit responds positively and signifi cantly to changes in the policy interest rate. the impact started to be signifi cant in the fi fth month and remained signifi cant until about the 25th month. in contrast, bank loan does not respond signifi cantly to changes in the overnight rates. however, the monetary policy indicator seems to respond positively and signifi cantly to changes in bank loan, implying that the monetary authority reacts to changes in bank loan. bank loan, therefore, can be termed as the “trigger variable” for monetary policy. with respect to the responses of the macroeconomic variables to changes in bank deposit and bank loan, data for the whole sample period indicates that the only signifi cant response is that of real output (denoted by ipi) to changes in bank loan, where output increases following a positive innovations in bank loan. this fi nding implies that bank loan is important in infl uencing monetary policy direction and economic output in the short run. ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 47 to gain some insights of the changes in the relationship among the variables in the period before and after the asian fi nancial crisis in 1997/1998, the whole sample period is being divided into the pre-crisis and post-crisis periods. figure 2 shows the irf results for the pre-crisis period. as shown in the fi gure, there seems to be a puzzling result when bank deposit responds negatively signifi cant to changes in the policy interest rate. there is also a signifi cant negative response of onr to changes in bank deposit, implying in the pre-crisis period, monetary policy has primarily been used in respond to changes in deposit. bank deposit, however, is insignifi cant in affecting other macroeconomic variables in the short run in the pre-crisis period. turning to bank loan, clearly, bank loan is insignifi cant in the implementation of monetary policy in the pre-crisis period as refl ected by the insignifi cant responses of either bank loan to onr and onr to bank loan. however, changes in bank loan are shown to have signifi cant impacts on ipi (ipi responds positively and signifi cantly to changes in bl), cpi (cpi responds positively and signifi cantly to changes in bl) and rer (rer responds negatively and signifi cantly to changes in bl). figure 3 provides the irf results for the post-crisis period. during this period, bank loan seems to play a lesser role in the transmission process of monetary policy, with the exception of bank deposit which seems to respond positively signifi cant to changes in onr in the fi rst threeto four-month period. despite this, bank deposit seems to be signifi cant in infl uencing cpi as the irf shows that the cpi respons signifi cantly positive to changes in bank deposit. there are still evidences that bank loan is signifi cant in infl uencing output and a clear indication that bank loan is signifi cant in infl uencing real exchange rate as shown by the positive and signifi cant irf. figure 1: generalized impulse-response functions, entire sample-1989:1 to 2006:12 -.02 -.01 .00 .01 .02 .03 5 10 15 20 25 30 35 response of lbd to onr -.02 -.01 .00 .01 .02 .03 5 10 15 20 25 30 35 response of lbl to onr -.4 -.2 .0 .2 .4 .6 .8 5 10 15 20 25 30 35 response of onr to lbd -.4 -.2 .0 .2 .4 .6 .8 5 10 15 20 25 30 35 response of onr to lbl -.02 -.01 .00 .01 .02 .03 .04 .05 5 10 15 20 25 30 35 response of lipi to lbd -.02 -.01 .00 .01 .02 .03 .04 .05 5 10 15 20 25 30 35 response of lipi to lbl -.002 -.001 .000 .001 .002 .003 .004 5 10 15 20 25 30 35 response of lcpi to lbd -.002 -.001 .000 .001 .002 .003 .004 5 10 15 20 25 30 35 response of lcpi to lbl -2 -1 0 1 2 3 5 10 15 20 25 30 35 response of rer to lbd -2 -1 0 1 2 3 5 10 15 20 25 30 35 response of rer to lbl ht tp :// ijb f.u um .e du .m y 48 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 figure 2: generalized impulse-response functions pre-crisis period-1989:1 to 1996:12 figure 3: generalized impulse-response functions post-crisis period-1999:1 to 2006:12 -.04 -.03 -.02 -.01 .00 .01 .02 .03 .04 .05 5 10 15 20 25 30 35 response of lbd to onr -.04 -.02 .00 .02 .04 .06 5 10 15 20 25 30 35 response of lbl to onr -0.8 -0.4 0.0 0.4 0.8 1.2 5 10 15 20 25 30 35 response of onr to lbd -0.8 -0.4 0.0 0.4 0.8 1.2 5 10 15 20 25 30 35 response of onr to lbl -.03 -.02 -.01 .00 .01 .02 .03 .04 .05 5 10 15 20 25 30 35 response of lipi to lbd -.03 -.02 -.01 .00 .01 .02 .03 .04 .05 5 10 15 20 25 30 35 response of lipi to lbl -.006 -.004 -.002 .000 .002 .004 .006 .008 5 10 15 20 25 30 35 response of lcpi to lbd -.006 -.004 -.002 .000 .002 .004 .006 .008 5 10 15 20 25 30 35 response of lcpi to lbl -2 -1 0 1 2 3 5 10 15 20 25 30 35 response of rer to lbd -2 -1 0 1 2 3 5 10 15 20 25 30 35 response of rer to lbl -.03 -.02 -.01 .00 .01 .02 5 10 15 20 25 30 35 response of lbd to onr -.010 -.005 .000 .005 .010 .015 .020 5 10 15 20 25 30 35 response of lbl to onr -.08 -.04 .00 .04 .08 .12 .16 .20 5 10 15 20 25 30 35 response of onr to lbd -.08 -.04 .00 .04 .08 .12 .16 .20 5 10 15 20 25 30 35 response of onr to lbl -.02 -.01 .00 .01 .02 .03 .04 .05 5 10 15 20 25 30 35 response of lipi to lbd -.02 -.01 .00 .01 .02 .03 .04 .05 5 10 15 20 25 30 35 response of lipi to lbl -.006 -.004 -.002 .000 .002 .004 .006 5 10 15 20 25 30 35 response of lcpi to lbd -.006 -.004 -.002 .000 .002 .004 .006 5 10 15 20 25 30 35 response of lcpi to lbl -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2 1.6 5 10 15 20 25 30 35 response of rer to lbd -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2 1.6 5 10 15 20 25 30 35 response of rer to lbl ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 49 t ab le 3 : v ar ia nc e d ec om po si ti on , q 1: 19 89 t o q 2: 20 06 ( co n ti n u ie d ) h or iz on (m on th ly ) e xp la in ed b y sh oc k s in b d o n : b d o n r b d o n r b d ip i b d c p i b d r e r 1 10 0. 00 0 0. 00 0 0. 34 5 99 .6 55 0. 26 0 99 .1 87 0. 02 7 95 .6 73 0. 10 4 89 .1 56 2 99 .3 82 0. 46 8 0. 62 4 98 .2 12 0. 42 0 93 .9 80 0. 39 0 95 .9 22 0. 13 4 84 .2 38 3 98 .3 05 0. 87 6 0. 60 2 96 .5 54 0. 98 3 92 .8 47 0. 54 2 95 .6 88 0. 10 9 82 .5 06 6 91 .4 32 3. 86 0 0. 63 0 91 .2 26 1. 73 2 87 .5 46 0. 94 2 92 .2 31 0. 27 2 78 .4 25 9 81 .2 46 7. 77 7 0. 55 5 85 .0 72 2. 08 5 81 .7 04 1. 49 2 85 .9 87 1. 16 1 74 .7 82 12 70 .5 38 11 .6 77 0. 49 9 79 .7 09 2. 13 9 75 .6 65 2. 19 0 79 .0 42 2. 61 7 71 .3 29 16 57 .7 76 16 .1 91 0. 47 4 74 .0 70 2. 00 4 68 .2 50 3. 23 9 70 .5 80 5. 03 5 67 .2 82 19 49 .8 84 18 .9 62 0. 50 5 70 .7 75 1. 86 8 63 .4 31 4. 01 3 65 .3 14 6. 92 0 64 .8 18 23 41 .5 65 21 .9 22 0. 64 4 67 .3 03 1. 70 1 58 .0 43 4. 90 2 59 .7 43 9. 20 0 62 .3 78 27 35 .3 84 24 .1 95 0. 92 4 64 .6 46 1. 56 4 53 .7 75 5. 55 0 55 .5 76 10 .9 85 60 .8 31 31 30 .8 65 25 .9 25 1. 34 8 62 .6 41 1. 46 6 50 .4 97 5. 93 7 52 .5 07 12 .1 92 59 .9 72 36 26 .9 08 27 .4 95 2. 02 1 60 .8 66 1. 42 5 47 .5 27 6. 10 4 49 .8 25 12 .9 80 59 .4 85 t a b le co n ti n u es o n t h e n ex t p a g e ht tp :// ijb f.u um .e du .m y 50 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 h or iz on (m on th ly ) e xp la in ed b y sh oc k s in b l o n : b l o n r b l o n r b l ip i b l c p i b l r e r 1 10 0. 00 0 0. 00 0 0. 05 6 99 .9 44 1. 20 3 98 .3 72 0. 28 7 95 .8 30 0. 01 1 89 .2 36 2 99 .2 15 0. 03 6 0. 26 1 98 .8 71 5. 43 3 92 .0 37 0. 24 7 96 .5 67 0. 27 2 84 .4 68 3 96 .9 92 0. 12 3 0. 38 6 97 .2 31 5. 62 6 91 .0 18 0. 19 8 96 .5 18 0. 27 5 82 .9 94 6 88 .5 79 0. 46 8 1. 65 1 91 .9 34 6. 47 6 85 .5 60 0. 13 7 93 .2 14 0. 16 7 79 .4 29 9 80 .4 68 0. 81 3 3. 12 7 85 .7 03 6. 07 8 80 .1 33 0. 31 9 86 .9 95 0. 21 6 76 .4 57 12 73 .5 09 1. 14 7 4. 41 7 80 .2 82 5. 57 9 74 .5 98 0. 70 6 80 .1 58 0. 43 6 73 .7 77 16 65 .6 59 1. 62 1 5. 57 1 74 .6 30 5. 16 4 67 .7 80 1. 38 9 72 .0 47 0. 92 4 70 .6 65 19 60 .6 09 2. 02 9 6. 01 3 71 .3 62 5. 03 0 63 .3 11 1. 92 8 67 .1 62 1. 38 4 68 .6 93 23 54 .8 27 2. 66 5 6. 18 3 67 .9 68 4. 94 1 58 .2 76 2. 59 0 62 .1 65 2. 05 7 66 .5 49 27 50 .0 17 3. 41 8 6. 08 0 65 .4 40 4. 83 5 54 .2 88 3. 13 6 58 .5 57 2. 73 9 64 .9 28 31 46 .0 60 4. 28 1 5. 88 8 63 .6 15 4. 66 8 51 .2 78 3. 54 8 55 .9 66 3. 38 5 63 .7 69 36 42 .1 19 5. 48 5 5. 69 0 62 .1 28 4. 39 8 48 .6 76 3. 88 7 53 .7 18 4. 10 1 62 .8 47 t ab le 3 : v ar ia nc e d ec om po si ti on , q 1: 19 89 t o q 2: 20 06 ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 51 t ab le 4 : v ar ia nc e d ec om po si ti on , p re -c ri si s p er io d m 1: 19 89 t o m 12 :1 99 6 (c o n ti n u ed ) h or iz on (m on th ly ) e xp la in ed b y sh oc k s in b d o n : b d o n r b d o n r b d ip i b d c p i b d r e r 1 10 0. 00 0 0. 00 0 5. 54 5 94 .4 55 0. 45 4 99 .4 34 0. 46 0 99 .2 55 0. 47 5 84 .3 10 2 99 .3 64 0. 10 3 38 .9 42 60 .0 34 0. 53 8 97 .8 22 0. 43 1 96 .9 72 2. 56 1 79 .8 96 3 96 .4 82 0. 45 7 41 .1 70 55 .8 41 2. 65 1 93 .2 93 2. 44 7 95 .0 82 1. 82 4 75 .4 03 6 84 .3 02 4. 04 7 50 .1 65 42 .3 03 5. 01 4 82 .8 99 4. 68 8 93 .3 94 1. 88 9 63 .0 31 9 72 .8 78 8. 54 1 56 .5 76 31 .3 93 6. 00 5 71 .5 84 7. 04 9 87 .7 70 4. 54 3 57 .0 64 12 64 .8 07 12 .6 02 60 .6 28 23 .6 61 5. 68 9 63 .8 43 7. 34 7 80 .0 11 5. 80 0 51 .9 44 16 56 .8 32 17 .0 83 61 .9 42 17 .6 92 5. 18 6 58 .0 17 5. 96 1 70 .9 03 5. 42 2 46 .8 36 19 51 .7 99 19 .6 92 60 .9 10 15 .5 67 4. 94 7 55 .5 62 4. 93 1 66 .3 94 5. 39 7 45 .0 01 23 45 .9 53 22 .0 36 58 .6 65 14 .5 41 4. 69 5 53 .3 93 4. 27 4 63 .0 10 5. 97 7 43 .8 22 27 41 .1 17 23 .0 35 56 .5 09 14 .5 63 4. 51 3 51 .6 33 4. 43 7 61 .5 50 6. 80 7 43 .0 59 31 37 .2 76 22 .8 59 54 .8 09 15 .0 11 4. 44 1 49 .9 13 5. 17 9 61 .1 64 7. 60 7 42 .3 95 36 33 .6 82 21 .5 20 53 .3 36 15 .7 10 4. 55 2 47 .6 65 6. 50 9 61 .4 46 8. 41 8 41 .6 56 t a b le co n ti n u es o n t h e n ex t p a g e ht tp :// ijb f.u um .e du .m y 52 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 h or iz on (m on th ly ) e xp la in ed b y sh oc k s in b l o n : b l o n r b l o n r b l ip i b l c p i b l r e r 1 10 0. 00 0 0. 00 0 0. 18 2 99 .8 18 10 .1 65 89 .8 28 1. 60 6 97 .3 23 2. 07 3 82 .2 20 2 99 .5 20 0. 44 5 2. 73 2 95 .0 84 19 .4 63 79 .3 61 0. 92 8 94 .2 61 2. 68 1 79 .9 61 3 96 .1 36 0. 28 0 2. 56 9 91 .9 83 19 .5 38 78 .0 16 0. 86 4 92 .9 24 3. 87 9 75 .4 76 6 87 .6 32 0. 11 5 2. 39 3 83 .1 83 22 .7 35 73 .2 50 0. 79 5 93 .1 52 7. 52 6 66 .0 53 9 82 .5 61 0. 28 3 6. 55 7 66 .6 15 24 .0 22 69 .7 73 1. 47 2 93 .6 40 8. 40 9 62 .7 71 12 79 .8 34 0. 67 8 12 .7 31 49 .3 85 24 .7 35 66 .7 12 3. 07 8 92 .8 36 8. 41 1 62 .3 04 16 77 .9 82 1. 44 8 19 .7 35 33 .3 00 25 .5 56 63 .0 39 6. 84 1 89 .4 35 8. 72 6 61 .7 22 19 77 .2 60 2. 13 2 23 .3 59 26 .1 69 26 .2 73 60 .5 79 10 .8 43 85 .2 47 9. 37 3 60 .6 57 23 76 .7 03 3. 08 5 26 .5 05 20 .8 43 27 .4 44 57 .6 59 17 .3 42 78 .0 05 10 .5 33 58 .8 04 27 76 .3 10 3. 99 8 28 .3 72 18 .3 15 28 .8 66 55 .0 68 24 .3 89 69 .8 15 11 .7 41 56 .9 65 31 75 .8 83 4. 80 3 29 .4 31 17 .3 17 30 .5 00 52 .7 21 31 .1 34 61 .7 23 12 .7 99 55 .4 48 36 75 .1 62 5. 59 7 30 .0 47 17 .1 47 32 .7 43 50 .0 39 38 .3 38 52 .8 11 13 .7 85 54 .1 43 t ab le 4 : v ar ia nc e d ec om po si ti on , p re -c ri si s p er io d m 1: 19 89 to m 12 :1 99 6 (c o n ti n u ed ) ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 53 t ab le 5 : v ar ia nc e d ec om po si ti on s: p os tc ri si s p er io d19 99 :1 t o 20 06 :1 2 (c o n ti n u ed ) h or iz on (m on th ly ) e xp la in ed b y sh oc k s in b d o n : b d o n r b d o n r b d ip i b d c p i b d r e r 1 10 0. 00 0 0. 00 0 0. 00 4 99 .9 96 1. 24 3 98 .2 62 7. 98 6 92 .0 13 0. 96 6 87 .7 88 2 93 .7 27 5. 86 0 0. 29 2 98 .8 63 3. 19 6 93 .5 30 15 .7 91 80 .8 25 0. 58 9 88 .4 18 3 91 .9 22 6. 84 2 0. 56 9 97 .9 05 3. 25 8 92 .6 10 18 .2 09 74 .9 57 0. 47 3 88 .5 77 6 87 .8 08 5. 40 8 1. 09 7 93 .3 06 2. 92 3 82 .8 50 23 .8 31 62 .5 61 0. 24 5 87 .7 42 9 80 .9 74 3. 69 8 1. 53 6 87 .5 99 2. 38 0 70 .1 12 28 .9 28 55 .3 33 0. 19 6 86 .7 57 12 72 .9 20 2. 84 9 2. 03 9 82 .5 29 2. 16 5 61 .0 52 33 .8 95 50 .9 80 0. 24 8 85 .8 48 16 63 .0 13 2. 63 1 2. 88 2 77 .7 00 2. 21 3 53 .5 56 39 .7 50 47 .3 78 0. 40 2 84 .8 53 19 56 .9 31 2. 79 2 3. 64 0 75 .2 94 2. 40 3 49 .9 37 43 .1 40 45 .4 09 0. 54 9 84 .2 67 23 50 .5 97 3. 14 3 4. 77 7 73 .0 97 2. 79 3 46 .5 36 46 .0 30 43 .1 30 0. 76 1 83 .6 67 27 45 .8 93 3. 50 2 6. 00 8 71 .4 92 3. 29 7 44 .0 49 47 .1 68 41 .0 03 0. 97 7 83 .2 26 31 42 .3 84 3. 81 4 7. 28 1 70 .0 63 3. 88 8 42 .0 70 46 .9 71 38 .9 97 1. 18 9 82 .8 97 36 39 .1 75 4. 12 3 8. 86 5 68 .1 67 4. 71 6 40 .0 03 45 .5 92 36 .7 29 1. 44 2 82 .5 83 t a b le co n ti n u es o n t h e n ex t p a g e ht tp :// ijb f.u um .e du .m y 54 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 h or iz on (m on th ly ) e xp la in ed b y sh oc k s in b l o n : b l o n r b l o n r b l ip i b l c p i b l r e r 1 10 0. 00 0 0. 00 0 0. 00 5 99 .9 95 0. 47 7 99 .1 81 0. 48 1 99 .2 39 2. 54 6 87 .8 04 2 97 .3 15 0. 85 4 0. 87 0 97 .9 28 4. 51 7 91 .7 02 0. 79 1 96 .6 51 6. 60 2 83 .6 97 3 96 .7 22 0. 71 6 1. 42 8 96 .7 12 3. 90 6 92 .5 65 0. 56 7 96 .2 84 11 .3 66 79 .4 28 6 95 .7 53 0. 70 2 4. 28 5 91 .9 18 4. 72 0 88 .4 96 0. 88 8 96 .2 52 27 .6 97 63 .4 68 9 94 .7 70 1. 19 2 7. 76 9 87 .0 86 8. 94 5 77 .8 39 1. 45 6 95 .4 84 39 .3 52 52 .3 59 12 93 .7 15 1. 65 0 10 .8 73 83 .1 89 14 .0 18 67 .2 95 1. 44 9 94 .3 70 46 .3 96 45 .6 14 16 91 .8 88 2. 10 8 13 .6 77 79 .8 06 19 .4 04 56 .5 93 1. 12 0 92 .4 22 51 .4 96 40 .5 87 19 90 .0 90 2. 35 0 14 .8 32 78 .3 12 22 .0 88 50 .8 91 1. 15 6 90 .5 43 53 .5 36 38 .4 74 23 87 .1 55 2. 56 1 15 .4 96 77 .1 04 24 .2 78 45 .4 00 1. 84 0 87 .6 19 55 .0 03 36 .8 44 27 83 .7 72 2. 67 3 15 .5 87 76 .2 56 25 .4 17 41 .4 46 3. 09 7 84 .5 38 55 .7 00 35 .9 66 31 80 .1 56 2. 71 4 15 .4 34 75 .4 09 25 .9 21 38 .4 26 4. 62 9 81 .6 03 56 .0 08 35 .4 85 36 75 .5 77 2. 70 0 15 .1 76 74 .1 13 26 .0 43 35 .4 42 6. 57 8 78 .3 83 56 .1 34 35 .1 61 t ab le 5 : v ar ia nc e d ec om po si ti on s: p os tc ri si s p er io d19 99 :1 t o 20 06 :1 2 ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 55 c. results based on the variance decomposition analysis the variance decomposition analysis is used to asses the dynamic interactions between the monetary policy indicator and the bank balance sheet items. at the same time, we provide the variance decomposition of the macroeconomic variables to show the contribution of bank deposits and loans in explaining the variations in the macroeconomic variables. the results from the variance decomposition analysis (vda) are shown in tables 3, 4 and 5. in general, these results further substantiate our earlier fi ndings based on the irfs. table 3 provides the vda results for the entire sample period. the results show that the innovations in onr are highly signifi cant in accounting for the variations in bank deposit. in particular, innovations in onr account for around 27 percent of the forecast error variance of deposit at the 36-month horizon. in contrast, onr only accounts for about 6 percent of the forecast error variance of bank loan at the 36-month horizon. bank deposit therefore, is a signifi cant variable in transmitting monetary policy effects to the economy. bank loan, however, is shown as a rather important variable to trigger appropriate monetary policy action. innovations in bank loan account for around 6 percent of the forecast error variance of onr compared to bank deposit which contribute only around 2 percent of the forecast error variance of onr. the role of bank deposit in explaining the forecast error variance of the objective variables is only signifi cant in real exchange rate, where around 13 percent of the forecast error variances in rer are explained by the variations in bank deposit. the contribution of bank loan in explaining the forecast error variances of the objective variables are rather insignifi cant, ranging from 3 to 6 percent. statistics on the tables 4 and 5 are referred to now. for the pre-crisis period, the vda shows that innovations in onr are signifi cant in accounting for the forecast error variance of bank deposit, contributing to a maximum of 23 percent at the 27-month horizon. innovations in onr, however, are small in accounting for the forecast error variances of bank loan, at around 6 percent maximum. innovations in bank deposit and loan are signifi cant in accounting for the forecast error variances of onr, accounting for around 60 percent and 30 percent, respectively. as for the objective variables, bank loan seem to be signifi cant in accounting for the variances in ipi (33 percent of the forecast error variances), cpi (38 percent of the forecast error variances) and rer (around 14 percent of the forecast error variances). there are clear changes in the relationship in the relationship of the variables in the post-crisis period compared to the pre-crisis period. in particular, innovations in onr account for smaller contribution in accounting for the forecast error variances in both bank deposit (around 7 percent maximum) and bank loan (around 3 percent maximum). innovations in bank loan remains high in explaining the forecast error variances of onr at around 16 percent maximum, while that of bank deposit decline to around 9 percent. in this period, there is an obvious contribution of the innovations in bank deposit in accounting for the ht tp :// ijb f.u um .e du .m y 56 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 forecast error variances of cpi (around 47 percent maximum), and signifi cant contribution of the innovations in bank loan in accounting for the forecast error variances in rer (around 56 percent maximum) and ipi (around 26 percent maximum). 5. conclusion in conclusion, the study shows that both bank loans and deposits play crucial role in the monetary transmission process in the economy. based on the empirical evidence, there is a clear link between the bank loans and deposits and the real economic activity. the credit is a relevant and important channel for monetary transmission in the economy. in view of the important role played by bank lending in the economy, this study re-emphasises the importance of ensuring the stability of banking system. not only that a healthy and stable banking system ensures an effective implementation of monetary policy, it is indeed a pre-condition towards the overall economic stability. in view of the heavy dependency of the economy on bank lending and the critical role that it plays towards economic stability, it is also good to diversify this dependence away from banks to fi nancial markets. author statement: the paper was submitted by salina hj. majid, assistant professor in the department of economics, international islamic university, kuala lumpur, malaysia. e-mail: ksalina@iiu.edu.my. the co-author is sabri abdul majid, assistant professor in the same university. references altunbas, y., fazylov, o., and molyneux, p. (2002). evidence on the bank lending channel in europe. journal of banking and finance, 26(11), 2093-2110. azali, mohamed (2003). transmission mechanism in a developing economy: does money or credit matter? (2nd ed.). university putra malaysia press. bahmani-oskooee, m., and bohl, m.t. (2000). german monetary unifi cation and the stability of the german m3 money demand function. economics letters, 66, 203-208. bahmani-oskooee, m., and wing ng, r.c. (2002). long-run demand for money in hong kong: an application of the ardl model. international journal of business and economics, 1(2), 147-155. bank negara malaysia. monthly statistical bulletin (various issues). bank negara malaysia. (1994). money and banking in malaysia: 35th anniversary edition 1959-1994 (4th ed.). bank negara malaysia. (1999). the central bank and the fi nancial system in malaysia: a decade of change (1st ed.). bernanke, b. s. (1983). non-monetary effects of the fi nancial crisis in the propagation of the great depression. american economic review, 73, 257-276. ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 57 bernanke, b.s., and alan s. b. (1988). credit, money and aggregate demand. aea papers and proceedings, 78(2), 435-39. bernanke, b.s. (1992). the federal funds rate and the channels of monetary transmission. american economic review, 82(4), 901-921. brissimis, s. n., kamberouglou, n. c., and simigiannis, g. t. (2001). is there a bank lending channel of monetary policy in greece? evidence from bank level data. european central bank. working paper (series no. 104). brunner, k., and allan, h. m. (1972). money, debt, and economic activity. journal of political economy, 80(5), 951-77. brunner, k. (1988). money and credit in the monetary transmission process: is it money or credit, or both, or neither? aea papers and proceedings, 78(2), 446-450. brunner, k. (1990). money supply. chapter 9. in elsevier’s handbook of monetary economics. christiano, l., and massimo, r. (2001). money growth monitoring and the taylor rule. national bureau of economic research. working papers (no. 8539). dale, s., and andrew, g. h. (1995). interest rates and the channels of monetary transmission: some sectoral estimates. european economic review, 39, 1611-1626. dotsey, m., and christopher, o. (1994, winter). m2 and monetary policy: a critical review of the recent debate. federal reserve bank of richmond. economic quarterly, 41-49. driscoll, j. c. (2004). does bank lending affect output? evidence from the u.s. states. journal of monetary economics, 51(3), 451-471. ehrmann, m., et al. (2001). financial systems and the role of banks in monetary policy transmission in the euro area. european central bank. working paper (series no. 105). farinha, l., and marques, c. r. (2001). the bank lending channel of monetary policy: identifi cation and estimation using portugese micro bank data. european central bank. working paper (series no. 102). feldstein, m., and james, h. s. (1993). the use of monetary aggregates to target nominal gdp. national bureau of economic research. working papers (no. 4304). garretsen, h., and swank, j. (1997). the bank lending channel in the netherlands: the impact of monetary policy on households and fi rms. de nerlandesche bank. working papers (no. 567). garretsen, h. (1998). the transmission of interest rate changes and the role of bank balance sheets: a var analysis for the netherlands. journal of macroeconomics. 20(2), 325-339. gertler, m., and gilchrist, s. (1993). the role of credit market imperfections in the monetary transmission mechanism: arguments and evidence. scandinavian journal of economics, 95(1), 43-65. gertler, m., and gilchrist. (1994). monetary policy, business cycles and the behavior of small manufacturing fi rms. quarterly journal of economics, 109(2), 309-340. ht tp :// ijb f.u um .e du .m y 58 the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 hallsten, k. (1999). bank loan and the transmission mechanism of monetary policy. sveriges riksbank. working paper (series no. 73). hess, g.d., and richard, d.p. (1993). comparing interest-rate spread and money growth as the predictors of output growth: granger causality in the sense granger intended. journal of economics and business, 45(3-4), 247-268. hubbard, g. r. (1994). is there a credit channel for monetary policy? national bureau of economic research. working papers (no. 4977). ibrahim, m. h. (2001). financial factors and the empirical behaviour of money demand: a case study of malaysia. international economic journal, 15(3), 55-72. ibrahim, m. (2002, november). money, credit, asset prices and monetary transmission mechanism: evidence from malaysia. paper presented at the 8th convention of the east asian economic association, kuala lumpur. ibrahim, m. (2005). sectoral effects of monetary policy: evidence from malaysia. asian economic journal, 19(1), 83-102. jansen, d., and chang, j. c. (2005). the effect of monetary policy on bank lending and aggregate output: asymmetries from non-linearities in the lending channel. annals of economics and finance, 6, 129-153. kakes, jan. (1998). monetary transmission and bank lending in the netherlands. som research institute, university of gronigen. research report (no. 98c30). kassim, s. (2006). bank lending and monetary policy in malaysia (doctoral dissertation, international islamic university malaysia). kashyap, a., and s., j. c. (1994). the impact of monetary policy on bank balance sheets. national bureau of economic research. working papers (no. 4821). kashyap, a., and stein, j. c. (1995). monetary policy and bank lending. national bureau of economic research. working papers (no. 4317). kashyap, a., and stein, j. c. (1997a). what do a million banks have to say about the transmission of monetary policy. national bureau of economic research. working papers (no. 6056). kashyap, a., and stein, j. c. (1997b). the role of banks in monetary policy: a survey with implications for the european monetary union. federal reserve bank of chicago, economic perspectives, 2-18. kuppers, m. (2002). curtailing the black box: german banking groups in the transmission of monetary policy. european economic review, 45, 19071930. laurenceson, j., and chai, j. c. h. (2003). financial reform and economic development in china. cheltenham, uk: edward elgar. mateut, s., spiros, b., and paul, m. (2003). trade credit, bank lending and monetary policy transmission. royal economic society annual conference, (no. 149). mccallum, b. t. (1990). could a monetary base rule have prevented the great depression? journal of monetary economics, 26(1), 3-26. ht tp :// ijb f.u um .e du .m y the role of bank loans and deposits in the monetary transmission mechanism in malaysia: 37-59 59 mccallum, b. t. (1999). recent developments in monetary policy analysis: the roles of theory and evidence. journal of economic methodology, 6(2), 171-198. mccallum, b. t. (2001). should monetary policy respond strongly to output gaps? american economic review, 91(2), 258-262. mccallum, b. t. (2004). monetary policy in economies with little or no money. pacifi c economic review, 9(2), 81-92. mccallum, b. t., and edward, n. (2005). targeting vs. instrument rules for monetary policy. national bureau of economic research. working papers (no. 10612). meltzer, a. (1969). money, intermediation and growth. journal of economic literature, 7(1), 27-56. meltzer, a. (1970). is there an optimal money supply? a discussion. journal of finance, 25(2), 450-53. meltzer, a. (2001, july). money and monetary policy: an essay in honor of darryl francis. review, 23-32. nelson, e. (2003a). the future of monetary aggregates in monetary policy analysis. journal of monetary economics, 50(5), 1029-1059. nelson, e. (2003b). money and the transmission mechanism in the optimizing is-lm specifi cation. center for economic policy research. discussion papers (no. 3898). pesaran, m.h., shin, y., and smith, r. j. (1996). testing for the existence of a long-run relationship. dae working paper (no. 9622). department of applied economics, university of cambridge. pesaran, m.h. (2001). bounds testing approaches to the analysis of level relationships. journal of applied econometrics, 6, 289–326. pesaran, m.h., and pesaran, b. (1997). working with microfi t 4.0: interactive econometrics analysis. oxford university press. scholnick, b. (1996). asymmetric adjustment of commercial bank interest rates: evidence from malaysia and singapore. journal of international money and finance, 15(3), 485-496. suzuki, t. (2001). is the lending channel of monetary policy important in australia? the australian national university. working papers in economics and econometrics (no. 400). tan, e.c. (1997). money demand amid fi nancial sector developments in malaysia. applied economics, 29(9), 1201-1215. tang, t. c. (2000). commercial bank lending and economic growth in malaysia: an empirical studies. borneo review, 11(1), 60-71. tang, t. c. (2002). banks lending and economic growth in malaysia: revisited. jurnal analisis, 10(1), 69-87. vaithilingam, s., guru, b. k., and shanmugam, b. (2003). bank lending and economic growth in malaysia. journal of asia-pacifi c business, 5(1). ht tp :// ijb f.u um .e du .m y ht tp :// ijb f.u um .e du .m y 43 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 how to cite this article: mohd aziz, n. i., & kassim, s. (2021). investment decision making towards investing in islamic unit trusts amongst women. international journal of banking and finance, 16(1), 43-54. https://doi.org/10.32890/ijbf2021.16.1.3 investment decision making towards investing in islamic unit trusts amongst women 1nor izzati mohd aziz & 2salina kassim iium institute of islamic banking and finance international islamic university malaysia, kuala lumpur 2corresponding author: ksalina@iium.edu.my received: 14/8/2020 revised: 30/9/2020 accepted: 25/10/2020 published: 30/1/2021 abstract the differences in skill expectations and knowledge are amongst factors that contribute to the variances when men and women choose financial products. women are claimed to be risk averse, somewhat insecure, lacking self-confidence and interest in financial investment products. nevertheless, the evolving working and educational environment has changed this stereotype, wherein women are increasingly more educated, knowledgeable, and are more exposed to investments. gender equality between men and women in terms of the volume of investments has turned into a requirement that boosts economic growth with people being the most important factor to contribute through investing habit. this cannot be realised if only men are involved in investment. financial institutions aim certain figures of targeted investment volume each year. as such, this study identified the factors of women investments, particularly in islamic unit trusts. http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 44 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 a total of 201 respondents were selected via the convenience sampling technique. the likert scale questionnaires were analysed using smart pls software. as a result, several investment behaviours, including investment objectives, return expectations, awareness, and risk attitude, emerged as factors that influenced women investors to choose unit trusts. the study outcomes are beneficial in devising effective strategies that may attract women to invest in islamic unit trust. besides, certain agencies should conduct campaigns or seminars on financial literacy for this potential sector, so as to enhance their knowledge in investing in islamic unit trusts. keywords: gender equality, financial investment, risk tolerance, islamic unit trusts. jel classification: g11, g14, g17, g20, g40 introduction unit trust or mutual fund is an indirect financial investment tool that has been growing at a rapid pace and supports a range of complex needs within the financial industry. this investment vehicle is known as islamic unit trust when the diversified portfolios are invested in a portfolio of halal stocks that comply with shariah rules. islamic unit trusts have recorded tremendous growth since 2001, which reflect the acceptance of this type of investment in malaysia. it has attracted muslim investors, particularly small investors aiming to fulfil the needs to invest based on shariah principles. islamic unit trusts allow retail investors with similar financial objectives to accumulate savings and pool their money and invest in a well-diversified portfolio of assets, such as cash, bonds, shares, and properties; whichever included in the portfolio. such portfolio diversifications can minimise risks. the malaysian government, along with the support of bank negara malaysia, gives a strong commitment through its capital market master plan 2 2011-2020 to promote an effective investment management industry, in order to stimulate economic growth. as a muslim majority country, malaysia has vast opportunities to contribute and play an active role in developing its islamic economy. malaysia desires to become the global halal hub for financial markets 45 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 by year 2020. with 75 percent achievement recorded in 2017, the halal industry is expected to contribute as high as 87 percent to the country’s gdp in 2020, as projected by its related government agency; the halal industry development corporation. the halal industry operates across a number of sectors, including food, clothing, tourism, and finance. by enhancing the development of islamic capital market (icm), islamic finance is exposed to multiple opportunities to gain market share in the halal industry realm. malaysia’s icm has been under the control of kuala lumpur shariah index (klsi), which was introduced in 1999, primarily to accelerate and boost icm. the total islamic funds accounted for 28.5 percent in year 2017, whereas the net asset value (nav) of conventional and unit trust funds increased from rm432 billion in 2018 to reach higher by the end of 2019. although this figure grew over the years, the capital market master plan 2 target of nav for islamic unit trusts was rm158 billion for 2020. it is essential for islamic unit trusts to undergo detailed and clear efforts to inculcate the investment culture. the population in malaysia is predominantly muslims, which offers a huge potential to this country in developing shariah-based islamic products, including islamic unit trusts. in the present investment industry, women have been claimed to be not very receptive toward investments. the statistics showed that women participation in investment was as low as 30 percent when compared with men, which evidences a huge gap in light of gender equality. the deputy minister of the women, family, and community development ministry hannah yeoh in 2020 asserted that women should embark on the path of stock investment to grow their financial wealth and achieve greater financial freedom. in the effort to generate better product innovations in future, the r&d centre of each unit trust conducted regular research on the determinants of reasons for people to buy unit trust products. insufficient data have “proven the acceptance of women investing in unit trusts”. despite the features of the products, there are other factors that influence behaviour toward investing. based on a different approach, this study focused on internal factors and behaviour toward investment. more precisely, this study looked into the factors for women to invest or otherwise in islamic unit trusts, in comparison with men. as such, this study concentrated on unit trust management, fundi size performance, corporate disclosure, and performance of fund managers. 46 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 studies concerning islamic unit trusts in malaysia are scarce, as the compilation of literature review have been mostly based on conventional unit trusts. this serves as a motivation to assess the acceptance of islamic unit trusts amongst women so as to identify if the level of their investment is encouraging. literature review annamalah et al., 2019 claimed that investment behaviour is related to the investors’ attitude in the process of consideration between the alternatives people have. investors with positive attitude directly create positive intention and preferences to invest (ali et al., 2014). they feel good, useful, and beneficial when investing. the behaviour is based on several factors in order to make the right investment decisions for choosing specific investments. in a study about investors’ decision in unit trust investment, they seemed to be more influenced by financial status, risk-taking behaviour, and sources of information. those with highly stable financial status, excess income, and better jobs had more awareness to choose to invest in unit trusts. with money surplus on a monthly basis, some of them relied on unit trust knowledge shared by their work mates familiar with investing. hence, they were aware of many unit trust functions (xian & hassan, 2019). although it is not difficult to measure demographic characteristics, correlations between gender and other demographic factors cannot be assumed without conducting a survey. (stoker et al., 2012) discovered that gender has an important role in management positions. investors will look into the gender of fund managers when ascertaining that the volume of investment invested in mutual funds is managed by one with high self-confidence and a risk taker. a good fund manager is believed to be amongst men. one of the many factors connected to gender variance is the risk-taking behaviour between male and female. (karim et al., 2016) found that male working adults in kuala lumpur, malaysia displayed more risk tolerance toward choosing riskier assets, such as mutual funds or unit trusts. women are not necessarily of the best managers because they hesitate to take risks and appear to lack confidence when partaking in competition. they exert low confidence when making investment decisions. their willingness to take risks is flexible across age and lifecycle (mehta & shah, 2012). investors less than 30 years old prefer 47 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 high return with low risks, while those above 50 years would choose low risk regardless of the return. hence, the related risk factors must be examined as they vary based on investors’ demographic profiles, despite some studies reporting that risk factors have no impact on the selection of unit trusts (kamil et al., 2016). the literature depicts the opposite, as it claims that participation of females in unit trust is more strongly encouraged than males (cannivate, 2018). women and men, generally, make different decisions and display different attitudes toward unit trusts, in a way where females are better at investments than males are (begum & rahman, 2016). for women, it is not good to let overconfidence/ emotional responses drive their investment decision, while men are overconfident in their ability to actively manage their portfolios as they believe that they are in control of the outcomes, more than they really do. men tend to feel an extra urge to buy more in good times to capture more gain. nonetheless, everyone has differing views regardless of gender. no investor prefers either male or female fund managers because investors generally place trust on the manager to make decisions and to analyse the performance of mutual funds (shafee, 2018). a specialist often spreads the risk through careful diversification because the basic nature of unit trust is to carry low-level risks. dharani et al. (2014) described that safety of funds is a priority for choosing investment amongst investors. young adults below 30 years of age and married women exhibited higher tendency to buy investment funds. for investors, an investment needs to be part of their monthly financial planning. they need to put aside a fraction of their monthly income for investment and expenditure purposes. since 2006, a number of studies began highlighting the lack of information about islamic unit trusts. the studies have urged for improvements in financial literacy amongst people. lack of financial literacy and information may lead non-shariah investors to opt for better alternative investments other than islamic unit trusts. the tendency to rely on various sources seemed to affect the information capacity of investors as they gain information from the mass media, family, relatives, friends, and islamic unit trust agents (hasna & salleh, 2018). 48 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 it is a general expectation that gender influences the value which leads to decision making. women are more confident to just keep their money safe in the bank as they believe that investing in unit trusts takes a longer timeframe to achieve capital gain returns. this stems from lack of knowledge in financial planning, hence their reluctance to participate in unit trusts. similarly, bakar et al. (2015) found that women would rather choose a fixed price unit trust investment with low risk, such as skim amanah saham bumiputra (asb). knowledge barrier hinders one from investing as they do not know that islamic unit trusts offer diversified portfolios and the investments adhere to shariah guidelines (hasna & salleh, 2018). methodology the study respondents comprised malaysian women who had already invested in islamic unit trusts, as well as potential users of islamic unit trusts. the distributed questionnaires gathered data regarding demographic profiles and the factors that influence decision making to invest in islamic unit trusts, including risk-taking behaviour, investment behaviour, and barriers to invest. behaviour of investors toward investment may differ, depending on varied investment goals and objectives, awareness levels, and expected return from the investment. risk-taking behaviour and other barriers, such as lack of information, confidence to save money only in banks, and poor affordability to invest, are some factors that have been highlighted in past studies. these factors prevent women investors from investing in islamic unit trusts. figure 1 illustrates the framework that represents the factors that contribute to the specific variables, which influence the direct relationship with the dependent variable. figure 1. constructs and items embedded in the questionnaire. 4 the performance of mutual funds (shafee, 2018). a specialist often spreads the risk through careful diversification because the basic nature of unit trust is to carry low-level risks. dharani et al. (2014) described that safety of funds is a priority for choosing investment amongst investors. young adults below 30 years of age and married women exhibited higher tendency to buy investment funds. for investors, an investment needs to be part of their monthly financial planning. they need to put aside a fraction of their monthly income for investment and expenditure purposes. since 2006, a number of studies began highlighting the lack of information about islamic unit trusts. the studies have urged for improvements in financial literacy amongst people. lack of financial literacy and information may lead non-shariah investors to opt for better alternative investments other than islamic unit trusts. the tendency to rely on various sources seemed to affect the information capacity of investors as they gain information from the mass media, family, relatives, friends, and islamic unit trust agents (hasna & salleh, 2018). it is a general expectation that gender influences the value which leads to decision making. women are more confident to just keep their money safe in the bank as they believe that investing in unit trusts takes a longer timeframe to achieve capital gain returns. this stems from lack of knowledge in financial planning, hence their reluctance to participate in unit trusts. similarly, bakar et al. (2015) found that women would rather choose a fixed price unit trust investment with low risk, such as skim amanah saham bumiputra (asb). knowledge barrier hinders one from investing as they do not know that islamic unit trusts offer diversified portfolios and the investments adhere to shariah guidelines (hasna & salleh, 2018). methodology the study respondents comprised malaysian women who had already invested in islamic unit trusts, as well as potential users of islamic unit trusts. the distributed questionnaires gathered data regarding demographic profiles and the factors that influence decision making to invest in islamic unit trusts, including risk-taking behaviour, investment behaviour, and barriers to invest. behaviour of investors toward investment may differ, depending on varied investment goals and objectives, awareness levels, and expected return from the investment. risk-taking behaviour and other barriers, such as lack of information, confidence to save money only in banks, and poor affordability to invest, are some factors that have been highlighted in past studies. these factors prevent women investors from investing in islamic unit trusts. figure 1 illustrates the framework that represents the factors that contribute to the specific variables, which influence the direct relationship with the dependent variable. risk-taking behaviour investment behaviour investment barrier investment decision making 49 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 table 1 constructs and the items in the questionnaire no. variables items 1. investment decision making on islamic unit trusts 1. i prefer investing in fixed/term deposit accounts and investment accounts offered by islamic banks 2. i prefer investing in unit trust because of the low transaction cost. 2 risk taking 1. i prefer to invest in unit trust because it carries a lower level of risk compared with other similar investments. 2. i am aware that the level of risk will determine the level of return on investment. 3. i will consider the risk of each type of investment before buying financial products. 4. i took my personal risk assessment test to understand the instruments that suit my risk-taking ability. 3 investment behaviour • investment objectives • awareness on investment • expected return 1. i am interested in increasing wealth through investing. 2. i am aware that investment can increase my income. 3. i am aware that financial institutions offer online transfer services (just click) for easy investment transaction. 4. i am aware it is good to diversify income into various investments for future returns. 5. i have my target of an expected return. 6. i have set the return factor to be the priority for investing. 4. investment barrier • financial stability • financial knowledge 1. even if i consider buying investment products, i still have surplus money at the end of the month. 2. i am even prepared to cut my expenses for saving and investment. 3. i don’t think that the required principal amountto invest is too high. 4. i am very confident that i have a very good level of financial knowledge. 5. i understand that diversification is buying various assets and combining them in a portfolio in order to reduce risks. 50 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 findings table 2 shows the demographic profiles of the respondents, which is composed of all women who have already invested in islamic unit trusts, as well as potential users who prefer investing in islamic unit trusts. from the 201 respondents, only 44 respondents (22%) had invested in unit trusts, while 78 percent of the respondents either chose another investment or were uninterested in any investment product. most of the respondents were married, 26-39 years old, possessed degrees, and earned rm3,001 – rm5,000 a month. table 2 demographic profiles frequency percent marital status single married 54 147 26.8% 73.2% age under 26 26 – 39 40 – 59 over 60 11 165 24 1 5.5% 82.1% 11.4% 1% educational level spm/stpm/diploma degree masters professional degree 16 123 55 7 8% 61.1% 27.4% 3.5% employment salaried retired business housewife 170 3 21 7 84.6% 1.3% 10.8% 3.3% income less than rm1,500 rm1,500 – rm3,000 rm3,001 – rm5,000 rm5,001 – rm10,000 rm10,000 – rm15,000 more than rm15,000 25 61 108 2 3 2 12.4% 30.3% 53.7% 0.9% 0.9% 1.8% investmen in unit trust yes no 44 157 21.9% 78.1% as shown in table 3, the cronbach’s alpha value for decision making exceeded 0.5, which is listed in the acceptable range. the acceptable 51 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 value must be equal or higher than 0.7 to indicate investment barrier, investment behaviour, and risk attitude, which resulted in 0.75, 0.81, and 0.72, respectively. the purpose of cronbach’s alpha is to measure internal consistency, reliability, and validity of data. for better internal consistency, it has been suggested to generate composite reliability (cr) to indicate the weighting rate from various categories. the cr is more reliable than cronbach’s alpha, and it helps to enhance reliability and validity. the values tabulated in table 3 exceeded 0.7, except for investment decision making. additionally, all the variables were qualified and fulfilled the condition of average variance extracted (ave), which is acceptable above 0.5. this verified the convergent validity, and no measurement error was detected in the variance. table 3 summary of test criteria and reliability measures cronbach alpha composite reliability ave (a) (b) (c) (d) investment barrier (a) 0.75 0.83 0.50 0.56 investment behaviour (b) 0.81 0.87 0.52 0.83 0.54 decision making (c) 0.51 0.76 0.64 0.54 0.71 0.45 risk attitude (d ) 0.72 0.82 0.54 0.75 0.74 0.73 0.55 the discriminant validity was determined using fornel and larcker (1971) by comparing the square root of each ave in the diagonal with the correlation coefficients (off-diagonal) for each construct in the relevant rows and columns. discriminant validity refers to the extent a measurement is able to assess the concept of variables that is unrelated. the results presented in table 3 show that the discriminant validity was not well established with having values below 0.5. this signifies that the structural model has moderate validation and only fits the model moderately. after confirming the reliability and validity aspects (see table 3), the path coefficient was identified to generate the significance of parameter estimates; in order to estimate the standard errors and to evaluate the structural model. the summary of structural model analysed using sub-samples of “bootstrapping” is presented in table 4. the significant level was set at 10 percent, which means that when p -value is greater than the significant level, the result is unsupported. the p -values of investment barrier were lower than the significant 52 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 level, thus not supporting the relationship. nevertheless, the p-values of investment behaviour and risk attitude exceeded the significant level and appeared significant to decision making. path coefficient table 4 summary of structural model standard dev. t-stats p-values results investment barrier 0.072 1.199 0.231 supported investment behaviour 0.091 2.588 0.010 supported risk attitude 0.075 4.397 0.000 supported the analysis elaborates in detail the reasons why some women prefer investing in islamic unit trusts, while some do not. the analysis of items embedded in the questionnaire using statistical software revealed the reasons and causes of both attitude and behaviour exerted by the respondents towards investing in islamic unit trusts. the following section interprets the analysis and concludes the issue highlighted in this research paper. conclusion behaviour, such as investment objective, awareness, and return, were evaluated in this study to determine the influential factors in investment decision. women, similar to men, are keen in increasing wealth through investing as they believe that investment can increase income. for concerned women, they are aware that it is indeed a good decision to diversify their income into various investments for future returns. to date, many are aware that financial institutions offer online transfer services (just a click away) for easy investment transaction. the study outcomes showed that those earning between rm3,000 and rm5,000 were aware about the benefits of investing in unit trusts. women from the middle-income group accounted for 29.5 percent, which is half from the total respondents who were knowledgeable about diversifying money. meanwhile, only 4.5 percent from the lowincome group invested in unit trusts. this emphasises the need to expose those from the low-income group to the fact that the principal 53 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 money required for investment is relatively smaller. nonetheless, 15.9 percent of the respondents had strongly agreed that unit trust is to cater to the low-income group, while 4.5 percent had no idea about the statement. they still believed that unit trust is only meant for those with a high income. risk attitude displayed a significantly positive impact on decision making. this study focused only on women, which is known to possess low-risk tolerance during investment decision. the respondents agreed that they were willing to choose unit trust because it poses lower risk, when compared with other investments, such as investment account offered by islamic banks or investing in the stock market or buying shares. they are particular about the risk posed by each investment product, apart from expecting that the risk taken would determine the level of return. this study found that 22.7 percent of the respondents had set the return factor as the priority for investing. they already had their target of return to achieve each year. not surprisingly, some had taken the personal risk assessment in order to test and to understand the instruments that suit their risk-taking ability. here, they are able to identify the investment products that suit their target in an affordable manner. in conclusion, the barriers that hinder one from investing were financial stability and financial knowledge. apart from surplus money at the end of the month or confidence to invest, many other factors seemed to dictate investing decision, such as behaviour and risk attitude “that are more crucial factors to be considered. unit trust is becoming a popular alternative for investment that can boost economic growth. therefore, the related institutions and agencies should take effective measures in attracting individuals to invest regardless of gender. acknowledgment this research received no specific grant from any funding agency. references ali, s., zani, r. m., & kasim, k. (2014). factors influencing investors’ behavior in islamic unit trusts: an application of theory of planned behavior. journal of islamic economics, banking and finance, 10(2), 183-201. 54 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 43–54 annamalah, s., raman, m., marthandan, g., & logeswaran, a. k. (2019). an empirical study on the determinants of an investor’s decision in unit trust investment. economies journal, 7(80), 4-23. bakar, r. a., nawawi, a., saiful, a., & puteh, a. (2015). determinants of lower bumiputera participation in the variable-price. international academic research journal of social science, 1(1), 58-66. begum, n. n., & rahman, s. (2016). an analytical study on investors preference towards mutual fund investment: a study in dhaka city, bangladesh. international journal of economics and finance, canadian center of science and education, 8(10), 184-191. cannivate, m. (2018). why women are better at investing. http:// www.forbes.com. dharani n., inbalakshmi, m., murugapandi, j. (2014). investment pattern of working women in dindigul district. european academic research, 2(7), 9073-9082. hasna, m. a., & salleh, m. c. m. (2018). comparative analysis on epf investors’ knowledge, experience, and behavior towards investing in islamic and conventional unit trusts: the case of malaysia. journal of islamic economics, banking and finance, 14(1), 125-145. kamil, n. m., subramaniam, m., ali, h. e., musah, m. b., & alex, a. (2016). factors influencing the selection of unit trust funds among malaysian retail investors. journal of islamic accounting and business research, 9(2), 155–170. karim, b. a., wenceslas, c. n., & shukri, m. h. m. (2016). risk tolerance level among working adults in kuala lumpur malaysia. review of accounting and finance, 1(1), 99–109. mehta & shah. (2012). preference of investors for indian mutual funds and its performance evaluation. pacific business review international, 5(6), 62-76. shafee, n. b. b. (2018). mutual fund investment decision by malaysian investor. international journal of academic research in business and social sciences, 8(11), 607–615. stoker, j. i., velde m. v. d., & lammers, j. (2012). factors relating to managerial stereotypes: the role of gender of the employee and the manager and management gender ratio. journal of business and psychology, 7(1), 31–42. xian, y. l., & hassan, h. h. (2019). awareness of unit trust among young professionals in malaysian private sector. international journal of recent technology and engineering, 7(5), 201–211. leverage and ipo pricing: evidence from malaysia : 1-19 1the international journal of banking and finance, vol. 15. number 1, 2020: 1-19 1 how to cite this article: ong, c. z mohd-rashid, r., & taufil-mohd, k. n. (2020). leverage and ipo pricing: evidence from malaysia. international journal of banking and finance, 15(1), 1-19. https://doi.org/10.32890/ijbf2020.15.1.9929 leverage and ipo pricing: evidence from malaysia 1chui zi ong 2rasidah mohd-rashid 3kamarun nisham taufil-mohd school of economics, finance and banking, universiti utara malaysia, malaysia 1corresponding author: cztracy.1012@gmail.com; 2m.rasidah@uum.edu.my; 3kamarun@uum.edu.my a r t i c l e i n f o article history: received 22 september 2019 revised 2 october 2019 accepted 7 october 2019 published 31 march 2020 keywords: fixed-price mechanism, initial public offerings, leverage, offer price, information asymmetry. jel code: g10, g11, g13 a b s t r a c t the pricing of ipos is a challenging task among underwriters as they require resources from firms. contrary to the non-financial information presented in a prospectus to set an offer price, pre-ipo accounting information could arguably influence ipo offer price. this study aims to investigate the relationship between leverage and ipo offer price. a crosssectional ordinary least square (ols) regression was implemented to investigate the relationship between leverage and offer price based on a sample of 129 malaysian ipos issued between january 2009 and december 2018. as a result, it was proven that leverage was negatively related to offer prices. accordingly, it was proposed in the findings that fit, which issued higher leverages prior to ipo listing, often posed high financial risks. subsequently, underwriters and issuers set a lower price for ipos to 2 the international journal of banking and finance, vol. 15, no 1, 2020 : 1-19 1. introduction art or science is the possible pricing of ipos. from the aspect of science, it involves the use of a quantitative model to determine the approximate worth of organisations. for instance, underwriters employ valuation model (relative or absolute method) to determine the approximate intrinsic value of firms (how et al., 2007; kim & ritter, 1999). in terms of the art aspect, underwriters determine offer prices by assessing the market conditions of ipos and identifying the potential demand for ipo shares. specifically, underwriters and issuers determine firms’ offer prices through the pricing mechanism. in emerging markets, specifically malaysia, the majority of ipos are issued through a fixedprice mechanism (tajuddin, mohd-rashid, abdullah, & abdul-rahim, 2015). in this case, underwriters and issuers set the offer price of ipos without taking the demands of potential investors into account. in this case, potential investors would face challenges in determining the firms’ values as they are informed asymmetrically. nevertheless, the book-building mechanism has been more prevalent since 2002 despite the small number of organisations employing bookbuilding in pricing ipos. moreover, the book-building mechanism in malaysia is adopted in a hybrid form, which comprises a combination of fixed-price and book-building mechanisms. underwriters employ fixed-price mechanism in ipo pricing and allocation of shares to retail investors. meanwhile, the ipo issued through book-building incorporates its valuation based on institutional investors. theoretically, the ipo pricing should reflect its intrinsic value to encourage aftermarket buying and ensure a steady increase in price. it has been indicated in previous studies that fixed-price ipos are underpriced as compared to book-built ipos. in the case of fixed-price ipos, investors who receive share allocations ought to earn higher initial returns. this perception has become a concern among underwriters. despite the issuers’ attempt of increasing ipo price in order to raise the proceeds from the ipos, the underwriters are inclined to set a lower ipo price. subsequently, low ipo price would be more appealing among potential investors, and a full subscription of ipo shares could be guaranteed. therefore, setting appropriate ipo offer price would be challenging for underwriters and issuers, specifically in fixed-price mechanism compensate for a higher degree of information asymmetry among retail investors. among the implications of this study’s findings include investor concerns on accounting information, especially leverage upon determining ipo value and ipo investment. leverage and ipo pricing: evidence from malaysia : 1-19 3 markets where the offer price is determined before investors’ demands or opinion on the firms’ prospects are acquired. for this reason, diverse opinions are formed among investors towards the organisations’ values in the ipo market, resulting in the emergence of information asymmetries issues (carey, fang, & zhang, 2016). the investors attempt to determine the essential factors of ipo pricing as their confinement in “lemon” issue is not a favourable situation (rock, 1986). with that being said, this article attempts to identify several relevant informative resources in determining ipo offer prices. it could be seen that prior to the recent malaysian study, firms’ offer prices were determined according to specific public information. specifically, mohdrashid, masih, abdul-rahim, and che-yahya (2018) found that issuers offered their ipos at discounted prices, provided if the firms were issued by prestigious underwriters and individuals who made high share allocations to institutional investors. lower ipo prices offered by issuers and underwriters often attracted potential investors to subscribe to shares. contrary to information which is publicly available in the firms’ prospectuses, no study has been conducted on the influence of pre-ipo accounting information on offer prices in the relatively high information asymmetry of the malaysian ipo market. therefore, this article attempts to provide further insights from mohd-rashid et al.’s (2018) study and findings from the literature about the capital structure of firms, specifically in terms of firms’ offer prices. other than the study by mohd-rashid et al. (2018), previous studies were conducted on the factors of ipo pricing and evaluation. as a result, accounting information (aggarwal et al., 2009; cotter, goyen, & hegarty, 2005; lai & lo, 2012; sahoo & rajib, 2012), signalling variables (keasay & mcguinness, 1991; leland & pyle, 1977), and comparable firms of price-earnings (cotter et al., 2005; how et al., 2007; kim & ritter, 1999) were found to be the significant factors. however, no conclusive evidence was found regarding the influence of accounting information on ipo pricing in the malaysian market. therefore, it is required for this study to determine the informational role of accounting variables on ipo offer price in the context of the malaysian market. the malaysian accounting standards board (masb) introduced the implementation of the malaysian financial reporting standards (mfrs) framework, effective 1 january 2012. the framework implementation often enhances the transparency and credibility of financial reports. it could also be said that the prices are determined based on information acquired by underwriters and issuers in the fixed-price regime. for this reason, there may be differences in price valuation based on the accounting information reported in the pro-forma balance sheets of ipo prospectuses. this paper aims to provide further insights from the existing works of literature by identifying the accounting information incorporated into the proforma balance sheets from ipo prospectuses. this information, for example the 4 the international journal of banking and finance, vol. 15, no 1, 2020 : 1-19 information regarding leverage, was not focused in previous studies. leverage represents the amount of debt raised by firms before ipos. provided that debt ratio is the indicator of a firm’s leverage, high-leverage firms possess higher uncertainties and financial risk to the market (sahoo & rajib, 2012). there is also a high possibility for ipos to be overpriced. this phenomenon results in the lack of interest among institutional investors to subscribe to ipos which issue greater debts. meanwhile, a higher share portion of the high-leveraged ipos would be allocated to retail investors. to illustrate this point, they might be confined in the “lemon” issue (beatty & ritter, 1986; rock, 1996). therefore, high-leveraged firms are assumed in this study to be risky. therefore, issuers and underwriters should determine lower offer prices so that retail investors could achieve high risk-adjusted returns in the aftermarket. however, no studies have been conducted on the relationship between pre-ipo leverage and offer price in the malaysian market. as the majority of malaysian ipo prices are determined through fixed-price mechanism which displays higher information asymmetries (yong, 2015), this study provides further insights on the investigation regarding the degree of the pre-ipo leverage influence on ipo offer price after other related factors are controlled. the present study found a negative significant coefficient on debt ratio after the estimation of ipo offer price. this finding indicated that underwriters and issuers determined high-leveraged ipos at lower offer prices, thereby contributing to existing works of literature. to illustrate this point, the explanation of the results was based on information asymmetry theory (rock, 1986). provided that the firms issuing greater debts prior to ipo listing posed greater financial risks, high-leverage firms offered negative information to potential investors. in this case, institutional investors prevented the subscription of firms with greater information asymmetry. therefore, underwriters and issuers placed discounts on ipo offer price to compensate for the higher extent of information asymmetry among retail investors. this study consists of particular procedures. section 2 illustrates the works of literature related to pre-ipo leverage and the developed hypothesis. section 3 presents data, methodology, and the variables used in this study including the model specification. this is followed by section 4, which provides the results of analyses and discussion. section 5 presents, the conclusion and implications of this study. 2. literature review investors are usually provided with information on ipo offer price. rock (1986) highlighted that the issue of “lemon” is prevalent in the ipo market as leverage and ipo pricing: evidence from malaysia : 1-19 5 institutional investors do not attempt to subscribe to ipos offered with high prices. however, retail investors generally receive the allocation of shares either from firms which provide higher information asymmetries or small firms offered at high prices. therefore, this article highlights that retail investors should obtain public information from the firms’ prospectuses to prevent the issue of “lemon”. this is because the investors have a chance of using the information to decide on share subscription. in respect of accounting information, this study could enhance understanding among investors on the informational role of leverage from the firms’ prospectuses as a signal device in offer prices before ipo listing. to attract investors into the ipo market, issuers and underwriters should place discounts on offer prices. however, the amount of compensation should be based on the degree of information asymmetry. to avoid the issue of “lemon” among investors, specifically retail investors, this study aims to reduce information asymmetry by identifying the potential signals of leverage in a firm’s prospectus to elaborate on its offer price. to explain the informational role of pre-ipo leverage on offer prices, this article presents several pieces of evidence on the link between pre-ipo leverage and ipo anomalies. su (2004) reported that ipos which issued greater debts posed a higher degree of underpricing in the chinese market. it was implied that highly leveraged-ipos posed higher financial risks. therefore, retail investors receive compensation from the exposure to information asymmetry. this finding was supported by studies from akyol, cooper, meoli and vismara (2014), chen, wang, li, sun and tong (2015), and huang, chiang, lin and lin (2017), who explained that higher leverage could increase the risk of financial distress. furthermore, although the leverage issue was for capital raising purposes, highleverage firms could not deliver their superior qualities to the market. meanwhile kim, pukthuanthong-le and walker (2007) focused on a sample of high and low technology firms in the u.s. market. as a result, a positive coefficient of preipo leverage was found when underpricing was estimated. this coefficient was due to the volatile operating cash flows of high-technology firms, besides their higher intangible assets (intellectual properties and patents) and low profitability. in this case, it was challenging for issuers to make repayment to debtors on time. additionally, investors must be compensated due to their exposure to high information asymmetry in high-tech firms. in this regard, based on winner’s curse hypothesis, issuers should offer ipos with low offer prices to compensate for higher information asymmetry on high leveraged ipos. this assumption is in agreement with studies by cotter et al. (2005), lai and lo (2012), reber and vencappa (2016) and sahoo and rajib (2012). compared to higher pre-ipo leverage which poses a higher extent of information asymmetry, it is arguable that the higher leverage issued by the organisations serves as a better signal to the market. kim et al. (2007) found a 6 the international journal of banking and finance, vol. 15, no 1, 2020 : 1-19 negative coefficient on pre-ipo leverage in a sample of low-technology u.s. ipos when underpricing was estimated. this coefficient could be explained through the capital structure signalling theory. furthermore, firms have a preference for internal to external financing, such as debt, to raise funds for financing growth activities rather than equity financing. this situation is known as pecking-order hypothesis (myers & majluf, 1984). this hypothesis indicates that high leverage of firms could provide positive information to potential investors about their value. therefore, based on the signalling theory, higher leverage issued by firms prior to ipos often incurred high offer prices due to superior prospects of the firms. as mentioned earlier, the majority of the prices of malaysian ipos are determined by a fixed-price mechanism, indicating that investors’ views on firms’ prospects are not incorporated in setting the offer price (tajuddin et al., 2015). a significant presence of information asymmetry on ipos is implied in the aftermarket (yong, 2015). furthermore, leverage is claimed to be the determinant of financial survival (ahmed sheikh & wang, 2011) if high debts are issued by firms prior to ipo listing. as a result, there is a high possibility for firms to face financial distress, indicating their most disadvantageous qualities to the market. besides, investors would not desire to subscribe to high-risk ipos. based on the arguments on pre-ipo leverage, the information asymmetry theory was implemented in this study (rock, 1986) to assume that underwriters often undervalue high-leveraged malaysian ipos. for this reason, issuers tend to set a lower ipo price so that investors are compensated from their exposed information asymmetry. the following is the hypothesis of the present study: = there is a significant relationship between pre-ipo leverage and offer price. 3. methodology a sample consisting of 168 ipos listed in bursa malaysia from 2009 to 2018 was employed in this study. the year of research, 2009 was selected for the elimination of sub-prime crisis effects to occur. notably, this crisis occurred in 2008. the contents of the data comprised several variables, such as offer price, pre-ipo accounting data, offer size, underwriter’s reputation, and ipo market conditions. all of the data were manually extracted from the organisations’ prospectuses. prior to hypothesis testing analysis, the present study highlighted the particular criteria for data collection. however, an exception was applied on ipos with special types of offers, such as restricted offer-of sale to eligible employees leverage and ipo pricing: evidence from malaysia : 1-19 7 and indigenous (bumiputra) investors, tender offer, and special issues (mohd rashid, abdul-rahim, & yong 2014). similarly, ipos from selected industries sectors, namely, finance, real estate investment trusts (reits), exchange-traded funds (etf), and special purpose acquisition companies (spac) were excluded due to different presentation formats in financial statements. the exclusion of such data resulted in 150 ipos as the final total sample in this study, which represented 86.31% of the overall ipos listed from 2009 to 2018. the distribution of the sample of ipos is illustrated in table 1. table 1. selection process of ipos’ sample from january 2009 to december 2018. description number of ipos number of ipos before exclusion 168 (-) total number of ipos issued through restricted offer-tosale, tender offer and special issues 6 (-) total number of ipos from finance sector 2 (-) total number of ipos from reits sector 5 (-) total number of ipos from etf sector 3 (-) total number of ipos from spac sector 6 total number of final ipo sample 150 3.1 dependent variable and independent variable the dependent variable of this study was the offer price of ipo (op) reported in the prospectus. the offer price of ipo functioned as the reference for potential investors’ decision-making on share subscription. it is claimed in this article that firms with high offer prices are generally large companies with higher information asymmetry, while those with low offer prices are usually small companies with lower information asymmetry. in this case, potential investors could perceive the indicator of the firms’ qualities in various ways. in this study, leverage was the main variable which acted as an indicator in setting offer prices. they were obtained from the pro-forma of the firms’ balance sheet. in respect of the firms’ leverage, debt ratio (leverage) was used in this study. it was measured in the form of short and long-term total liabilities of the firms over the firms’ total assets. sahoo and rajib (2012) highlighted that leverage acts as a signal device of the firms’ financial strength. underwriters often undervalue firms which obtain higher debts than equity, as high-leverage firms are perceived to have the most disadvantageous qualities. in line with 8 the international journal of banking and finance, vol. 15, no 1, 2020 : 1-19 winner’s curse hypothesis, it is suggested that firms with high debt ratio can be identified as high-risk firms. as a solution, issuers and underwriters should set lower offer prices to compensate for the exposure to information asymmetry among retail investors. 3.2 control variables in order to capture the influence of leverage on appropriate ipo pricing, six control variables were employed in this study. the first control variable of this study was the return on equity (roe), which acted as the indicator of the firms’ profitability. given that firms with greater profitabilty represented “good news” to investors, underwriters would determine the high roe of the firms at high offer prices as this feature is regarded as the firms’ better qualities, based on signalling theory (allen & faulhaber, 1989; grinblatt & hwang, 1989; welch, 1989). the second control variable was the firm’s size (lnoffsz) through the total amount of shares issued. given the firm’s size as a proxy of information asymmetry was justified by mohd-rahsid et al. (2018) and yung and zender (2010), large and mature firms possessed fewer uncertainties as compared to small and young firms, and hence capable of influencing ipo offer price. meanwhile, the third control variable was the underwriter’s reputation (undwr), which was obtained from ipo prospectuses and bloomberg. this variable was referred to the annual change in underwriters’ market shares. given the prestigious work of underwriters as promoters to attract investors to subscribe ipos was justified by mohd-rashid et al. (2018) whereby the offer price should be priced lower by underwriters to compensate investors for a higher degree of information asymmetry. the fourth control variable in this study was board types (board), which was computed as a dummy variable. there are two main types of boards on bursa malaysia—main market and ace market. as discussed earlier, yong (2015) highlighted that ipos listed in the main market possessed lower levels of uncertainties compared to ipos in the ace market. therefore, underwriters should set a higher offer price for the ipos listed in the main market. the next control variable was the dummy of pricing mechanism (book) used in pricing ipos. although the majority of malaysian ipos was determined through a fixedprice method, it was known in this study that small entities of ipos were priced through book-building mechanisms, contributing to varied findings with fixedprice ipos. according to kutsana and smith (2004), book-built ipos possessed fewer uncertainties due to soliciting investors’ demands during roadshows. accordingly, this could reduce information asymmetry among the book-built ipos and thus influenced the underwriters’ decision on setting offer prices leverage and ipo pricing: evidence from malaysia : 1-19 9 finally, this study controlled ipo market conditions (ipomkt) when ipo pricing. the market conditions indicated investors’ confidence in the stock market performances. the measurement by kooli and suret (2004) was adopted, where ipo market conditions was categorised as ‘‘hot’’ and ‘‘cold’’ markets. investors had higher optimism and confidence with ipos issued in the hot market (helwege & liang, 2004). due to its potential for ipo success, the determined offer price is expected to be higher. 3.3 model specification to examine the effect of accounting information on the offer price in the malaysian market, ordinary least square (ols) regression model was employed in the present study. the cross-sectional regression model is constructed as follows: (1) a summary of the variables’ measurements in eq. (1) is provided in table 2. table 2. variables and measurements (continued) 12 the international journal of banking and finance, vol. number , 2019: according to kutsana and smith (2004), book-built ipos possessed fewer uncertainties due to the soliciting investors’ demands during roadshow. accordingly, this could reduce information aymmetry among the book-built ipos and thus influencing the underwriters’ decision on setting offer prices. finally, this study controlled ipos market condition (ipomkt) when ipo pricing. the market condition indicated investor’s confidence in the stock market performances. the measurement by kooli and suret (2004) was adopted, where the ipo market condition was specified into ‘‘hot’’ and ‘‘cold’’ markets. investors had higher optimism and confidence with the ipos issued in the hot market (helwege & liang, 2004). due to its potential for ipo success, offer price is expected to be determined higher. 3.3 model specification to examine the effect of accounting information on offer price in the malaysian market, the ordinary least square (ols) regression model was employed in the present study. the cross-sectional regression model is constructed as below: opi = β0 + β1leveragei + β2roei + β3lnoffszi + β4undwri + β5boardi + β6 booki + β7 ipomkti + εi (1) the summary of the variables’ measurements in eq. (1) was provided in table 2. table 2: variables and measurements. variables definition opi offer price of ipo leveragei leveragei = total liabilities prior to incorporation of proceeds total asset prior to incorporation of proceeds roei roei = net income prior a year of ipo listing shareholder equity prior a year of ipo listing lnoffszi lnoffszi = ln(number of shares issued at the ipo x op) 13 leverage and ipo pricing: evidence from malaysia undwri the measurement based on megginson and weiss (1991) is described in the below formula: undwri = undti totundti where, undti = total underwriting amount for ith underwriters in the listing year, and totundti = total underwriting amount in ith listing year. underwriters is considered as prestigious if the amount of proceeds raised by underwriters is greater than the average of the total underwriting amount in the particular year of listing. thereby, a dummy variable takes a value of one if an ipo was issued by prestigious underwriter. undwri = 1 =ipo underwritten by prestigious underwriter 0 =ipo underwritten by less reputable underwriter boardi boardi = 1 =ipo listed in main market 0 = ipo listed in ace market booki booki = 1=ipo issued by book-building mechanism 0 =ipo issued by fixed-price mechanism ipomkti ipos are listed in hot market when the sum of the ipo offers exceeds the average of the sample issue size of the year listing. henceforth, a dummy variable takes a value of one if an ipo was issued in hot market. ipomkti = 1 =ipo listed in the hot market 0 =ipo listed in the cold market 4. analysis of results prior to further empirical analysis, data cleansing process was conducted to eliminate the outliers of the ipo samples. accordingly, 129 malaysian ipos were present for further empirical analysis. table 3 presents the results of descriptive statistics on variables. table 3: descriptive statistics. notes: ipo sample size (n) = 129, starting from january 2009 to december 2018. min= minimum value; maximum= maximum value; s.d= standard deviation. variables mean median min max s.d offer price (rm) 0.963 0.680 0.120 6.500 0.901 debt ratio 0.443 0.456 172.6 0.833 0.199 return on equity (percent) 27.068 24.824 3.297 89.187 14.596 proceeds (rm millions) 302 43.750 7.321 6370 833 12 the international journal of banking and finance, vol. number , 2019: according to kutsana and smith (2004), book-built ipos possessed fewer uncertainties due to the soliciting investors’ demands during roadshow. accordingly, this could reduce information aymmetry among the book-built ipos and thus influencing the underwriters’ decision on setting offer prices. finally, this study controlled ipos market condition (ipomkt) when ipo pricing. the market condition indicated investor’s confidence in the stock market performances. the measurement by kooli and suret (2004) was adopted, where the ipo market condition was specified into ‘‘hot’’ and ‘‘cold’’ markets. investors had higher optimism and confidence with the ipos issued in the hot market (helwege & liang, 2004). due to its potential for ipo success, offer price is expected to be determined higher. 3.3 model specification to examine the effect of accounting information on offer price in the malaysian market, the ordinary least square (ols) regression model was employed in the present study. the cross-sectional regression model is constructed as below: opi = β0 + β1leveragei + β2roei + β3lnoffszi + β4undwri + β5boardi + β6 booki + β7 ipomkti + εi (1) the summary of the variables’ measurements in eq. (1) was provided in table 2. table 2: variables and measurements. variables definition opi offer price of ipo leveragei leveragei = total liabilities prior to incorporation of proceeds total asset prior to incorporation of proceeds roei roei = net income prior a year of ipo listing shareholder equity prior a year of ipo listing lnoffszi lnoffszi = ln(number of shares issued at the ipo x op) 10 the international journal of banking and finance, vol. 15, no 1, 2020 : 1-19 4. analysis of results prior to further empirical analysis, data cleansing process was conducted to eliminate outliers in the ipo samples. accordingly, 129 malaysian ipos were present for further empirical analysis. table 3 presents the results of descriptive statistics on variables. table 3. descriptive statistics variable mean median min max s.d offer price (rm) 0.963 0.680 0.120 6.500 0.901 debt ratio 0.443 0.456 172.6 0.833 0.199 return on equity (percent) 27.068 24.824 3.297 89.187 14.596 proceeds (rm million) 302 43.750 7.321 6370 833 note: ipo sample size (n) = 129, starting from january 2009 to december 2018. min= minimum value; maximum= maximum value; s.d= standard deviation. it was found that the average ipo offer price was rm 0.96, with a standard deviation of rm 0.90. therefore, a high dispersion in offer prices was indicated. based on the report by mohd-rashid et al. (2018) from 2000 to 2014, the average value of ipo offer price was lower than the rm0.96. abdulrahim and che-embi (2013) reported that this value was lower than rm1.11 13 leverage and ipo pricing: evidence from malaysia undwri the measurement based on megginson and weiss (1991) is described in the below formula: undwri = undti totundti where, undti = total underwriting amount for ith underwriters in the listing year, and totundti = total underwriting amount in ith listing year. underwriters is considered as prestigious if the amount of proceeds raised by underwriters is greater than the average of the total underwriting amount in the particular year of listing. thereby, a dummy variable takes a value of one if an ipo was issued by prestigious underwriter. undwri = 1 =ipo underwritten by prestigious underwriter 0 =ipo underwritten by less reputable underwriter boardi boardi = 1 =ipo listed in main market 0 = ipo listed in ace market booki booki = 1=ipo issued by book-building mechanism 0 =ipo issued by fixed-price mechanism ipomkti ipos are listed in hot market when the sum of the ipo offers exceeds the average of the sample issue size of the year listing. henceforth, a dummy variable takes a value of one if an ipo was issued in hot market. ipomkti = 1 =ipo listed in the hot market 0 =ipo listed in the cold market 4. analysis of results prior to further empirical analysis, data cleansing process was conducted to eliminate the outliers of the ipo samples. accordingly, 129 malaysian ipos were present for further empirical analysis. table 3 presents the results of descriptive statistics on variables. table 3: descriptive statistics. notes: ipo sample size (n) = 129, starting from january 2009 to december 2018. min= minimum value; maximum= maximum value; s.d= standard deviation. variables mean median min max s.d offer price (rm) 0.963 0.680 0.120 6.500 0.901 debt ratio 0.443 0.456 172.6 0.833 0.199 return on equity (percent) 27.068 24.824 3.297 89.187 14.596 proceeds (rm millions) 302 43.750 7.321 6370 833 leverage and ipo pricing: evidence from malaysia : 1-19 11 from 1999 to 2008. notably, the maximum value for offer price amounted to rm6.50, as shown by lotte chemical titan holdings berhad listed in 2017. conversely, pasukhas group berhad listed in 2012 reported that the minimum value for offer price amounted to rm 0.12. in respect of explanatory variable, it was found in this study that an organisation named cypark resources berhad listed in 2010 issued the highest debt ratio at 0.833. meanwhile, the lowest debt ratio at 172.6 was found in this study, represented by ea holdings berhad listed in 2010. furthermore, the average debt ratio amounted to 0.443, indicating that approximately 44.48 per cent of total assets were used for financing prior to ipo listing. the average value of debt ratio was slightly higher than 0.416 according to rashid, ibrahim, othman, and see (2012) from 2004 to 2008. meanwhile, 0.412 of debt ratio was found by tajuddin, abdullah, and taufil-mohd (2016) from 2005 to 2014. nevertheless, the average debt ratio reported in this study was 0.51 lower, as reported by ammer and ahmad-zaluki (2016) from 2002 to 2014 in the malaysian ipo market. in respect of control variables, it was found that the average value for return on equity was 27.07 per cent, indicating that firms were capable of using their assets to generate profits at an average of 27.07 per cent to shareholders. furthermore, the return on equity was at its minimum value of 3.30 per cent, while its maximum value amounted to 89.19 per cent. there was an average value of rm 302 million for the ipo proceeds, as reflected by the total amount raised by firms for ipos in ringgit malaysia. this value included the highest proceeds of rm 6370 million and the lowest proceeds of rm 7.321 million. notably, this average figure was higher than the figure reported by mohd-rashid et al. (2014) from 2000 to 2012. it was also higher than the figure reported by abdul-rahim and che-embi (2013) from 1999 to 2008 for the malaysian ipo market. table 4 illustrates the correlation matrix of all variables. based on the table, it was found that most of the variables possessed a correlation lower than 0.7, with four pairs of the relationship as exceptions. the four relationship pairs included op and lnoffsz, op and book, lnoffsz and book, and lnoffsz and ipomkt, which presented positive coefficients of: 0.807, 0.706, 0.804, and 0.706, respectively. the occurrence of high coefficients indicated possible strong correlation between the two variables. further analysis was conducted on the two variables using vif (variance inflation factors) prior to regression analysis. the result showed that the vif values for these variables were constantly lower than 9.52. this was a value below the 10.00 cut-off point (gujarati, 2003; hair, black, babin, & anderson, 2010). this value indicated the absence of multicollinearity issue in these variables. to ensure the reliability of the regression models and their results, diagnostic tests, such as the identification of autocorrelation issue (durbin-watson test) and heteroskedasticity test (white’s test) was performed to validate ols assumptions. 12 the international journal of banking and finance, vol. 15, no 1, 2020 : 1-19 table 4. pearson’s correlation matrix results between variables. op 1. 2. 3. 4. 5. 6. 1. leverage -0.090 2. roe 0.038 0.076 3. lnoffsz 0.807 0.293 -0.057 4. undrw 0.412 0.219 0.159 0.487 5. board 0.497 0.091 -0.120 0.512 0.276 6. book 0.706 0.241 -0.076 0.803 0.434 0.292 7. ipomkt 0.520 0.252 -0.083 0.706 0.394 0.129 0.656 note: op = offer price; leverage = total liabilities (short-term + long-term liabilities) over total assets; roe= return on equity; lnoffsz = natural logarithm of offer size; board = dummy variable of market board; book= dummy variable of book-building method issue; ipomkt= dummy variable of ipo market conditions; the numbers in the column headings correspond to the number of variables in the row headings. table 5 presents a comparison between high and low debt ratios in terms of mean differences. specifically, the high debt ratio group displayed an average offer price of rm 0.92, while the low debt ratio group recorded an average offer price of rm 1.10. furthermore, it was found that issuers determined the ipos with low debt ratios at higher offer prices. meanwhile, the level of leverage was highly diverse across high and low debt ratio groups. specifically, the high debt ratio group possessed a mean leverage value of 67.4 per cent, while the mean leverage value of the low debt ratio group amounted to 20.1 per cent. the findings shown in table 5 proved that the high return of equity and large shares offering were associated with high debt ratio ipos. moreover, prestigious underwriters showed a preference for issuing ipos with high debt ratios. firms issued through the book-building mechanism and those issued in the hot market possessed high debt ratios. there was a high mean difference for offer price, leverage, returns on ipo market conditions between the high and low debt ratio groups. the difference in equity, firm size, reputation of underwriters, dummy book-building, and dummy was at the level of 10 per cent or lower, with the dummy of market board as an exception. leverage and ipo pricing: evidence from malaysia : 1-19 13 table 5. mean difference between high and low debt ratio ipos. debt ratio mean difference t-test mann-whitney test variable high low t-statistics z-statistics op 0.916 1.097 -0.181 0.812* -1.850* leverage 67.4 20.1 47.3 25.625*** -7.550*** roe 0.306 0.245 0.061 1.824* -1.936* lnoffsz 18.590 17.536 1.054 3.312*** -3.220*** undrw 0.530 0.260 0.270 2.494** -2.412** board 0.740 0.670 0.070 0.666 -0.668 book 0.290 0.050 0.240 2.904*** -2.771*** ipomkt 0.320 0.080 0.240 2.739*** -2.629*** note: ***, ** and * denote statistical significance at the 1 per cent, 5 per cent and 10 per cent levels, respectively. according to fama and french’s (1993), this study segregated the high and low debt ratios by taking ipos with the lowest debt ratio quartile (30 per cent lowest from the sample) and represented them as low-debt ratio portfolio whereas ipos in the highest debt ratio quartile (30 per cent highest from the sample) to be represented as high-debt ratio portfolio. table 6 illustrates the findings from the cross-sectional regressions regarding the estimation of ipo offer price through leverage and profitability variables. no autocorrelation issue was present after the implementation of the correction by the newey-west method as the dublin-watson value amounted to 1.7959. the adjusted r-squared value for the regression model was 68.58 per cent. it was inferred from this figure that all explanatory variables could elaborate on the variations in ipo offer price in the model, which amounted to at least 68.58 per cent. additionally, the f-statistics values for the regression model were significant, and the goodness-of-fit of models (p<0.01) were validated. the coefficients of predicting variables involved an explanatory variable known as debt ratio. these coefficients were placed in the respective directions as predicted in the hypothesis presented in table 5. the directions of all the coefficients of predicting variables in all models were as per prediction, with the ipo market condition (ipomkt) as the exception. however, the debt ratio did not affect the offer price set as the results obtained were insignificant. in respect of leverage variable, a negative coefficient was reported on debt ratio (leverage) when the offer price was estimated. therefore, hypothesis h was accepted in this study. it was indicated from this finding that high debt ratios in the firms’ balance sheets led underwriters and issuers to set a low offer price, 1 14 the international journal of banking and finance, vol. 15, no 1, 2020 : 1-19 while the low debt ratios prompted them to set a high offer price. this finding could be explained through information asymmetry theory (rock, 1986). table 6. regression results for pre-ipo leverage model for 129 malaysian ipos, listed from january 2009 to december 2018. variable coefficient t-statistics p-value leverage -0.7110 -1.8131* 0.0723 roe 0.2048 0.7752 0.4398 lnoffsz 0.4040 3.7621*** 0.0003 undrw 0.0371 0.3928 0.6951 board 0.2177 1.7067* 0.0904 book 0.5090 3.1622*** 0.0020 ipomkt -0.1577 -0.9110 0.3641 c -6.2684 -3.7559*** 0.0003 adjusted 0.6858 f-statistics 40.9057*** dublin-watson 1.7959 note: autocorrelation and heteroskedasticity problems are determined by conducting dublin-watson test and white test, respectively followed with correction by newey-west covariance estimator. the highlighted adjusted values with the symbols: *, ** and *** represent significance at levels of 0.1, 0.05, and 0.01, respectively. moreover, high-leverage firms were exposed to higher financial risks, such as financial distress (wang et al., 2015). high financial risks in firms were associated with higher information asymmetries, offering their most disadvantageous qualities to the market (su, 2004). subsequently, even when the organisations appeared in public, institutional investors would not be interested in subscribing to them. this was due to their awareness that high-leveraged firms would be overpriced after the ipo listing. in contrast, retail investors would receive share allocations from high-leveraged ipos as they were not informed about the firms’ prospects. as most malaysian ipos were determined by a fixedprice mechanism, underwriters and issuers offered high-leveraged ipos at low offer prices to compensate for investors who were trapped in "curses". these findings were in agreement with reber and vencappa (2016) and peng and wang (2007). leverage and ipo pricing: evidence from malaysia : 1-19 15 significant results were found between three control variables and ipo offer price. specifically, firm size (lnoffsz) had a significant positive association with ipo offer price. therefore, it was indicated that a high number of new shares offered by large firms displayed lower information asymmetries. these findings were in agreement with yung and zender (2010) and mohd-rashid et al. (2018). subsequently, higher ipo prices were set to indicate the firms’ qualities. furthermore, it was suggested from the results about board types (board) in the regression model that the ipos listed in the main market were related to significant high offer prices. according to yong (2015), the ipos listed in the ace market were exposed to high information asymmetries due to insufficient information on the firms’ track record and challenges in securing conventional sources of finance. for this reason, underwriters set higher offer prices for the ipos listed in the main market. it was indicated from the results for dummy book-building (book) that the ipos priced by the book-building mechanism were positively related to offer prices. this finding was supported by kutsana and smith (2004), who proposed that book-built ipos possessed low information asymmetries as investors’ demands were acquired by underwriters during road shows. as for malaysia, large and developed firms used book-building mechanisms to issue ipos for improvement in the credibility of firms’ signal qualities. therefore, issuers set high offer prices for book-built ipos. other control variables, such as return on equity (roe) and underwriter reputation (undwr) were positively related to ipo offer price. however, they displayed insignificant results. the positive coefficient on roe suggested that issuers and underwriters often set high offer prices for firms with high roe prior to ipo listing. therefore, it was indicated that investors had a preference for the subscription of ipos with higher profitability as the signal of firms was more superior. in respect of underwriter reputation (undwr), the positive signs indicated that high-priced ipos were offered, provided if they were issued by prestigious underwriters. in this case, prestigious underwriters were employed to reduce information asymmetries according to certification theory (carter & manaster, 1990). such findings contradicted with mohd-rashid et al. (2018) who reported a negative coefficient on underwriter reputation when estimating offer price. another control variable, namely ipo market conditions (ipomkt) had a negative insignificant relationship with ipo offer price. optimistic investors were supposed to garner interests in ipos issued in the hot market, resulting in a high offer price being set. however, the negative sign indicated underwriters and issuers’ expectations that investors would be interested in the subscription of ipos with low offer price. therefore, issuers arranged their listing in the hot market while promoting low offer price of ipos to investors. 16 the international journal of banking and finance, vol. 15, no 1, 2020 : 1-19 5. conclusion determining an appropriate offer price is a challenging task for underwriters and issuers. the concerns of issuers and underwriters about accounting variable based on pro-forma financial statements in ipo prospectuses might influence offer prices compared to non-financial information. this study aims to investigate the influence of pre-ipo leverage on offer price by employing a sample of 130 ipos listed on bursa malaysia. this study began from january 2009 to december 2018, which was the duration after the sub-prime crisis in 2008. based on the cross-sectional regression analysis, a negative coefficient of pre-ipo leverage was found when the ipo offer price was estimated. therefore, it was indicated that organisations which issued high debts prior to ipo listing posed a higher extent of financial risks. these risks were associated with higher information asymmetries. subsequently, institutional investors would not be interested in the ipo subscription despite the public appearance of the firms. consequently, retail investors received the allocation of shares from high-leveraged ipos. in the case of malaysian ipos, their cost was mostly determined by fixed-price mechanism. additionally, underwriters and issuers offered high-leveraged ipos at low offer price to compensate for investors who were trapped in “curses”. the implication of this study could be seen from the findings regarding accounting information, specifically the leverage variable that investors need to consider when valuing ipos and making decisions about share subscription. this study’s findings have also resulted in guidelines for regulators when approving offer prices of listed firms. these guidelines would help ensure the effectiveness of offer prices in protecting shareholders’ interests. it is recommended that future studies focus on other influencing factors of ipo offer price through the introduction of lock-up provision. 6. acknowledgements the authors would like to acknowledge their gratitude to the ministry of higher education, malaysia for funding from the fundamental research grant scheme (frgs/1/2018/ss01/uum/02/7) (s/o code: 14203). references abdul-rahim, r., & che-embi, n.a. (2013). initial returns of shariah versus non-shariah ipos: are there any differences? jurnal pengurusan, 39, 3750. leverage and ipo pricing: evidence from malaysia : 1-19 17 aggarwal, r., bhagat, s., & rangan, s. (2009). the impact of fundamentals on ipo valuation. financial management, 38(2), 253–284. ahmed sheikh, n., & wang, z. (2011). determinants of capital structure: an empirical study of firms in manufacturing industry of pakistan. managerial finance, 37(2), 117–133. akyol, a., cooper, t., meoli, m., & vismara, s. (2014). do regulatory changes affect the underpricing of european ipos? journal of banking and finance, 45, 43–58. allen, f., & faulhaber, g. (1989). signalling by underpricing in the ipo market. journal of financial economics, 23, 303–323. ammer, m.a., & ahmad-zaluki, n.a. (2016). the effect of disclosure regulation on the bias and accuracy of management earnings forecasts in malaysian ipo prospectuses. journal of financial reporting and accounting, 15(1), 59–77. beatty, r., & ritter, j. (1986). investment banking, reputation, and the underpricing of initial public offerings. journal of financial economics, 15(1–2), 213–232. carey, p., fang, v., & zhang, h. f. (2016). the role of optimistic news stories in ipo pricing. journal of international financial markets, institutions & money, 41, 16–29. carter, r., & manaster, s. (1990). initial public offerings and underwriter reputation. the journal of finance, 45(4), 1045–1067. chen, y., wang, s., li, w., sun, q., & tong, w. (2015). institutional environment, firm ownership, and ipo first-day returns: evidence from china. journal of corporate finance, 32, 150–168. cotter, j., goyen, m., & hegarty, s. (2005). offer pricing of australian industrial initial public offerings. accounting and finance, 45, 95–125. fama, e. f., & french, k. r. (1993). common risk factors in the returns on stocks and bonds. journal of financial economics, 33(1), 3–56. grinblatt, m., & hwang, c. y. (1989). signalling and the pricing of new issues. the journal of finance, 44, 393–420. gujarati, d. n. (2003). basic econometrics. new york: mcgraw hill. hair, j. f., black, w. c., babin, b. j., & anderson, r. e. (2010). multivariate data analysis: a global perspective. london, u.k.: pearson education. helwege, j., & liang, n. (2004). initial public offerings in hot and cold markets. journal of financial and quantitative analysis, 39(3), 541–569. how, j., lam, j., & yeo, j. (2007). the use of comparable firm approach in valuing australian ipos. international review of financial analysis, 16, 99–115. huang, h., chiang, m., lin, j., & lin, y. (2017). fixed-price, auction, and bookbuilding ipos: empirical evidence in taiwan. finance research letters, 22, 11–19. 18 the international journal of banking and finance, vol. 15, no 1, 2020 : 1-19 keasey, k., & mcguinness, p. (1991). prospectus earnings forecasts and the pricing of new issues on the unlisted securities market. accounting and business research, 21(82), 133–145. kim, j., pukthuanthong-le, k., & walker, t. (2007). leverage and ipo underpricing: high-tech versus low-tech ipos. management decision, 46(1), 106–130. kim, m., & ritter, j.r. (1999). valuing ipos. journal of financial economics, 53, 409–437. kooli, m., & suret, j. (2004). the aftermarket performance of initial public offerings in canada. journal of multinational financial management, 14(1), 47–66. kutsana, k., & smith, r. (2004). why does book building drive out auction methods of ipo issuance? evidence from japan. review of financial studies, 17(4), 1129–1166. lai, p., & lo, c. (2012). the empirical study of ipo in hong kong china: with the initial offer price to the ipo reflected its fundamentals information significantly. international journal of business and social science, 3(19), 23–41. leland, h. e., & pyle, d. h. (1977). information asymmetries, financial structures and financial intermediation. journal of finance, 32, 371– 387. megginson, w., & weiss, k. (1991). venture capitalist certification in initial public offerings. the journal of finance, 46, 879–903. mohd-rashid, r., abdul-rahim, r., & yong, o. (2014). the influence of lockup provisions on ipo initial returns: evidence from an emerging market. economic systems, 38(4), 487–501. mohd-rashid, r., masih, m., abdul-rahim, r., & che-yahya, n. (2018). does prospectus information matter in ipo pricing? journal of islamic accounting and business research, 9(4), 514–530. myers, s., & majluf, n. (1984). corporate financing and investment decisions when firms have information that investors do not have. journal of financial economics, 13, 187–221. peng, y., & wang, k. (2007). ipo underpricing and flotation methods in taiwan — a stochastic frontier approach. applied economics, 39, 2785–2796. rashid, a. a., ibrahim, m. k., othman, r., & see, k. f. (2012). ic disclosure in ipo prospectuses: evidence from malaysia. journal of intellectual capital, 13(1), 57–80. reber, b., & vencappa, d. (2016). deliberate premarket underpricing and aftermarket mispricing: new insights on ipo pricing. international review of financial analysis, 44, 18–33. rock, k. (1986). why new issues are underpriced. journal of financial economics, 15(1), 187–212. leverage and ipo pricing: evidence from malaysia : 1-19 19 sahoo, s., & rajib, p. (2012). determinants of pricing ipos: an empirical investigation. south asian journal of management, 19(4), 59–87. su, d. (2004). leverage, insider ownership, and the underpricing of ipos in china. journal of international financial markets, institutional and money, 14, 37–54. tajuddin, a. h., mohd-rashid, r., abdullah, n. a. h., & abdul-rahim, r. (2015). an empirical examination of over-subscription in the malaysian ipo market. international journal of economics and management, 9(s), 81–102. tajuddin, a. h., abdullah, n. a. h., & taufil-mohd, k. n. (2016). does growth opportunity matter in explaining the oversubscription phenomena of malaysian ipo? procediasocial and behavioral sciences, 219, 748– 754. welch, i. (1989). seasoned offerings, imitation costs, and the underpricing of initial public offerings. the journal of finance, 44, 421–449. yong, o. (2015). what is the “true” value of an initial public offering? journal of scientific research and development, 2(10), 78–85. modelling and forecasting volatility in the gold market 48 international journal of banking and finance, volume 9 (number 1), 2012: pages 48-80 modelling and forecasting volatility in the gold market stefan trück and kevin liang macquarie university, australia _____________________________________________ abstract we investigate the volatility dynamics of gold markets. while there are a number of recent studies examining volatility and value-at-risk (var) measures in financial and commodity markets, none of them focuses on the gold market. we use a large number of statistical models to model and then forecast daily volatility and var. both insample and out-of-sample forecasts are evaluated using appropriate evaluation measures. for in-sample forecasting, the class of tarch models provide the best results. for out-of-sample forecasting, the results were not that clear-cut and the order and specification of the models were found to be an important factor in determining model’s performance. var for traders with long and short positions were evaluated by comparing failure rates and a simple ar as well as a tarch model perform best for the considered back-testing period. overall, most models outperform a benchmark random walk model, while none of the considered models performed significantly better than the rest with respect to all adopted criteria. key words: gold markets, volatility, forecasting, value-at-risk, backtesting jel classification: g17, c22, g32 ____________________________________________ 1. introduction the recent global financial crisis has highlighted the need for financial institutions to find and implement appropriate models for risk quantification. hereby, in particular value-at-risk (var) and volatility estimates were subject to significant changes during 2007-9 financial turmoil in comparison to normal market behaviour. further, as the risk in equity and bond markets was increasing, there was a particular interest of investors to increase their positions in the gold market. this study evaluates the trück and liang: forecasting volatility in the gold market effectiveness of various volatility models with respect to forecasting market risk in the gold bullion market. while there is a stream of literature examining performance of models for volatility and var, this is a pioneer study to particularly focus on the gold market. despite the important role gold plays for risk management and hedging in financial markets, there has been relatively little literature on the estimation of volatility of gold. exceptions include the studies by mills (2003), tully and lucey (2006), canarella and pollard (2008), morales (2008) and jun (2009). generally, the gold market has a significant and unique role in financial markets as a safe haven that is also used for hedging and diversification. while there is no theoretical reason why gold is referred to as a safe haven asset, historical evidence suggests that investments in the gold market spikes during times of turmoil in other financial markets. one explanation could be that it is one of the oldest forms of money and was traditionally used as an inflation hedge. moreover, gold is often uncorrelated or even negatively correlated with other types of assets. this is an important quality that allows gold to act as a diversification asset in portfolios, since a more globalised market has led to the increase in correlation among other assets. this became also evident during the financial crisis of 2007-2009 where the negative effect of one market readily flowed into other markets, yet the gold market remained relatively unscathed during this period of turbulence. so far there has been no study using volatility and var modelling in the spot gold and gold futures markets. gold market research has concentrated on the role of gold as a hedging or diversification tool, in particular as a safe haven during market crashes. this study examines various models that can be used in forecasting volatility, to evaluate their respective performance. finding appropriate models for volatility is of interest for several reasons: firstly, it is an integral factor of derivative security pricing, for example, in the classic black-scholes model or alternative option pricing formulas. secondly, as a representation of risk, volatility plays an important role in an investor’s decision making process. volatility is not only of great concern for investors but also policy makers and regulators who are interested in the effect of volatility on the stability of financial markets in particular and the whole economy in general. finally, volatility estimation is an essential input in many var models, as well as for a number of applications in a firms market risk management practices. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 the remainder of the paper is set up as follows. section 2 provides a brief review on the global gold market and studies on volatility modelling of financial markets in general and gold markets in particular. section 3 provides an overview on the data and techniques used in this study. in particular various models for volatility forecasting and evaluating model performance are reviewed. empirical results of the study are reported in section 4 while section 5 concludes. 2. gold markets and volatility models 2.1 the gold market gold has been used throughout history as a form of payment and has been a standard for currency equivalents to many economic regions or countries. in spite of its historical monetary significance, a free functioning world market only came of age in recent times. before 1971, the gold standard was mostly used in various times in history, where domestic currencies have been backed by gold. the system existed until 1971, when the us stopped the direct convertibility of the united states dollar to gold, effectively causing the system to break down. since then, a global market for gold in its own right developed, remaining open around the clock and open to a range of derivative instruments. the market for gold consists of a physical market in which gold bullions and coins are bought and sold and there is a paper gold market, which involves trading in claims to physical stock rather than the stock themselves. physical gold is generally traded in the form of bullions. the bullion market serves as a conduit between larger gold suppliers such as producers, refiners and central banks and smaller investors and fabricators. the bullion market is essentially a spot market, but is complemented by the use of forward trading for the hedging of physical positions. since 1919, the most widely accepted benchmark for the price of gold is known as the london gold fixing, a twice-daily (telephone) meeting of representatives from trück and liang: forecasting volatility in the gold market five bullion-trading firms. 1 furthermore, there is active gold trading based on the intra-day spot price, derived from gold-trading markets around the world as they open and close throughout the day. the key prices in the london bullion market are the spot (fixings) price, the forward price and the lease rate. the spot (fixings) price is a daily clearing or fix price obtained by balancing purchases and sales ordered through its members. the forward price (gofo) is the simultaneous purchase and sales price of gold forward contracts of various lengths. generally, the gofo rate is expressed as an annual percentage. finally, the lease rate refers to short-term loans denominated in gold and is expressed as an annualized interest rate. since 1971 the price of gold has been highly volatile, ranging from a high of us$850 on january 21, 1980, to a low of us$252.90 on june 21, 1999. the period from 1999 to 2001 marked the so-called brown bottom after a 20-year bear market. prices increased rapidly from 1991, but the 1980 high was not exceeded until 2008 when a new maximum of $865.35 was set on january 3, 2008. another record price was set on march 17, 2008 at $1023.50. in the second half of 2009, gold markets experience renewed momentum upwards due to increased demand and a weakening us dollar. overall, since april 2001, the gold price has more than tripled in value against the us dollar. 2.2 factors influencing gold prices as mentioned above, gold has a unique place in financial markets. of all the precious metals, gold is the most popular as an investment. investors generally buy gold as a hedge or safe haven against any economic, political, social or currency-based crises. these crises include investment market declines, burgeoning national debt, currency failure, inflation but also scenarios like war or social unrest. as in any commodities, the price of gold is ultimately driven by its supply and demand. however, unlike other resources, hoarding and disposal plays a much bigger role in price formation because most of the gold ever mined still exists and is potentially able to enter the market for the right price. given the huge quantity of stored gold, 1 all five members bank of nova scotia–scotiamocatta, barclays bank plc, deutsche bank ag, hsbc bank usa, na and société générale are market making members of the london bullion market association (lbma). international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 compared to the annual production, the price of gold is mainly affected by changes in sentiment, rather than changes in the actual annual production. also macroeconomic factors such as low real interest rates can have an effect on gold price. if the return on bonds, equities and real estate is not adequately compensating for risk and inflation, then the demand for gold and other alternative investments such as commodities increases. an example of this is the period of stagflation that occurred during the 1970s which led to an economic bubble forming in precious metals. financial market declines such as the 2007-9 global financial crisis usually leads investors to look for alternative and less volatile investment opportunities for their funds. it will also increase the need for investors to hedge their portfolios to minimise their risk in case of further decline. the demand for gold and, thus, its price increase, empirically is due to the role of gold as a safe haven in times of crises. this is one of the major reasons to drive gold prices to new highs throughout the post-financial crisis period. central banks and the international monetary fund (imf) also play an important role in determining the gold price. at the end of 2004 central banks and official organizations held 19 percent of all above-ground gold as official gold reserves. thus, they have a significant influence on the gold market not only as a major buyer and seller. also, speculation on their future gold holding levels can also be a driving factor. recently, the assumption that central banks around the world will increase their gold reserve levels as a hedge against the falling us dollar has also contributed to the rise of gold prices. the performance of gold bullion is often compared to stocks. however, they are fundamentally different asset classes. gold is regarded by some as a store of value (without growth) whereas stocks are regarded as a return on value. stocks and bonds perform best in periods of economic stability and growth, whereas gold is seen as the asset to hold in times of uncertainty and crisis. throughout history there has been a cyclical run with long periods of stock outperformance followed by long periods of gold outperformance. over the long term, equity markets have been able to outperform gold overall. trück and liang: forecasting volatility in the gold market 2.3 volatility models within the last three decades various approaches to volatility modelling have been suggested in the econometric and financial literature. in the following we will provide a brief overview of developments in the literature starting with the autoregressive conditional heteroskedasticity (arch) models (engle, 1982). bollerslev (1986) introduced the generalised arch (garch) model. the latter is often utilised in financial market studies. the general idea is to predict the current period’s variance by forming a weighted average of a long term average, the forecasted variance from last period, and information about volatility observed in the previous period. if the return is unexpectedly large either in the upward or the downward direction, then the trader will increase the estimate of the variance for the next period. this model is also consistent with the volatility clustering often seen in financial returns data, where large changes in returns are likely to be followed by further large changes. since the introduction of these models, they have been widely used in volatility modelling and forecasting. researchers such as french et al. (1987) and akgiray (1989) utilised garch models to capture the behaviour of stock market price volatilities. argiray (1989) compared the garch (1,1) model to other historical estimation methods and found that the garch (1,1) model outperformed its competitors. many extensions of the garch model have been introduced in the literature since: e.g. garch-in-mean (garch-m) models (engle et al., 1987), egarch models (nelson, 1991), threshold arch (tarch) and threshold garch (tgarch) (glosten, jaganathan, and runkle, 1993; zakoïan, 1994) and power arch (parch) models (ding et al., 1993) just to name a few. a number of studies have focused on optimal model specification and the performance of various garch models in financial markets providing no clear-cut results. hansen and lunde (2005) carried out comprehensive testing of 330 variants of arch type models on their performance in estimating volatility in exchange rates and stock returns. the study found that the garch (1,1) model outperforms other models in estimating exchange rate volatilities but underperforms in estimating stock returns. mcmillan et al. (2000) tested a set of ten volatility estimation models including international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 random walk, moving average and garch models in forecasting uk stock market returns at different frequencies. they found that the performance of each model varied depending on the length of frequencies, the series as well as the type of loss function being applied. the random walk model outperformed others at the monthly frequency, while garch and moving average models were superior using daily forecasts. brooks and persand (2002, 2003) examine various arch and garch type models with respect to volatility forecasting. they report that, while the forecasting performance of the models depended on the considered data series and time horizon, the overall most preferred model is a simple garch(1,1). this is also consistent with many other studies such as e.g. bollerslev et al. (1992). on the other hand, braisfold and faff (1996) evaluate volatility models in forecasting stock returns, and find that none of the models significantly outperforms the others. recently, also a stream of literature has emerged focusing on modelling and forecasting volatility with respect to the quantification of value-at-risk (var). as pointed out by jorion (1996), var plays a substantial role in managing risks for financial institutions. the importance of the var measure is further highlighted by regulators in the basel committee on banking supervision. 2 the performance of volatility models with respect to appropriate quantification of var has been investigated by danielsson and de vries (2000): conditional parametric methods such as the garch model significantly underpredict the var of u.s. stock returns. giot and laurent (2001, 2003) investigate volatility models for both negative and positive returns, with the latter representing risk for short position holders. they find that skewed asymmetric arch models using the student t distribution perform best with respect to risk quantification. sadorsky (2006), investigating oil price volatility, tested a great variety of volatility models by evaluating the forecasting performance using different var measures. his findings suggest that while no model could consistently outperform the others, a garch model as well as a tgarch performed quite well for modelling and forecasting the volatility and risk of oil prices. tully and lucey (2006) examine various macroeconomic influences on gold using models including the asymmetric power garch model (apgarch) for spot 2 the basle committee on banking supervision announced in april 1995 that capital adequacy requirements for commercial banks are to be based on var. trück and liang: forecasting volatility in the gold market and futures prices over a 20 year period, paying special attention to periods of stock market crashes. their results suggest that the price of gold is significantly influenced by the u.s. dollar while during periods of financial crises an apgarch model performs best with respect to volatility. mills (2003) investigates the statistical behaviour of daily gold prices, and finds that price volatility scaling with long-run correlations is important while gold returns are characterised by short-run persistence and scaling with a break point of 15 days. canarella and pollard (2008) apply power garch model to the london gold market fixings to investigate long memory features as well as conditional volatility behaviour of the returns. they find that apgarch models were able to adequately capture long memory in returns and that market shocks have strong asymmetric effects: conditional volatilities of gold prices are affected more by good news (positive shocks) than bad news (negative shocks). morales (2008) discusses volatility spill-over effects between precious metal markets using garch and egarch techniques. gold was found to be influenced by prices of other precious metals, but there was little evidence to suggest other precious metals influencing gold prices. 3. data and models 3.1 the data the data for this study are daily pm gold fixing prices on the london bullion market available from the official the london bullion market association website (www.lbma.org.uk). the market is a wholesale over-the-counter (otc) market for gold and silver. the fixings are the internationally published benchmarks for precious metals. the gold fixing is conducted twice a day by five gold fixing members, at 10:30 am and 3:00 pm. this study will use the daily pm fixings price released at 3:00 pm as quoted in usd. the data cover 2508 observations from 4 january 1999 to 30 december 2008. the time series exhibits a number of price shocks, e.g. during the periods around september 11, 2001, the beginning of the iraq war in 2003 as well as the global financial crisis in 2008. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 for the observed gold fixing prices pt, the daily log-returns are calculated as rt = ln (pt/pt-1). table 1 provides a summary of descriptive statistics for the considered return series. table 1: summery statistics of gold-fixing log returns gold fixing mean 0.044204 median 0.045767 maximum 7.005954 minimum -7.971887 std. dev. 1.143359 skewness -0.053361 kurtosis 8.533989 jarque-bera 3200.230 probability 0.000000 sum 110.8201 sum sq. dev. 3276.017 observations n = 2507 we observe that the mean and median of daily returns are positive indicating that overall gold prices were increasing during the considered time period. the magnitude of the average return (0.044%) is very small in comparison to its standard deviation (1.14%). further, the large kurtosis of 8.53 indicates the leptokurtic characteristics of daily returns. figure 1: time series and distribution of returns for gold fixing prices, 1999 to 2008 -12 -8 -4 0 4 8 500 1000 1500 2000 2500 r_daily 0 100 200 300 400 500 600 700 -8 -6 -4 -2 0 2 4 6 8 f re qu en cy r_daily trück and liang: forecasting volatility in the gold market obviously, the series has a distribution with tails that are significantly fatter than those of a normal distribution. this indication of non-normality is also supported by the jarque and bera (1980) test statistic, which rejects the null hypothesis of a normal distribution at all levels of significance. figure 1 provides a plot of the time series for the daily log-returns as well as a histogram of the return distribution. the figures indicate heteroscedasticity and volatility clustering for the return series that also exhibits a number of rather isolated extreme returns caused by unforeseen events or shocks to the gold market. we further test for stationarity of the return series using the augmented dick fuller (1979) (adf) and phillips perron (1988) (pp) unit root tests. the adf test is set to a lag length 0 using the schwarz information criterion (sic) and the pp test is conducted using the bartlett kernel spectral estimation method. results are reported in table 2, and indicate that for both tests the null hypothesis of a unit root is rejected. so the return series gold fixing prices can be considered to be stationary. table 2: results for augmented dickey-fuller and phillips-perron unit root tests for gold fixing return series (4 january 1999 -30 december 2008) t-statistic prob.* augmented dickey-fuller test statistic -49.24794 0.0001 adj. t-stat prob.* phillips-perron test statistic -49.29521 0.0001 3.2 considered models in the following, a variety of models is introduced for volatility modelling and forecasting of the daily returns. we will follow several studies in the literature, see e.g. sadorsky (2006), and measure the volatility of gold by its squared daily return: 𝜎�𝑡2 = 𝑟𝑡2 (1) thus, most of the models will be evaluated with respect to their ability to model and forecast the volatility measured by the squared return of the gold fixings price. the first model to be considered in the empirical analysis is a random walk model international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 (rw). if the volatility of gold market returns follows a random walk, the best forecast for the next period’s volatility is the volatility observed in the current period: 𝜎�𝑡+12 = 𝑟𝑡2 (2) this random walk model will be used as a benchmark model for the out-of-sample performance of the estimated models. the second standard class of models to be considered are historical mean (hm) models. in these models, the forecast for the volatility of the next period is the average of all previous volatilities. in particular, if 𝜎�𝑡2 is a random variable, which is uncorrelated with other observable variables and if 𝜎�𝑡2 is uncorrelated with its own past values, then the population mean can be considered as the optimal forecast. defining 𝜎𝑡2 = 𝑟𝑡2, the hm model can be denoted by 𝜎�𝑡+12 = 1 𝑇 � 𝜎𝑡−𝑖 2𝑇−1 𝑖=0 (3) a popular alternative to the hm model is the m-period moving average (ma) model. the forecast for the next period is based on the average of the last m observations. a value for m has to be determined. we decided to use moving averages of length m=20, 40 and 120 days, corresponding to about one month, two months and six months. the ma(m) model can be denoted by:. 𝜎�𝑡+12 = 1 𝑚 � 𝜎𝑡−𝑖 2𝑚−1 𝑖=0 (4) the next model we consider is the exponentially weighted moving average model (ewma). it forecasts the future volatility by applying weighting factors which decrease exponentially. that is, the method gives higher weights to more recent observations while still not discarding older observations entirely. it is calculated as the weighted average of the estimated volatility 𝜎�𝑡2 for day t (made at the end of day t1) and the value of volatility 𝜎𝑡2 observed on day t: 𝜎�𝑡+12 = 𝛼𝜎�𝑡2 + (1 − 𝛼)𝜎𝑡2 (5) the smoothing parameter α governs how responsive the forecast is to the most recent daily percentage change. generally, α lies between 0 and 1, and the process becomes a rw for α =0. a popular choice for the parameter α is based on j.p. trück and liang: forecasting volatility in the gold market morgan’s riskmetrics (1995) where it is suggested that α = 0.94 provides forecasts of the variance rate closest to the actual variance rate for a range of different market variables. an alternative is an ordinary least squares (ols) model. the relationship between volatility on day t and day t+1 is described based on a linear relationship: 𝜎�𝑡+12 = �̂�0 + �̂�1𝜎�𝑡2 (6) the parameter estimates are then determined by ols estimation. the model can be extended to an autoregressive (ar) model of order p where the current volatility is a linear function of the last p observations for the volatility. we implement a model of order p = 5 such that we estimate an ar(5) model that can be described by the following equation: 𝜎�𝑡+12 = �̂�0 + �̂�1𝜎�𝑡2 + �̂�2𝜎�𝑡−12 + �̂�3𝜎�𝑡−22 + �̂�4𝜎�𝑡−32 + �̂�5𝜎�𝑡−42 (7) we also consider a weighted moving average of disturbance terms model (mad) where the volatility in period t+1 is modelled as a function of the lagged values of the disturbance term εt. similar to the ar model, we decided to use a mad model of order 5 that can be described by the following equation: 𝜎�𝑡+12 = 𝛼�0 + 𝛼�1ε𝑡 + 𝛼�2ε𝑡−1 + 𝛼�3ε𝑡−2 + 𝛼�4ε𝑡−3 + 𝛼�5ε𝑡−4 (8) we decided to also use an autoregressive moving average (arma) or box-jenkins model that includes both an autoregressive (ar) and a moving average (mad) component. a simple arma(1,1) can then be described by the following equation: 𝜎�𝑡+12 = �̂�0 + �̂�1𝜎�𝑡2 − 𝛼�1ε𝑡 (9) since the introduction of autoregressive conditional heteroscedasticity (arch) models by engle (1982), the arch and even more the related garch (bollerslev, 1986) model have become standard tools for examining the volatility of financial variables. the model has proven to be very useful in capturing heteroskedastic behaviour or volatility clustering without the requirement of higher order models in various financial markets, see e.g. choudhy (1996) or sadorsky (2006). in a garch (1,1) model the conditional variance equation can be denoted by international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 ℎ𝑡 = 𝜔 + 𝛼𝜀𝑡−12 + 𝛽ℎ𝑡−1 (10) while the equation for the conditional mean is 𝑟𝑡 = 𝜋 + 𝜀𝑡, 𝜀𝑡~𝑁(𝑜, ℎ𝑡) (11) such that the one day ahead variance forecast can be expressed as: ℎ�𝑡+1 = 𝜔� + 𝛼�𝜀�̂�2 + �̂�ℎ�𝑡 (12) a popular extension of the garch (1,1) model is also the garch in mean (garch-m) model that was first proposed by engle et al. (1987). the garch-m model includes the conditional variance in the specified equation for the conditional mean. this allows for so-called time varying risk premiums. chou (1988) suggests that the dynamic structure of the conditional variance can be captured more flexibly by a garch-m model, using the following specification for the conditional mean: 𝑟𝑡 = 𝜋 + 𝛿ℎ𝑡 + 𝜀𝑡, 𝜀𝑡~𝑁(𝑜, ℎ𝑡) (13) another extension of standard arch and garch models has been suggested by glosten et al. (1994) and hentschel (1994): threshold arch (tarch) and garch (tgarch) models, which are popular in describing return asymmetry. large negative returns are often followed by a substantial increase in volatility such that the tarch and tgarch models distinguish between negative and positive returns. the tgarch model that will be considered in the empirical analysis treats the conditional standard deviation as a linear function of shocks and lagged standard deviations (hentschel, 1994) and is denoted by: ℎ𝑡 = 𝜔 + 𝛼𝜀𝑡−12 + 𝛽ℎ𝑡−1 + 𝛾𝜀𝑡−12 𝐷𝑡−1 (14) where 𝐷𝑡−1 is equal to 1 if εt < 0, and zero otherwise. obviously, in this model, 𝜀𝑡−12 > 0 , and 𝜀𝑡−12 < 0 will have different effects on the conditional variance. if 𝛾 ≠ 0, there is asymmetry in the model. if 𝛾 > 0, the occurrence of bad news will increase volatility and there is evidence of a leverage effect. trück and liang: forecasting volatility in the gold market 3.3 performance evaluation measures to evaluate the performance of the considered models, we apply a variety of measures such as mean squared error (mse), root mean squared error (rmse), mean absolute deviation (mad), mean absolute percentage error (mape) and the theil u statistic. the mse quantifies the difference between predicted and actually observed values by considering the squared difference between these two quantities: 2 2 2 1 1 ˆ( )t t t t mse t σ σ = = −∑ (15) the rmse is simply the root of mse and has the advantage of being measured in the same unit as the forecasted variable: 2 2 2 1 1 ˆ( )t t t t rmse t σ σ = = −∑ (16) the mae is also measured in the same unit as the forecast, but gives less weight to large forecast errors than the mse and rmse: 2 2 1 1 ˆ| |t t t t mae t σ σ = = −∑ (17) the mape measures the forecast quality independent of the unit of measurement of the variable. the measure might be less useful when the actual values of 2tσ are close to zero, because in this case the mape will take on large values even if the errors are fairly small in magnitude. another drawback of the mape is that if there are zero values (which may happen for daily squared returns) there will be a division by zero. we still decided to examine the results using the mape measure that can be denoted by: 2 2 2 1 ˆ| |100 t t t t t mape t σ σ σ= − = ∑ (18) we also investigate the forecasting performance using the theil u statistic that examines the rmse measure of a forecast against a naïve one step ahead forecast. if international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 the theil u statistic is smaller than 1, the tested forecast model outperforms the naïve model: if the u statistic is larger than 1, the naïve forecast is the better model. note that in our analysis we decided to use the rw model as the naïve benchmark model for forecasting. 𝑈 = 𝑅𝑀𝑆𝐸(𝑓𝑜𝑟𝑒𝑐𝑎𝑠𝑡) 𝑅𝑀𝑆𝐸(𝑛𝑎𝑖𝑣𝑒𝑓𝑜𝑟𝑒𝑐𝑎𝑠𝑡) (19) while the above forecasting quality measures are useful for providing different performance measures on applied models, they do not statistically test if the models are significantly different or better from another. therefore, we will also apply the diebold-mariano (1995) test (dm) to compare the predictive ability between two forecasting models. the null hypothesis of the test is that the predictive ability of two forecasting models is the same. in our empirical analysis, we are particularly interested whether our forecast models are able to significantly outperform a simple rw model such that the considered models are tested against the rw model using a simple t-test, see e.g. diebold (1998). thus, the null hypothesis of equal performance of the models is rejected when the test-statistic 𝐷 = 𝑑 � 𝜎� √𝑛⁄ (20) with 𝑑 = 1 𝑛 ∑ (𝑒1𝑡2 − 𝑛 𝑡=1 𝑒2𝑡2 ) (21) yields significant values. in the empirical analysis we will restrict ourselves to oneperiod-ahead forecasts only. note that the test could also be applied to k-step-ahead forecasts, see e.g. diebold and mariano (1995). the authors point out that the test tends to be less accurate for small sample sizes and k-step-ahead forecasts. however, these issues are unlikely to affect our empirical analysis due to a comparably large sample size and the use of one-period-ahead forecasts only. trück and liang: forecasting volatility in the gold market 4. empirical results 4.1. in-sample forecasting performance in this section, we compute the one-step-ahead volatility forecasts using the models described in the previous section. for the in-sample analysis, the data are divided into three sub-periods: sub-period 1 from 28th jun 1999 dec 2004, which is a period of slightly increasing gold prices over 6 years; sub-period 2 from jan 2005 dec 2007, which is a period of substantially increasing gold prices over 3 years; and sub-period 3 from jan 2008 dec 2008, a period of very volatile gold prices. for in-sample forecasting, all observations within the period are used to estimate the models, and the forecasting results are compared to the actual values. the complete results including the ranking for each measure in each sub-period are presented in tables 3, 4 and 5, respectively. obviously, the forecast error statistics, mse, rmse and mae depend on the scale of the dependent variable and the differences between the actual and forecasted volatility. on the other hand, the mape does not depend on the scale of the variable: given a perfect fit of a model, the mape would be zero while there is no upper bound for the mape. for all measures, it can be concluded that the smaller the error statistic, the better the forecasting ability of a model. in order to facilitate the comparison of the models, for each performance criteria also the relative ranks are provided. in the first sub-period, the price fluctuations were relatively low with a general upward trend. only one structural break occurred after the 11 september 2001 attack. the shock lasted only a short period of time and did not have long-lasting effects on the volatility. the results show that for the mse, mae, rmse and theil u criteria the tarch model yields the best results with respect to in-sample one period ahead forecasts. the model further ranks second for the mape criterion. overall, most test statistics are consistent in ranking the forecasting performance of the considered models, with the exception of the mape measure. however, as discussed earlier, the mape can be unreliable in cases where the denominator contains the value of zero as it was the case in our evaluation where results were removed when the daily change and volatility were zero. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 48 table 3: in-sample results for sub-period 1 rw hm ma(20) ma(40) ma(120) ols ar(5) mad(5) arma(1,1) ewma garch(1,1) garch-m tarch mse 10.231 7.617 7.937 7.798 7.615 6.774 6.708 6.718 7.615 7.472 7.907 7.258 5.713 rank 13 10 12 11 9 4 2 3 8 6 11 5 1 rmse 3.199 2.760 2.817 2.793 2.760 2.603 2.590 2.592 2.760 2.733 2.812 2.694 2.390 rank 13 10 12 11 9 4 2 3 8 6 11 5 1 mae 1.209 1.054 1.064 1.058 1.066 1.029 1.014 1.018 1.065 1.032 1.100 1.064 0.990 rank 13 6 8 7 9 4 2 3 9 5 12 8 1 mape 8264.1 11071.5 8413.4 10260.2 10723.3 10263.7 9707.6 9913.2 10722.3 9114.0 9892.5 9904.0 8288.1 rank 1 13 3 9 12 10 5 8 11 4 6 7 2 theil u 1.000 0.863 0.881 0.873 0.863 0.814 0.810 0.810 0.863 0.855 0.879 0.842 0.747 rank 13 9 12 10 8 4 2 3 7 6 11 5 1 dm t-stat -1.217 -1.082 -1.140 -1.228 -1.893 -1.912 -1.903 -1.228 -1.313 -1.082 -1.457 -2.348 rank 9 12 10 8 4 2 3 7 6 11 5 1 p-value 0.112 0.140 0.127 0.110 0.029 0.028 0.029 0.110 0.095 0.140 0.073 0.010 table 3 in-sample forecast results for sub-period 1 (june 1999 december 2004) for examined models and the considered performance evaluation measures mse, rmse, mae, mape, theil u as well as results for diebold-mariano test with rw model as benchmark. trück and liang: forecasting volatility in the gold market table 4: in-sample results for sub-period 2 rw hm ma(20) ma(40) ma(120) ols ar(5) mad(5) arma(1,1) ewma garch(1,1) garch-m tarch mse 15.203 8.347 7.704 7.706 7.829 8.153 7.595 7.762 7.558 7.607 7.558 7.682 6.624 rank 13 12 5 6 8 9 3 7 2 4 2 6 1 rmse 3.899 2.889 2.776 2.776 2.798 2.855 2.756 2.786 2.749 2.758 2.748 2.772 2.574 rank 13 12 5 6 8 9 3 7 2 4 2 6 1 mae 1.785 1.325 1.426 1.422 1.424 1.487 1.467 1.481 1.418 1.421 1.434 1.443 1.314 rank 13 4 9 7 8 12 10 11 5 6 8 9 1 mape 14790.9 16439.6 15475.7 15515.4 19616.2 22234.2 18682.8 19663.1 17411.6 16083.5 17535.2 17679.1 9491.3 rank 4 8 5 6 11 13 10 12 9 7 8 9 1 theil u 1.000 0.741 0.712 0.712 0.718 0.732 0.707 0.715 0.705 0.707 0.705 0.711 0.660 rank 13 12 5 6 8 9 3 7 2 4 2 6 1 dm t-stat -2.819 -3.113 -3.093 -3.038 -2.932 -3.055 -3.033 -3.173 -3.162 -3.138 -3.184 -3.567 rank 12 6 7 9 11 8 10 3 4 5 2 1 p-value 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.001 0.002 0.001 0.000 table 4 in-sample forecast results for sub-period 2 (january 2005 december 2007) for examined models and the considered performance evaluation measures mse, rmse, mae, mape, theil u as well as results for diebold-mariano test with rw model as benchmark. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 table 5: in-sample results for sub-period 3 rw hm ma(20) ma(40) ma(120) ols ar(5) mad(5) arma(1,1) ewma garch(1,1) garch-m tarch mse 102.833 61.154 50.452 49.695 51.953 53.201 48.378 48.168 49.789 50.017 49.822 51.133 44.288 rank 13 12 8 4 10 11 3 2 5 7 6 9 1 rmse 10.141 7.820 7.103 7.049 7.208 7.294 6.955 6.940 7.056 7.072 7.058 7.151 6.655 rank 13 12 8 4 10 11 3 2 5 7 6 9 1 mae 5.758 4.668 4.205 4.060 3.820 4.509 4.360 4.410 4.083 4.129 4.016 4.148 3.812 rank 13 12 8 4 3 11 9 10 5 6 4 7 2 mape 11637.6 6670.9 11419.8 10527.5 8212.6 13501.3 12257.8 12197.5 10654.3 10654.9 9977.4 9912.7 10973.9 rank 10 1 9 5 2 13 12 11 6 7 4 3 8 theil u 1.000 0.771 0.700 0.695 0.711 0.719 0.686 0.684 0.696 0.697 0.696 0.705 0.656 rank 13 12 8 4 10 11 3 2 5 7 6 9 1 dm t-value -9.769 -12.394 -12.498 -11.948 -11.765 -12.467 -12.699 -12.549 -12.534 -12.391 -12.475 -13.876 rank 12 8 6 10 11 5 2 3 4 9 7 1 p-value 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 table 5 in-sample forecast results for sub-period 3 (january 2008 december 2008) for examined models and the considered performance evaluation measures mse, rmse, mae, mape, theil u as well as results for diebold-mariano test with rw model as benchmark. trück and liang: forecasting volatility in the gold market the different ranking according to the mape criterion is also indicated by the fact that while the benchmark rw model ranks last for all other statistics, it is the preferred model according to the mape. the performances of the other garch type models vary substantially depending on the considered evaluation criterion. interestingly, the garch (1,1) model performs poorly and is ranked in the bottom quartile of all tested models. the garch-m model performs better, ranking 5th out of the 13 models in most statistics. the simple ar(5) and mad(5) models also perform well, producing the second and third lowest values for the considered error statistics. other arma type models including the ols and arma(1,1) rank in the middle, while models based on past averages such as the hm, and the various ma models do not perform very well. however, they still outperform the rw model for most of the considered criteria. the theil u statistics are less than 1 for all considered models being indicative of the models performing better than the random walk model. on the other hand, the results for the dm test are not so clear-cut. only the ols, ar, mad and tarch models are significantly better than the benchmark random walk model at the 0.05 significance level. so during the period with low volatility from june 1999 and december 2004, all models seem to provide better in-sample forecasts than the rw model, while only 4 out of 12 significantly outperform the rw model according to the dm test. we will see that during the more volatile periods, the dm test provides more significant results. we could conclude that during periods of low volatility in the gold market, it seems difficult to significantly outperform the rw model using a parametric model. overall, while the tarch model had the lowest forecast errors of all models, one may also consider ar, ma and ols type models when forecasting in periods of low volatility. as mentioned above, during the second sub-period from january 2005 and dcember 2007, the gold market experienced a major boom, with gold almost doubling in value. overall, the forecast errors for this period are slightly higher than that of subperiod 1. once again the tarch model turns out to be the best model considering the different performance criteria. also the arma(1,1) and garch(1,1) models provide good results and they rank among the best three models for several of the performance criteria. interestingly, for this period, different performance measures provide quite international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 different rankings: theil’s u allocates more weight to large deviations while mae and mape provide a different weighting of the performance errors. thus, the conditional variance type models garch(1,1) and garch-m models were selected by mse, rmse, and theil’s u as the second and sixth best models. in fact, these perform worse against the absolute error measures of mae and mape. the mse and theil’s u also indicate that the ewma, ar(5) and arma models perform well. the rw is once again the worst performing model, ranking last for all statistics except mape. the dm values for this period are all highly significant even at the 0.01 level, indicating that most models are able to significantly outperform the rw benchmark in this period. this is also confirmed by u statistic where all models yield lower values than in the first sub-period. the u values range from 0.26 to 0.34 indicating that even the worst performing model (hm) is still significantly better than the rw benchmark. overall, the results for the second sub-period suggest that predictive models with conditional volatility like tarch, garch and garch-m seem to perform quite well during this period of significant increases in the gold price. the third sub-period from january to december 2008 also includes the advent of the global financial crisis, when various financial markets as well as the gold market exhibited a long period of extreme volatility. generally, one would expect this period being the most difficult for volatility prediction. this is confirmed by both mse and mae-based criteria yielding clearly higher values than for the previous two subperiods. for example, the mse is five times higher than during the first and second sub-period while the mae increases by roughly 200 percent. also for the third subperiod, mse, rmse and u favour the tarch model as yielding the best predictions, while the ar(5) and mad(5) rank second and third. for these criteria, the random walk model is the worst performing model, followed by the hm model. also the mae measure gives indication of superiority of the tarch model over the others. however, for this criterion, the ar and mad models perform rather poorly and only rank ninth and tenth. again, the two worst performing models are the rw and hm model. the dm test show that for the third sub-period all models were able to significantly outperform the rw model at the 0.01 level. results for theil’s u are trück and liang: forecasting volatility in the gold market similar to the second sub-period indicating that the models provide substantially smaller rmse than the rw model for the volatile third sub-period. overall, we conclude that for in-sample fit, the tarch model can be considered as the most appropriate, ranking first for almost all of the examined performance measures and sub-periods. 4.2. out-of-sample forecasting results in the following we report the results for an out-of-sample analysis of the models by comparing one-step-ahead volatility for the most volatile period from july 1, 2008 to december 30, 2008. a recursive window approach is used. for the recursive window approach, the initial estimation date is fixed and the models are estimated using all observations available up to the initial estimation date. it is an iterative procedure, where in each time step, the estimation sample is augmented to include one additional observation in order to re-estimate the volatility forecast for the next day. again, results are benchmarked against a rw model. note that despite its simplicity, particularly in out-of-sample forecasting the random walk model is often considered as a benchmark model that is difficult to beat: for example, stock and watson (1998) examine various us macroeconomic time series and suggest the rw model to perform best amongst a number of competing models. the out-of-sample results for the different models are provided in table 6. our results for the mse criterion suggest that the ma(40) model provides the most accurate forecasts while the ewma model ranks seconds. interestingly, similar to the considered in-sample periods, the rw model proved to be the worst amongst the examined models also for out-of-sample forecasting. it ranked last with respect to the mse criterion and provided predictions significantly less accurate than most of the considered models. another feature of the results is that there are only relatively small differences with respect to mse among the ten best models: the mse for the ma(40) model is 83.29 while the mad(5) model provides a mse of 90.08. with respect to mae, we observe the smallest error for the ma(120) model. the hm and mad models, also perform well, ranking second and third, respectively. the benchmark rw model is substantially less accurate than the other models. again the marginal difference between the first and tenth ranked model is comparably small. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 overall, as could be expected, the out-of-sample errors are higher than the comparable numbers for the in-sample results. for mse, there is an average increase by approximately 30 percent while for mae, the average errors increase by approximately 80 percent in comparison to the in-sample results for the third subperiod. again choosing the mape statistics provides slightly different findings and suggests the hm model as the best model followed by the ols model. once again the rw model performs worst. according to the u statistic, all models performed better than the rw model. since u is based on the mse measure, the results suggest the ma(40) followed by the ewma model as being most appropriate. generally the differences between the results for all models are rather small, since the values for theil’s u range from 0.70 to 0.78. results are confirmed by the dm test indicating that all models are significantly better than the benchmark rw model at the 0.01 level of significance. considering the different model types we find that ma models performed clearly better in out-of-sample than in-sample forecasting, in particular the ma(40) model. yet again there was no clear-cut outperformance of this model class with the ma(20) ranking fourth and the ma(120) ranking eighth. the hm model was the second worst according to the mse, but it is one of the better models according to the mae and mape. for the garch models it is noteworthy that the tarch model, which provided superior in-sample fit for all three sub-periods, yielded one of the worst results for out-of-sample forecasting. on the other hand, in terms of the magnitude of mse and mae, the difference with the best model was still rather small. the arma models rank second to eleventh across the different measures indicating the importance of the right choice of the order of the coefficients. in summary, we conclude that there are only small differences with respect to the out-of-sample forecast performance between the considered models. the ma(40) could be considered the best model based on the mse and u measures. other models that have performed well are the arma(1,1) and the ewma model. furthermore, despite their generally good performance in the in-sample periods, for the considered out-of-sample period the garch models did not perform that well. in particular the tarch model, that was the clear winner when in-sample volatility predictions were trück and liang: forecasting volatility in the gold market considered, only ranked between 9 and 13 across the measures. overall, there are no significant differences between the models and the rankings based on each performance measure are quite different. we conclude that, for the out-of-sample forecasting, it is hard to choose an overall winner. we will now extend our analysis by examining the different models with respect to risk quantification. in particular, we investigate and report their performance in forecasting value-at-risk (var). 4.3 value at risk analysis in this section, we examine the proposed models with respect to adequate var quantification in an out-of-sample forecasting study. for a given portfolio, probability and time horizon, var is defined as a threshold value of the probability that the markto-market loss of the portfolio over the given time horizon exceeds this value at a given probability level. in or analysis, following giot and laurent (2001, 2003), we evaluate the var forecasts from the perspective of both long and short position traders. the empirical models discussed in the previous sections are used to estimate and forecast the volatility 𝜎�𝑡. the var for a given level of significance α can then be determined as var = zα 𝜎�𝑡. in our analysis we consider one day 95% and 99% var and, thus, set α equal to 0.05 and 0.01. the corresponding failure rates and var violations are then computed by comparing the one period ahead forecasts of var with the actual observed returns in the out-of-sample period with 127 observations. we define the number of violations for long traders as being equal to the number of times the negative observed return on a particular day exceeds the one-day-ahead var forecasts. correspondingly, the number of violations for a short position is the number of times the positive return is larger than the determined var forecast for that particular day. the failure rate is the percentage of violations occurring in the out-of-sample period. if the model has been specified correctly, the failure rate should approximately be equal to the theoretical number of exceptions at the chosen var level, see e.g. kupiec (1995); christoffersen (1998); christoffersen and diebold (2000) or hull (2007). the results for the calculated var forecasts for long and short positions in the gold market are provided in table 7 and 8. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 table 6: out-of-sample forecast results for the period july 1, 2008 to december 30, 2008 table 6 out-of-sample forecast results for the period july 1, 2008 to december 30, 2008. results are reported for examined models and the considered performance evaluation measures mse, rmse, mae, mape, theil u as well as results for diebold-mariano test with rw model as benchmark. rw hm ma(20) ma(40) ma(120) ols ar(5) mad(5) arma(1,1) emwa garch(1,1) garch-m tarch mse 172.870 105.245 84.649 83.291 87.972 101.224 86.081 90.080 84.249 83.812 86.942 86.352 88.238 rank 13 12 4 1 8 11 5 10 3 2 7 6 9 rmse 13.148 10.259 9.200 9.126 9.379 10.061 9.278 9.491 9.179 9.155 9.324 9.293 9.394 rank 13 12 4 1 8 11 5 10 3 2 7 6 9 mae 8.124 5.433 5.966 5.722 5.401 5.654 5.673 5.546 5.681 5.844 5.569 5.685 5.855 rank 13 2 12 9 1 5 6 3 7 10 4 8 11 mape 10101.6 2165.3 10040.5 9429.3 7006.9 3936.8 6460.6 5060.6 8460.3 9513.5 7889.5 7747.1 13112.8 rank 12 1 11 9 5 2 4 3 8 10 7 6 13 theil u 1.000 0.780 0.700 0.694 0.713 0.765 0.706 0.722 0.698 0.696 0.709 0.707 0.714 rank 13 12 4 1 8 11 5 10 3 2 7 6 9 dm t-value -1.764 -2.432 -2.457 -2.281 -1.997 -2.399 -2.257 -2.437 -2.459 -2.308 -2.327 -2.726 rank 12 5 3 9 11 6 10 4 2 8 7 1 p-value 0.040 0.008 0.008 0.012 0.024 0.009 0.013 0.008 0.008 0.011 0.010 0.003 trück and liang: forecasting volatility in the gold market given the confidence levels of 95% and 99% and a total of 127 days during the outof-sample period, one would expect approximately 6.35, respectively, 1.27 var exceptions. to test for the appropriateness of the considered var models, we apply a test that is based on the actual number of observed exceptions versus the expected number of exceptions, see e.g. hull (2007). the test uses a binomial distribution such that given a true probability p of an exception, the probability of the var level being exceeded m or more days is: ( ) ! (1 ) ! ! n k n k k m n p p k n k − = − − ∑ (22) a similar reasoning applies to the case where the number of var violations m is lower than the expected number of exceptions. the probability of m or less exceptions is: ( )0 ! (1 ) ! ! m k n k k n p p k n k − = − − ∑ (23) based on these quantities it is easy to derive p-values for a correct var model specification given the number of exceptions that were actually observed. we find that the random walk model performs rather poorly both for the 95% and 99% var. for the long position, we observe 18, respectively 16 var exceptions corresponding to a failure rate of 14.2% and 12.6% that is substantially higher than the expected 5% and 1% under the assumption of a correct model specification. similar results are obtained for holding a short position where the fraction of var exceptions is approximately 11% and 9.4%, respectively. thus, as indicated by the pvalues, for both 95% and 99% var levels, the model is significantly rejected. while most of the models provide clearly less var violations than the rw model, only few of them are not rejected by the test for at least one of the two considered confidence levels. the hm and ols model also significantly underestimate the risk, and yield too many exceptions for both long and short positions in particular at the 0.01 level. on the other hand, the three ma models yield a very small number of var violations, but the estimates are too conservative. as indicated in table 7, for the long position, each ma model only yields one exception international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 tables 7: out-of-sample results for periods 2 rw hm ma(20) ma(40) ma(120) ols ar(5) mad(5) arma(1,1) ewma garch(1,1) garch-m tarch long position α = 5% (expected violations 6.35) violations 18 12 1 1 1 9 3 5 2 2 2 2 4 failure rate 0.142 0.094 0.008 0.008 0.008 0.071 0.024 0.039 0.016 0.016 0.016 0.016 0.031 p-value 0.0002 0.0113 0.0114 0.0114 0.0114 0.1046 0.1163 0.3860 0.0442 0.0442 0.0442 0.0442 0.2338 long position α = 1% (expected violations 1.27) violations 16 8 0 0 1 7 2 4 1 0 1 1 2 failure rate 0.126 0.063 0.000 0.000 0.008 0.055 0.016 0.031 0.008 0.000 0.009 0.008 0.016 p-value 0.0000 0.000 0.2790 0.2790 0.6370 0.0000 0.1352 0.0093 0.6370 0.2790 0.6370 0.6370 0.1352 tables 8: out-of-sample results for period 3 rw hm ma(20) ma(40) ma(120) ols ar(5) mad(5) arma(1,1) ewma garch(1,1) garch-m tarch short position α = 5% (expected violations 6.35) failures 14 12 0 0 2 9 1 1 0 0 1 0 1 failure rate 0.110 0.094 0.000 0.000 0.016 0.071 0.008 0.008 0.000 0.000 0.008 0.000 0.008 p-value 0.0018 0.0113 0.0015 0.0015 0.0044 0.1046 0.0114 0.0114 0.0044 0.0044 0.0114 0.0044 0.0114 short position α = 1% (expected violations 1.35) failures 12 9 0 0 1 6 1 0 0 0 0 0 1 failure rate 0.094 0.071 0.000 0.000 0.008 0.047 0.008 0.000 0.000 0.000 0.000 0.000 0.008 p-value 0.0000 0.0000 0.2790 0.2790 0.6370 0.0003 0.6370 0.2790 0.2790 0.2790 0.2790 0.2790 0.6370 table7 & 8: number of var violations for the considered models assuming a short position in gold. results are reported for 1-day 95% and 99%-var. the p-values refer to a test based on hull (2007) where the null hypothesis is an appropriate specification of the var model. bold letters indicate models that were not rejected at the 5% significance level. trück and liang: forecasting volatility in the gold market 59 at the 95% var level leading to a rejection of the models even at the 0.10 significance level. almost the same results are obtained for holding a short position in the gold market where the 95%-var estimates are also too conservative, so all ma models are rejected. note however, that the models are not rejected for the 99%-var level since only a very small number of exceptions are expected at this level. similar results are obtained for the arma, ewma and two models with conditional variance garch(1,1) and garch-m model. these models only yield two exceptions at the 95% level and zero or one exception at the 99% level for a long position: for a short position, only the garch(1,1) model yields one exception at the 95% confidence level. the var estimates of these models are too conservative for the considered time period such that all models are rejected at the 5% significance level. the mad(5) model gives too many exceptions at the 95% confidence level for a long position in gold, while it performs reasonably well at the 99% level for short positions. the best results – at least for long positions are obtained for the ar(5) model and again for the threshold conditional volatility tarch model. these models seem to provide adequate one-day-ahead risk forecasts for long positions and cannot be rejected for any of the considered confidence levels. considering short positions, the models seem to provide estimates that are overly conservative and yield only one exception at the 95% and no exception at the 99% confidence level. still, given the reasonable performance of the ar(5) and garch models for long positions, they could be considered as being most appropriate in terms of providing var forecasts. overall, we conclude that there was no clear winner with respect to providing one-day ahead value-at-risk forecasts. 5. summary and conclusions in this paper we investigate the modelling of volatility dynamics of gold market returns in london. gold markets are usually considered as a safe haven and investments into this class of assets have been very popular, in particular, since the global financial crisis. therefore, appropriate models for volatility dynamics in these markets are of great interest to both investors and hedgers. while there are a number international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 of recent studies examining volatility and value-at-risk (var) measures in financial and commodity markets, none of them focuses in particular on the gold market. compared to the numerous studies on volatility modelling and forecasting focused on equity and commodity markets in general, we provide a pioneering study on the volatility of this important market. we contribute to the literature by using a large number of statistical approaches in order to model and forecast the daily volatility and value-at-risk in the gold spot market. hereby, we distinguish between different time horizons including a sub-period of continuously but only slightly increasing gold prices, a sub-period of substantially increasing gold prices and, finally, a sub-period of high volatility in the gold market. both in-sample and out-of-sample forecasts are evaluated using appropriate forecast evaluation measures. for in-sample forecasting, the class of tarch models provided the best results among the tested models. interestingly, the performance of a garch (1,1) model, that is generally supported by empirical studies for volatility modelling in financial markets (akgiray, 1989; franses and van dijk, 1996), was only ranked in the middle of all models in our study. for out-of-sample forecasting, results were not that clearcut and the order and specification of the models was found to be an important factor in determining the model’s performance. var for traders with long and short positions were evaluated by comparing actual var exceptions to theoretical rates. for this task a simple ar as well as a tarch model performed best for the out-ofsample period. we also find that most models were able to significantly outperform a benchmark random walk model both in the in-sample and the out-of-sample forecasting. however, none of the considered models performed significantly better than the rest with respect to all of the considered criteria. the out-of-sample period from july to december 2008 that has been tested in this study was one of the most volatile periods in the history of financial markets. as a result, the behaviour of the daily returns might be significantly different to previous periods and, also, possibly future periods. thus, models that perform well in the considered out-of-sample period may well underperform in future periods, particularly when market conditions change. second, though the study attempts to comprehensively investigate the volatility in the gold market by the means of using various models, it still only covered a small number of models available in this area. trück and liang: forecasting volatility in the gold market for example, for models with conditional volatility, only three of the most widely used garch models were considered, leaving out a huge number of other garch model extensions. the flaws of var as a measure of risk along with the effectiveness of alternative risk measures such as expected shortfall, have been pointed out in the literature by e.g. artzner et al. (1999). we leave the investigation of these issues to future work. author information: stefan trück is a professor of finance in the department of applied finance and actuarial studies and co-director of the centre for financial risk at macquarie university. email: stefan.trueck@mq.edu.au. kevin liang is a graduate student of macquarie university. he has extensive professional experiences in the finance industry and is currently working as a credit risk analyst. email: kzyliang@yahoo.com.au. references akgiray, v. (1989). conditional heteroskedasticity in time series of stock returns: evidence and forecasts. the journal of business, 62(1): 55-80. artzner, p., delbaen, f., eber, j.-m., & heath, d. (1999). coherent measures of risk. mathematical finance, 9(3): 203-228. bollerslev, t. (1986). generalized autoregressive conditional heteroskedasticity. journal of econometrics, 31(3) : 307-327. bollerslev, t., chou, r. y., & kroner, k. f. (1992). arch modeling in finance: a review of the theory and empirical evidence. journal of econometrics, 52(12) : 5-59. brooks, c. (1998). predicting stock index volatility: can market volume help? journal of forecasting, 17(1) : 59-80. brooks, c., & persand, g. (2002). models choice and value at risk performance. financial analysts journal, 58(5) : 87-97. brooks, c., & persand, g. (2003). volatility forecasting for risk management. journal of forecasting, 22(1) : 1-22. canarella, g., & pollard, s. k. (2008). modelling the volatility of the london gold market fixing as an asymmetric power arch. the journal of applied finance, 14(5) : 17-43. chou, r. y. (1988). volatility persistence and stock valuation: some empirical evidence using garch. journal of applied econometrics, 3(4) : 279-294. christoffersen, p. f. (1998). evaluating interval forecasts. international economic review, 39(4) : 841-862. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 mailto:stefan.trueck@mq.edu.au mailto:kzyliang@yahoo.com.au christoffersen, p. f., & diebold, f. x. (2000). how relevant is volatility forecasting for financial risk management. review of economics and statistics, 82(1) : 1222. danielsson, j., & de vries, c. g. (2000). value-at-risk and extreme returns. annales d'economie et de statistique(60) : 11. diebold, f. x. (1998). elements of forecasting. cincinnati, ohio: south western college publishing. diebold, f. x., & mariano, r. s. (1995). comparing predictive accuracy. journal of business and economic statistics, 13(3) : 253-263. dimson, e., & marsh, p. (1990). volatility forecasting without data-snooping. journal of banking and finance, 14(2-3) : 399-421. ding, z., granger, c. w. j., & engle, r. f. (1993). a long memory property of stock market returns and a new model. journal of empirical finance, 1(1) : 83-106. engle, r. f. (1982). autoregressive conditional heteroskedasticity with estimates of the variance of united kingdom inflation. econometrica, 50(4) : 987-1007. engle, r. f., lilien, d. m., & robins, r. p. (1987). estimating time varying risk premia in the term structure: the arch-m model. eonometrica, 55(2) : 391407. franses, p. h., & van dijk, d. (1996). forecasting stock market volatility using (nonlinear) garch models. journal of forecasting, 15: 229-235. french, k. r., schwert, g. w., & stambaugh, r. f. (1987). expected stock returns and volatility. journal of financial economics, 19(1) : 3-29. giot, p., & laurent, s. (2003). market risk in commodity markets: a var approach. energy economics, 25(5) : 435-457. giot, p., & laurent, s. (2003). value-at-risk for long and short trading positions. journal of applied econometrics, 18(6) : 641-663. glosten, l. r., jagannathan, r., & runkle, d. e. (1993). on the relation between the expected value and the volatility of the normal excess return on stocks. the journal of finance, 48(5) : 1779-1801. hansen, p. r., & lunde, a. (2005). a forecast comparison of volatility models: does anything beat a garch (1,1). journal of applied econometrics, 20(7) : 873889. henriksson, r. d., & merton, r. c. (1981). on market timing and investment performance. ii. statistical procedures for evaluating forecasting skills. journal of business, 54(4) : 513-533. hentschel, l. (1995). all in the family, nesting symmetric and asymmetric garch models. journal of financial economics, 39(1) : 71-104. hull, j. c. (2007). risk management and financial institutions: pearson prentice hall. jarque, c. m., & bera, a. k. (1980). efficient tests for normality, homoskedasticity and serial independence of regression residuals. economics letters, 6(3) : 255259. jorion, p. (1996). risk2: measuring the risk in value at risk. financial analysts journal, 52(6) : 47-56. jun, j. h. (2009) global financial crisis and gold market. working paper. kupiec, p. (1995). techniques for veryfying the accuracy of risk management models. journal of derivatives, 3(2) : 73-84. mcmillan, d., speight, a., & apgwilym, o. (2000). forecasting uk stock market volatility. applied financial economics, 10(4) : 435-448. trück and liang: forecasting volatility in the gold market morales, l. (2008). volatility spillovers on precious metals markets: the effects of the asian crisis. paper presented at the european applied business research conference (eabr), salzburg, austria. j.p. morgan., riskmetrics monitor, fourth quarter, 1995. nelson, d. b. (1991). conditional heteroskedasticity in asset returns: a new approach. econometrica, 59(2) : 347-370. sadorsky, p. (2006). modeling and forecasting petroleum futures volatility. energy economics, 28(4) : 467-488. stock, j. h., & watson, m. w. (1998). a comparison of linear and non-linear univariate models for forecasting macroeconomic time series. nber working paper, 6607. tully, e., & lucey, b. m. (2007). a power garch examination of the gold market. research in international business and finance, 21(2) : 316-325. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 3 estimating the early exercise premium of american put index options the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 31 estimating the early exercise premium of american put index options ako doffou sacred heart university, united states of america abstract this paper examines empirically the value of early exercise by testing the ability of two american put valuation models to predict the early exercise premium for the s&p 100 american put options. an accuracy test and a quality test are performed on (1) the macmillan and barone-adesi and whaley model, and (2) the carr, jarrow and myneni model. the test results show that early exercise premium is significant regardless of moneyness. moreover, consistent with the theory, the value of early exercise is significantly negatively related to moneyness and interest rates and significantly positively related to time to maturity and to the volatility of the underlying index. both american put valuation models examined do not fully capture the value of early exercise embedded in american put prices. jel classification: g13 key words: option prices, early exercise, moneyness, s&p american put option 1. introduction the value of early exercise or early exercise premium is the difference in price between an american option and an otherwise identical european option. it is difficult to estimate the early exercise premium because american and european options rarely have the same state variable or underlying asset. using transaction data on both american and european foreign currency options traded at the philadelphia stock exchange, jorion and stoughton (1989) directly computed the actual value of early exercise for foreign currency options. the average value of the premium found was about two percent of the option price. they also tested the geske and johnson (1984) american option pricing model by regressing observed early exercise premium over model predicted early exercise premium. they concluded that the geske-johnson model tracks well the variation in ijbf 32 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 the true early exercise premium. with the exception of foreign currencies, other state variables driving both the american option and otherwise identical european option do not exist. hence, estimation of early exercise premium from direct comparison is not possible for options whose underlying assets are not currencies. shastri and tandon (1986) and whaley (1986) propose an alternative approach to estimate the value of early exercise for futures options by computing the difference between the price given by an american option pricing model and that given by the black-scholes model. using the geske-johnson futures options valuation model, shastri and tandon (1986) showed that the value of early exercise is significant only for in-the-money options with a fairly long time-tomaturity. whaley (1986) found identical result that early exercise premium for out-of-the –money futures options is negligible. because of the significant and systematic pricing biases inherent in most american put option valuation models, using any model dependent technique to estimate the early exercise premium is problematic. brenner and galai (1986) proposed the use of the put-call parity relationship to estimate the value of early exercise for american options. given the observable prices for put options, call options and the underlying stock, they derive the implied risk free interest rate from the put-call parity arbitrage condition and use this risk-free rate to estimate the value of early exercise. using ibm option data from the chicago board options exchange (cboe) from june 1977 to august 1978, brenner and galai found an average early exercise premium of 0.9 per cent for at-the-money put options and 3.5 per cent for in-the-money put option. zivney (1991) takes a similar but modified approach to estimate early exercise premium as the difference between the american option price and the european option price derived from put-call parity. the data used consist of the s&p 100 index options for which harvey and whaley (1992) found the dividend stream to be distinctly discrete and early exercise for both calls and puts to occur frequently. estimates of early exercise premium are analyzed separately for cases where calls are in-the-money and for cases where puts are in-the-money. zivney found an average value of early exercise of 10 percent for put options and 3.5 per cent for call options on the s&p 100 index. this approach does not pick up the misspecification of the stochastic process for stock prices inherent in option pricing models and suffers from inherent biases induced when estimating a put-call parity implied riskless rate. moreover, the estimated values for early exercise for put options were contaminated by those for call options, making it impossible to separate the two. loudon (1990) compares the predictive ability of the put-pricing model developed by macmillan (1986) and barone-adesi and whaley (1987) to that of the black-scholes european model. he finds systematic pricing errors in the american put option pricing model with the american model put prices being significantly closer to market prices than are the black-scholes european model prices. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 33 mispricing by american put valuation models may arise when these models misspecify the dynamics of stock prices and the behavior of other option parameters, and when the early exercise feature is not properly accounted for. dueker and miller (2003) measure directly the early exercise premium using american and european s&p 500 index options. using mid-points from bid-ask quotes, dueker and miller find that the early exercise premium for the american s&p 500 index put options is between 7.97% and 10.86% of the option price and the estimated premium is most likely biased. this result is consistent with dueker and miller (1996) in which an adjustment to liquidity biases reduces the value of early exercise premium for the s&p 500 index options from $0.22 to $0.19. carr, jarrow, and myneni (1992), herein refereed to as cjm model, show that the early exercise premium is an exact solution to the black-scholes or (bs) american put option price with moving boundary conditions in contrast to the approximations by macmillan (1986) and barone-adesi and whaley (1987) or mbaw model. hence, the cjm model gives theoretically better results than the mbaw model which is a quadratic approximation of the black-schole’s partial differential equation. the results from previous studies seem to indicate that existing american put option pricing models do not fully capture the market value of early exercise. this paper examines empirically the value of early exercise and tests the ability of the cjm and mbaw american put valuation models to predict the value of early exercise for american put options. the market values of early exercise are derived using rhim and kim (2000) methodology which does not rely on any particular option pricing model. assuming that the european put-call parity holds and that investors are rational, the european put value embedded in an american put price can be derive from the put-call parity relationship. the market values of early exercise are then easily obtained as the difference between the observed market prices of american put options and the derived european put option values. using the s&p 100 index options transaction prices , this study finds that the values of early exercise are statistically significant for in-the-money and at-the-money put options. to get reliable estimates of the values of early exercise, the accuracy and quality of the mbaw and cjm american put valuation models are tested. the accuracy of each model is tested by comparing model-predicted values to observed market values. pricing biases are measured as absolute differences between market prices and theoretical prices predicted by the mbaw and the cjm models. to further investigate the pricing biases, they are decomposed into two components: mispricing of the early exercise premium and mispricing of the european option price component caused by the misspecification of the blackscholes formula. the quality of an american put valuation model is tested by examining how well the model captures the true value of early exercise premium. the results are consistent with previous work and confirm that american put option valuation models do not fully capture the value of early exercise. 34 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 the rest of the paper is organized as follows. section 2 introduces the model. the test methodologies are explained in section 3. section 4 describes the data set used. the empirical results are analyzed in section 5. finally, section 6 concludes the paper. 2. the model the theoretical value of early exercise can be estimated as the difference between the option price generated by an american option pricing model and the price of a corresponding european option obtained from the black-scholes model. this approach implicitly assumes that both american and european option pricing models generate correct prices. prior empirical studies show that both american and european option pricing models misprice actual market prices. the mispricing by theoretical models may arise from a misspecification of the stock price dynamics. to avoid problems caused by model misspecification, an estimation method that does not rely on any particular option pricing model is used. it is assumed that option markets are efficient and therefore that the european putcall parity holds. it is assumed further that investors are rational in that holders of american call options do not prematurely exercise their call options when no dividends are to be paid on the underlying stock until option maturity. based on these assumptions, the european put-call parity relationship is reconstructed by replacing a european call with an american call as follows: p e + s = c a + ke-rt (1) where c a and p e are respectively the value of an american call option and the value of a european put option with the same strike price k and the same maturity t. from equation (1) above, we derive the market value of a european put option: p e = c a s + ke-rt (2) the early exercise premium embedded in the american put option, eep mkt , can be obtained by subtracting the european put value imputable to put-call parity from the observed market price p a of the american put option: eep mkt = p a (c a s + ke-rt ) (3) the four parameters that affect the option price also affect the option early exercise premium. these parameters are the option moneyness s/k, the volatility of the underlying stock/index price σ, the time to maturity of the option t, and the risk-free interest rate r. these parameters are picked up by two major factors that directly affect the value of early exercise. the first factor is the critical stock price which triggers immediate exercise of the put option. the other factor is the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 35 the time value of money. the critical stock price, s c , is the maximum stock price at which it is optimal to exercise the put option immediately. this critical stock price is reached when the time value is zero or close to zero so that the put price is equal to its intrinsic value. hence, the critical stock price is obtained by solving the following equation k s c = p(s c ,k,t,r,σ) (4) the critical stock price depends on all the option pricing parameters except the current stock price. a rise in the critical stock price leads to a greater value for the early exercise premium because it makes a put option more likely to be in the money and therefore more likely to be exercised early. the value of early exercise is also affected by the interest to be earned on the strike price that would be received in the event the put option is exercised early. the interest earned on the strike price is the difference between the strike price and the present value of the strike price or k(1-e-rt). because the time value of money on interest income is a function of the strike price, interest rate, and the time to maturity, a greater time value of money creates a greater value of early exercise. a decrease in moneyness due to a stock price increase will decrease the value of early exercise. using the black-scholes formula, it can be shown that the partial derivative of the critical stock price with respect to the strike price in equation (4) is exactly equal to which is strictly positive. hence, a higher strike price increases the critical stock price and therefore increases the value of early exercise. moreover, an increase in the strike price increases the interest income that can be earned on the strike price when the put option is exercised. an increase in the volatility of the underlying stock price makes the put option more likely to be in the money and therefore increases the value of early exercise. an increase in the time-to-maturity increases the value of the put option as well as the time value of money and therefore increases the value of early exercise. when interest rates increase, the present value of the strike price decreases, making put options less valuable. hence, as interest rates increase, it is more likely for put options to be out of the money, and therefore less likely for put options to be exercised early. moreover, as interest rates increase, the time value of money increases. therefore, the early exercise premium for a put option may increase, decrease, or remain constant when interest rates increase. in the paragraph that follows, the test methodologies for both the mbaw and cjm models are discussed. 3. test methodologies 3.1 macmillan, barone-adesi and whaley model and tests both american and european option prices satisfy the black-scholes partial differential equation. it follows that the early exercise premium which is the difference in prices between an american option and an otherwise identical n(d 2 ) n(d 1 ) 36 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 ( )( )+= 2 1 2 1 8 1 2 1 2 2 1 ttre rrr y . y c ea s s pp += for css > , and skpa = for css < ( )( )[ ]cc sdn y s 11= , european option must also satisfy the same partial differential equation. using a quadratic approximation technique, a solution for an american put valuation formula p a is (5) where p e is the black-scholes european put price, , and the put option strike price is k. the critical stock price below which it is optimal to exercise the put option immediately is s c which is obtained by solving the following equation (6) assuming that the prices generated by both the mbaw and the blackscholes models are correct, the theoretical value of early exercise, eep mdl , for an american put option can be estimated as follows (7) where mdla p is the mbaw model american put option price and mdle p the black-scholes european put option price. the theoretical value of early exercise given in equation (7) above is compared to the actual value of early exercise estimated directly from observed market put prices. the difference between model-predicted values of early exercise and actual observed market values of early exercise reveals the extent to which the american put option pricing models considered do not capture the early exercise premium embedded in american put option prices. there are two critical problems in testing any option pricing model. the first problem is that a test of any option pricing model is a joint test of market efficiency and model validity. the second problem is that stock price volatility is not directly observable. to solve the first problem, a filtered sample is used which excludes all observations that violate the no-arbitrage boundary conditions for american put-call parity. to get a proper estimate of the stock price volatility, the mbaw model is assumed to generate a correct price for a put option when the put option is at-the-money. the implied volatility is derived from the mbaw model by fitting the model to the at-the-money put option prices. the resulting implied volatility is then used to predict remaining put prices. it follows that the validity of the tests conducted here lies on the accuracy of the volatility estimation. the mbaw model is tested in two ways. first, the accuracy of the model is tested by comparing the model predicted prices to the observed market prices. ( )+= cec spsk mdlmdl eamdl ppeep = the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 37 mdla p mkta p mdla p mkt ap second, the quality of the mbaw model is tested by investigating how well this american put valuation model captures the actual value of early exercise. in measuring the accuracy of the mbaw model, its pricing biases pb am are computed as follows (8) where is the observed market value of the american put option and the mbaw model price. the mbaw model implicitly assumes that the black-scholes model correctly prices the european portion of an american put option. it follows that pricing biases from the mbaw model may arise from the misvaluation of early exercise premium and from the misspecification of the black-scholes formula. hence, the mbaw pricing biases can be decomposed as follows (9) where pe bs is the pricing error due to the black-scholes model misspecification and pe eex the pricing error associated with the misvaluation of early exercise premium. the pricing error related to the black-scholes model misspecification is evaluated as the difference between the black-scholes european put price, , and the market value of a european put option, , derived from european put-call parity as follows (10) the pricing error caused by the misvaluation of early exercise premium is assessed as the difference between the actual market value of early exercise, eep mkt , and the model predicted value of early exercise, eep mdl , pe eex = eep mkt eep mdl (11) the quality or performance of the mbaw model is evaluated by examining how well this model captures the actual market value of early exercise. the degree to which the mbaw model misprices the true market values of early exercise, dm, is given by dm = (1-eep mdl /eep mkt )(100). (12) to further evaluate the quality of the mbaw model, market values of early exercise premium are regressed over model-predicted values of early exercise premium. the regression equation is eep mkt = λ 1 + λ 2 (eep mdl ) + ω (13) mdlmkt aaam pppb −= eexbsam pepepb += mdlmkt eebs pppe −= 38 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 under the null hypothesis that the model-predicted early exercise premium perfectly forecasts the true market value of early exercise premium, the intercept coefficient, λ 1 , should equal zero and the slope coefficient, λ 2 , should equal one. 3.2 carr, jarrow and myneni model and tests the carr, jarrow, and myneni (1992) model or cjm model shows the early exercise premium as an exact solution to the black-scholes or bs model american put option price with moving boundary conditions as opposed to the approximation approach of the mbaw model. consequently, the cjm model gives theoretically better results than the mbaw model which is a quadratic approximation of the black-scholes partial differential equation. however, the cjm model has a singularity problem at time zero (i.e., option at maturity) and therefore has no known analytic solution. if an option is at expiration, the numerical approximation is the only alternative to the cjm exact solution. in this paper, the performance of the mbaw model and the cjm model are compared. the tests undertaken for the cjm model are exactly the same as those described above for the mbaw model. the mbaw model provides computing efficiency even though the cjm model is theoretically superior. whether or not the cjm model performs significantly better than the mbaw model to offset its relative computing time inefficiency is an empirical question not yet answered 4. the data daily transaction data for put and call options on the s&p100 index (oex) traded on the chicago board of options exchange (cboe) are used. the data set spans over a ten-year period from january 3, 1995 to january 3, 2005. the data consist of the time and price of every transaction in which the price changed from the previous transaction for the ten-year period considered. the initial data set contains 362,847 oex index options. all oex put-call pairs that meet the requirements that follow are selected and are used to compute early exercise premium. both put and call options in a put-call pair are options on the same underlying stock index, with the same strike price and the same maturity. bid and-ask data were discarded since no transactions were conducted at those prices. only contracts with maturities between one and four months are retained. longer maturities were too thinly traded, and shorter maturities were too close to the maturity date to contain useful information about early exercise premium. to avoid days with thin trading, at least 20 call and 20 put transactions are required for any given day’s data to be retained. transactions in at least four strike classes for calls and four for puts are required to ensure a range of “moneyness” sufficient to provide a good picture of the distribution of early exercise premium. deep-out-of-the-money call and put options defined as those with prices less than 5 cents, are deleted from the sample. for these options, the bid-ask spread is a big proportion of their time value. therefore, their implied volatilities would be highly sensitive to the bid-ask bounce. put and call option prices in a retained the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 39 put-call pair must satisfy the no-arbitrage boundary conditions for american put-call parity. finally, the early exercise premium of a put option cannot be greater than the put option price. all of the above criteria eliminate 243,906 put and call options from the sample, leaving only a total of 118,941 put-call pairs for the estimation of early exercise premium. the daily risk-free interest rate is computed from the prices of treasury bills maturing close to the maturity of the option. the tests conducted use two volatility measures. the historical volatility (hv) is obtained with daily index returns for the previous 180 trading days. the implied volatility (iv) is derived using the methodology proposed in whaley (1982). 5. empirical results the actual market value of early exercise for american put options is given by equation (3). while previous empirical studies estimate the value of early exercise based on model-predicted prices, this study estimates early exercise premium from actual market prices. table 1 shows the mean market values of early exercise for american put options on oex index for the period 01/03/1995 – 01/03/2005, for all options and for all degrees of moneyness as percentage of the put option market prices. for all options, the average value of early exercise is $7.436 or 9.55 percent of the put option market price. the mean market value of early exercise for in-the-money put options ($11.013) is greater than that for at-the-money put options ($5.237) which in turn is greater than that for out-ofthe money put options ($4.411). the average value of early exercise is 3.427 percent for all out-of-the-money put options compared to 14.622 percent for all in-the-money options. this finding is consistent with the fact that out-of-themoney options contain a negligible value of early exercise. numbers in parentheses are mean market values of early exercise expressed as a table 2 displays the model predicted mean values of early exercise for american put options on oex index for the period 01/03/1995 – 01/03/2005. model predicted values of early exercise are estimated as the difference between (1) the theoretical american put option price given by the macmillan, baroneadesi and whaley (mbaw) model or the carr, jarrow and myneni (cjm) model, and (2) the black scholes european put option price. the values in parentheses are the mean values of model-predicted early exercise expressed as a percent of the put price given by the model. the computations based on implied volatility show that the mbaw model predicts a mean early exercise premium of $4.42 for in-the-money put options and $0.33 for out-of-the-money put options, while the cjm model predicts a mean exercise premium of $4.62 for in-the-money put options and $0.29 for out-of-the-money put options. these models predicted mean values of early exercise are substantially lower than the mean market values of early exercise computed in table 1, an indication that the mbaw and the cjm american valuation models undervalue the early exercise premium. 40 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 this result is sustained when the model predicted mean values of early exercise are computed based on historical volatility. table 2 also shows that computations based on historical volatility lead to a wider gap for the mean value of early exercise between the in-the-money options and out-of-the-money options. table 1: mean market values of early exercise for american put options on oex index: 01/03/1995 – 01/03/2005. number of observations mean market value of early exercise in-the-money 52,025 11.013 (14.622) at-the-money 19,699 5.237 (10.829) out-of-the-money 47,217 4.411 (3.427) all 111,941 7.436 (9.550 consistent with equation (3), market values of early exercise premium are defined as observed market price of american put option minus the european put value of the option derived from put/call parity. the average bid and ask yield quotations on treasury bills that expire closest to the option’s expiration date are used as estimates for the risk-free rate of interest. the values in parentheses are mean market values of early exercise expressed as a percentage of the put option market price. table 2: model predicted mean values of early exercise for american put options on oex index: 01/03/1995 – 01/03/2005. number of observations mbaw h.v cjm h.v mbav i.v cjm i.v in-the-money 52,025 5.78129 (7.12458) 5.93864 (7.41797) 4.42129 (5.23872) 4.61518 (5.53892 at-themoney 19,699 0.69173 (3.61192) 0.72479 (3.76726) 0.71026 (2.59809) 0.73148 (2.73949) out-of-themoney 47,217 0.25164 (3.20478) 0.22471 (2.19897) 0.32812 (1.69968) 0.29036 (1.44648) all 118,941 2.74320 (4.98674) 2.80680 (4.74151) 2.18177 (3.39646) 2.25510 (3.45067) mbaw stands for the macmillan, barone-adesi and whaley model values, while cjm stands for the carr, jarrow and myneni model values. model predicted values of early exercise are estimated as the difference between (1) the theoretical american put option price given by the mbaw model or the cjm model, and (2) the black scholes european put option price. h.v. and i.v. are historical volatility and implied volatility respectively. the values in parentheses are mean values of model-predicted early exercise expressed as a percent of model put price. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 41 to test the relationship between early exercise premium and option parameters, market values of early exercise are regressed on four option parameters: option moneyness measured by the stock price over the strike price, time to maturity, volatility of the underlying stock/index price, and interest rates. the results of these regressions are displayed in table 3. table 3: market values of early exercise, eep mkt , are regressed on put option moneyness, time to maturity, volatility of the stock price and interest rate: the signs of the coefficient estimates are consistent with the theory. as the moneyness defined by s / k increases, the put option is more likely to be out-of-the money and therefore less valuable. hence, the early exercise premium decreases as the moneyness increases which is consistent with a negative sign for the β 1 coefficient. as the time to maturity increases, a put option value increases, making the early exercise premium more valuable. hence, the sign of the β 2 coefficient is positive. as the volatility increases, a put option price increases, making the early exercise premium more valuable. therefore, the sign of the β 3 coefficient is positive. as interest rates increase, the present value of the strike price decreases, making it more likely for the stock price to be above the strike price. hence, as interest rates increase, the put option will more likely be out-of-themoney, making the early exercise premium less valuable. this corresponds to a negative sign for the β 4 coefficient. the value of early exercise premium, eep mkt , is computed as the difference between (1) the price of the american put option observed in the market and (2) the price of the european put option derived from put/call parity. implied volatility measures are used to estimate this regression model. the time-to-maturity is derived as the number of days to maturity divided by 365 days. the daily risk-free interest rate is computed from the prices of treasury bills maturing close to the maturity of the options. the numbers in parentheses are t-statistics. the signs of the coefficient estimates are consistent with the theory. as the moneyness defined by s / k increases, the put option is more likely to be outof-the-money and therefore less valuable. hence, the early exercise premium decreases as the moneyness increases which is consistent with a negative sign for the β 1 coefficient. as the time to maturity increases, a put option value increases, making the early exercise premium more valuable, consistent with a positive sign for the β 2 coefficient. as the volatility increases, a put option price 21 ++= k s eepmkt (time to maturity) + 3 (volatility) + 4 (interest rate) + number of observations 1 2 3 4 2radj 118,941 0.03865 (-10.2735) 0.00239 (-21.3814) 0.34862 (87.5975) 0.00186 (17.9108) 0.05271 (41.3829) 0.3685 42 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 increases, making the early exercise premium more valuable, consistent with a positive sign for the β 3 coefficient. as interest rates increase the value of a put option decreases making the early exercise premium less valuable consistent with a negative sign for the β 4 coefficient. the value of early exercise premium, eep mkt , is computed as the difference between (i) the price of the american put option observed in the market and (ii) the price of the european put option derived from put/call parity as given in equation (3). implied volatility measures are used to estimate this regression model. the time to maturity is measured as the number of days to maturity divided by 365 days. the daily risk-free interest rate is computed from the average bid and ask yield quotations on treasury bills maturing closest to the maturity of the option. the numbers in parentheses are tstatistics. overall, the regression results show that the early exercise premium is significantly positively related to time to maturity and volatility and significantly negatively related to meneyness and interest rates. the empirical results of the accuracy test of the valuation models used are reported in tables 4. the accuracy of the model is tested by comparing modelpredicted values to observed market values. table 4 shows the mean absolute percentage option pricing error by moneyness. this statistic is defined as the difference between the model price and the observed market price of the put option divided by the market price of the put option. the mbaw and cjm models perform more accurately under implied volatility (iv) than under historical volatility (hv), consistent with previous work in the finance literature. the black-scholes model (bs) performs better under historical volatility than under implied volatility. american put valuation models attempt to predict american put prices by approximating the value of early exercise. for in-the-money put options, market prices reflect mainly their intrinsic values. for out-of-the-money put options, market prices consist mainly of their time values. hence, out-ofthe-money options are subject to more pricing errors than in-the-money options. this assertion is supported by the mean absolute percentage option pricing error by moneyness reported in table 4. for the mbaw and cjm models with iv, the mean absolute percentage pricing error is 2.05 percent and 2.00 percent respectively, for all in-the-money options and 5.66 percent and 5.74 percent respectively, for all out-of-the-money put options. this result is consistent with blomeyer and johnson (1988). the models with iv price more accurately than the same models with hv. because both the mbaw and cjm models implicitly assume that the blackscholes model correctly prices european values of american put options, pricing biases reported for these models inherently contain pricing errors resulting from the black-scholes model’s misspecification of the stock price dynamics. for the black-scholes model, table 4 reports a mean absolute percentage pricing error of 7.81 percent and 6.42 percent using implied volatility and historical volatility respectively for in-the-money put options. for out-the-money options, the numbers are 19.45 percent and 17.27 percent under iv and hv respectively. hence, the black-scholes model’s misspecification is more pronounced in pricing out-of-the-money put options than in pricing in-the-money put options. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 43 table 4: option models pricing biases: mean absolute percentage option pricing error by moneyness. number of observations mbaw (h.v) cjm (h.v) bs (h.v) mbaw (h.v) cjm (i.v) bs (i.v) in-themoney 52,025 4.2831 4.1937 6.4228 2.0543 2.0075 7.8109 at-themoney 19,699 7.8694 7.8429 11.6481 4.5712 4.3918 15.1404 outof-themoney 47,217 12.2573 12.4521 17.2707 5.6628 5.7404 19.4461 all 118,941 8.0426 8.0765 11.5946 3.9036 3.8843 13.6437 the mean absolute percentage option pricing errors are computed for the macmillan (1986) and barone-adesi & whaley (1987) model (mbaw), the carr, jarrow and myneni (1992) model (cjm), and the black-scholes (1973) model (bs). these computations are done using both historical volatility (h.v) and implied volatility (i.v). the mean absolute percentage option pricing error is defined as the absolute value of the difference between the model price and the observed market price divided by the market price of the put option table 5: mispricing of early exercise premium for american put options on oex index: 01/03/1995 01/03/2005. number of observations mbaw (h.v) cjm (h.v) bs (h.v) mbaw (h.v) in-themoney 52,025 5.2317 (48.3105) 5.0744 (43.9843) 6.5917 (52.7835) 6.3978 (47.2425) at-themoney 19,699 4.5453 (71.3251) 4.5122 (68.4893) 4.5267 (71.8931) 4.5055 (68.5547) outof-themoney 47,217 4.1594 (83.4691) 4.1863 (84.5722) 4.0829 (82.0743) 4.1206 (83.2172) all 118,941 4.6923 (65.4271) 4.6257 (63.1082) 5.2538 (67.3314) 5.1804 (66.0452) the mispricing of early exercise premium is computed for the macmillan, baroneadesi and whaley model (mbaw) and, the carr, jarrow and myneni model (cjm). the mispricing is the difference between the market value of early exercise premium and the model predicted value of early exercise premium. the computations are performed using both historical volatility (h.v.) and implied volatility (i.v.). the values in parentheses are mean values of the mispricing expressed as a percent of the market value of early exercise premium. 44 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 the quality of an american put valuation model is tested by examining how well the model captures the true values of early exercise premium. mispricing of early exercise premium can be assessed by comparing modelpredicted values of early exercise premium to market values of early exercise premium. the computations are performed using both historical volatility (hv) and implied volatility (iv). table 5 shows the mispricing of early exercise premium by moneyness for american put options on oex index for the period 01/03/1995– 01/03/2005. the values in parentheses are mean values of the mispricing expressed as a percent of the market value of early exercise premium. for all options, only about 33% of the market value of early exercise is captured by the mbaw model with iv. for out-of-the-money options, more than 82% of the market value of early exercise is not captured by this model, whereas nearly 53% of the early exercise premium is not accounted for when the option is in-the-money. this large mispricing of early exercise premium for put options regardless of moneyness seems to explain the significantly high pricing biases reported in table 4. table 6: regression results: the market value of early exercise premium, eep mkt , is regressed over the theoretical value of early exercise premium, eep mdl , for put options on oex index: 01/03/1995 – 01/03/2005. the market values of early exercise premium are defined as observed market price of american put option minus the european put value of the option derived from put/call parity. the theoretical value of early exercise premium is computed as the difference between (1) the model price estimated using either the mbaw model or the cjm model, and (2) the black-scholes model price. implied volatility and historical volatility are designated by (i.v.) and (h.v.) respectively. the values in parentheses are t-statistics computed by setting λ 1 = 1 and λ 2 = 1. ( )++= mdlmkt eepeep 21 model all observations 1 2 2radj mbaw (h.v.) 118,941 -1.68735 (-6.9381) 0.62476 (49.5137) 0.70135 cjm (h.v.) 118,941 -1.62067 (-3.8527) 0.62926 (27.3643) 0.70120 mbaw (i.v.) 118,941 -0.98684 (-6.7985) 0.54234 (36.5764) 0.73694 cjm (i.v.) 118,941 -0.91073 (-5.2941) 0.54298 (22.3641) 0.73158 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 45 to investigate further the quality of the mbaw and the cjm models with both measures of volatility, market values of early exercise premium is regressed over model-predicted values of early exercise premium. under the null hypothesis that the model-predicted early exercise premium perfectly forecasts the true market value of early exercise premium, the intercept and slope coefficients are expected to equal zero and one respectively. the regression results appear in table 6. the values in parentheses are t-statistics computed by setting the intercept and slope coefficients to zero and one respectively. the statistically significant t-values reported for both coefficients suggest that the null hypothesis is rejected. the explanatory power of the regression as provided by the adjusted-r2 indicates that more than 70% and more than 73% of the variation in the market value of early exercise premium can be explained by the two models under hv and iv respectively. overall, the results from the quality test suggest that both american put valuation models do not properly capture the value of early exercise embedded in american put prices. the ability of these models to capture the value of early exercise is particularly weak for out-of-themoney put options. 6. conclusions this paper examines empirically the value of early exercise for american put options. market values of early exercise premium for put options on s&p 100 index are directly derived from observable market prices of put options using the european put-call parity relationship. for the period 01/03/1995 – 01/03/2005, the mean market value of early exercise was about $7.44 per option or 9.55% of the option price. consistent with prior empirical evidence, early exercise premium for out-of-the-money put options is not significant. consistent with the theory, the value of early exercise found is significantly positively related to the time to maturity and to the volatility of the underlying index, and significantly negatively related to moneyness and interest rates. the quality of the american put option pricing models developed by (1) mcmillan (1986) & barone-adesi and whaley (1987) and (2) carr, jarrow and myneni (1992) are tested by investigating the proportion of early exercise premium captured by the model-predicted prices. the results indicate that for all options, on average, about 35% of the market value of early exercise is captured by either the mbaw model or the cjm model. hence, the mbaw and the cjm american put valuation models do not fully capture the value of early exercise embedded in american put prices. these models ability to capture the value of early exercise is particularly weak for out-of-the-money put options. assuming that the mbaw and cjm models correctly price at-the-money put options, a performance test is conducted comparing model-predicted prices to observed market prices. the result is that both american valuation models significantly misprice options regardless of moneyness. these pricing biases are mainly due to the misvaluation of early exercise premium and the misspecification of the black-scholes model. the results of this study lead to conclude that the 46 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 mbaw and the cjm american put valuation models predictions of put option market prices are acceptable. author statement: aku doffu is an associate professor of finance, john f. welch college of business, sacred heart university, united states. e-mail: adoff21@ yahoo.com. this paper was reviewed anonymously, and then edited to conform to the format of the journal. the author wishes to record his appreciation to its editorial office for the invitation to submit this paper when it was presented at the 2007 fma conference. the author takes full responsibility for the remaining errors. references barone-adesi, giovanni, and robert e. whaley, 1987, “efficient analytic approximation of american option values,” journal of finance 42: 301320. black, fischer, and myron s. scholes, 1973, “the pricing of options and corporate liabilities,” journal of political economy 81: 637-654. brenner, menachem, and dan galai, 1986, “implied interest rates,” journal of business 59: 493-507. carr, peter, robert jarrow, and ravi myneni, 1992, “alternative characterizations of american put options,” mathematical finance 2: 87-105. dueker, michael, and thomas w. miller jr., 2003, “directly measuring early exercise premiums using american and european s&p 500 index options,” the journal of futures market 23: 287-325. dueker, michael, and thomas w. miller jr., 1996, “market microstructure effects on the direct measurements of the early exercise premium in exchange-listed options,” federal reserve bank of st. louis. geske, robert, and h. e. johnson, 1984, “the american put valued analytically,” journal of finance 39: 1511-1524. harvey, campbell r., and robert e. whaley, 1992, “market volatility prediction and the efficiency of the s&p 100 index option market,” journal of financial economics 31: 43-73. jorion, philippe, and neal m. stoughton, 1989, “an empirical investigation of the early exercise premium of foreign currency options, journal of futures market 9: 365-375. loudon, g. f., 1990, “american put pricing: australian evidence,” journal of business finance & accounting 17: 297-321. macmillan, l. w., 1986, “an analytic approximation for the american put option,” advances in futures and options research 1: 141-183. rhim, jong c., and yong h. kim, 2000, “an estimation of early exercise premium for american put options,” global business and finance review 5: 13-30. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 31-47 47 shastri, kuldeep, and kishore tandon, 1986, “an empirical test of a valuation model for american options on futures contracts,” journal of financial and quantitative analysis 21: 377-392. whaley, robert e., 1986, “valuation of american futures options: theory and empirical tests,” journal of finance 41: 127-150. zivney, terry l., 1991, “the value of early exercise in option prices: an empirical investigation,” journal of financial and quantitative analysis 26: 129-138. international journal of banking and finance 9-1-2008 estimating the early exercise premium of american put index options ako doffou recommended citation the international journal of banking and finance, vol. 15. number 2, 2020: 1-24 1 how to cite this article: mohd ariffin, m. s., wan-hussin, w. n., & abdul malak, s. s. d. (2020). disclosure of agm minutes on corporate website: an assessment of malaysian listed companies. international journal of banking and finance, 15(2), 1-24. https://doi.org/10.32890/ ijbf2020.15.2.1 disclosure of agm minutes on corporate website: an assessment of malaysian listed companies 1mohd shazwan mohd ariffin 2wan nordin wan-hussin othman yeop abdullah graduate school of business, universiti utara malaysia, malaysia 3siti seri delima abdul malak tunku intan safinaz school of accountancy, universiti utara malaysia, malaysia 1corresponding author: mohd_shazwan1@oyagsb.uum.edu.my; 2wannordin@uum.edu.my; 3seridelima@uum.edu.my a r t i c l e i n f o article history: received 3 march 2020 revised 30 april 2020 accepted 30 may 2020 published 31 july 2020 keywords: agm minutes, shareholder activism, grey directors, malaysian, corporate website. jel code: g30 a b s t r a c t disclosure of annual general meeting (agm) minutes on the corporate website is considered as additional disclosure. consequently, listing requirements states compulsory disclosure of corporate information. in this regard, the awareness of compulsory disclosure of information among all malaysian listed companies in 2017 was below expectations. despite the role of agm minutes as documented proof of meeting, the management had been reluctant to disclose agm minutes on the corporate website unless demanded by the shareholders. this paper focuses on previous mandatory disclosure of discussions on key matters after 2 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 agm minutes among 261 listed companies based on the financial year ended december 31, 2016. the findings indicated that shareholder activism and the role of grey directors exerted influence on the disclosure of agm minutes on the corporate website. undoubtedly, management and investors have contributed toward voluntary disclosure in line with the government’s role to enhance shareholder rights. besides, grey directors’ contribution towards information disclosure has remained relatively unexplored. 1. introduction slack and shrives (2010, p. 84) claimed that researchers and professionals strongly believe that the demand for voluntary disclosures by stakeholders have been increasing for the last two decades. recently, a majority of researchers argued that there has been less effort given to papering between corporate governance and voluntary disclosures. in fact, there have not been many studies regarding agm minutes on corporate websites even though it is acknowledged that financial reporting and disclosures are significant for a firm’s performance and good governance especially towards its investors (healy & palepu, 2001, p. 406). the intention of this paper is to highlight the importance of agm minutes on corporate websites. even though there has been research regarding agms such as formalities and speeches at agms (catasús & johed, 2007); the format of agm minutes (nyqvist, 2015); the way top management conducts agms and accesses the accounts (carrington & johed, 2007) and the procedure of agms (apostolides, 2010), there is still a lack of studies which focuses on the disclosure of agm minutes on corporate websites. thus, this study intends to explore the influence of shareholder participation and grey directors on agm minutes displayed on corporate websites. according to the minority shareholders watch group (mswg), “one of the best practices which mswg advocated over the years is the publication of annual general meeting (agm) minutes on the listed issuer’s website in a timely manner. it was aimed that all publicly listed companies (plc) will take the disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 3 initiative to publish a meaningful summary of key matters arising at the agm, which, among others, should include pertinent questions raised at the agm and the board’s responses as well as the full attendance list of the directors at the agm” (website mswg). so far, studies conducted by samaha, khlif, and hussainey (2015) on voluntary disclosures focused on corporate social responsibility (csr), intellectual capital, directors’ remuneration, dividend policies, forward-looking information, risk disclosures, and internet disclosures. despite the role of agm minutes as prima facie evidence of documentation, there is a noticeable lack of studies on the disclosure of agm minutes on corporate websites. it is noted that an agm can be stated as the place where all members are called to generate forums for discussions in accordance with company law which provides valuable opportunities for promoting organizational accountability (cordery, 2005, p. 2). in other words, the agm should be a place for members to support the closing of the accounting year (catasús & johed, 2007: p. 172). in malaysia, agm occurs once a year whereby company executives (top management) together with the board of directors (non-executive and independent directors) and shareholders/investors meet. the focus point of this paper is aligned with section 143(1) of the previous companies act 1965 whereby all companies established in malaysia (public or private) are required to convene its first agm within eighteen (18) months from the date of incorporation of the company in malaysia (malaysia companies act 1965, amended 1999). however, effective january 31, 2017, due to the transition period from the companies act 1965 to the new companies act 2016, there was a major change regarding the convening of an agm. beginning january 31, 2017, section 340(1) of the companies act 2016 states that “every public company shall hold an agm in every calendar year besides other meetings held during that period, to transact the following business: the laying of audited financial statements and the reports of the directors a) and auditors; the election of directors in place of those retiring;b) the appointment and the fixing of the fee of directors; andc) any resolution or other business of which notice is given in accordance d) with this act or the constitution.” in summary, the companies act 1965 stressed the importance of the agm to be held once in every calendar year either by the circulation of resolution or physical agm regardless whether they are public or private companies. in contrast, the companies act 2016 (ca 2016) only required publicly listed companies (plcs) to hold physical agms once in every calendar year. this is 4 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 no longer applicable for private companies in malaysia (companies act, 2016 on section 340(1): p. 320)1. it is notable that the implementation of the ca 2016 is a benchmark towards good governance and applicable to all sections except section 241 and division 8 of part iii. interestingly, the new companies act 2016 highlighted that under section 340(3) of ca 2016, only public companies whether listed or non-listed need to convene a physical meeting within eighteen (18) months from the date of incorporation of the company in malaysia. with the updated main market listing requirements, the outcome of the agm which consists of the total voting results will be made available to the public on the day the meeting is conducted. meanwhile, for companies that applied, they are required to ensure that their corporate websites are active and accessible to the public2. in fact, the company is solely responsible in ensuring that the link to company announcements appears in its corporate website(s). based on the regulations, section 343 of the ca 2016 states that agm minutes is a formal record of proceedings at a meeting and is considered prima facie evidence. thus through the agm minutes, management and investors will know the list of participants during the meeting, the agenda discussed, questions raised by the audience, time frame of the meeting, total voting of resolution(s), and outcome of the meeting which can benefit existing and potential shareholders who are unable to attend or are not involved in the event as mentioned in the best practice guide on agms for listed issuers published by maicsa-bursa malaysia berhad in 2016 (maicsa, 2016). the rest of this paper is organised as follows; section 2 briefly presents the literature review and hypotheses development; section 3 discusses the research methodology; section 4 highlights the results and analysis, and section 5 presents the conclusion and contributions of this paper. 2. literature review currently, there is no accepted standard guideline regarding agm minutes even though generally there are a few formats of minutes prepared and drafted in malaysia. each type and format of minutes depends on specific usage where it fulfills certain objectives (foo, 2019, p. 25). to date, the securities commission of malaysia, institutional investor council of malaysia, and the minority shareholders watch group have launched the agm corporate governance checklist for shareholders on february 5, 2020 which includes the importance of agm minutes to enhance a better perspective of corporate governance (securities commission of malaysia, 2020, p. 16). collectively, the agm minutes acts as evidence of a written record and must be audited by a licensed auditor, yearly. in fact, the chairman used disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 5 the agm minutes as confirmation of a true record of the previous meeting (the malaysian association of the institute of chartered secretaries and administrators [maicsa], 2016). however, there is an issue regarding the need for agm minutes to be disclosed on corporate websites which had been non-mandatory in malaysia. the compulsory disclosure of agm minutes on corporate websites was effected in 2016 with the amendment of the main market listing requirements chapter 9-continuing disclosure paragraph 9.21(2) (b) which stipulates that “a listed issuer must publish the following information on its website such as ...summary of the key matters discussed at the agm, as soon as practicable after the conclusion of the agm” (website bursa malaysia). the gap in the literature is related to the reluctance of the plcs to publish their agm minutes on their corporate websites even though bursa malaysia berhad has released a letter to the board of directors to comply with paragraph 9.21(2) for both main markets and ace markets. this can be seen as the mswg noted that only the top 100 listed companies published their agm minutes on their corporate websites from 2017 onwards instead of all listed companies in malaysia. there has been a lot of studies which focused on the meeting process (apostolides & boden, 2005; nyqvist, 2015), implementation of corporate governance (apostolides, 2010), conduct of meetings (carrington & johed, 2007; schwartz-ziv & weisbach, 2013), and criteria on information of corporate website disclosures (kelton & yang, 2008; martherly & burton, 2005). however, there is a lack of studies on the disclosure of agm minutes on corporate websites in malaysia and other countries. the publication of agm minutes on corporate websites can be considered as one of the new prospects in voluntarism that has been given less attention by researchers, either locally or globally. in this paper, the voluntary disclosure is extended by examining the agm minutes on corporate websites in malaysian plcs with the financial year ended december 31, 2016 and focused on companies listed under the main market. apart from examining the disclosure of agm minutes on corporate websites, this paper also addressed the impact of shareholders and grey directors’ activism (through voting percentage) on such disclosures. in 2018, bursa malaysia made it compulsory for all malaysian plcs to provide a corporate website and publish key matters discussed at the agm as soon as possible after the conclusion of the meeting (bursa saham, 2018, p. 2). 2.1 trend of agm minutes on corporate website table 1 shows that 48 percent of the top 100 malaysian plcs published their agm minutes in 2016 which is slightly better compared to the period from 2013 6 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 to 20153. it indicated the willingness of the companies to disclose agm minutes on their corporate websites which reflected good governance (rita & salleh, 2016). moreover, the malaysia-asean corporate governance report (2015) revealed that calls by mswg for a voluntary publication of agm minutes on corporate websites was well received by most malaysian plcs. most of the listed companies agreed to publish agm minutes within 30 days after the conclusion of their agm event (malaysia-asean corporate governance report, 2015). table 1. table extract of minority shareholder watch group for all top 100 malaysian plcs between 2012 and 2016 other disclosure 2016 (%) 2015 (%) 2014 (%) 2013 (%) 2012 (%) companies with board charters 94 97 80 70 38 companies with code of ethics 90 77 68 57 58 companies which published agm minutes 48 37 26 7 1 companies which published m&a 41 28 22 11 8 companies which disclosed individual director remuneration 43 33 35 39 34 companies with dividend policies 41 38 35 38 34 companies with whistle blowing policies 86 70 51 48 42 companies with corporate responsibility policies 94 93 97 94 90 companies which disclosed training attended by each of its directors 80 65 59 65 56 annual financial report released within four months 99 99 82 81 81 *source: minority shareholder watch group media release (2016). in comparison, table 2 reported specific agm minutes disclosed on corporate websites before and after the implementation of the ca 2016. the table indicated that there was an increase in disclosures of agm minutes from disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 7 nine percent (2016) to 37 percent (2017). the huge jump in percentage was due to initiatives enforced by the regulatory bodies that required all listed companies either the top 100 malaysian plcs or non-top 100 malaysian plcs to publish their agm minutes or key matters on their corporate websites as soon as possible after their meetings. in accordance with the ca 2016, which began in 2017, several voluntary disclosure items such as the establishment of dividend policies, whistle blowing policies, implementation of corporate responsibility policies, disclosure of training attended by individual directors, and the release of annual financial reports (fy) within four months became either not available or mandatory as stipulated by the bursa malaysia listing requirements. table 2. table extract of minority shareholder watch group for all listed companies between 2016 and 2017 other disclosure 2017 (%)* 2016 (%) companies with board charters 86 82 companies with code of ethics 60 57 companies which published agm minutes 37 9 companies which published m&a n/a n/a companies which disclosed individual director remuneration 11 10 companies with dividend policies n/a n/a companies with whistle blowing policies n/a n/a companies with corporate responsibility policies n/a n/a companies which disclosed training attended by each of its directors n/a n/a annual financial report released within 4 months n/a n/a *source: malaysia-asean cg report (2017). note: (*) n/a represents not available. previous studies described shareholder activism as how shareholders exert influence by way of their involvement and engagement with company management through monitoring and control mechanisms based on governance 8 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 structure (akmal musa & ismail, 2016; gillan & starks, 2000). empirical evidence has divided shareholder activism into public pension funds, coordinated groups of investors, and individual investors (gillan & starks, 2000, p. 277– 278). to reiterate, the need for active participants especially in corporate policy has affected managers’ interests. in a study involving 130 active investors of firms from the united states of america showed that managers tend to react toward activism based on disclosure information (bourveau & schoenfeld, 2017, p. 30). as such, passive shareholders focused on additional disclosures tend to vote against resolutions if a firm does not fulfill their preferences (bourveau & schoenfeld, 2017: p. 3). on the other hand, activist shareholders are more likely to become proactive for the sake of their investors. the manager is aware that once the company increases its vulnerability, there is more susceptibility in terms of exposure to activism (klein & zur, 2009, 2011). unfortunately, even though shareholder activism by institutional investors has been increasingly prominent over the last few years, studies which investigate the effects of such activism towards voluntary disclosures are relatively limited (gillan & starks., 2000). in accordance with de jong, mertens and roosenboom (2006). shareholders’ voting at general meet and roosenboom (2006), shareholder activism during a meeting on dominant shares tends to influence agency cost and maximize shareholders’ wealth. this paper can contribute to the literature as not many studies are focused on the emergence of shareholder activism (voting percentage) especially in terms of the numbers in favor or against the resolution for voluntary disclosure of agm minutes on corporate websites. 2.2 issue with shareholder activism van der elst (2012) studied whether there was a relationship between shareholder rights and shareholder activism in annual general meetings. the result of a comparison among five countries showed no significant relationship between the number of items or the importance of items to be voted at general meetings and the attendance of shareholders. however, it was highlighted that shareholders have the power to attend and support items in the agenda during agms. there is also the issue of free-riding or apathetic shareholders during agms which suggests the need for a more in-depth analysis regarding shareholder activism (van der elst, 2012, p. 63). bourveau and schoenfeld, (2017) who studied the relationship between voluntary disclosure and shareholder activism suggested that, broadly, shareholder activism plays a crucial determining role in voluntary disclosures. the evidence also suggests that managers are more receptive to disclosures once shareholders are active in their firm. disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 9 van der elst (2012) identified shareholders as the owners of the company. in essence, shareholders are the supreme and final decision makers of the company even though they only own company shares and not the company management. currently, the power of shareholders is only limited to voting on issues governed by the law or to allocate decision-making power in the general meeting. this limitation of power leads to active participation and the tendency to vote for or against resolutions at general meetings namely, through having share vote percentage (van der elst, 2012, p. 40–41). previous studies on malaysian companies in 2001–2002 regarding the impact of family-controlled companies and board composition towards corporate transparency indicated that companies with more affiliated directors brought significant changes to corporate disclosures (wan-hussin, 2009). generally, majority shareholders tend to dominate the decision-making process instead of minority shareholders (catasús & johed, 2007, p. 178). this is because other than attending general meetings, shareholders also exercise their power through voting resolution (voting percentage). traditionally, in malaysia, most of the dominant voting process was either by a show of hands or by poll (ca 2016 section 146)4. however, since the new ca 2016, bursa malaysia requires all resolutions in general meetings to be voted by poll. the proposal to exercise voting by poll is also recommended by the latest malaysian code on corporate governance (mccg 2017-practice 12.3). does share voting influence the decision to disclose agm minutes on corporate websites? as mentioned earlier, agm minutes on corporate websites are still relatively new in malaysia. however, in the united kingdom, it has already established the united kingdom code of corporate governance as “... when, in the opinion of the board, a significant proportion of votes have been cast against a resolution at any general meeting, the company should explain when announcing the results of voting what actions it intends to take to understand the reasons behind the vote result...” (council, 2012). therefore, the importance of the number of votes (voting percentage) whether in favor or against the resolution can have a significant impact on the disclosure of agm minutes on corporate websites. based on the statement, it is predicted that shareholder participation can influence the disclosure of agm minutes on corporate websites. thus, the following hypothesis is posited: h1: there is a positive relationship between shareholder participation and disclosure of agm minutes on corporate websites. 2.3 emerging grey directorship previous studies showed that the function of grey directors is mainly to receive less attention compared to other directors. the agency theory suggests that 10 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 agency conflicts happen due to the separation of ownership and control in a firm. through this theory, the directors bear responsibility for the company’s future value in disseminating private information to the public (jensen & meckling, 1976). grey directors can be defined as directors that have social ties with company management such as a former director, possess shares in the company, and provide beyond monitoring and advisory functions of the boardroom (hsu & wu, 2014). according to baysinger and butler (1985), grey directors comprise a mixture of executive and independent directors who provide proper administrative and multiple tasks (as cited in hsu & wu, 2014). it is also reported that companies with a large proportion of grey directors in the board composition can hinder corporate failures such as in the united kingdom (hsu & wu, 2014). borokhovich, boulton, brunarski and harman (2014) found that grey directors with shares in a company could jeopardise their position as directors of the company. based on a sample from 370 top executives and board chairs from 1978 to 2007, borokhovich et al. (2014) noted that a firm’s succession plan is influenced by the power of grey directors. the results also revealed that grey directors’ ownership helps to align company interest with that of the shareholders. to date, it has been debated based on empirical evidence whether the grey director tends to act as advisor instead of monitoring the boardroom (vicknair, hickman & carnes, 1993). unlike executive or independent directors, the grey director plays a crucial role as independent advisor and decision maker of the company. thus, it is posited that grey directors could affect management strategy in voluntary disclosure. a paper by ibrahim and angelidis (1995) revealed that outside directors (which include grey directors) tend to influence company performance because of their strong mutual relationship with the stakeholders as these directors know their expectations and are able to entertain investors’ demands. the result of the paper indicated that outside directors provide greater corporate responsibility but a weaker orientation towards the economic performance of companies. it is predicted that grey directors can influence the disclosure of agm minutes on corporate websites. hence, the second hypothesis is as follows: h2: there is a positive relationship between grey directors and disclosure of agm minutes on corporate websites. 3. methodology in this paper, we selected companies that have annual reports made publicly available in their corporate websites and the bursa malaysia website. we chose disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 11 2016 as the financial year ended for our sample because the preferred year ended by the plcs is either in june or december. hence, on january 31, 2017, the implementation of the new ca 2016 superseded the previous acts together with the latest malaysian code on corporate governance (mccg, 2017) in april 2017 and enhancement of the bursa malaysia securities berhad listing requirements subsequent to the release of the mccg (corporate governance guide 3, 2017). the initial sample was 274 companies. we excluded 13 companies due to missing data or unfulfilled criteria. the final sample with the required data was 261 companies listed under the main market of the bursa malaysia website. apart from collecting annual reports, notices of annual general meetings, outcomes of general meetings and corporate governance (cg) reports, we also collected board profiles, company profiles and other data from the bloomberg database, thomson reuters datastream database and respective corporate websites. as mentioned earlier, this paper was conducted based on companies listed in the main market of bursa malaysia for the financial year ended december 31, 2016. this paper also applied secondary data as the main data collection. the information on company characteristics and directors’ profiles were collected, coded and extracted from annual reports published under bursa announcements. in addition, the disclosure of agm minutes on the corporate website was downloaded, analyzed and extracted from each company’s corporate website. 3.1 measurement of variables the measurement of variables in this paper, adopted from studies by kassim, ishak, and manaf (2013, p. 322) were proxies based on company characteristics and directors’ profiles namely market sector, business activity, directors’ profiles and related governance information. the asean cg scorecard served as a benchmark because it provides details and rigorous methodology against international best practices to assess corporate governance performance among asean countries consisting of malaysia, indonesia, singapore, thailand, vietnam, and the philippines. the scorecard helps investors and public issuers to compare and contrast their corporate information for a transparent investment decision-making process. besides, according to the asean corporate governance scorecard country reports and assessments from 2011 to 2014, it encourages foreign direct investments among the six participating asean member countries. 3.2 definition of each variable 3.2.1 dependent variable—agm minutes on corporate website a detailed scoring index of this paper was adopted from the checklist used by haniffa and cooke (2005) with minor changes to fit the context of malaysian 12 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 plcs. a level disclosure of agm minutes was collected by observing each malaysian publicly listed corporate website under the column: “corporate governance”, “media release”, “investor relations” or its related section(s) to identify whether the companies published their agm minutes or not. besides, we also emailed to each firm that did not publish its agm minutes to confirm the non-disclosure of agm minutes on its corporate website. the dependent variable, vd of this paper is the level of voluntary disclosure of agm minutes measured by score (0) if there is no disclosure of agm minutes on the corporate website and score (1) for otherwise. 3.2.2 independent variable—shareholder activism for the independent variables in this paper, data were collected from annual reports of each corporate website, cg reports, bloomberg database and thomson reuters datastream database of the sampled firms. with regard to shareholder participation, our paper indicated participation by measuring the resolution with share votes (unit) against the agenda of the meeting divided by the total number of shares (unit). percentage share votes against the resolution indicate the visibility of shareholder activism during the agm event. therefore, shareholder participation, sa_vote % in this paper was measured by way of share voting which was calculated with share votes (unit) divided by the total number of shares (unit). share votes (unit) was measured based on the outcome of the meeting from the bursa announcement by identifying the higher total number of votes for and against each resolution (shareholder participation ratio at agm). meanwhile, the total number of shares (unit) was collected from the total number of the issued share capital of the companies as per financial year ended december 31, 2016. shareholder participation, sa vote (%)= share votes (unit)/ total number of shares (unit)*100. 3.2.3 independent variable—grey directors the second independent variable was the grey director. the details of the measurement for a firm assigned with grey directors, bod_grey% was measured by classification: “executive” represented the position as executive director of the company; “independent non-executive” represented the position as an independent non-executive director of the company and “grey” to represent the position as grey director/non-independent non-executive director of the company. for directors with bod_grey%, by using the scoring index, the total number of grey directors of each firm was combined, recorded and coded by the total number of grey directors in each firm. disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 13 3.2.4 control variable—board size according to kassim et al. (2013), previous studies had shown that large companies preferred to establish various diversifications in business such as sime darby group of companies, fgv group of companies (formerly known as felda global ventures group of companies), kpj group of companies and genting group of companies. theoretically, large size companies, require large capital injection which leads to the diversity of its investors, bodsize. therefore, the role of the board of directors such as executive, non-executive or grey director with a single tier or two-tier leadership could influence disclosure of the agm minutes on the corporate website. a large size company requires a large boardroom to control and monitor the performance of the company. hence, a mixture of board diversity and position could provide better decision-making. from a negative perspective, it is argued that the large number in board size could create chaos and a tendency for personal benefit instead of fulfilling company objectives. from a positive perspective, diversity in the board of directors such as the presence of independent non-executive directors could help the management and owners (investors) to aim for profit maximization. this is because independent non-executive directors work for the company and provide checks and balances in decision-making. besides that, the more independent the directors the more voluntary information disclosed by the company. due to this reason, bodsize is expected to influence voluntary disclosure of agm minutes. 3.3 empirical models the following model was developed to achieve the research objectives used to examine the disclosure of annual general meeting on the corporate website: where, vd means voluntary disclosure of agm minutes on the corporate website, independent variables represented by bod grey% to indicate grey directors, sa vote% to denote shareholder participation and bodsize to mean board size. details of the operational definitions of the variables, measurements and sources are shown in table 3. vd = 0+ 1 sa_vote% + 2 bod_grey% + 3 bodsize + i 14 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 table 3. operational definition of each variable of the study variable(s) proxy indicator measurement source vd disclosure of agm minutes on corporate website dv • indicate voluntary disclosure index • (1=disclose agm minutes on corporate website, and 0=otherwise) corporate website sa_vote% shareholders’ activism through voting percentage iv • share votes (unit)/ total number of shares (unit)*100 annual report, bursa outcome, corporate website bod_ grey% grey director iv • step 1: classification grey director (non-independent non-executive director/outside director) • step 2: total number of grey directors/ total number of directors*100 annual report, datastream, bloomberg bodsize board size cv • indicated based on the total number of directors for each firm annual report, datastream, bloomberg indicator(s): dv denotes dependent variable, iv denotes independent variable and cv denotes control variable of the study. 4. analysis of results this section reports on the general findings, from descriptive statistics followed by pearson partial correlation, one-sample t-test and regression results. it also aims to show the relationship between the dependent and independent variables in the study. the results confirm the two hypotheses posited. the data is analyzed based on the methodology described earlier. this chapter reported the findings with discussions and suggestions for future research. this chapter arranged with a summary of general findings, disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 15 future research recommendations and the conclusion. the result will be analysed based on the three research questions (rqs) and research objective (ros) using spss software. the chapter aims to shows the association between dependent variables and independent variables. the data acquired will be analysed based on the methodology explained in earlier chapters. figure 1. distribution of agm minutes on corporate website among malaysian public listed companies. from figure 1, based on 261 companies listed under the main market of bursa malaysia, it was found that 78 companies (29.89%) did not publish agm minutes on the corporate website (indicated by “0”) whereas 183 companies (70.11%) disclosed the agm minutes on the corporate website (indicated by “1”). the willingness of the listed companies to publish agm minutes on the corporate website reflected their alignment with the initiatives of the mswg to promote good governance through transparency and accountability5. table 4. descriptive statistics on agm minutes on corporate websites, grey directors, shareholder activism and board size. minimum maximum mean std. deviation vd 0.00 1.00 0.70 0.45 sa_vote% 2.76 99.66 58.95 20.53 bod_grey% 0.00 60.00 19.56 16.63 bod size 4.00 15.00 7.51 1.85 17 4. analysis of results this section reports on the general findings, from descriptive statistics followed by pearson partial correlation, one-sample t-test and regression results. it also aims to show the relationship between the dependent and independent variables in the study. the results confirm the two hypotheses posited. the data is analyzed based on the methodology described earlier. this chapter reported the findings with discussions and suggestions for future research. this chapter arranged with a summary of general findings, future research recommendations and the conclusion. the result will be analysed based on the three research questions (rqs) and research objective (ros) using spss software. the chapter aims to shows the association between dependent variables and independent variables. the data acquired will be analysed based on the methodology explained in earlier chapters. figure 1. distribution of agm minutes on corporate website among malaysian public listed companies. from figure 1, based on 261 companies listed under the main market of bursa malaysia, it was found that 78 companies (29.89 percent) did not publish agm minutes on the corporate website (indicated by “0”) whereas 183 companies (70.11 percent) disclosed the agm minutes on the corporate website (indicated by “1”). the willingness of the listed companies to publish agm minutes on the corporate website reflected their alignment with the initiatives of the mswg to promote good governance through transparency and accountability5. 16 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 from table 4, based on the sample of 261 companies, grey directors were in the range of 0 to 60 percent per company, with a mean of 19.56 and a standard deviation of 16.63. share vote percentage ranged from 2.76 percent to 99.66 percent with a mean of 58.95 and a standard deviation of 20.53. board size, ranged from 4 to 15 directors, with a mean of 7.51 and a standard deviation of 1.85. interestingly, there were several companies with more than half of its composition dominated by grey directors. even though there were companies with no grey directors from the table, surprisingly there were companies with more than eight grey directors such as bintulu port holdings berhad and pba holdings berhad. the awareness of the role played by grey directors concurred with results by abdulmalik, ahmad and aliyu (2015) in that the proportion of grey directors in the boardroom was significantly and positively related to accrual and real earnings management compared to independent directors. this result also supported the argument that the existence of grey directors could influence disclosure of agm minutes which was seldom explored by previous researchers. meanwhile, for shareholder participation via proxies by shareholder vote percentage at agm events, the results showed that the minimum was 2.76 percent and the maximum was 99.66 percent in terms of shareholders exercising their voting rights. therefore, frontken berhad had a lower shareholder vote whereas bimb berhad, mmc corporation berhad, osk holdings berhad, classic scenic berhad, petronas gas berhad, and sp setia berhad had higher shareholder votes during agm events. this reflected the assertiveness of the shareholders as to how the management should utilize their shareholdings for profit maximization. the active participation of the shareholders or by way of the presence of a corporate representative (proxy) was in accordance with the mccg 2017 in encouraging more shareholders’ attendance during agm events. overall, for board size, public listed companies have a board size range of four to 15 directors. for instance, pba holdings berhad has 15 directors comprising nine grey directors and six independent non-executive directors wholly owned by the state government. the composition of board size acts as a control variable that may influence disclosure of agm minutes on the corporate website among malaysian plcs. the main reason is because having a mixture of directors of different characteristics, work experiences and level of education could lead to more willingness to disclose agm minutes on corporate websites as encouraged by mswg and bursa malaysia. accordingly, table 5 of the partial correlation shows the dependent variable, vd; independent variables, sa_vote% and bod_grey% and the control variable, bodsize. both the two independent variables produced a strong disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 17 correlation of 0.206** and 0.211**, respectively (significant 2-tailed). in fact, the control variable also provided a good correlation with 0.124* (significant 2-tailed) in this paper. table 5. pearson partial correlation variable vd sa_vote % bod_grey% bodsize vd 1.00 sa_vote% 0.206** 1.00 bod_grey% 0.211** 0.261** 1.00 bodsize 0.124* 0.343** 0.260** 1.00 notes: ** indicated the significant at the 0.05 level (2-tailed). tables 6 shows the results based on one-sample t-test. the one-sample t-test was used to confirm the assumption that the dependent variables, vd01 were measured either at internal or ratio level, that the data were independent and not related, with no significant outliers and that the dependent variables should be normally distributed. thus, there was a significant difference in the mean (vd, sa_vote%, bod_grey% and bod size) between the sample and overall population and that the average mean for vd was 0.70, sa_vote% was 58.95, bod_grey% was 19.56 and bodsize was 7.51 compared to the overall population average. table 6. one-sample t-test t-value mean 95 % confidence interval of the difference lower upper vd01 24.698 0.701 0.650 0.760 sa_vote% 46.385 58.954 56.452 61.457 bod_grey% 18.992 19.560 17.532 21.588 bodsize 65.494 7.506 7.280 7.730 based on table 7, the r square of the paper was 0.007 and the adjusted r square was 0.059 which was slightly low due to minimum variables tested in this paper. the multiple regression showed that while disclosure of agm minutes on corporate website was constant, vd, both the independent variables, sa_ vote% and bod_grey% produced highly significant results, 0.002 and 0.001, 18 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 respectively. these results supported h1 and h2 that inside forces (such as grey directors) do influence the disclosure of agm minutes on corporate websites and outside forces (such as shareholder participation during agm) also provided significant impact on the disclosure of agm minutes on corporate websites. however, the control variable, bodsize did not have a significant impact on the relationship with vd01, sa_vote%, and bodgrey%. thus bodsize did not influence the disclosure of agm minutes on corporate websites. table 7. regression results of study unstandardized coefficients b std. error standardized coefficients beta tstatistics sig. (constant) 2.896 0.004 sa_vote% 0.003 0.001 0.153 2.343 0.020 bod_grey% 0.005 0.002 0.164 2.582 0.010 bod size 0.007 0.016 0.029 0.451 0.653 adjusted r2 0.059 f ratio 6.41 significance 0.00 based on the results in section 4, preliminary analyses were conducted to ensure no violation of the assumptions of normality, linearity, multicollinearity and homoscedasticity. r square value was 1 percent of the variance by including board size as a control variable which was relatively low. the statistically significant contribution, was indicated by the sig. f change value for the line (.000). in addition, the anova table indicated that the model was significant (f(261) = 6.41 < .001). meanwhile, the coefficients table, showed that the sig. column, three variables, sa_vote% (beta = 0.15), sa_vote% (beta = 0.16) and bodsize (beta = 003) made a unique contribution towards voluntary disclosure of agm minutes on corporate websites. as shown, control measures were statistically non-significant, with grey directors recording a higher beta value followed by shareholder participation and board size. the pivotal role of shareholder participation towards voluntary disclosure concurred with de jong et al. (2006) in that shareholders become proactive in exercising their rights as company owners. in addition, as mentioned by bourveau and schoenfeld (2017), the more shareholders utilize their voting rights, the more it may affect the voluntary disclosure information, in this case, the agm minutes on corporate websites. despite the few studies on grey directors, hsu and wu (2014) stressed that grey directors are also part of the directorship of companies. by playing disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 19 their role as outsider directors, grey directors are free from the management of companies and therefore are able to influence company decisions as suggested by angelidis and ibrahim (1995). the role of grey directors should be measured in a broader perspective as grey directors serve not only to fit the composition of directorship but they can also influence board directorship in aligning shareholder demands and needs. thus, h1 there is a positive relationship between shareholder participation and disclosure of agm minutes on corporate websites. accepted h2 there is a positive relationship between grey directors and disclosure of agm minutes on corporate websites. accepted 5. conclusion in a nutshell, this paper examined the relationship between voluntary disclosures of agm minutes on the corporate website based on a sample of malaysian public listed companies for the financial year ended december 31, 2016. it revealed that disclosures in agm minutes in 2016 were relatively low, at 47 percent compared to other disclosures from all listed companies. this suggests that several factors may have influenced the willingness of company management to disclose their agm minutes on the corporate website. there are several noteworthy contributions in this paper. to the best of our knowledge, this paper is the first to highlight the importance of voluntary disclosure of agm minutes on the corporate website. next, this paper has also examined the link between shareholder proxies (by way of shareholder participation voting at agm events) and grey directors on the disclosure of agm minutes on the corporate website. besides, this paper offers some insights on shareholder participation and grey directors using data from a developing country, malaysia. this paper in particular, complemented previous research which emphasized the importance of grey directors in influencing voluntary disclosures. this paper also noted that meaningful disclosure of agm minutes on the corporate website in malaysia was relatively low in 2016 during the period of study. subsequently, the new companies act followed by the malaysian code on corporate governance and the corporate governance guide 3 came into effect in 2017. eventually based on the minority shareholder watch group media released in 2017, almost 78 percent of the top 100 plcs disclosed their 20 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 agm minutes on the corporate website. the disclosures were aligned with bursa malaysia requirements in paragraph 9.21(2) (b) on the disclosure of agm key matters after a meeting is held. the findings showed that grey directors and shareholder activism (voting percentage) did influence the level of agm minutes of the disclosure. the results supported the government initiatives in introducing the malaysian code for institutional investors 2014. as such, this paper has emphasized the importance of grey directors and shareholders which are relatively unexplored. future studies should be conducted on extensive factors which lead to the disclosure of agm minutes on the corporate website as the way forward in promoting good corporate governance. 6. acknowledgements the authors would like to thank the staff of the minority shareholders watch group for the necessary support and assistance. we would also like to thank the editor and independent reviewers of the international journal of banking and finance for their constructive comments and suggestions. endnotes 1 malaysia companies act 2016. retrieved from https://www.ssm.com.my/ pages/ legal_ framework/document/act%20777%20reprint.pdf (section 340(2) on annual general meeting). 2 fullarticle. retrieved from https://www.bursamalaysia.comsites/5bb54b e15f36ca0af339077a/assets/5c10c71ef67d7c7c0037bfc7/icn_1_2018. pdf (issuers communication corporate website for listed issuers (icn 1/2018). 3 minority shareholder watch group or mswg was established in 2000 as one of the government’s initiative to protect the interests of minority shareholders through shareholder activism by encouraging good governance amongst public listed companies with the objective of raising shareholder value over time. 4 section 332(1) of the ca 2016 and paragraph 8.29a(1) of the bursa malaysia listing requirements stated that any resolution by plc laid in the notice of agm must be voted by poll. 5 on july 18, 2018, bursa malaysia berhad released a letter to the board of directors (listed issuers/corporations) on compliance with paragraph/rule 9.21 of the main/ace market listing requirements. disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 21 references abdulmalik, s., ahmad, a. c., & aliyu, u. b. (2015). financial reporting quality: the role of independent and grey directors, board continuous training and internal audit function. ipbj, 7(2), 40–57. akmal musa, m., & ismail, s. e. (2016). shareholder activism, institutional shareholders and agency problems in malaysia. global journal of business and social science review, 4(3), 30–36. apostolides, n. (2010). exercising corporate governance at the annual general meeting. corporate governance: the international journal of business in society, 10(2), 140–149. apostolides, n., & boden, r. (2005). cedric the pig: annual general meetings and corporate governance in the uk. social r e s p o n s i b i l i t y journal, 1(1/2), 53–62. bank, a. d. (2016). asean corporate governance scorecard country report and assesment 2014. baysinger, b. d., & butler, h. n. (1985). corporate governance and the board of directors: performance effects of changes in board composition. journal of law, economics & organization, 1(1), 101–124. borokhovich, k. a., boulton, t. j., brunarski, k. r., & harman, y. s. (2014). the incentives of grey directors: evidence from unexpected executive and board chair turnover. journal of corporate finance, 28, 102–115. bourveau, t., & schoenfeld, j. (2017). shareholder activism and voluntary disclosure. review of accounting studies, 22(3), 1307–1339. carrington, t., & johed, g. (2007). the construction of top management as a good steward: a paper of swedish annual general meetings. accounting, auditing & accountability journal, 20(5), 702–728. catasús, b., & johed, g. (2007). annual general meetings—rituals of closure or ideal speech situations? a dual analysis. scandinavian journal of management, 23(2), 168–190. cordery, c. j. (2005). the annual general meeting as an accountability mechanism. retrieved from https://papers.ssrn.com/sol3/papers. cfm?abstract_id=1286930 corporate governance guide 3 (2017). retrieved from https://bursasustain. bursamalaysia.com/dropletdetails/resources/corporateg o v e r n a n c e guide. council, f. r. (2012). the uk corporate governance code. london, september. de jong, a., mertens, g., & roosenboom, p. (2006). shareholders’ voting at general meetings: evidence from the netherlands. journal of management and governance, 10(4), 353–380. 22 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 foo, k. p. k. (2019). corporate voice-minutes of meetings part 2, 25.retrieved from http://maicsa.org.my/interactive/ejournal/ejournal. aspx?ejournal=1810. gillan, s. l., & starks, l. t. (2000). corporate governance proposals and shareholder activism: the role of institutional investors. journal of financial economics, 57(2), 275–305. haniffa, r. m., & cooke, t. e. (2005). the impact of culture and governance on corporate social reporting. journal of accounting and public policy, 24(5), 391–430. healy, p. m., & palepu, k. g. (2001). information asymmetry, corporate disclosure, and the capital markets: a review of the empirical disclosure literature. journal of accounting and economics, 31(1–3), 405–440. hsu, h. h., & wu, c. y. h. (2014). board composition, grey directors and corporate failure in the uk. the british accounting review, 46(3), 215– 227. ibrahim, n. a., & angelidis, j. p. (1995). the corporate social responsiveness orientation of board members: are there differences between inside and outside directors? journal of business ethics, 14, 405–410. jensen, m. c., & meckling, w. h. (1976). theory of the firm: managerial. journal of financial economics, 3, 305–360. kassim, a. a. m., ishak, z., & manaf, n. a. a. (2013). board effectiveness and company performance: assessing the mediating role of capital structure decisions. international journal of business and society, 14(2), 319. kelton, a. s., & yang, y. w. (2008). the impact of corporate governance on internet financial reporting. journal of accounting and public policy, 27(1), 62–87. klein, a., & zur, e. (2009). entrepreneurial shareholder activism: hedge funds and other private investors. the journal of finance, 64(1), 187–229. klein, a., & zur, e. (2011). the impact of hedge fund activism on the target firm’s existing bondholders. the review of financial studies, 24(5), 1735–1771. malaysian association of the institute of chartered secretaries and administrators maicsa. (2016). best practice guide on agms for listed issuers. retrieved from http://www.maicsa.org.my/ download/publication/ publication_agm.pdf. main market and ace market on issuers communication corporate website for listed issuers (icn 1/2018). retrieved from http://www. bursamalaysia.com/misc/system/assets/25717/icn_1_2 0 1 8 % 2 0 % 2 0 guidanceon corporatewebsitedisclosure.pdf. main market listing requirement (2016). disclosure of agm minutes on corporate website: an assessment of malaysian listed companies: 1-24 23 malaysia companies act. (1965, amended 1999). kuala lumpur: international law book services. malaysia companies act. (2016). kuala lumpur: international law book services. malaysia-asean corporate governance report (2015). retrieved from https:// www.mswg.org.my/sites/default/files/mswg_2015_aseancg_ scorecard_web_version_0.pdf. matherly, c. m., & burton, h. a. (2005). an analysis of corporate website disclosures. management accounting quarterly, 6(2), 26. mort, s. (1991). the minutes—a guide for company secretaries and managers. england: brookfield. nyqvist, a. (2015). the corporation performed: minutes from the rituals of annual general meetings. journal of organizational ethnography, 4(3), 41–355. poulsen, t., strand, t., & thomsen, s. (2010). voting power and shareholder activism: a paper of swedish shareholder meetings. corporate governance: an international review, 18(4), 329–343. rita, b., & salleh, h. (2016). report of agm practices by malaysian companies. malaysia-asean cg report 2015. kuala lumpur, malaysia. samaha, k., khlif, h., & hussainey, k. (2015). the impact of board and audit committee characteristics on voluntary disclosure: a metaanalysis. journal of international accounting, auditing and axation, 24, 13–28. schwartz-ziv, m., & weisbach, m. s. (2013). what do boards really do? evidence from minutes of board meetings. journal of financial economics, 108(2), 349–366. securities commission malaysia (2020). annual general meeting corporate governance checklist for shareholders. retrieved from http:www. maicsa.org.my/media/4980/technical_announcements_ 200206 _2.pdf. slack, r., & shrives, p. (2010). voluntary disclosure narratives: more research or time to reflect? journal of applied accounting research, 11(2),84–89. the malaysian association of the institute of chartered secretaries and administrators. (2016). best practice guide on agms for listed issuers.77–78. retrieved from http://www.maicsa.org.my/ download/ publication/publication_agm.pdf. van der elst, c. (2012). shareholder rights and shareholder activism: the role of the general meeting of shareholders. belgrade law review. vicknair, d., hickman, k., & carnes, k. c. (1993). a note on audit committee independence: evidence from the nyse on “grey” area directors. accounting horizons, 7(1), 53. 24 the international journal of banking and finance, vol. 15, no 2, 2020 : 1-24 wan-hussin, w. n. (2009). the impact of family-firm structure and board composition on corporate transparency: evidence based on segment disclosures in malaysia. the international journal of accounting, 44(4), 313–333. usury (riba) and the place of bank interest in islamic banking and finance the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 1 usury (riba) and the place of bank interest in islamic banking and finance m. raquibuz zaman ithaca college, united states of america abstract this paper examines the concept of usury or riba as was understood at the time of the prophet of islam and his contemporaries in mecca and medina, and what differing interpretations of the term developed in succeeding centuries in muslim populated countries of the world. it gives a brief summary of the concept of usury in judaism and christianity and how this term is equivalent of riba in islam. it demonstrates that riba and interest are not synonymous terms, and that what islam forbids is usury and not interest. it asserts that, although some interests are usurious, the claim by the contemporary islamic banking and financial institutions, ibfis, that these institutions are “islamic” because the term interest is not used in their transactions, is misleading at best. it ends with the proclamation that true ibfis are not only feasible, but also are inevitable to serve the needs of the muslims around the world. key words: profit sharing, usury, interest, prohibition of usury, islamic banking jel classification: g12, g21 1. introduction islamic banking and financial institutions, ibfi, have been in operation in some countries for around four decades, yet in most muslim countries where islamic as well as conventional banks operate, the ibfi are relatively very small, and command even smaller market share. it is not that the majority of the muslim citizens are ambivalent about islam, but they do not either believe in the islamicity of the ibfi, or that these institutions do not serve their needs. there are good reasons for skepticism on both counts. the basic impetus for the creation of the ibfi came from the belief of their modern proponents that all interest, irrespective of type and magnitude, are ijbf 2 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 usurious (riba, in islamic lexicon). in order to avoid the term “interest rate” the ibfi resurrected some financial practices that existed at the time of the prophet mohammad and the early centuries of islam. the islamic jurists continue to approve practices that were evolved from the earlier institutions and instruments without a critical examination to determine whether or not they meet the islamic precepts of justice and fair play – some say efficiency and serve the needs of the public in a muslim country. the objective of this study is to show the real historical meaning of the term usury or riba in islam, and why riba should be interpreted as usury and not ordinary interest rates. it examines how riba has come to be identified with interest at modern times, and how some muslim scholars has incorporated this limited or narrower interpretation in forming the ibfi. a second objective is to show that not only islam but judaism and christianity also struggled with the concept of usury for centuries. finally, the study briefly describes the operation of the contemporary “islamic” banking and financial institutions and shows how these can be changed and restructured to operate on truly islamic principles. 2. the concept of riba (usury) in islam 2.1 prohibition of riba according to the holy quran: “o you who believe! devour not usury, doubling and quadrupling (the sum lent). observe your duty to god, that ye may be successful.”(chapter iii, verse 130). “o you who believe! observe your duty to god, and give up what remain (due to you) from usury, if you are (in truth) believers. and if you do not, then be warned of war (against you) from god and his messenger. and if you repent, then you have your principal. wrong not, and you shall not be wronged.”(chapter ii, verses 278-279). “those who swallow usury cannot rise up save as he arise whom the devil has prostrated by (his) touch. that is because they say: trade is just like usury; whereas god permits trading and forbids usury. he unto whom an admonition from his lord comes, and (he) refrains (in obedience thereto), he shall keep (the profits of) that which is past, and his affair (henceforth) is with god. as for him who returns (to usury)—such are rightful owners of the fire. they will abide therein.”(chapter ii, verse 275). “god has blighted usury and made almsgiving fruitful. god loves not the impious and guilty.”(chapter ii, verse 276). the arabic word for usury is riba (pronounced as rēē’ba). according to the holy quran (chapter 3, verse 130, quoted above), riba is defined as “doubling and quadrupling (the sum lent).” this is the only definition of riba available from the quran. all the other verses (all of which were quoted above) admonish the believers to refrain from usury and avoid god’s punishment. judaism and christianity, also ask their followers to refrain from usurious transactions to avoid burning in hell fire (homer 1997: 69-81). the classic commentators of quran, such as, al-tabari (d. 923 a.d.), zamakhshari (d. 1144 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 3 a.d.) and ibn kathir (d. 1373 a.d.) argued that the quranic verse 3:130 on riba essentially talks about riba-al-jahiliya (i.e., usury practiced in the pre-islamic period in arabia). al-tabari in his jami (4:49) points out that riba in the preislamic period referred to doubling and redoubling of the principal amount lent in commodities over a period of time. at the due date, if the borrower failed to repay the amount borrowed, the loan was extended for another year but the borrower had to pay double the quantity originally lent. if, at the end of the second year, the borrower again failed to repay the loan, the loan was extended again for double the quantity of the second year to be repaid next year. thus, in two years’ time, the insolvent borrower would owe the original lender four times the quantity lent to him at the outset. al-zamakhshari, in his al-kashshaf (p. 234, date of publication unknown), presents a similar interpretation pointing out that even a small debt could consume all the wealth of a debtor because of repeated doubling of the unpaid loan. ibn al-arabi, (d. 1240 a.d.) in his ahkam al-quran (no date, v. 1: 241), also gives a similar interpretation of riba, that is riba al-jahaliya. prophet muhammad in his farewell hajj, clearly proclaimed that “. . . riba aljahiliya is null and void from this day and that riba of abbas bin abdul-muttalib has been waived.” ibn al-arabi (op. cit.: 241) states that: “riba was well known among the arabs. a person would sell something on a deferred payment basis and upon maturity the creditor would ask: ‘will you pay (now) or add an amount to the bill’?” (see also, zaman, m. r., and movassaghi, h. 2002). those who equate riba with interest seek support from al-jassas (d. 981 a.d.), who claimed that pre-islamic arabia practiced a form of riba where money was lent at a predetermined sum over the principal amount (ahkam alquran, v. 1: 465). however, there is no historical evidence to suggest that riba al-jahiliya also consisted of transactions that were similar to modern loans on interest (suhail 1936, tantawi 1989, afzal 1996, et al.). thus, the basis for equating all forms of interest, whether they double or quadruple the principal or whether they are a small fraction of the principle, must be sought elsewhere. let us look at the verified traditions or sayings (hadith) of prophet muhammad, which are the second major source of islamic (shari’ah) jurisprudence. 3. riba in hadith literature and their interpretation by the muslim jurists (fuqaha) the traditions (i.e., hadith) of prophet mohammad on riba are essentially confined to riba al-fadal dealing with the borrowing or lending of commodities e.g., gold, silver, etc. but returned/exchanged at the same point in time. these “hand to hand” transactions do not make any sense—why would anyone borrow, i.e., exchange gold for gold or for silver, unless it is part of a barter transaction, which is perfectly legitimate form of trade under islam. more about this later in the paper. 4 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 the sahih al-bukhari (1976, volume iii) cites a considerable number of hadith dealing with riba al-fadal. a very few hadith—indeed, if any specific one—can be found in the verified collection of hadith (1) on riba alnasia, i.e., collection of fixed and predetermined percentages of return on the amount of money lent, popularly known as “interest.” then one wonders what is the basis for regarding all kinds of interest as usurious? is it the meaning of the arabic word riba itself which lends to this type of assertion? the word riba apparently means “an excess or addition—i.e., an addition over and above the principal sum that is lent” (m. m. ali, no date, page 323, quoted from lane’s arabic-english lexicon). if the muslim jurist are referring to interest as usury on the basis of this literal meaning of riba, then naturally one wonders why god almighty used the terms “doubling” and “quadrupling” (the sum lent) as usury in chapter 3 verse 130 (quoted above), and why there was no further clarification of this verse or by the prophet. some of the muslim writers (see for example, afzalur-rahman, 1976, volume iii, pages 1-189) who claim interest is riba, have no problems in explaining the verses of the quran and the hadith, as they simply translate the term riba as interest and then assert it is forbidden (ibid., chapter 4). it appears as though they start with an axiom that interest is riba and, as such, forbidden. there is no other basis. it is instructive to note here that while the quran refers to riba in the context of loan (debt), the hadith literature cites it in the context of sales, and there is no mention of loan (qard) or debt (dayan). one such hadith that is often quoted, and is known by the epithet “six commodities hadith” relates to the time around the khyber expedition—well before the prohibition on riba was imposed by god’s revelation (afzal, op. cit., 3) – was as follows (it was narrated by abu sa’id al-khudri on the authority of the prophet): “gold is to be paid for by gold, silver by silver, wheat by wheat, barley by barley, dates by dates, salt by salt, like by like, payment being made hand to hand. he who made an addition to it, or asked of an addition, in fact dealt in usury. the receiver and the giver are equally guilty.” (muslim, no date, v. iii: 834). this particular injunction said to have made by the prophet to stop the cheating of the muslims by the defeated hypocrites of medina who were trying to get back the valuables they had to surrender as part of the khyber treaty. muawiah, the fifth caliph, a very learned man of his time, questioned the validity of this hadith and also calling such transactions riba (suhail1936). it is important to note here that when one borrows one commodity to be returned later, it is no longer “hand to hand”—there is an elapsed time and, as such, it comes under the purview of riba al-nasia, i.e., deferred payment. if we were to accept the “six commodities hadith” then all transactions that did not involve the use of money (i.e., barter trade) would be usurious. this is unacceptable from the view point of rational thinking as well as religious and moral principles. the muslim jurists, depending on how transactions are worded, consider nasia transactions as installment payments at a higher price than the original (salam sale) or pre-determined mark up on the original price to be paid the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 5 over time (murabaha transaction), and, therefore, permitted. others contend that all nasia transactions to be usurious. ironically, ibfi not only consider these transactions to be permissible, but make most of their profit from such practice! about 70 percent of assets in ibfi are murabaha assets, while in the case of conventional banks, two-thirds of assets are murabaha (loan) contracts. two concepts, illa (efficient cause) and hikma (rationale) need to be considered. the fukaha use illa to describe a transaction to be dealing with riba. for example the quranic verse lakum ru’usu amwalikum (i.e., “you shall be entitled to the return of your principal.” 2: 279) is used for illa, ignoring the remainder of the verse, la tazlimuna wa-la tuzlamun (i.e., “you will do no wrong, and neither will you be wronged.”), which provides the hikma. if the first part of the verse was taken literally, then the ibfi policy of profit/loss sharing cannot be accepted. the following excerpt from afzal (1996) is quite instructive: “in many case the ‘illa’ approach appears superficial and devoid of moral considerations. for example, coins like ‘fals’ did not involve riba in maliki, and shafi’i schools. thus 100 ‘fals’ (which were used as a unit of currency but were not made of gold or silver) could be exchanged for 200 either on the spot or on a deferred delivery basis. (why today’s fiat money could also be not counted in this category, is a good question to ask). commodities which were countable like apples and eggs did not involve riba, and hence could be exchanged less for more according to hanafis. a piece of cloth could be exchanged for two pieces of the same quality and measure since it was neither measurable, or weighable, or gold or silver, or foodstuff.” let us now go back to the issue of riba and interest. the contention of this author is that the two most important sources of the shariah (i.e., the quran and the hadith) do not provide us with clear guidance as to whether or not interest is riba (i.e., usury). some of the fuqaha or muslim jurists are of the opinion that interest is riba and, as such, it is prohibited. this view can be acceptable if it is solely based on the argument that the word riba has only one meaning, and that is “an addition over and above the principal sum lent.” all other arguments against interest (i.e., rates of interest which are only fractions of the principal sum) such as provided by muslim writers or scholars are, at best, superfluous and self-serving. if riba means any addition over money lent out, then no one can loan out money to someone else or an organization to share in profits or losses. in other words, one cannot be an inactive participant in business or trade ventures and share profit because he/she cannot get back money which is in “addition to the sum lent.” just because one is ready to accept losses, does not entitle one to claim a part of the profit because he/she is a partner in capital. if riba really meant “an addition to sum lent,” then how did the traders during the life of the prophet share their profits and losses with their aged and/or female partners who could not physically participate in the long journeys that such trades required? mudaraba or joint ownership does not necessarily mean that every single shareholder actively participates in the decision making process as well as the 6 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 implementation of the decisions. if islam permits mudaraba then it is difficult to understand how riba can mean, “an addition to the sum lent.” once riba is defined as “an addition to the sum lent,” all arguments for joint enterprises, cooperative ventures (as not all members of the coops can participate in the decision making process or in implementation of those decisions), interest-free banking, and/or islamic monetary policies become redundant and all these institutions become forbidden for the muslims. the so-called interest-free banks that operate in some of the muslim countries are really indirect participants in riba, in its literal sense. the practice of payments of commissions to the depositors or charging of variable or fixed fees from the borrowers, or lending less than the contracted amount while recovering the full amount, cannot be anything less than riba—i.e., “an addition to the sum lent”. on the one hand, the ibfi claim interest is riba and, as such, forbidden, while on the other, they are dealing in interest, but calling it profit/loss sharing. the fukaha, it seems, have developed quite narrow interpretations of the term riba to suit their own line of thinking. some just substitute “interest” for the term riba. among them are siddiqui 2006, 1987 and 1986; chapra 1985; naqvi 1993; faridi 1991; wohles-scharf 1983; and anwar 1987, who argue that ibfi cannot deal with interests. whereas, sanhuri 1954-1959; al-saud 1985; tantawi 1989; salus 1991; suhail 1936; yousuf ali 1946 (in his commentaries on quran 2:275 and 2:324); shah 1967; rahman 1980; and afzal 1996, make a clear distinction between riba and interest. 4. the concept of usury in judaism and christianity it appears that judaism allowed usury to be extracted from non-jews only. nelson (1969) quotes deuteronomy xxiii: 20, thus: “from him, it says there, demand usury, whom you rightly desire to harm, against whom weapons are lawfully carried. upon him usury is legally imposed. on him whom you cannot easily conquer in war, you can quickly take vengeance with the hundredth. from him exact usury whom it would not be a crime to kill. he fights without a weapon who demands usury: he who revenges himself upon an enemy, who is an interest collector from his foe, fights without a sword. therefore, where there is the right of war, there also is the right of usury.” (nelson 1969, p. 4). nelson continues to quote st. ambrose’s interpretation of deuteronomy, thus, “the law forbids you under any circumstances to exact usury from your brother,” who “is your sharer in nature, co –heir in grace, every people, which, first, is in faith, then under the roman law.” (ibid., 4). with the beginning of the crusade in the eleventh century, the teaching of ambrose gave the christians an excuse to demand interest from muslims, even though it was against the teaching of jesus not to extract usury from anyone (endnote 2). the jews in europe got the cue from this to charge usury from their christian debtors (ibid., 6). however, christianity in medieval and renaissance europe, in general, condemned usury, and started to make a distinction between the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 7 usury and interest. homer (1977) provides an excellent expose on the history of interest rates. he states that, “it was from exceptions to the canon law against usury that the medieval theory of interest slowly developed. compensation for loans was not licit if it was a gain to the lender, but became licit if the compensation was not a net gain, but rather a reimbursement for loss or expense… it was often a compensation or penalty for delayed repayment of loan.”(homer 1977, p. 73). this is similar to the riba transactions at the pre-islamic arabia, ribaal-jahalia. the church doctrine on usury began to be modified around the sixteenth century when use of credit to finance wars and trade became prevalent. theological controversies continued while the rise of protestant ethics and capitalism made the use of interest in financial transactions quite common and acceptable to the public first in germany and then after the pope’s edit in 1751, elsewhere thus laying the foundation of early banking. however, theologians continued to urge for regulation of interest rates—a rate of 5 to 6 percent interest on loans was not to be considered usurious (ibid., 80). it was not until 1950 that pope pius xii declared that the banking system dealing with interest was permissible for the catholics, because the bankers earned their livelihood honestly (ibid., 81). 5. riba, usury and interest: a recapitulation from the short brief above on the position of judaism and christianity on usury and interest, it appears that islam’s position is more similar to that of the latter. over the centuries both judaism and christianity changed their position, and came to accept that interest is not usury, provided it is regulated so that no party is unduly exploited. the texts by nelson and homer, referred above, are two good sources for detailed information about the evolution of the jewish and christian position on usury and interest. islam encourages muslims to earn a living whether by working in agriculture, industry or trade, etc. as some of these professions require more capital than any one individual can muster, it allowed cooperatives and joint ventures among its followers. when a group of people directly or indirectly engages in joint ventures, it is not always possible for everyone to equally participate in all aspects of such ventures. yet islam allowed its followers to share in the profits and losses of these enterprises. if riba meant “an addition to the sum lent,” islam could not permit such institutions. it is logical, therefore, to argue that the muslim scholars should distinguish between fair loan (qard-ihasna) and usurious loan (qard-i-saiya) and rearrange their entire thinking on capital, interest, banking and monetary policies. it has become a fashion to condemn publicly all western monetary institutions as anti-islamic. controlling interest rates through various interest rate ceilings and monetary policies, some of these countries run much superior, and socially equitable, financial institutions than some of the so-called islamic states. the western societies are not all unmindful of the consequences of usury, and some of them have laws prohibiting usury. muslims may have much to learn 8 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 from their experience. it would be proper to end this discourse on riba and interest with the following hadith: “narrated aisha, (may god be pleased with her) the prophet (pbuh) said, ‘why do some people impose conditions which are not present in the allah’s laws? whoever imposes such a condition as is not in allah’s laws, then that condition is invalid even if he imposes one hundred conditions, for allah’s conditions are more binding and reliable.”(sahih al-bukhari, volume iii, chapter 69, number 364). 6. ibfi and their modus operandi the financial instruments used by the ibfi are numerous. according to raphaeli (2006), the most frequently used ones are: “wadi’ah (safekeeping), mudharabah (profit sharing), murabahah (cost plus), ijarah (leasing), qardh al-hassan (benevolent loan), and musharakah (joint venture)”. wadi’ah is a non-interest bearing demand deposit (checking account) similar to the ones offered by commercial banks around the globe (although since 1980s, a small interest is paid on this account). it is also a non-interest bearing savings deposit account, but with a promise of giving hiba (gift) whenever the bank has “profits” to share! in practice though, ibfi claim to offer “hiba” which are a tad higher than the interests on savings account in conventional banks, to attract deposits. in practice, ordinary depositors get nothing and accept that as will of god. mudharabah (profit sharing) account is a form of time deposit that cannot be withdrawn before the contractual time is elapsed. the bank as the intermediary (mudharib) takes no risk when it lends money to the borrowers. the borrowers sign contracts that specify the principal amount borrowed and how much would be paid back, all predetermined. since usually the “cost” of the loan is collected when the loan is given, i.e., the amount of the loan minus the “cost” is advanced, usually such loans are more expensive to the borrower than the ordinary commercial loans offered by conventional banks! if the borrower defaults on the loan, the loss is borne by the depositors to the bank. if the loan is repaid early, the borrower does not receive any part of the “cost” that the bank collected at the origination of the loan! murabaha (cost plus) transactions, a very profitable business for ibfi is no different in character than that of a finance company. the ibfi buys the merchandise, adds an amount for “profit” and sells it to the buyer, to be paid back in equal installments. even if the borrower would attempt to pay back the loan ahead of time, the payments are not reduced. for all practical purposes, murabaha loans have implicit rate of fixed interest rates that are higher than on installment loans from a conventional financial institutions. ijarah (leasing) or ijarah waqtina (lease and purchase), is another way of enriching the ibfi at the cost of the borrower. the lessee cannot refinance or change the contract in view of changes in the market conditions. any break in the contract goes in favor of the ibfi. thus, the lessee can lose the equity in the the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 9 property after paying payments for a number of years, if for some reason s/he can no longer honor the contract! qardh al-hassan (goodwill loan) is given in good faith—the borrower is obliged to pay back the amount lent only, without any obligation to pay anything extra. it is quite obvious that not many loans would be given by a financial institution that is open for business of making money. musharakah (joint venture) in which an ibfi and clients jointly finance a venture by issuing ‘participation certificates’ that can be bought and sold in the market. this type of joint venture is commonly known as sukuk, and the ‘participation certificates’ are traded like bond. apparently, the market for sukuk mushroomed to around $41 billion (raphaeli 2006) fueled by middle eastern oil money that is being invested in building booms and other infrastructure. it is not clear how returns on such instruments are determined but, the returns are predetermined nonetheless (siddiqi 2006). if the past is any indication, it would be based on returns from similar instruments issued by “non-islamic” entities and would simply be labeled as profit sharing. it would be interesting to watch what happens when an ibfi fails to meet its obligations on sukuk to the investing public. the ibfi, at this period of bounties from oil exporting countries, seem to be doing financially alright. hardly a day goes by when islamic banking does not make into the pages of financial papers, e.g., the financial times, the wall street journal, and the economist (see for example, prystav 2006, diwany 2006, tett, and johnson 2006, tett 2006, the economist 2006, davies 2006, lane 2006, saigol and tett 2006, bokhari 2006, and others shown in the reference page). 7. assessing the operations of the ibfi the brief review of the financial instruments above lends to an obvious question, in what ways do the so-called islamic banks and financial institutions qualify to be called islamic? apart from the claims that the ibfi follow islamic “principles”, there is no difference in their operations than the conventional financial institutions, cfi, except for the fact that the ibfi’s principal thrust is towards maximizing their own coffers at the cost of depositors and the borrowers. it can be demonstrated that the ibfi use religious slogans to misrepresent the common people that these are functioning to promote welfare of the believers. in reality, most of their loans cost lot more than what the conventional banks charge. while the cfi are subject to regulations and are accountable for their actions, the ibfi are probably more prone to corruption and mismanagement, all they need is a review from their shari’ah council (in cases where the nbfi are outside the supervision of central banks), that the transactions do not involve dealing with interest rates. the ibfi frequently place their customers’ deposits into short term instruments of conventional banks abroad to earn interests so that they can present their own depositors shares of their interest earnings as “profits” and/or, advance hiba (gift) to them! 10 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 in reality the ibfi are deposit-taking finance companies if outside the central bank purview. here in lies their attractiveness to the major financial institutions of the developed west. it is not that the citibank or hsbc suddenly developed fervor for “islamic” banks and financial institutions. where else they can find easy way of making money without any risk or responsibility to their clients, and without being subject to regulations. after all, the ibfi have been using their facilities to earn short term income (read, ‘interest’) and delve into futures, options, and other speculative transactions! the ibfi, as institutions, seem to attract a fraction of the business of financial transactions in countries where cfi are allowed to operate. the total assets of all ibfi are miniscule (see the 2005-2006 annual report of the islamic development bank) compared to the cfi in the muslim countries, and compared to some of the regional banks’ assets in the us. it will not take long for the muslims to figure out that the sukuk transactions are nothing more than institutional bonds, but with mismatched risk-return relationship. it appears that no matter how the transactions are structured by the ibfi, as long as these are made into “contract” documents, these are termed as islamic (siddiqi 2006). 8. streamlining the ibfi the ibfi should be reorganized into three distinct operational units. one unit should deal strictly with commercial and consumer banking, and the second one with investment banking, and the third one with commercial financing. the sources and the uses of funds of these institutions should be clearly delineated to avoid misuse of funds and public trust. commercial banking: the sources of the fund are depositors’ money and the equity financed by the bank shareholders, the share of the latter as directed by national and international standards for commercial banks. the depositors should earn flexible rates of return based on the overall costs of and returns for the bank during a quarter, or such time by which the bank can estimate its net returns. all deposits ordinary savings accounts, or time deposits should earn flexible rates, subject only to some adjustments for risks. similarly, certificates of deposits, cds, should also earn flexible returns, unlike in modern commercial banks. in the past some modern banks have run into heavy disintermediation during rapid changes in the interest rates, and in some cases, went bankrupt when long term fixed loans at low rates could no longer be financed through costly deposits. the islamic principles of justice and fair play require that no party should take advantage of the other. only flexible rates of interest/return can assure that. insurance is an integral part of an islamic economic system. depositors’ money, up to a certain limit, must be insured by some national system of insurance in a manner similar to the federal deposit insurance company of the us, or elsewhere. the depositors need to have some assurance that their funds would not be completely lost in case the bank faces serious economic problems. this is also compatible with islamic injunction of safeguarding the principal. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 11 the uses of funds should be limited to commercial and consumer loans, e.g., accounts receivable financing, inventory financing, line of credit, installment loans for short and medium terms, and mortgage financing. all loans should also be on flexible cost basis, i.e., the rates charged for these loans must be adjusted periodically as suggested above, based on returns to banks as calculated say every three months. the fdi inflows from foreign commercial banks would increase with such a system since they will no longer be tied with fixed rate commitments. the business community and the consumers in the muslim countries will prefer the new system to what exist in their communities now. investment banking: the principal source of funds for this institution should come from the shareholders’ equity. this can be supplemented by borrowing through the money market. such a bank should not be allowed to use depositors’ money. its earnings will be fee based for intermediation services, and flexible returns from its investments. the activities of the investment bank should also be clearly monitored to avoid questionable practices and transactions. the role the investment bankers played in the enron disaster in the us needs to be averted. commercial financing: at present the ibfi are essentially finance companies. their practices are contrary to what they are supposedly trying to avoid. the murabaha transactions are fixed interest-bearing instrument in reality. no matter how quickly one pays back the loan, one has to pay the full extent of the so-called ‘mark up’, making them usurious (two-fold three fold increases in repayments). this and the instrument of ijarah muntatahia bittamleek should all be modified and standardized, removing the need for shari,ah councils, and be based on flexible rates only. while the recent trend in the world of banking and finance is mergers and consolidation of institutions and functions, the idea of proposing a break up of ibfi into three separate functional areas may appear to be far fetched. yet, this possibly is a very sound way to learn the ropes of running a flexible rate based banking and financial system. from the practice of determining, more or less arbitrarily, the ‘profit’ or ‘mark up’ rate, the new institutions will have to institute a system of accounting to monitor very closely costs of and returns from operations so that they can set competitive rates for their customers. this will be a very difficult task, but must be undertaken in order to ensure that all parties are treated equitably. it appears that the ibfi have hitherto faced very little supervision and scrutiny from the banking authorities of the countries where they operate. it was this lack of centralized supervision that led to the massive failure of the first nbfi in egypt just few years after its founding in 1963. this needs to be remedied. not only these should be brought under the supervision of the monetary authorities, especially the central banks, but also be subjected to regular examinations by the regulatory authorities, results of which should be made available to the public. the ibfi disclose very little information to the public, and this keeps the depositors, the borrowers, and the investors in the dark about their financial health. this calls for improved governance standards 12 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 and compliance monitoring. the restructured ibfi must fulfill their fiduciary responsibilities wholeheartedly. at present the ibfi lacks uniform standards and policies within and across national boundaries. there are attempts to bring in common accounting and financial practices by the bharain-based accounting institute and malaysia-based ifsb. these need to be aggressively instituted, subject to the overall banking and finance policies of individual states. there will not be any meaningful role for the ever-present shari,ah councils in the reorganized ibfi, and these should be disbanded in due course of time with improved supervision structures. banking and financial operations should be left on the shoulders of those who are well trained and well versed in the management of financial services and instruments. 9. conclusion it is important to realize that banking and financial institutions are engaged in a trade (tijara) that deals with the management of money and financial securities and services, and they do not operate like “money lenders” or “loan sharks.” the depositors place their deposits either for a short or a longer period, seeking some returns to offset the loss of purchasing power if kept ‘under the mattress’, or to augment their future income, while the borrowers borrow to meet the needs for funds for business and/or consumption that are vital to them and are willing to pay a reasonable fee for the service. it should be borne in mind that banking/ financial institutions are only intermediaries and should act accordingly, and that the bank officials must be held accountable for their actions and should be held responsible for willful negligence. author statement: this is an invited contribution to the sixth volume of the journal. it is contributed by a notable scholar, raquibuz zaman. dr zaman is the charles a. dana professor of finance & international business and chair of the department of finance and international business, ithaca college, new york. his scholarly publications may be accessed in www.ithaca.edu/faculty/ zaman. e-mail: zaman@ithaca.edu. this paper was reviewed anonymously, and accepted after that as representing a point of view consistent with the ongoing historical debate on usury and interest. end notes 1. this author checked sahih al-burkhari, sahi muslim, mishkat al-masabih, and a manual of hadith (see references at the end of the paper) for various hadith related to usury, debts, and general business transactions. 2. apparently jesus objected to deuteronomy’s position about extracting usury from enemies. in luke vi: 35: the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 13 “but love ye your enemies, and do good, and lend, hoping for nothing again; and your reward shall be great, and ye shall be the children of the highest: for he is kind unto the unthankful and to the evil.” nelson 1964, p.8). references afzal, o. (1996) “riba: usury or interest or both”, a conference paper for the islamic chamber of commerce and industry (icci), san jose, california, november, 7-9. afzal-ur-rahman, (1976). economic doctrines of islam, vol. iii, lahore, pakistan: islamic publications, ltd. ahmad, khurshid (editor), (1980). studies in islamic economics, , jeddah, saudi arabia: international centre for research in islamic economics, king abdulaziz university, and the islamic foundation: united kingdom ali, maulana m., a manual of hadith, (no date of publication). lahore, pakistan: the ahmadiyya anjuman ishaat islam. al-jassas, a.r. (no date), ahkam al-quran, v. 1, istanbul, turkey: 1916. al-saud, a.m. (1985), “bain al-faida wa al-riba,” al-shuruq al-islami, april,1820. al-tabari, a.j.m. (no date), jami’ al-bayan ‘an ta’wil ay al-quran, english translation of the abridged version by cooper, j., new york: oxford university press 1987. al-zamakhshari, m.i.u (no date), al-kashshaf ‘an haqa’iq al-tanzil wa-‘uyun al-aqawil fi wujuh al-ta’wil. ali, a.y. (1946), the holy qur’an: text, translation and commentary, washington, d.c.: the american international printing company (no printing date). anwar, m. (1987), modelling interest-free economy: a study in macroeconomics and development, herndon, virginia: the international institute of islamic thought. bokhari, f. (2006), “oil wealth paves way for islamic mba,” the financial times, monday, december 18: 10. bokhari, f., and oakley, d. (2006), “islamic finance gets ready to spread,” the financial times, november 24: 29. bokhari, f. and felsted, a. (2007), “insurance: takaful cover worth billions set to take off,” the financial times, january 18: 17. chapra, m.u. (1985), towards a just monetary system, london: the islamic foundation. croft, j. (2007), “retail products: world’s banks respond to growing muslim demand,” the financial times, january 18: 17. diwany, t.e., (2006), “how the banks are subverting islam’s ban on usury,” the financial times, july 14: 11. economist, the, (2006), “islamic finance: calling the faithful,” the economist, december 9th, 77. 14 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 faridi, f.r. (1991), essays in islamic economic analysis, edited volume, new delhi: genuine publications (p) ltd. homer, s. (1977) a history of interest rates (second edition), new brunswick, new jersey: rutgers university press. ibn al-‘arabi (no date), ahkam al-quran, beirut: dar-al-ikzat al-‘arabia, 1968, v. 1:24. islamic development bank (2006), annual report 2005-2006, jeddah, saudi arabia, november 1. johnson, s. (2006), “how to hedge and abide by sharia,” the financial times, august 16, 14. khalaf, r., tett, g., and oakley, d. (2007), “investment routes: eastern promise turns to western delight,” the financial times, january 18: 17. lane, k. (2006), “islamicbond market becomes global by attracting nonmuslim borrowers,” the wall street journal, november 16, c1 and c6. mackintosh, j. (2007), “hedge funds: sharia-compliant funds aim to test the islamic appetite,” the financial times, january 18: 17. muslim, i. (1990) sahih muslim (volume iii), translated into english by siddiqi, a.h., lahore, pakistan: sh. muhammad ashraf. naqvi, s.r. (1993), history of banking and islamic laws, karachi, pakistan: hayat academy. nelson, b. (1969), the idea of usury from tribal brotherhood to universal otherhood, (second edition, enlarged), chicago: the university of chicago press. oakley, d. (2007), “pioneer islamic-compliant tracker launched,” the financial times,” january 9: 31. oakley, d. (2007), “sukuk: shock fades as deals break records,” the financial times, january 18: 17. prystay, c. (2006), “mqalaysia seeks role as global player after nurturing islamic bond market,” the wall street journal, august 9, c1 and c4. raphaeli, n.(2006), “islamic banking—a fast-growing industry,” inquiry and analysis series—no. 297, the middle east media research institute, september 29. rahman, f. (1980), major themes of the quran, chicago: bibliotheca islamica. sahih al-bukhari, vol. iii, (1967). translated by dr. muhammad muhsin khan, , ankara, turkey: hilal yayinlari. sahih muslim, vol, iii, (1976). translated by abdul hamid siddiqi, lahore, pakistan: sh. muhammad ashraf. saigol, l. and tett, g. (2006), “investors tap into wave of islamic bonds,” the financial times, september 18: 17. saleem, m. (2007), “islamic finance has much to learn from the west,” the financial times, january 19, 13. salus, a.a. (1991), al-rad ‘ala kitab mufti misr, cairo: dar al-manar alhadithah. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 1-15 15 sanhuri, a. (1954-1959), masadir al-haq fi al-fiqh al-islami, 6 parts in 2 volumes (no other information available). siddiqui, m. n. (1976). banking without interest (2nd edition), lahore, pakistan: islamic publication, ltd siddiqui, m. n. (1978). some aspects of the islamic economy (2nd edition),lahore, pakistan: islamic publications, ltd. siddiqui, m. n. (1972). the economic enterprise in islam, lahore, pakistan: islamic publications, ltd. siddiqui, m.n. (1986), model of an islamic bank, chicago: kazi publications. siddiqi, m. n. (2006). “islamic banking and finance in theory and practice: a survey of state of the art,” islamic economic studies, 13 (2): 1-48. suhail, i. (1936), haqiqat al-riba (no other information available). tantawi, m.s., et. al. (1989), arbah al-bunuk baina al-halal wa al-haram, cairo: dar al-ma’arif. tett, g. (2006), “secondary trading in islamic bonds promisse earthly riches,” the financial times, july 14: 20. tett, g. (2006), “shariah-cmpliant finance: banks create muslim ‘windows’ as islamic banking expands its niche,” the financial times, june 2, 6. tett, g. (2006), “london gains greater role in expanding sharia market,” the financial times, june 2: 6. timewell, s. (1998) “a market in the making,” the banker, february 1998, london wohlers-scharf, t. (1983), arab and islamic banks: new business partners for developing countries, paris: oecd development centre studies. zaman, m.r., and movassaghi, h.(2002), “interest-free islamic banking: ideals and reality,” in ghosh, d. and ariff, m. edited regional financial markets: issues and strategies, westport, connecticut: praeger publishers, chapter 14. zaman, m.r., and movassaghi, h., (2001), “islamic banking: a performance analysis,” journal of global business, vol. 12 (no. 22): 31-38. international journal of banking and finance 9-1-2008 usury (riba) and the place of bank interest in islamic banking and finance m. raquibuz zaman recommended citation stochastic frontier analysis of indonesian firm efficiency: a note 74 international journal of banking and finance, volume 8 (number 2) 2011: pages 74-91 stochastic frontier analysis of indonesian firm efficiency: a note t. handono e. prabowo and emilyn cabanda sanata dharma university, indonesia and regent university, united states ____________________________________________________________ abstract this research attempts to model performance measurement for the firms listed on indonesia stock exchange (idx) using the stochastic frontier approach. there are 121 firms analyzed over the period of 2000-05 with 726 pooled observations. we also test whether firm’s age, size, market share, manufacturing classifications and time period have effects on the technical inefficiency of the manufacturing sector. our findings reveal that the average technical efficiency of the tested firms is 0.7149, which is below the efficiency frontier: factors that affect inefficiency are found and explained. our research has offered notable original contributions to performance measurement and provides insights on managerial decision making on operational performance of listed firms in an increasingly competitive indonesian economy. keywords: stochastic frontier analysis, performance measurement, efficiency, manufacturing sector, indonesia stock exchange jel classification: g21 _____________________________________________ 1. introduction prior research on the indonesian economy used stochastic frontier analysis (sfa) for evaluating a firm’s performance but on other than manufacturing sectors: agriculture (see daryanto, battese, and fleming (2002)) on technical efficiencies of rice farmers in west java; public and private sectors (viverita and ariff (2004); viverita and ariff (2006)); commercial banks (abidin and cabanda (2007); electronics manufacturing plants (palangkaraya and yong (2006); manufacturing sector’s labor growth (jacob and los (2006)); and consumer industry (probowo and cabanda (2010). however, those few studies on indonesian manufacturing firms are not prabowo and cabanda: indonesian firm efficiency listed on the stock market, except the study of probowo and cabanda (2010). this present research is an extension of the study of probowo and cabanda (2010) but covers all three manufacturing classifications listed on idx. this research attempts to fill the gap in existing performance literature on the behavior of listed manufacturing firms in a highly volatile emerging stock market namely indonesia. this paper can also serve as an added contribution to the literature on performance measurement by introducing a frontier model as an alternative measure to a widely-used conventional accounting model to measure firm’s performance. in addition, this research also provides significant empirical contributions to the performance literature in general, and offers specific managerial implications that can be helpful in the decision making of these firms. this research is also prompted by the competitive environment putting pressure on the manufacturing industry in much more open emerging economy: indonesia’s is a case in point as this economy was restructured by the imf and world bank in 1998-2001, and is responding to competition. some firms sought to acquire others to consolidate resources and, through merger, some firms rose to the status of global corporations. to survive in such increasingly competitive environments, manufacturing firms seek to continuously improve their efficiency and productivity performance to sustain long-term growth and profitability. the sector studied is one of the most important sectors listed on indonesia stock exchange (idx). in this new reformed era of high competition, it is important to determine the operational performance of this sector as one of the paramount factors that contributes to the growth of indonesia’s economy. scholars and practitioners alike have been looking for the right measurement tools to evaluate the overall performance of any industries. several studies had been conducted using the sfa approach on performance measurement for manufacturing sectors in other countries. wei, tahman and tan (2004) examined an alternative measure: the rate of technical efficiency change in singapore manufacturing sector. rodriguez and mini (2000) in their study on the manufacturing sector of the philippines found that efficiency and size of firms are positively correlated, and larger establishments are more efficient. lundvall and battese (2000) examined efficiency of kenyan manufacturing firms. kathuria (2001) conducted an efficiency analysis of indian manufacturing firms. kim and han international journal of banking and finance, vol. 8, iss. 2 [2011], art. 5 (2001) applied a stochastic frontier approach to korean manufacturing industries and showed that technical efficiency had a significant positive effect on its productivity growth. in another study, söederbom and teal (2001) examined three dimensions of the performance of firms in ghana’s manufacturing sector. the findings of these previous studies will be later compared to the new empirical findings derived in this research. our research attempts to model performance measurement for the firms listed on the idx. this research has three specific objectives: (1) determine the stochastic frontier measures on labor, inventory, fixed assets, and capital on total sales; (2) test whether firm’s age, size, market share, manufacturing classifications, and time period have effects to the technical inefficiency of the sector; and (3) test whether there is a significant difference among technical efficiency (te) scores of classifications. new findings will offer significant and new empirical contributions to the performance management field. the rest of the paper is structured as follows. section 1 discusses the state of the indonesia’s manufacturing sector. the economic and regulatory environment is described in section 2. data, variables and the model are presented in section 3 as part of the methodology. section 4 presents new findings and our discussion while the conclusion and managerial implications research is in the last section. 2. overview of sector studied from the late 1970s, indonesia experienced a rapid economic growth which was sustained over the next three decades. the economy was transformed from highly dependent on agriculture in 1960s into one in which this sector’s contribution was more than a quarter of the gross domestic product (gdp) in the mid-1990s. from 1973 to 1980, the value of indonesian export was dominated by oil/gas and timber (60 per cent). later on, as more and more processing plants developed domestically, the share of semi-processed goods in total exports rose steadily and the in the mid 1990s became one of the most important foreign exchange earners. the 1997 financial crisis turned the economic miracle into shambles. by january 1998 the currency had depreciated by 80 per cent, while the economy contracted sharply to 51 per cent of gdp at its trend growth. with the loss of valuable times, as the confidence of public and investor continued to evaporate, prabowo and cabanda: indonesian firm efficiency the crisis that was relatively mild in october 1998 continued to deepen when financial crisis led to political one. the currency continued to slide and the crisis had serious consequences; output contracted by 51 per cent of gdp, and us $ 238.60 billion estimated cost of the crisis (widianto et al., 2000). the severe economic contraction in 1998 was slightly reversed in 1999, when the economy grew again, though at a miniscule rate of 0.8 per cent. rupiah was stabilized around rp 9,000 per us dollar since november 2002 – a far cry from its 3,000 rupiah to dollar during pre-crisis times. the appreciation of rupiah from around 15,000 rupiah along with the availability of food supply has held inflation in check. measured by consumer price index (cpi), inflation reached its peak in 1998 at 82 per cent per annum. the inflation rate in 2001 was 11.2 per cent, 10.0 per cent (2002), 5.1 per cent (2003), 6.4 per cent (2004), 17.1 per cent (2005), and less than 10 per cent (2011). the inflation rate was higher in 2005 due to government decision to increase gas and oil prices by 100 per cent in 2005 (bps statistic indonesia, 2006). from 2000 through 2003, economic growth was mainly driven by private and public consumption, while fixed investment, just like in the preceding years after the crisis, remained sluggish. as a result of sluggish investment growth, the investment to gdp ratio in 2003 dropped to 17.8 per cent in 2003, the lowest level since the early 1970s. during the late soeharto era, the investment to gdp ratio was around 30 per cent. however, in 2004 for the first time after the asian crisis, gdp growth just exceeded 5 per cent. this time growth was not only driven by consumption, but also by investment, the growth of which for the first time after the crisis grew at double digit at 15.7 per cent. export growth at 8.5 per cent was also higher than in 2002 and 2003. during the first and second quarters of 2005 fixed investment continued its double-digit growth (wie, 2006). the manufacturing sector accounts for an increasing share of gdp. the manufacturing sector accounted for an estimated 27.6 per cent of gdp in 2001, 27.8 percent (2002), 28.0 per cent (2003), 28.36 per cent (2004), and 28.1 per cent (2005) of gdp: it is close to a third in 2011. the growth rates were 3.8 per cent (2001), 5.3 per cent (2002), 5.3 per cent (2003), 6.4 per cent (2004), and 4.6 per cent (2005). the sector contributes the highest contribution to indonesian gdp growth from the year 2001 to 2005 (bps-statistic indonesia, 2006). with this, international journal of banking and finance, vol. 8, iss. 2 [2011], art. 5 the financial sector has responded well with its own rapid growth and rehabilitation to a healthy state. in the late 1980s and the 1990s, indonesia implemented policies designed to move toward a freer, more market-oriented financial system. indonesia deregulated its financial sector in 1988-1989. there were 56 listed companies before the deregulation of the financial sector in 1988-1989. one year later (1990), there were 123. subsequently, there were 349 listed firms as of december 2005. in the manufacturing sector, there are 127 firms listed on jakarta stock exchange (jsx). the jsx changed its name to indonesia stock exchange (idx) in december 2007. these firms listed are categorized into three classifications: basic industry (48 companies), consumer goods industry (38 companies), and miscellaneous industry (41 companies). 3. data, variables and methodology 2.1 data sample this research covers 121 out of the total 127 manufacturing firms listed on idx from 2000 to 2005: due to data unavailability for recent periods, 2005 financial reports are the latest available. a pooled data of 726 represent the panel data for the current analysis. data were gathered from audited annual financial reports of manufacturing firms from securities and exchange commission (bapepam) and idx. this research include all the three listed manufacturing classifications: basic industry (47 companies), consumer goods industry (36 companies), and miscellaneous industry (38 companies). all financial data were adjusted for inflation, using the consumer price index (cpi) with a base year of 1993 prices. 2.2 variables there are four (4) inputs used: (1) labor, (2) inventory, (3) fixed assets, and (4) capital (see probowo and cabanda, 2010; kathuria, 2001;wei koh, et al., 2004; and mojo, 2007). the one output is total sales (nakajima,1998; chirwa 2001; probowo and cabanda, 2010). other zvariables used are age, size, market share, manufacturing classifications, and time period (see lundvall and battese, 2000; biggs and srivastava,1996, viverita and ariff, 2006, tybout, 2000, diaz and sanchez, 2008; and probowo and cabanda, 2010). prabowo and cabanda: indonesian firm efficiency 2.3 stochastic frontier analysis model we attempt to propose a model for technical inefficiency effects in a stochastic frontier production function for panel data of the listed manufacturing firms to estimate the trans-log stochastic production function over the time period. provided the inefficiency effects are stochastic, the model permits the estimation of both technical change in the stochastic frontier and time-varying technical inefficiencies. table 1: variables and definitions input variables labor salaries and wages are a proxy for labor inventory inventory includes raw materials, work-in-process, auxiliary materials, finished goods, and spare parts. fixed assets fixed assets include plant, property and equipment, land, transportation equipment, office equipment. capital stockholders’ equity as proxy to capital is the amount received from investors in exchange for stock. output variable total sales total sales indicate the total amount of sales received by the firm for the sale of its products. z-variables age age is the length of period a firm has been operating to produce and sell products. size total assets as proxy to size. market share market share is the ratio of sales to total sales of manufacturing sector. manufacturing classifications manufacturing classifications are basic industry, consumer goods industry, and miscellaneous industry. time period time period of 2000 to 2005 source: probowo and cabanda (2010). battese and coelli (1995) provided the stochastic frontier production function for panel data: )exp( itititit uvxy   (1) where yit denotes the production at the t-th observation (t = 1,2, …,t) for the i-th firm (i = 1,2, …,n); it x is a )1( xk vector of values of known functions of inputs of production and other explanatory variables associated with the i-th firm at the t-th observation; β is a )1( xk vector of international journal of banking and finance, vol. 8, iss. 2 [2011], art. 5 unknown parameters to be estimated; it v s are assumed to be iid ),0( 2 v n  random errors, independently distributed of the it u s; it u s are non-negative random variables, associated with technical inefficiency of production, which assumed to be independently distributed, such that it u is obtained by truncation (at zero) of the normal distribution with mean  it z and variance, 2  ; it z is a )1( xm vector of explanatory variables associated with technical inefficiency of production of firms over time; and  is a )1(mx vector of unknown coefficients (battese and coelli (1995). to characterize the stochastic frontier production of the listed manufacturing sector firms, this research applies a trans-log stochastic production function. applying the battese and coelli’s (1995) model, equation (2) presents the empirical log-linear form for this research: )(lnln)ln(lnlnlnlnln 6 2 543210 itititititititit fiilkfiy   )(lnln)ln()(lnln)(lnln 10 2 987 ititititititit kffliki   2 1413 2 1211 )ln()ln(ln)ln()(lnln itititititit llkklf   itit uv  (2) where: yit represent total sales of the manufacturing firm i-th at the t-th year of observation; it i represent inventory of the manufacturing firm i-th at the t-th year of observation; it f represent fixed assets of the manufacturing firm i-th at the t-th year of observation; it k represent capital of the manufacturing firm i-th at the t-th year of observation; it l represent labor of the manufacturing firm i-th at the t-th year of observation; 1  represents the natural log of inventory ( it i ); 2 represents the natural log of fixed assets ( it f ); 3 represents the natural log of capital ( it k ); 4 represents the natural log of labor ( it l ); 5 represents the natural log of inventory ( it i ) 2 ; 6 represents the natural log of inventory ( it i ) x the natural log of fixed assets ( it f ); 7 represents the natural log of inventory ( it i ) x the natural log of capital ( it k ); 8 represents the natural log of inventory ( it i ) x the natural log of labor ( it l ); prabowo and cabanda: indonesian firm efficiency 9 represents the natural log of fixed assets ( it f ) 2 ; 10 represents the natural log of fixed assets ( it f ) x the natural log of capital ( it k ); 11 represents the natural log of fixed assets ( it f ) x the natural log of labor ( it l ); 12 represents the natural log of capital ( it k ) 2 ; 13 represents the natural log of capital ( it k ) x the natural log of labor ( it l ); 14 represents the natural log of labor ( it l ) 2 ; it v s assumed to be iid ),0( 2 v n  random error, independently distributed of the it u ; and it u are non-negative random variable. furthermore, battese and coelli (1995), specified the technical inefficiency effect, it u , in the stochastic frontier model as shown in equation (3): )()()( 3210 itititit emarketsharsizeageu   (3) itit wtimeperiodclass  )()( 54  where ageit represents the number of operation years of the manufacturing firm i-th at the t-th year of observation; sizeit represents the total assets of the manufacturing firm i-th at the t-th year of observation; it emarketshar represents sales of the manufacturing firm i-th at the t-th year of observation divided by total sales of the manufacturing sector; classit represents the classification of the manufacturing firm i-th at the t-th year of observation; time periodit represents the time period of the manufacturing firm i-th at the t-th year of observation (2000 – 2005); and it w is defined by the truncation of the normal distribution with zero mean and variance. the stochastic frontier production function may investigate a firm’s technical efficiency and may also identify factors for the technical inefficiency effects of the manufacturing sector firms. the computer software known as frontier 4.1 by tim coelli was used to derive all empirical findings in this research. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 5 4. empirical findings the value of the generalized likelihood-ratio (lr) statistics for the parameters in the stochastic production function for sales is shown in table 2. the null hypothesis that that the cobbdouglas functional form is a correct functional form to represent the data in indonesia’s listed sector is significantly rejected. therefore, the trans-log model is chosen based on the lr value of 155.59. this is greater than the critical value of 18.30 based on a chi-square distribution table, tested at 5 per cent probability level. the null hypothesis that there is no technical inefficiency effect in the model is also significantly rejected, based on the lr value of 546.26, implying that inefficiency effect is present in the model. table 2: generalized likelihood-ratio tests of null hypotheses for parameters in the stochastic frontier production function for sales null hypotheses, ho lr value critical value* decision 4,3,2,1,0 ij  (cobb-douglas function) 155.59 18.30 reject 0 543210   (no inefficiency effects) 546.26 13.40 reject *critical values are obtained from the appropriate chi-square distribution, except for the test of hypothesis involving 0 for technical inefficiency effects (kodde and palm, 1986). 4.1 panel i findings to determine the stochastic effects of labor, inventory, fixed assets, and capital on total sales, results are shown in table 3. the estimated coefficients of four inputs for the sector are reported in panel i. there are five coefficients out of 14 that are significantly different from zero at the 5 per cent probability level. one direct effect, three squared terms and one cross product have coefficients significantly different from zero. these findings support the rejection of the cobb-douglas model: this is not an adequate representation of the sector. inventory, among the four inputs, remains the single most significant predictor of sales output (efficiency), with an estimated elasticity of 0.7182. prabowo and cabanda: indonesian firm efficiency table 3: the maximum-likelihood estimates of parameters of the translog stochastic frontier production function for sales a significant positive effect (0.7182) on technical efficiency. the positive effect implies that the manufacturing sector firms’ efficiency increases as more inventory utilized. variables parameters coefficient estimates t-ratio i. production frontier constant 0  4.0894 4.464** ln l (labor) 1  -0.2799 -1.256 ln i (inventory) 2  0.7182 3.573** ln f (fixedassets) 3  0.1511 0.920 ln k (capital) 4  -0.0241 -0.123 (ln l) 2 5  0.1118 5.377** ln l x ln i 6  -0.1559 -4.488** ln l x ln f 7  0.0581 1.853 ln l x ln k 8  -0.0448 -1.380 (ln i) 2 9  0.0521 2.862* ln i x ln f 10  -0.0085 -0.302 ln i x ln k 11  0.0067 0.191 (ln f) 2 12  -0.0165 -1.013 ln f x ln k 13  -0.0086 -0.404 (ln k) 2 14  0.0295 2.214* ii. inefficiency effects constant 0  -24.6063 -12.757** age 1  0.1938 8.292** size 2  0.1854e-06 2.241* market share 3  -0.7265 -4.122** classification 4  3.0547 5.737** time 5  0.0893 0.520 iii. variance parameters 222 uvs   7.9012 6.849** 22 / su   0.9809 297.503** log-likelihood ratio 546.260 *** mean technical efficiency 0.7149 * significant at 5 percent level (p< 0.05). ** significant at 1 percent level (p < 0.01). *** critical value is 13.40 for 7 d.f as for table 1 of kodde and palm (coelli and battese, 1998) for technical inefficiency effects. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 5 overall, constant ( 0  ) is statistically significant (4.0894). this finding suggests that the joint effects of four predictors of technical efficiency in this sector are positive and significant, in general, while individual effects of one or more variables are not statistically significant. labor shows a negative effect (-0.2799) but is statistically insignificant. this finding is consistent with the findings of wei koh et al. (2004) and gholami, moshiri, and yong (2004); they found out that technical efficiency decreases as more labor inputs are used. inventory coefficient has the estimated coefficient for fixed assets (0.1511), which is positive, but the effect is insignificant. lastly, capital (-0.0241) is found to have a negative but insignificant effect on efficiency, suggesting that efficiency declines when more capital is injected. this result supports the finding of lundvall and battese (2000) on kenyan industry. this results are indicative of the sector’s lack luster productivity. 4.2 panel ii findings to further test whether firm’s age, size, market share, classifications, and time period have effects on technical inefficiency of the sector, and the findings are shown in table 3, panel ii. overall, the joint effect of five z-variables on the technical inefficiency is significant, where the constant is -24.6063. the estimated coefficient associated with age (0.1938) is positive and statistically significant, suggesting that older firms are technically inefficient than younger firms perhaps due to the latter adopting newer technology. size is also found to have a positive significant effect on technical inefficiency, which is a normal results. this finding is consistent with the results of biggs et al. (1996) that larger firms are technically inefficient than smaller firms. meanwhile, market share is found to have a negative effect on technical inefficiency and is statistically significant. this finding supports tybout (2000) and diaz and sanchez (2008) that firms with higher market shares demonstrate market power and are technically efficient compared to firms with lower market shares. moreover, classifications show a positive effect on technical inefficiency and the coefficient is significant. this finding suggests that basic and consumer classifications are technically inefficient than miscellaneous type. lastly, time has a positive effect: an indication that technical inefficiency is present in production over time. this prabowo and cabanda: indonesian firm efficiency finding is in line with chirwa (2001) on the manufacturing sector in malawi that, on average, technical efficiencies decline over time. 4.3 panel iii findings the variance parameters, 222 uvs   and 22 / su   , are all positive and significant. the estimate for  (gamma) is close to unity (0.981) and very high. this result indicates that much of the variation in the composite error term is due to inefficiency effects (and not simply random errors) in this sector’s data. this finding supports the previous result of hill and kalirajan (1993) on small-scale indonesian garment producers. lastly, the mean technical efficiency is 71.49 per cent for the sector. on average, this sector produces 71.49 per cent of the total sale output that could be theoretically produce with the same combinations of inputs by a fully-efficient firm: of course this is the theoretical limit, which is not possible, given firms in any economy operate with some slack because of cyclical changes in demand for their outputs. this further suggests that sector needs to increase their sale output by 28.51 per cent to attain the optimal efficiency level. 4.4 technical efficiency analysis the 121 firms used in this analysis are classified into three (3) categories: basic industry, consumer industry, and miscellaneous industry. the companies’ technical efficiency data (2000 – 2005) are provided in tables 4 and 5. the average technical efficiency scores of basic industry, consumer industry and miscellaneous industry are 0.703, 0.705, and 0.739, respectively. the overall mean technical efficiency of the manufacturing industries is 0.715. the highest average of technical efficiency was obtained by tbms (0.904) in the basic industry and the lowest average of technical efficiency was 0.375 (pyfa) in the consumer industry. the lowest average of standard deviation in technical efficiency was in miscellaneous industry (0.093). kruskal-wallis test was used to test whether there is a significant difference among technical efficiency scores of manufacturing sector classifications. we found that there is no statistically significant difference (0.178) in technical efficiency scores of the three international journal of banking and finance, vol. 8, iss. 2 [2011], art. 5 classifications (basic, consumer and miscellaneous). table 5 presents the descriptive statistics of the three industry classifications. our sfa model appears to have the same statistical results for efficiency scores among three classifications. therefore, basic, consumer and miscellaneous industry classifications seem to be operating at the same efficiency level. table 4: firm average technical efficiency scores for manufacturing classifications basic consumer miscellaneous firm te firm te firm te intp 0.560 ades 0.533 acap 0.702 smcb 0.573 aqua 0.884 asii 0.831 smgr 0.700 ceka 0.579 auto 0.790 arna 0.801 davo 0.844 bram 0.694 ikai 0.445 fast 0.819 gjtl 0.784 mlia 0.610 indf 0.766 gdyr 0.766 almi 0.782 myor 0.732 admg 0.809 bton 0.706 mlbi 0.644 hexa 0.790 ctbn 0.672 ptsp 0.771 inds 0.686 inai 0.766 psdn 0.685 inta 0.727 jksw 0.580 shda 0.728 lpin 0.685 jprs 0.830 sklt 0.785 nips 0.784 lmsh 0.826 sttp 0.762 pras 0.800 lion 0.563 sipd 0.822 smsm 0.694 pico 0.655 smar 0.779 turi 0.910 tbms 0.904 suba 0.709 untr 0.809 tira 0.645 tbla 0.805 pafi 0.700 akra 0.887 ultj 0.595 hdtx 0.746 budi 0.815 bati 0.644 rdtx 0.611 clpi 0.820 rmba 0.808 mytx 0.789 ltls 0.780 ggrm 0.767 doid 0.732 sobi 0.793 hmsp 0.766 esti 0.611 unic 0.775 dvla 0.692 indr 0.816 akpi 0.756 inaf 0.682 bima 0.487 amfg 0.719 kaef 0.723 ricy 0.686 apli 0.735 klbf 0.744 srsn 0.734 brna 0.760 merk 0.507 bata 0.724 dyna 0.662 pyfa 0.375 kbli 0.743 fpni 0.734 scpi 0.754 jecc 0.771 lmpi 0.610 sqbi 0.729 kblm 0.563 lapd 0.784 tspc 0.745 voks 0.834 sima 0.781 tcid 0.725 komi 0.814 smpl 0.604 mrat 0.610 intd 0.813 trst 0.747 unvr 0.689 mdrn 0.814 brpt 0.529 kici 0.415 koni 0.646 dsuc 0.704 kdsi 0.746 asgr 0.514 prabowo and cabanda: indonesian firm efficiency suli 0.656 mlpl 0.789 sudi 0.601 mtdl 0.882 tirt 0.661 fasw 0.754 inkp 0.627 tkim 0.637 spma 0.702 saip 0.712 dpns 0.505 ekad 0.787 inci 0.796 table 5: descriptive statistics of technical efficiency scores for manufacturing classifications basic consumer miscellaneous firm te firm te firm te mean 0.703 0.705 0.739 std deviation 0.102 0.113 0.093 min 0.445 0.375 0.487 max 0.904 0.884 0.910 number of firms 47 36 38 5. conclusion and managerial implications this research has modeled a performance measurement for an important sector that is drving the economic recovery in this vast country: the selected firms are listed on the indonesia stock exchange. we apply a stochastic frontier analysis. new findings derived from this study have offered notable original contributions to performance measurement and provides insights relevant to the managerial decision making on the operational performance of firms. first, our research provides new findings on the predictors of firms’ technical efficiencies and inefficiencies in the sector, using six years of combined firm-level accounting-financial and market data as well as other firm’s specific variables. the finding indicates that the cobbdouglas functional form was rejected for the indonesia’s manufacturing sector. this finding further suggests that the trans-log functional form is a more general functional form, which is used as would be an appropriate model in representing the data for the sector listed. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 5 second, this research helped to reject the null hypothesis that there is no inefficiency effect in the sector model. findings demonstrate that inefficiency effects are likely to be highly significant and are not simply random errors in the analysis of the value of output. this finding affirms previous studies covering different economies. lastly, our research has provides results relevant for managerial actions. the stochastic frontier model can be an alternative measure to the traditional ratio analysis when it comes to measuring performance of any firm: in banking this measure has been widely used for some 15 years to-date. the results from this model will be useful as a guide on corporate factor efficiency for stockholders (investors), managers, bankers and stakeholders of the indonesia’s business community in the evaluation of the operational performance and the behavior of listed firms as well as in identifying a specific factor that can affect the technical efficiency of firms. for the management of firms, this research also serves as a guide in making the right decisions based on the reported association of inputs and other firm’s specific variables to the firm’s technical efficiency as well as the inefficiency effects. for the investors, analysis and evaluation of a firm’s efficiency would provide better quality appraisal tool in making a business decision to either invest or not in a given sector, and to either buy or not to buy shares to maximizing their returns on investment. lastly, for creditors, the new empirical findings on a firm’s efficiency may provide insights for analyzing and evaluating a loan application to minimizing risks. for bankers, these results provide a clear means of identifying the level of risk from inefficiency of the firms in this sector so that correct credit decisions could be based on objective facts about inefficiency. a future extension of this research could be to analyze the sector as well as the financial firms listed on all asean stock exchanges to evaluate how technical efficiency has changed over time. in redesigning future studies, variables such as market capitalization and other market data may need to be considered. these are the present limitations of our research due to data unavailability at this time. other performance measurement tools such as linear programming techniques can also be utilized in future research for benchmarking performance. prabowo and cabanda: indonesian firm efficiency author information: dr emilyn cabanda is an associate professor in the school of global leadership & entrepreneurship at the regent university, virginia beach, vr, united states: email: ecabanda@regent.edu. t handono eko prabowo is at the faculty of economics of santa dharma university, yogyakarta indonesia: e-mail: thep_phd@yahoo.com. references abidin, z. and cabanda, e., (2007). frontier approaches to production efficiency of commercial banks in indonesia. manajemen usahawan indonesia, xxxvi (6): 9-14. battese, g.e., and coelli, t.j., (1995). a model for technical inefficiency effects in a stochastic frontier production function for panel data. empirical economics, 20: 325-332. biggs, s., and srivastava, h., (1996). technological capabilities and learning in african enterprises. rped case study series, washington, d.c, usa: the world bank. bps (2006). statistic of indonesia. bps, jakarta: biro pusat statistik. chirwa, e. w., (2002). privatization and technical efficiency: evidence from the manufacturing sector in malawi. african development review, 13 (2): 276-307. coelli, t., (1995). recent developments in frontier modelling and efficiency measurement. australian journal of agricultural economics, 39(3): 219-245. coelli,t., rao, d.s. p., o’donnell, c.j., and battese, g. e., (2005). an introduction to efficiency and productivity analysis, (second edition). new york, ny: springer science. daryanto, h., battese, g.e, and fleming, e.m., (2002). technical efficiencies of rice farmers under different irrigation systems and cropping seasons in west java. paper presented at the 2002 asia conference on efficiency and productivity growth, school of economics, university of new england, australia. diaz, m. a., and sanchez, r., (2008). firms’ size and productivity in spain: a stochastic frontier analysis, small business economics, 30 (3): 315-323. gholami, r., moshiri, s., yong t. l., sang (2004). ict and productivity of the manufacturing industries in iran. ejisdc, 19 (4): 1 – 19. hill, h., and kalirajan, k.p., (1993). small enterprise and firm-level technological efficiency in the indonesian garment industry. applied economic, 25: 1137-1144. jacob, j., and los, b., (2006). international technology spillovers and manufacturing performance in indonesia. doctoral thesis, eindhoven university of technology, eindhoven, netherlands. kathuria, vinish., (2001). foreign firms, technology transfer and knowledge spillovers to indian manufacturing firms: a stochastic frontier analysis. applied economics, 33: 625-642. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 5 kim, s., and han. g., (2001). a decomposition of total factor productivity growth in korean manufacturing industries: a stochastic frontier approach. journal of productivity analysis, 16, 269-281. kodde, da., and palm, f.c., (1986).wald criteria for jointly testing equality and inequality restrictions, econometrica 54: 1243-1248. lundvall, k., and battese, g. e., (2000). firm size, age, and efficiency: evidence from kenyan manufacturing firms. journal of development studies, 36 (3):146-163. mojo, m. i., (2007). total factor productivity in indonesian manufacturing: a stochastic frontier approach. global economic review, 36 (4): 321-342. nakajima, t., (1998). an index number method for estimating scale economies and technical progress using time series of cross section data: source of total factor productivity growth for japanese manufacturing 1964 – 1988. the japanese economic review, 49 (3): 310-334. palangkaraya, a., and yong, j., (2006). entry, exit and productivity of indonesian electronics manufacturing plants, accessed december 15, 2008, available at: http://melbourneinstitute.com/wp2006n08.pdf probowo, t.h.e., and cabanda, e., (2010). performance measurement model for the consumer industry listed on indonesia stock exchange: dea and sfa approaches. indonesian capital market review, vol. ii, no. 1: 51-62. rodriguez, e., and mini, f., (2000) technical efficiency indicators in a philippine manufacturing sector. international review of applied economic, 14 (4): 461-473. söederbom, m., and teal, f., (2001). are african manufacturing firms really inefficient? evidence from firm-level panel data. center for the study of african economies, university of oxford, uk, wps 2001, 14, 1-25. tybout, j. r., (2000). manufacturing firms in developing countries: how well do they do, and why? journal of economic literature, 38(3): 1144. viverita, r., and ariff, m., (2004). corporate financial performance and production efficiency: a study on indonesia’s public and private sector. paper presented at the 4 th international symposium of dea, aston university, uk. viverita, r., and ariff, m., (2006). firm size, age, and technical efficiency of indonesian firms, usahawan, 09 (xxxv): 16 – 20. widianto, bambang, choesni, tb, choesni, a., (2000). indonesia: the impact of the economic crisis on industry performance. national development planning agency, bappenas, indonesia. wie, thee kian., (2006). policies for private sector development in indonesia. adb institute discussion paper no. 46. prabowo and cabanda: indonesian firm efficiency http://melbourneinstitute.com/wp2006n08.pdf wei koh, s., tahman, s., and tan, r., (2004). stochastic frontier analysis of singapore manufacturing industries. the singapore economic review, 49 (1): 85-103. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 5 how to cite this article: lee, a.x., & hooy, c.w. (2020). state ownership and risk-taking behaviour: evidence from malaysia’s banking industry. international journal of banking and finance, 15(2), 65-94. https://doi.org/10.32890/ijbf2020.15.2.4 state ownership and risk-taking behaviour: evidence from malaysia’s banking industry 1ai-xin lee 2chee-wooi hooy school of management, universiti sains malaysia, malaysia 1aixinlee@student.usm.my; 2corresponding author: cwhooy@usm.my a r t i c l e i n f o article history: received 30 march 2020 revised 10 june 2020 accepted 19 june 2020 published 31 july 2020 jel code: g21, g28, g32, g34 keywords: state ownership, glics, risk-taking, corporate governance, board of directors, malaysia. a b s t r a c t this study investigates state ownership on risk-taking behaviour in malaysia’s banking industry. using the panel of malaysian commercial banks, this paper examines whether banks’ risk-taking is affected by malaysian government ownership through the five largest investment arms of malaysia (glics). the findings show that state-owned banks exhibit higher risk-taking behaviour compared to the private-owned banks in terms of loans. there is evidence that a higher degree of state ownership has a more significant impact on banks’ risk-taking behaviour. we also investigate the relationship with corporate governance mechanisms. the findings suggest that the composition of board of directors somehow plays a significant role in the governance of banks. the international journal of banking and finance, vol. 15. number 2, 2020: 65-94 65 66 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 1. introduction the banking sector plays a significant role in southeast asia's economic development for the past 40 years. an effective and secure financial system in the banking sector has attributed to an outstanding high record on east asia economic growth in the mid-90s (world bank, 1993). however, the asian banking industry has undergone extensive financial transformation where many banks were forced to consolidate after an overwhelming impact during the previous financial crises. as a result, the number of commercial banks operating in malaysia has reduced to 26. one common characteristic of previous financial crises incident is that banks have taken excessive risks and ended up with too many defaults. governmental bailouts have also been blamed for creating incentives for banks to undertake excessive risks that contributed to these financial crises. global financial development report (2019/2020) claimed that there was extensive government intervention in backing up the national banking systems at that time. there is also no exception in malaysia, where the malaysian government has bailed out well-connected ailing, debt-ridden enterprises through governmentlinked investment companies (glics) and government-linked companies (glcs) (gomez, padmanabhan, kamaruddin, bhalla & fisal, 2018). the act of government intervention to rescue financial institutions has direct and indirect economic costs that have long-lasting effects (global financial development report, 2019/2020). the government intervention is further concerned when its involvement in the banking sector has intensified after the financial crises, particularly asian countries that have greater government ownership involvement compared to other regions (hossain, jain & mitra, 2013). prior studies claimed that the presence of moral hazard behaviour induced by government explicit and implicit protection leads to excessive risk-taking behaviour since it does not have to bear the costs (jensen & meckling, 1976; global financial development report, 2019/2020; zhu & yang, 2016). excessive risk-taking behaviour in the banking industry could mark the onset of a banking crisis. severe consequences on excessive risk-taking can be seen in the incident of lehman brothers bankruptcy during the subprime crisis. they had taken excessive risks in creating loans, hence could not pay the excessive debts and faced unprecedented losses. in light of globalisation in the banking sector, the central bank of malaysia (bank negara malaysia) has taken the initiatives to strengthen malaysia’s economies and boost the role of banking sector as a key element of national economic growth to compete in a more liberalised and challenging environment. chin (2015) claimed that protracted affirmative action policies have led to the protectionism of local banks in malaysia that favoured state-owned banks. the policies enforced by the government after the recent global financial crisis has produced power shift which driven malaysia’s state-owned banks to seize state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 67 more opportunities where foreign banks were impotent and have to withdraw their business in malaysia. this reveals that the state authority is strong in the malaysian banking sector. in view of liberalisation in the banking sector, the risk exposure faced by banks nowadays has become greater with the expansion of the banks’ business scope and economic activities. this is predominantly important for commercial banks in malaysia since they are the largest and most significant providers of funds in the malaysian banking system. about 55.61 percent of the total loans in the banking system are derived from the household sector including mortgages, personal loans and hire purchase loans1. currently, there are 26 commercial banks of which eight of them are local commercial banks. these commercial banks were governed by the banking act 1973 and finance companies act 1969 under a single legislation. as indicated in the earlier paragraph on the malaysian government bailout during previous financial crises, glics, indeed, play a significant part in the nation’s economic development. a key characteristic of malaysia’s economy under glic-led framework is that government has substantial influence in intervening the economy through their ownership in business operations due to their political, social and economic interests (gomez et al., 2018; shleifer & vishny, 1994) and these glics are closely tied with government policies (abdul razak, 2011). following haggarty and shirley (1997) and oecd (2013), glics are defined as government-owned or -controlled investment companies that have a primary commercial objective. the malaysian government has a direct controlling stake on the management decisions by virtue of its share ownership through glcs or glics, that is, the ability to appoint senior management or make major decisions such as contract awards, strategy, restructuring, and financing, acquisitions and divestments. these glics constitute a significant part of the malaysian economic structure as they hold more than 50 percent of the whole market capitalisation and majority ownership of malaysia’s leading public listed firms. according to gomez et al. (2018), glics hold majority ownership of 35 public listed firms from malaysia’s top 100 companies. this study focuses on five major glics that hold majority ownership in local commercial banks. they are known as employees provident fund (epf), permodalan nasional berhad (pnb), khazanah nasional berhad (knb), lembaga tabung angkatan tentera (ltat) and lembaga tabung haji (lth). statistics show that epf owns 45.34 percent of rhb bank berhad, pnb owns 42.96 percent shares in malayan banking berhad, knb owns 29.71 percent shares in cimb bank berhad whereas ltat has contributed to 35.42 percent of ownership in affin bank berhad2. the existing literature on glics or state ownership were mainly focused on firm performance (see abdul razak, ahmad & aliahmed, 2008; abdul rahman & rejab, 2013; hamid, 2011; najid & rahman, 2011; taufil-mohd, md-rus & musallam, 2013). similarly, other studies conducted in the developed countries also shared such findings (see berger, clarke, cull, klapper & udell, 68 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 2005; cornett, guo, khaksari & tehranian, 2010; iannotta, nocera & sironi, 2007). however, the literature on the relationship between state ownership and risk-taking is scarce. risk-taking is one important dimension for corporate success and it has an impact on national economic development, particularly the banking sector where the risk undertaking of a bank not only has an enormous impact on banks’ profitability but the national economy as a whole. in the event of excessive risk-taking, the impact of bank failure can spread to other banks and perhaps hitting the entire nation as well as the global economy (stiglitz, 1993). understanding the peril of excessive risk-taking, only a small number of recent studies examined the impact of state ownership on risk-taking. notably, most of the extant studies were conducted in western countries and developed countries such as argentina, european countries, middle east and north africa (mena) countries and the united states (see brandao-marques, correa & sapriza, 2020; iannotta, nocera & sironi, 2013; lassoued, sassi & ben rejeb attia, 2016; uddin, 2016). the studies have also evolved into asian countries lately (see lee & hooy, 2020; vo, 2018; zhu & yang, 2016). although there have been increasing empirical studies conducted in western countries and developed countries, very little attention has been given to emerging market such as malaysia in examining the impact of state ownership on the banks’ risktaking. it is relatively important to explore this issue in malaysia where glics hold majority ownership in local commercial banks. prior study claimed that banks have a greater propensity to undertake risky projects if they have a direct link with the government (lassoued et al., 2016). this statement is further supported by the empirical findings where state ownership is found positively related to risk-taking (brandao-marques et al., 2020; iannotta et al., 2013; lassoued et al., 2016; zhu & yang, 2016). the existing literature also proves that state-owned banks (sobs) tend to have poorer performance, are less efficient and less profitable and incur greater credit risks compared to private-owned banks (pobs). they claimed that these differences between sobs and pobs are more eminent in countries that had greater government involvement in the banking system, particularly in southeast asia where sobs are found underperforming (cornett et al., 2010; naima, houda & mouna, 2016). on the contrary, some other studies found a negative relationship between state ownership and risk-taking (see vo, 2018) and the nonlinear relationship (see uddin, 2016). the flaw in the functioning of corporate governance (cg) mechanisms has also played part of the root in previous financial crises that indirectly encourage government intervention in the banking sector. gomez et al. (2018) argued that there is no effective body in malaysia that can monitor the governance and performance of state-owned enterprises (soes) although there is a wellstructured governance mechanism in many other countries’ soes. hence, having a sound cg mechanism in the malaysian banking system is crucial in state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 69 monitoring bank activities and investment decisions. liang, xu and jiraporn (2013) and oecd (2006) claimed that the board of directors play an important role in establishing a sound bank governance. the finding from prior studies provides evidence that firms with more effective governance are associated with a lower level of risk-taking, particularly board of directors who play the effective monitoring role in reducing excessive risk-taking behaviour (anderson, mansi & reeb, 2004; jiraporn, chatjuthamard, tong & kim, 2015; oecd, 2006). hence, the fiduciary duty carried out by the board of directors is deemed important in the banking sector than any other industry since any shortcomings of bank governance could lead to potential financial system failure. prior studies have highlighted the fact that banks tend to undertake excessive risk when there is state ownership in the bank. however, the existing studies do not examine cg mechanisms in moderating the impact of state ownership on the banks’ risktaking. hence, this further motivates the exploration of cg mechanisms in this study. this study adds value to the existing literature in twofold. firstly, this study contributes by being a pioneer study addressing state ownership through glics and their risk-taking behaviour in malaysia’s banking industry. malaysia, being one of the southeast asian countries, is the best teaching material for this research study. in addition, the malaysian commercial bank is the largest and most significant provider of funds in the malaysian banking system, the level of banks’ risk-taking could give a large impact on nation economic development. secondly, most of the existing literature has largely ignored the role of corporate governance mechanisms while examining this relationship. using malaysia data, we look at how banks’ risk-taking behaviour is affected by the corporate governance mechanisms through different compositions of the board of directors such as board independence, foreign director and female director. the results of this study are important in providing insights into the role of government intervention and bank decision making through their understanding of the relationship between state ownership and risk-taking behaviour. we find that the role of state ownership is significant in determining the banks’ risk-taking behaviour in malaysia. we also find that the composition of board of directors somehow plays a significant role in bank governance. this paper proceeds as follows. hypotheses development is discussed in section 2. data, variables and summary statistics are presented in section 3. then, the results of data analysis are shared in section 4 and the paper ends with the conclusion and implications in the last section (section 5). 2. hypotheses development three theories have been widely used in the study of the relationship between state ownership and risk-taking behaviour. firstly, the moral hazard theory. the 70 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 action of financial bailouts by institutions, i.e., the government, would encourage riskier behaviour in the future if those risk-taking parties believe that they are not going to be responsible for the consequences as a result of excessive risk-taking behaviour (krugman, 2009). besides, the “too big to fail” perspective also leads to moral hazard as they believe that the government would not be willing to let larger banks fail and hence, is more likely to bail them out. secondly, the agency theory. this theory focuses on the conflict of interest arises between the major and minor shareholders. shareholders who have dominating power have the incentives to affect the banks’ decisions by undertaking riskier investments (laeven & levine, 2009; shleifer & vishny, 1986). thirdly, the social lending theory. this theory shows the need of the government to accomplish social objectives. prior studies claimed that sobs tend to channel resources for socially beneficial projects that aim to increase job creations and developing nations where the social returns fail to be taken into account by the pobs (berger et al., 2005; stiglitz, 1993). in addition to the above theories, prior studies showed that state ownership encourages banks to take more risks. this is further supported by empirical findings where state ownership is positively associated with risk-taking. this leads sobs to hold less core capital, less profitability and incur greater credit risk than pobs (brandao-marques et al., 2020; iannotta et al., 2013; lassoued et al., 2016; zhu & yang, 2016). these findings are consistent with the agency theory where banks controlled by major shareholders are found to have a tendency to undertake more risk. since glics hold majority ownership in local commercial banks where the malaysian government has a direct controlling stake, this study examines whether the state ownership and sobs are associated with higher risktaking. three risk-taking measurements are employed in this study; the credit risk proxy by non-performing loans (npl) ratio, the capital adequacy proxy by capital adequacy ratio (car) and the liquidity risk proxy by liquidity ratio (lr). koudstaal and wijnbergen (2012) reported that the more troubled the loan portfolio, the greater the inclination for banks to take risks. a larger number of npls due to excessive risk-taking in lending could be used to mark the onset of a banking crisis (reinhart & rogoff, 2011; jensen & meckling, 1976). moreover, capital adequacy plays an important role for banks solvency and their protection from untoward events which arise as a result of liquidity risk as well as the credit risk that banks are exposed to in the normal course of their business (karim, hassan, hassan & mohamad, 2014; athanasoglou, brissimis & delis, 2008). insufficient equity capital has also been partly blamed in the global financial crisis and the asian financial crisis. moving on, lr could be used to measure a bank’s financial health and it presents a preliminary expectation regarding the solvency of a company (imbierowicz & rauch, 2014). a study from de haan and van den end (2013) suggested that the extended liquidity support by the central bank might have been an incentive for banks to reduce their liquidity state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 71 buffers. the subprime crisis that happened in 2008 has demonstrated how severely illiquidity can crystallise. to conclude, these three proxies are important in determining the banks’ risk-taking behaviour and therefore, are employed in this study. based on the theoretical perspectives and empirical finding from prior studies, we proposed our first hypothesis as follows: h1a: state-owned banks have higher risks than private-owned banks. h1b: the degree of ownership by malaysian government in banks significantly increases banks’ risk-taking behaviour. the flaw in the functioning of cg mechanisms has played part of the root in previous financial crises where there are massive government intervention and excessive risk-taking behaviour in the banking sector. shareholder-friendly cg is also found connected with higher risk-taking (anginer, demirguc-kunt, huizinga & ma, 2018). as indicated by gomez (2005), the issue of ownership and the act of controlling a bank or enterprise is a key issue in the most definition of cg. this issue is even more crucial in developing countries such as malaysia where the government has a dominant controlling stake and decision making in the national economy. hence, it is important to establish sound cg in the malaysian banking sector to ensure that the investment decisions or strategies made by banks are well-monitored to prevent excessive risk-taking behaviour. a good cg comes from the essential role and the fiduciary duty play by the board of directors (gomez, 2005). prior studies also showed that sound bank governance played by the role of the board of directors is relatively important in the banking sector as the effective monitoring roles by the board is connected with lower risk-taking (anderson et al., 2004; jiraporn et al., 2015; liang et al., 2013; oecd, 2006). thus, this study focuses on three cg mechanisms (board independence, foreign director, and female director) in moderating the impact of state ownership through glics on banks’ risk-taking. the earlier literature studies on corporate governance issues provide no definite findings on the role of independent directors. advocates show that organisations will become more effective and efficient in reducing agency issues and moral hazard with the presence of independent directors on board (rosenstein & wyatt, 1990; klein, 2002; nguyen & nielsen, 2010). however, another strand of literature argues that independent directors may lack sufficient knowledge of the firm-specific information and lead to sub-optimal decisions (raheja, 2005). in the case of state ownership, oecd (2006) suggested that an adequate number of independent directors on the board of sobs be included. this is to ensure that the decision made by the board is independent and does not interfere by the government. based on the previous studies, we developed the second hypothesis as follows: 72 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 h2: board independence significantly affects the impact of ownership by malaysian government on banks’ risk-taking behaviour. foreign directors are non-local directors who served on the board. the role of foreign director is ambiguous although many researchers have examined this in their studies for the past few decades. statement shows that foreign directors bring new technology and managerial expertise which brings better supervision by reducing information asymmetry and agency cost (ezat & elmasry, 2008; samaha, dahawy, hussainey & stapleton, 2012). berger, hasan and zhou (2009) asserted that minority foreign ownership takes positions on the board and “leverage” the positions to monitor and improve bank management. others contend that foreign directors are likely to be less familiar with national accounting standard and management methods, making it more difficult for foreign directors to appraise managerial performance or challenge managerial decisions. based on these vague statements, we examine the effect of foreign director in the malaysian banking industry with the third hypothesis developed as follows: h3: foreign director significantly affects the impact of ownership by malaysian government on banks’ risk-taking behaviour. recently, female directors have become the theme of corporate governance mechanisms that arouse wide concern in research worldwide. adams and ferreira (2009) found that female directors are more likely to join monitoring committees and improve firm performance by easing the weak governance in chinese listed firms. for example, corporate governance could be strengthened through improved monitoring and greater oversight of management by female directors. however, some other literature suggests that female directors in statecontrolled firms are required to divert part of their efforts to non-profit related political and social activities. in other words, female directors in state-controlled firms are more likely charged with social or political tasks. hence, it is vital for us to examine the role of female directors. the fourth hypothesis is developed as follows: h4: female director significantly affects the impact of ownership by malaysian government on banks’ risk-taking behaviour. 3. research methodology 3.1 sample banks we first identified and obtained the list of banks involved in this study from bank negara malaysia (bnm). we construct a balanced panel of 8 public listed state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 73 local commercial banks in malaysia from year 2011 to 2015. foreign-owned commercial banks are excluded in the sample due to data limitations. besides, the five major glics hold more than 50 percent of the whole market capitalisation and they have hold majority ownership in the local commercial banks. hence, our final sample of this study focuses on local commercial banks. by referring to annual reports released from the bursa malaysia website, relevant data such as government shares ownership, bank’s liquidity and profitability, bank’s level of risk, bank’s size, and bank’s revenue growth could be identified. hence, all the data employed in this study are hand collected. we have state-owned banks and private-owned banks in our sample. we define banks as state-owned banks if: (1) malaysian government owns the largest percentage of ownership through five major glics, namely epf, pnb, knb, ltat and lth; (2) the degree of government ownership is equal to or more than 40 percent. likewise, for the private-owned bank, it is defined as such if the private entity owns the largest percentage of ownership in the bank. 3.2 variables measurement 3.2.1 measures of state ownership state ownership is the independent variable that will give an impact on the banks’ risk-taking behaviour. in this study, we focus on malaysian government ownership through the five largest glics in malaysia. they are employees provident fund (epf), permodalan nasional bnd (pnb), khazanah nasional bhd (knb), lembaga tabung angkatan tentera (ltat) and lembaga tabung haji (lth). these investment institutions have also hold majority ownership in the local commercial banks. hence, we employ two state ownership measurements in this study. they are dummy of state ownership (dummyso) and degree of state ownership (so). dummy of state ownership takes a value of “1” if (1) malaysian government owns the largest percentage of ownership; (2) the total of ownership is more than or equal to 40 percent or “0” otherwise. besides, the degree of ownership by malaysian government is obtained by taking the percentage of shares held by these five investment institutions in the banks. 3.2.2 measures of risk-taking in this study, we would like to know the impact of the degree of ownership by malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). 74 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 3.2.3 moderating variables three board characteristics variables have been chosen to further investigate the effect of corporate governance aspects in the study of the relationship between 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) = 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) = 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 74 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-95 since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 3.2.3 moderating variables three board characteristics variables have been chosen to further investigate the effect of corporate governance aspects in the study of the relationship between 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) = 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) = 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜 𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 74 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-95 since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 3.2.3 moderating variables three board characteristics variables have been chosen to further investigate the effect of corporate governance aspects in the study of the relationship between 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) = 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) = 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜 𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 74 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-95 since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 3.2.3 moderating variables three board characteristics variables have been chosen to further investigate the effect of corporate governance aspects in the study of the relationship between 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) = 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) = 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜 𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 74 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-95 since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 3.2.3 moderating variables three board characteristics variables have been chosen to further investigate the effect of corporate governance aspects in the study of the relationship between 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) = 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) = 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the 9 malaysian government through glics in local malaysian commercial banks on their risk-taking behaviour. with carefully selecting the dependent variables of this study, we have come to three measurements of risk-taking employed in this study. they are non-performing loans (npl) ratio, capital adequacy ratio (car) and liquidity ratio (lr). since decades ago, the npl ratio has been widely used by researchers to study the factors behind and risk associated with this ratio. for instance, it has been used by jensen and meckling (1976), demirguc-kunt (1989), barr, seiford and siems (1994), and koudstaal and wijnbergen (2012). we expect banks with a greater percentage of malaysian government ownership would have a higher npl ratio as a result of a more risk-taking of a bank. npl ratio can be calculated by using the formula as shown below: 𝑁𝑜𝑛 − 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 (𝑁𝑃𝐿) 𝑅𝑎𝑡𝑖𝑜 = 𝑁𝑜𝑛 − 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑖𝑛𝑔 𝑙𝑜𝑎𝑛𝑠 𝑇𝑜𝑡𝑎𝑙 𝑙𝑜𝑎𝑛𝑠 empirical evidence suggests that capital requirements have a significant impact on deposit and lending behaviours of a bank (karim et al., 2014). prior researchers such as shehzad, haan and scholtens (2010) and zhu and yang (2016) have examined car in the study of state ownership and ownership concentration in their studies. in this paper, we use total capital ratio as a measure of capital adequacy. we expect banks with a greater percentage of malaysian government ownership would have a lower car, which implies that the bank may have insufficient capital to absorb any potential losses. car can be calculated by using the formula as shown below: 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝐴𝑑𝑒𝑞𝑢𝑎𝑐𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐶𝐴𝑅) = 𝑇𝑖𝑒𝑟 1 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 + 𝑡𝑖𝑒𝑟 2 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑅𝑖𝑠𝑘 𝑤𝑒𝑖𝑔ℎ𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 moreover, recent studies conducted by de haan and van den end (2013), imbierowicz and rauch (2014), and zhu and yang (2016) found that liquidity risk has a significant impact on bank default probability. basel iii has introduced liquidity coverage ratio (lcr) to strengthen banks’ liquidity profiles. however, due to data limitation, we adopt a static measure for banks’ liquidity using the standard balance sheet data since the detailed breakdown of off-balance sheets is not available. hence, the liquidity ratio is employed in this study. we expect that banks with a greater percentage of malaysian government ownership would have a lower lr as a result of higher risk-taking. lr can be calculated by using the formula as shown below: 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑅𝑎𝑡𝑖𝑜 (𝐿𝑅) 𝐿𝑖𝑞𝑢𝑖𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 𝐷𝑒𝑝𝑜 𝑠𝑖𝑡𝑠 𝑓𝑟𝑜𝑚 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑛𝑑 𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑖𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 3.2.3. moderating variables three board characteristics variables have been chosen to further investigate the state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 75 state ownership and bank’s risk-taking behaviour. they are known as the board independence, foreign director and female director. one of the most used board characteristics variable by previous researchers is board independence. the independence of directors is the central theme in governance issues. bursa malaysia main market listing requirements chapter 3: admission part 3.04 mentioned that (1) an applicant must ensure that at least 2 directors or 1/3 of the board of directors of the applicant, whichever is higher, are independent directors. (2) if the number of directors of the applicant is not 3 or a multiple of 3, then the number nearest 1/3 must be used. realising the importance of quality corporate governance, we introduce dummy as adopted in zhu, ye, tucker and chan (2016). theoretically, banks with board independence ratio more than or equal to 0.33 should take a value of “1” or “0” otherwise. however, we find that most of the sample banks in our study have made an effort in adhering to the above requirements by having more than 0.33 independent directors in their board structure. to study the differences in board independence composition in banks’ risk-taking behaviour, we have identified banks with board independence ratio more than or equal to 0.50 should take a value of “1” or “0” otherwise. this is to the extent of studying whether a larger proportion of independent directors have a greater impact on the banks’ risk-taking. other than board independence, the dummy of foreign directors on board is also employed as one of the moderating variables in studying the relationship between state ownership and bank’s risk-taking behaviour. following du, jian and lai (2017), we take a value of “1” if one or more directors come(s) from other countries or “0” otherwise. besides, one of the utmost popular issues examined by researchers liu, wei and xie (2014), gulamhussen and santa (2015) and zhu and yang (2016) in their research study was gender diversity. we would like to know whether female director helps in bank monitoring and reduce the likelihood of excessive risk-taking. hence, we employ the dummy of female directors as one of the moderating variables in this study. we take a value of “1” if there is at least one woman on the board of directors of the bank or “0” otherwise. 3.2.4 c o n t ro l variables there are three control variables employed in this study. they are: bank size, bank revenue growth and return on assets. the bank size is measured by the logarithm of the bank’s total assets. this has been adopted in the studies by anginer et al. (2018), brandao-marques et al. (2020), laeven and levine (2009), lassoued et al. (2016), and zhu and yang (2016) while investigating the bank’s risk-taking. the reason for controlling this variable is bank size may influence a bank’s level of risk-taking. there is a possibility that larger banks may have better risk diversification and greater capacity to absorb their risk-taking. 76 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 the revenue growth informs us how much additional annual revenue that a business can handle according to the resources in the balance sheet. bank revenue growth is adopted as one of the control variables since every bank has a different revenue growth rate and it is believed that large banks may have higher revenue growth as they are accessible to greater resources that are available. to increase the reliability of the result conducted, we control the bank revenue growth. the bank revenue growth can be calculated by taking the first difference of the logarithm of total operating income. the return on assets (roa) is another control variable employed in this study. the roa has been widely used in the previous studies while investigating the relationship between the bank’s state ownership and risk-taking, such as dong, liu, shen and sun’s (2016), and lassoued et al.’s (2016). the reason for controlling this variable is because the bank’s total assets may influence the incomes generated and it also gives an impact on the bank’s risk-taking. there is a possibility that the banks who owned more assets will earn higher roa and exposed to greater risks. 3.3 model based on the above variables, model 1 and 2 are the baseline models used in this study to regress on the relationship between state ownership and banks’ risk-taking behaviour. model 1 is used to test hypothesis 1a in which to capture the risk difference between the state-owned banks and private-owned banks. the degree of state ownership, so is introduced in model 2 to test the degree of state ownership through glics on the banks’ risk-taking behaviour, as shown in hypothesis 1b: in testing for the effect of corporate governance mechanisms on banks’ risktaking behaviour, we develop six models as follows: to test for the interaction effect of hypothesis 2, we add board independence variable in model 3 and model 4. model 3 is used to investigate the interaction between board independence and state ownership on banks’ risk-taking whereas the interaction between board independence and the degree of state ownership is tested in model 4: 11 growth as they are accessible to greater resources that are available. to increase the reliability of the result conducted, we control the bank revenue growth. the bank revenue growth can be calculated by taking the first difference of the logarithm of total operating income. the return on assets (roa) is another control variable employed in this study. the roa has been widely used in the previous studies while investigating the relationship between the bank’s state ownership and risk-taking, such as dong, liu, shen and sun’s (2016), and lassoued et al.’s (2016). the reason for controlling this variable is because the bank’s total assets may influence the incomes generated and it also gives an impact on the bank’s risk-taking. there is a possibility that the banks who owned more assets will earn higher roa and exposed to greater risks. 3.3. m o d e l based on the above variables, model 1 and 2 are the baseline models used in this study to regress on the relationship between state ownership and banks’ risk-taking behaviour. model 1 is used to test hypothesis 1a in which to capture the risk difference between the state-owned banks and private-owned banks. the degree of state ownership, so is introduced in model 2 to test the degree of state ownership through glics on the banks’ risk-taking behaviour, as shown in hypothesis 1b: risk�� = 𝛼 + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (1) risk�� = 𝛼 + 𝛽�𝑆𝑂�� + 𝛽�(𝑆𝑂 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (2) in testing for the effect of corporate governance mechanisms on banks’ risktaking behaviour, we develop six models as follows: to test for the interaction effect of hypothesis 2, we add board independence variable in model 3 and model 4. model 3 is used to investigate the interaction between board independence and state ownership on banks’ risk-taking whereas the interaction between board independence and the degree of state ownership is tested in model 4: risk�� = 𝛼 + 𝛽�𝐵𝑜𝑎𝑟𝑑_𝐼�� + 𝛽�(𝐵𝑜𝑎𝑟𝑑_𝐼 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (3) risk�� = 𝛼 + 𝛽�𝐵𝑜𝑎𝑟𝑑_𝐼�� + 𝛽�(𝐵𝑜𝑎𝑟𝑑_𝐼 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (4) to test for the interaction effect of hypothesis 3, we add foreign director variable in model 5 and model 6. model 5 is used to investigate the interaction between foreign director and state ownership on banks’ risk-taking whereas the interaction between foreign director and the degree of state ownership is tested in model 6: (1) (2) 11 growth as they are accessible to greater resources that are available. to increase the reliability of the result conducted, we control the bank revenue growth. the bank revenue growth can be calculated by taking the first difference of the logarithm of total operating income. the return on assets (roa) is another control variable employed in this study. the roa has been widely used in the previous studies while investigating the relationship between the bank’s state ownership and risk-taking, such as dong, liu, shen and sun’s (2016), and lassoued et al.’s (2016). the reason for controlling this variable is because the bank’s total assets may influence the incomes generated and it also gives an impact on the bank’s risk-taking. there is a possibility that the banks who owned more assets will earn higher roa and exposed to greater risks. 3.3. m o d e l based on the above variables, model 1 and 2 are the baseline models used in this study to regress on the relationship between state ownership and banks’ risk-taking behaviour. model 1 is used to test hypothesis 1a in which to capture the risk difference between the state-owned banks and private-owned banks. the degree of state ownership, so is introduced in model 2 to test the degree of state ownership through glics on the banks’ risk-taking behaviour, as shown in hypothesis 1b: risk�� = 𝛼 + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (1) risk�� = 𝛼 + 𝛽�𝑆𝑂�� + 𝛽�(𝑆𝑂 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (2) in testing for the effect of corporate governance mechanisms on banks’ risktaking behaviour, we develop six models as follows: to test for the interaction effect of hypothesis 2, we add board independence variable in model 3 and model 4. model 3 is used to investigate the interaction between board independence and state ownership on banks’ risk-taking whereas the interaction between board independence and the degree of state ownership is tested in model 4: risk�� = 𝛼 + 𝛽�𝐵𝑜𝑎𝑟𝑑_𝐼�� + 𝛽�(𝐵𝑜𝑎𝑟𝑑_𝐼 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (3) risk�� = 𝛼 + 𝛽�𝐵𝑜𝑎𝑟𝑑_𝐼�� + 𝛽�(𝐵𝑜𝑎𝑟𝑑_𝐼 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (4) to test for the interaction effect of hypothesis 3, we add foreign director variable in model 5 and model 6. model 5 is used to investigate the interaction between foreign director and state ownership on banks’ risk-taking whereas the interaction between foreign director and the degree of state ownership is tested in model 6: (3) state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 77 to test for the interaction effect of hypothesis 3, we add foreign director variable in model 5 and model 6. model 5 is used to investigate the interaction between foreign director and state ownership on banks’ risk-taking whereas the interaction between foreign director and the degree of state ownership is tested in model 6: to test for the interaction effect of hypothesis 4, we add female director variable in model 7 and model 8. model 7 is used to investigate the interaction between female director and state ownership on banks’ risk-taking whereas the interaction between female director and the degree of state ownership is tested in model 8: where riskit is the risk measures for bank i in year t. the risk measurements in the above regressions consist of npl ratio, car, and lr. dummysoit is the state ownership dummy for bank i in year t whereas soit is the state ownership percentage for bank i in year t. board_iit is the board independence dummy for bank i in year t. foreign_dit is the foreign director dummy for bank i in year t. female_dit is the female director dummy for bank i in year t. ltait, rgit, and roait are the control variables for bank i in year t. 4. results 4.1 descriptive statistics table 1 below provides the descriptive statistics for all the variables used in the model of this study. in the context of malaysia, banks have an average of 2.16 percent npl. car and lr are 15.18 percent and 17.38 percent respectively. for a comparison with similar studies conducted previously, zhu and yang (2016) found a substantially higher npl (2.54%) and lr (32.18%), whereas 11 growth as they are accessible to greater resources that are available. to increase the reliability of the result conducted, we control the bank revenue growth. the bank revenue growth can be calculated by taking the first difference of the logarithm of total operating income. the return on assets (roa) is another control variable employed in this study. the roa has been widely used in the previous studies while investigating the relationship between the bank’s state ownership and risk-taking, such as dong, liu, shen and sun’s (2016), and lassoued et al.’s (2016). the reason for controlling this variable is because the bank’s total assets may influence the incomes generated and it also gives an impact on the bank’s risk-taking. there is a possibility that the banks who owned more assets will earn higher roa and exposed to greater risks. 3.3. m o d e l based on the above variables, model 1 and 2 are the baseline models used in this study to regress on the relationship between state ownership and banks’ risk-taking behaviour. model 1 is used to test hypothesis 1a in which to capture the risk difference between the state-owned banks and private-owned banks. the degree of state ownership, so is introduced in model 2 to test the degree of state ownership through glics on the banks’ risk-taking behaviour, as shown in hypothesis 1b: risk�� = 𝛼 + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (1) risk�� = 𝛼 + 𝛽�𝑆𝑂�� + 𝛽�(𝑆𝑂 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (2) in testing for the effect of corporate governance mechanisms on banks’ risktaking behaviour, we develop six models as follows: to test for the interaction effect of hypothesis 2, we add board independence variable in model 3 and model 4. model 3 is used to investigate the interaction between board independence and state ownership on banks’ risk-taking whereas the interaction between board independence and the degree of state ownership is tested in model 4: risk�� = 𝛼 + 𝛽�𝐵𝑜𝑎𝑟𝑑_𝐼�� + 𝛽�(𝐵𝑜𝑎𝑟𝑑_𝐼 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (3) risk�� = 𝛼 + 𝛽�𝐵𝑜𝑎𝑟𝑑_𝐼�� + 𝛽�(𝐵𝑜𝑎𝑟𝑑_𝐼 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (4) to test for the interaction effect of hypothesis 3, we add foreign director variable in model 5 and model 6. model 5 is used to investigate the interaction between foreign director and state ownership on banks’ risk-taking whereas the interaction between foreign director and the degree of state ownership is tested in model 6: (4) (5) 12 risk�� = 𝛼 + 𝛽�𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 �� + 𝛽�(𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (5) risk�� = 𝛼 + 𝛽�𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷�� + 𝛽�(𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (6) to test for the interaction effect of hypothesis 4, we add female director variable in model 7 and model 8. model 7 is used to investigate the interaction between female director and state ownership on banks’ risk-taking whereas the interaction between female director and the degree of state ownership is tested in model 8: risk�� = 𝛼 + 𝛽�𝐹𝑒𝑚𝑎𝑙𝑒_𝐷�� + 𝛽�(𝐹𝑒𝑚𝑎𝑙𝑒_𝐷 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (7) risk�� = 𝛼 + 𝛽�𝐹𝑒𝑚𝑎𝑙𝑒_𝐷�� + 𝛽�(𝐹𝑒𝑚𝑎𝑙𝑒_𝐷 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (8) where riskit is the risk measures for bank i in year t. the risk measurements in the above regressions consist of npl ratio, car, and lr. dummysoit is the state ownership dummy for bank i in year t whereas soit is the state ownership percentage for bank i in year t. board_iit is the board independence dummy for bank i in year t. foreign_dit is the foreign director dummy for bank i in year t. female_dit is the female director dummy for bank i in year t. ltait, rgit, and roait are the control variables for bank i in year t. 3.4. d e s c r i p t i v e s t a t i s t i c s table 1 below provides the descriptive statistics for all the variables used in the model of this study. in the context of malaysia, banks have an average of 2.16 percent npl. car and lr are 15.18 percent and 17.38 percent respectively. for a comparison with similar studies conducted previously, zhu and yang (2016) found a substantially higher npl (2.54%) and lr (32.18%), whereas a lower car (13.26%) in china. similar statistics are found in shaban and james's (2018) study in indonesia. the statistics show higher npl (3.01%) and lr (25.75%), and lower car (11.91%). this enlightens that malaysia’s banks have substantially lower default loans and a higher level of capital to absorb any potential losses. however, malaysia’s banks are found to have less liquid assets. another interesting variable worth mention is state ownership (so). so has a range from 12.12 percent to 66.87 percent. on average, so owns 33.14 percent in malaysia’s banking industry. lassoued et al. (2016) found an average of 14.49 percent so in mena countries whereas an average of 11.09 percent so is found in (6) 12 risk�� = 𝛼 + 𝛽�𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 �� + 𝛽�(𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (5) risk�� = 𝛼 + 𝛽�𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷�� + 𝛽�(𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (6) to test for the interaction effect of hypothesis 4, we add female director variable in model 7 and model 8. model 7 is used to investigate the interaction between female director and state ownership on banks’ risk-taking whereas the interaction between female director and the degree of state ownership is tested in model 8: risk�� = 𝛼 + 𝛽�𝐹𝑒𝑚𝑎𝑙𝑒_𝐷�� + 𝛽�(𝐹𝑒𝑚𝑎𝑙𝑒_𝐷 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (7) risk�� = 𝛼 + 𝛽�𝐹𝑒𝑚𝑎𝑙𝑒_𝐷�� + 𝛽�(𝐹𝑒𝑚𝑎𝑙𝑒_𝐷 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (8) where riskit is the risk measures for bank i in year t. the risk measurements in the above regressions consist of npl ratio, car, and lr. dummysoit is the state ownership dummy for bank i in year t whereas soit is the state ownership percentage for bank i in year t. board_iit is the board independence dummy for bank i in year t. foreign_dit is the foreign director dummy for bank i in year t. female_dit is the female director dummy for bank i in year t. ltait, rgit, and roait are the control variables for bank i in year t. 3.4. d e s c r i p t i v e s t a t i s t i c s table 1 below provides the descriptive statistics for all the variables used in the model of this study. in the context of malaysia, banks have an average of 2.16 percent npl. car and lr are 15.18 percent and 17.38 percent respectively. for a comparison with similar studies conducted previously, zhu and yang (2016) found a substantially higher npl (2.54%) and lr (32.18%), whereas a lower car (13.26%) in china. similar statistics are found in shaban and james's (2018) study in indonesia. the statistics show higher npl (3.01%) and lr (25.75%), and lower car (11.91%). this enlightens that malaysia’s banks have substantially lower default loans and a higher level of capital to absorb any potential losses. however, malaysia’s banks are found to have less liquid assets. another interesting variable worth mention is state ownership (so). so has a range from 12.12 percent to 66.87 percent. on average, so owns 33.14 percent in malaysia’s banking industry. lassoued et al. (2016) found an average of 14.49 percent so in mena countries whereas an average of 11.09 percent so is found in 12 risk�� = 𝛼 + 𝛽�𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 �� + 𝛽�(𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (5) risk�� = 𝛼 + 𝛽�𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷�� + 𝛽�(𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (6) to test for the interaction effect of hypothesis 4, we add female director variable in model 7 and model 8. model 7 is used to investigate the interaction between female director and state ownership on banks’ risk-taking whereas the interaction between female director and the degree of state ownership is tested in model 8: risk�� = 𝛼 + 𝛽�𝐹𝑒𝑚𝑎𝑙𝑒_𝐷�� + 𝛽�(𝐹𝑒𝑚𝑎𝑙𝑒_𝐷 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (7) risk�� = 𝛼 + 𝛽�𝐹𝑒𝑚𝑎𝑙𝑒_𝐷�� + 𝛽�(𝐹𝑒𝑚𝑎𝑙𝑒_𝐷 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (8) where riskit is the risk measures for bank i in year t. the risk measurements in the above regressions consist of npl ratio, car, and lr. dummysoit is the state ownership dummy for bank i in year t whereas soit is the state ownership percentage for bank i in year t. board_iit is the board independence dummy for bank i in year t. foreign_dit is the foreign director dummy for bank i in year t. female_dit is the female director dummy for bank i in year t. ltait, rgit, and roait are the control variables for bank i in year t. 3.4. d e s c r i p t i v e s t a t i s t i c s table 1 below provides the descriptive statistics for all the variables used in the model of this study. in the context of malaysia, banks have an average of 2.16 percent npl. car and lr are 15.18 percent and 17.38 percent respectively. for a comparison with similar studies conducted previously, zhu and yang (2016) found a substantially higher npl (2.54%) and lr (32.18%), whereas a lower car (13.26%) in china. similar statistics are found in shaban and james's (2018) study in indonesia. the statistics show higher npl (3.01%) and lr (25.75%), and lower car (11.91%). this enlightens that malaysia’s banks have substantially lower default loans and a higher level of capital to absorb any potential losses. however, malaysia’s banks are found to have less liquid assets. another interesting variable worth mention is state ownership (so). so has a range from 12.12 percent to 66.87 percent. on average, so owns 33.14 percent in malaysia’s banking industry. lassoued et al. (2016) found an average of 14.49 percent so in mena countries whereas an average of 11.09 percent so is found in 12 risk�� = 𝛼 + 𝛽�𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 �� + 𝛽�(𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (5) risk�� = 𝛼 + 𝛽�𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷�� + 𝛽�(𝐹𝑜𝑟𝑒𝑖𝑔𝑛_𝐷 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (6) to test for the interaction effect of hypothesis 4, we add female director variable in model 7 and model 8. model 7 is used to investigate the interaction between female director and state ownership on banks’ risk-taking whereas the interaction between female director and the degree of state ownership is tested in model 8: risk�� = 𝛼 + 𝛽�𝐹𝑒𝑚𝑎𝑙𝑒_𝐷�� + 𝛽�(𝐹𝑒𝑚𝑎𝑙𝑒_𝐷 × 𝐷𝑢𝑚𝑚𝑦��)�� + 𝛽�dummy���� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (7) risk�� = 𝛼 + 𝛽�𝐹𝑒𝑚𝑎𝑙𝑒_𝐷�� + 𝛽�(𝐹𝑒𝑚𝑎𝑙𝑒_𝐷 × 𝑆𝑂)�� + 𝛽�so�� + 𝛽�lta�� + 𝛽� rg�� + 𝛽�roa�� + 𝜀�� (8) where riskit is the risk measures for bank i in year t. the risk measurements in the above regressions consist of npl ratio, car, and lr. dummysoit is the state ownership dummy for bank i in year t whereas soit is the state ownership percentage for bank i in year t. board_iit is the board independence dummy for bank i in year t. foreign_dit is the foreign director dummy for bank i in year t. female_dit is the female director dummy for bank i in year t. ltait, rgit, and roait are the control variables for bank i in year t. 3.4. d e s c r i p t i v e s t a t i s t i c s table 1 below provides the descriptive statistics for all the variables used in the model of this study. in the context of malaysia, banks have an average of 2.16 percent npl. car and lr are 15.18 percent and 17.38 percent respectively. for a comparison with similar studies conducted previously, zhu and yang (2016) found a substantially higher npl (2.54%) and lr (32.18%), whereas a lower car (13.26%) in china. similar statistics are found in shaban and james's (2018) study in indonesia. the statistics show higher npl (3.01%) and lr (25.75%), and lower car (11.91%). this enlightens that malaysia’s banks have substantially lower default loans and a higher level of capital to absorb any potential losses. however, malaysia’s banks are found to have less liquid assets. another interesting variable worth mention is state ownership (so). so has a range from 12.12 percent to 66.87 percent. on average, so owns 33.14 percent in malaysia’s banking industry. lassoued et al. (2016) found an average of 14.49 percent so in mena countries whereas an average of 11.09 percent so is found in (7) (8) 78 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 a lower car (13.26%) in china. similar statistics are found in shaban and james’s (2018) study in indonesia. the statistics show higher npl (3.01%) and lr (25.75%), and lower car (11.91%). this enlightens that malaysia’s banks have substantially lower default loans and a higher level of capital to absorb any potential losses. however, malaysia’s banks are found to have less liquid assets. another interesting variable worth mention is state ownership (so). so has a range from 12.12 percent to 66.87 percent. on average, so owns 33.14 percent in malaysia’s banking industry. lassoued et al. (2016) found an average of 14.49 percent so in mena countries whereas an average of 11.09 percent so is found in vietnam by vo (2018). these numbers illuminate that the government involvement in malaysia’s banking industry is fairly high. this further motivates the researcher to examine the impact of state ownership on banks’ risk-taking. table 1. descriptive statistics of variables variables obs. mean minimum maximum std. dev. risk measures npl 40 0.0216 0.0050 0.0533 0.0105 car 40 0.1518 0.1203 0.1774 0.0123 lr 40 0.1738 0.0492 0.4363 0.0876 ownership indicator dummyso 40 0.5000 0.0000 1.0000 0.5064 so 40 0.3314 0.1212 0.6687 0.1893 moderating variables board_i 40 0.7250 0.0000 1.0000 0.4522 foreign_d 40 0.8000 0.0000 1.0000 0.4051 female_d 40 0.7500 0.0000 1.0000 0.4385 control variables ta 40 0.2310 0.0361 0.7080 0.1750 rg 40 0.0302 -0.2441 0.3453 0.0753 roa 40 1.2450 0.5500 2.0000 0.3194 note: obs. = observations, std. dev. = standard deviation, npl = non-performing loans ratio, car = capital adequacy ratio, lr = liquidity ratio, dummyso = state ownership dummy, so = state ownership percentage, board_i = board independence dummy, foreign_d = foreign director dummy, female_d = female director dummy, ta = total assets (in trillions), rg = revenue growth and roa = return on assets. state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 79 4.2 s t a t e ownership and banks’ risk-taking table 2. state ownership and banks’ risk-taking (baseline model) variables npl car lr (1) (2) (3) (4) (5) (6) dummyso 0.0096** 0.0003 -0.0015 -0.0554*** 0.0195 -0.0059 (0.0138) (0.9810) (0.6569) (0.0043) (0.1451) (0.9650) dummyso* so 0.1201 0.2226* 1.2327 (0.2890) (0.0530) (0.2272) so -0.1455 -0.1533 -1.7126 (0.2071) (0.1471) (0.1345) lta -0.0055 -0.0020 0.0176*** 0.0098* 0.0602** 0.1228* (0.1037) (0.5886) (0.0011) (0.0593) (0.0185) (0.0600) rg -0.0096 -0.0152* 0.0142 0.01427 -0.0858 -0.1623 (0.1371) (0.0735) (0.1674) (0.1528) (0.5160) (0.4765) roa -0.0039 -0.0014 0.0040 0.01165 0.0812* 0.0610 (0.5459) (0.8822) (0.6131) (0.3292) (0.0652) (0.1854) constant 0.0832** 0.0634 -0.0508 0.0489 -0.4079** -0.8683 (0.0114) (0.1301) (0.2943) (0.3926) (0.0477) (0.1169) observation 40 40 40 40 40 40 r2 0.2886 0.3260 0.2991 0.3880 0.2126 0.3096 note: ***, **, * denote significance at 1%, 5% and 10% significant levels respectively. npl = non-performing loans ratio, car = capital adequacy ratio, lr = liquidity ratio, dummyso = state ownership dummy, dummyso * so = interaction of state ownership dummy and state ownership percentage, so = state ownership percentage, lta = logarithm of total assets, rg = revenue growth and roa = return on assets. table 2 presents the regression results for the baseline models, model 1 and 2. the regression includes all control variables and r2 is reported at the bottom of the table. column 1, 3, 5 in table 2 shows the sobs’ baseline result as compared with the pobs. based on column 1 in table 2, sobs are associated with significantly higher non-performing loans (npl) ratio of 0.0096 percent compared with the pobs, at 95 percent confidence level. the other two risktaking measurements, however, are not significant. hence, we do not find strong evidence to support hypothesis 1a. column 2, 4 and 6 is the model used to test the interaction of dummy and the degree of state ownership. the results show 80 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 that the interaction of dummy and state ownership only significant in terms of capital adequacy ratio (car). the significant positive coefficient of 0.2226 in column 4 suggests that a higher degree of state ownership tends to improve car. in addition, the results are found to have no significant impact on banks’ npls and lr. thus, hypothesis 1b is rejected. as for the bank size, the results reveal that larger banks tend to have higher capital adequacy and liquidity, hence lowering the risk-taking behaviour. this is according to the literature that larger banks have better risk diversification and able to reduce high risk-taking activities. 4.3 board independence and state ownership on banks’ risk-taking table 3 presents the regression result of the effect of board independence in the relationship between state ownership and banks’ risk-taking. the result shows that when more than half of the board of directors were independent directors, it has significant positive impacts on npl ratio and liquidity ratio (lr) with a coefficient of 0.0112 and 0.1856 at 95 percent confidence level and 99 percent confidence level respectively. table 3. the effect of board independence and state ownership on banks’ risk-taking variables npl car lr board_i -0.0093** (0.0135) 0.0004 (0.9061) -0.1273*** (0.0024) board_i * dummyso 0.0112** (0.0168) -0.0085 (0.2437) 0.1856*** (0.0000) dummyso 0.0051 (0.2381) 0.0051 (0.4389) -0.0674*** (0.0000) lta -0.0076* (0.0888) 0.0182*** (0.0001) 0.0284* (0.0891) rg -0.0145** (0.0476) 0.0153 (0.2258) -0.1563 (0.1315) roa 0.0018 (0.8283) 0.0030 (0.6030) -0.0002 (0.9957) constant 0.1049** (0.0169) -0.0562 (0.1894) -0.0925 (0.5024) observation 40 40 40 r2 0.3706 0.3273 0.4530 note: ***, **, * denote significance at 1%, 5% and 10% significant levels respectively. npl = non-performing loans ratio, car = capital adequacy ratio, lr = liquidity ratio, state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 81 board_i = board independence dummy, dummyso = state ownership dummy, board_i * dummyso = interaction of board independence dummy and state ownership dummy, lta = logarithm of total assets, rg = revenue growth and roa = return on assets. however, the board independence is found to have no significant impact on car. hypothesis 2 of this study could not be rejected as the board independence significantly affects the impact of ownership by malaysian government on banks’ risk-taking behaviour. evidently, the risk-taking measurements show that the two risk-taking proxies out of the three are significant. the impact of board independence and the degree of state ownership is presented in table 4. the finding in table 4 is consistent with table 3 where greater state ownership involvement in banks is associated with a statistically significant higher lr of 0.5133 percent. table 4. the effect of board independence and degree of state ownership on banks’ risk-taking variables npl car lr board_i -0.0111** (0.0476) 0.0020 (0.7130) -0.1960*** (0.0004) board_i * so 0.0237 (0.1018) -0.0195 (0.2498) 0.5133*** (0.0000) so 0.0146 (0.2556) 0.0196 (0.2483) -0.2491*** (0.0000) lta -0.0104* (0.0792) 0.0165*** (0.0003) 0.0130 (0.6215) rg -0.0118 (0.2786) 0.0160 (0.1517) -0.1545** (0.0430) roa -0.0012 (0.8911) 0.0063 (0.2855) -0.0257 (0.5562) constant 0.1369** (0.0163) -0.0451 (0.2525) 0.1541 (0.5269) observation 40 40 40 r2 0.3061 0.3252 0.4167 note: ***, **, * denote significance at 1%, 5% and 10% significant levels respectively. npl = non-performing loans ratio, car = capital adequacy ratio, lr = liquidity ratio, board_i = board independence dummy, so = state ownership percentage, board_i * so = interaction of board independence dummy and state ownership percentage, lta = logarithm of total assets, rg = revenue growth and roa = return on assets. 82 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 4.4 foreign director and state ownership on banks’ risk-taking table 5 presents the regression result of the effect of foreign directors on the relationship between state ownership and banks’ risk-taking. the result shows that sobs, on average, have a lower npl ratio compared to the pobs by a magnitude of 0.0230 percent when there is the presence of foreign director on board. however, the foreign director is found to have no significant impact on car and lr. since the findings of two risk-taking proxies out of the three are not significant, hence hypothesis 3 is rejected. the impact of foreign director and the degree of state ownership is presented in table 6. the finding in table 6 is consistent with table 5 where greater state ownership involvement in banks is associated with a statistically significant lower npl ratio of 0.0719 percent. table 5. the effect of foreign director and state ownership on banks’ risktaking variables npl car lr foreign_d 0.0168** (0.0345) 0.0152 (0.1930) 0.0805*** (0.0033) foreign_d* dummyso -0.0230*** (0.0007) -0.0169 (0.2376) -0.0641 (0.2883) dummyso 0.0313*** (0.0006) 0.0152 (0.2930) 0.0882 (0.1633) lta -0.0021 (0.3946) 0.0203*** (0.0000) 0.0716*** (0.0063) rg -0.0121 (0.1764) 0.0135** (0.0127) -0.0782 (0.4354) roa 0.0081 (0.4246) 0.0153 (0.3339) -0.0178 (0.6625) constant 0.0158 (0.2143) -0.1080*** (0.0000) -0.6863*** (0.0058) observation 40 40 40 r2 0.4348 0.3732 0.2545 note: ***, **, * denote significance at 1%, 5% and 10% significant levels respectively. npl = non-performing loans ratio, car = capital adequacy ratio, lr = liquidity ratio, foreign_d = foreign director dummy, dummyso = state ownership dummy, foreign_d * dummyso = interaction of foreign director dummy and state ownership dummy, lta = logarithm of total assets, rg = revenue growth and roa = return on assets. state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 83 table 6. the effect of foreign director and degree of state ownership on banks’ risk-taking variables npl car lr foreign_d 0.0279** (0.0171) 0.0241 (0.1972) 0.1188* (0.0660) foreign_d* so -0.0719*** (0.0011) -0.0515 (0.2391) -0.2306 (0.2641) so 0.0866*** (0.0011) 0.0506 (0.2361) 0.2043 (0.3008) lta -0.0020 (0.4287) 0.0196*** (0.0000) 0.0929*** (0.0093) rg -0.0128 (0.1948) 0.0140*** (0.0013) -0.0966 (0.4293) roa 0.0039 (0.6866) 0.0170 (0.2675) -0.0522 (0.1860) constant 0.0077 (0.5498) -0.1119*** (0.0000) -0.9005** (0.0168) observation 40 40 40 r2 0.3943 0.3731 0.2471 note: ***, **, * denote significance at 1%, 5% and 10% significant levels respectively. npl = non-performing loans ratio, car = capital adequacy ratio, lr = liquidity ratio, foreign_d = foreign director dummy, so = state ownership percentage, foreign_d * so = interaction of foreign director dummy and state ownership percentage, lta = logarithm of total assets, rg = revenue growth and roa = return on assets. 4.5 female director and state ownership on banks’ risk-taking table 7 presents the regression result of the effect of female director in the relationship between state ownership and banks’ risk-taking. the result shows that when there is presence of female director on board, it has a significant negative impact on lr with a coefficient of 0.0859 at 95 percent confidence level. however, the female director is found to have no significant impact on npl and car when it interacts with state ownership. on the other hand, the result in table 8 shows that greater state ownership involvement in banks is associated with a significantly higher car of 0.0287 percent. hypothesis 4 of this study is rejected as the evidence by the risk-taking measurements shows that the two risk-taking proxies out of the three are not significant. 84 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 4.6 findings from the regression results shown above, firstly, we find significant and positive relationships between the state ownership and non-performing loans (npl) ratio. as compared to the pobs, sobs are associated with a significantly higher npl ratio. this indicates that the sobs tend to take higher risks in providing loans. this finding is consistent with the social lending view where the sobs tend to focus on developing nations and allocating too much funds to projects that comes with high social returns and improving general welfare. however, we could see a significant positive impact on capital adequacy ratio (car) when there is interaction from the degree of state ownership. this finding suggests that a higher degree of state ownership tends to improve the level of capital in banks. on the other hand, we do not find significant impact on banks’ liquidity ratio (lr). this finding is unexpected and suggests that the malaysian government ownership in banks does not have a direct relationship with the banks’ liquidity risk. one possible explanation is that the state shareholders concern more on the banks’ solvency rather than the banks’ liquidity position. while the shareholders are pursuing on pecuniary interests, they may have overlooked the banks’ liquidity position. this might increase any potential financial distress in the future. secondly, we expand the study to examine the effect of corporate governance mechanisms in the relationship between state ownership and banks’ risk-taking through board independence. the results show that sobs with board independence ratio more than or equal to 0.50 are associated with higher npl ratio. consistent with the social lending view, the presence of ownership by the malaysian government significantly affects the decision making made by the board, incurring higher risk in lending out loans. although the ownership by the malaysian government has no direct impact on banks’ lr, it has impacts on banks’ lr through the channel of board independence. the evidence shows that independent directors in sobs attribute to lower liquidity risk. we further test and find that a higher degree of state ownership has a greater impact on lr. this finding is supported by rosenstein and wyatt (1990), klein (2002) and nguyen and nielsen (2010), suggesting that the intervention from malaysian government and outside directors has foster efficiency in supervising and advising functions in terms of liquidity assets and liabilities. hence, reduce the likelihood of liquidity risk. thirdly, we examine the effect of foreign director in the relationship between state ownership and risk-taking behaviour. the results show that the presence of foreign director in sobs is associated with lower credit risk due to the need to adhere to risk management policies and procedures designed and implemented by the banks. gillian and starks (2003) revealed that foreign director plays a more active role in upholding better firm-level governance in reducing information asymmetry and strengthening a firm’s transparency which may have an impact on the firm’s investment policy. moreover, we could see risk reduction to a greater extent when the malaysian government ownership is state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 85 higher. although we do not find evidence on the interaction of foreign director and state ownership on lr, we find similar finding in the individual effect of foreign director. the result suggests that the presence of foreign director tends to improve the banks’ liquidity. lastly, we examine the effect of female director and the results show that sobs are associated with higher liquidity risk with the presence of female director. however, there is evidence that banks’ capital adequacy tends to improve when there is a higher degree of intervention by the malaysian government. a possible explanation is the adoption of basel iii accords in malaysia. although the implementation only begins in 2013, the malaysian government has imposed strict regulations to ensure banks’ managers have attended to the requirements increment and started to adhere in placing a better capital prior to the full adoption of basel iii accords. being comply with stricter capital requirements, it shows that the car in malaysian banks can be further improved in years to come. in short, sobs’ risk-taking could be mitigated through cg mechanisms. although there is no strong evidence to support across all three risk-taking proxies in this study, we find that the banks’ liquidity is strengthened with the presence of independent directors. moreover, the credit risk is reduced with the presence of foreign director on board. this enlightens the fact that effective bank governance from the board of directors’ supervision is somehow crucial in mitigating malaysian banks’ risk-taking. table 7. the effect of female director and state ownership on banks’ risktaking variables npl car lr female_d -0.0002 (0.9433) -0.0110** (0.0104) 0.0804*** (0.0000) female_d * dummyso 0.0022 (0.5430) 0.0101 (0.1753) -0.0859** (0.0493) dummyso 0.0082 (0.1278) -0.0103* (0.0577) 0.0910*** (0.0006) lta -0.0061 (0.2693) 0.0231*** (0.0000) 0.0247 (0.4224) rg -0.0097 (0.1317) 0.0159* (0.0907) -0.0980 (0.5013) roa -0.0034 (0.6499) 0.0027 (0.6559) -0.0750* (0.0953) constant 0.0902* (0.0932) -0.1020*** (0.0100) -0.0784 (0.7815) observation 40 40 40 r2 0.2912 0.3645 0.2858 86 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 note: ***, **, * denote significance at 1%, 5% and 10% significant levels respectively. npl = non-performing loans ratio, car = capital adequacy ratio, lr = liquidity ratio, female_d = female director dummy, dummyso = state ownership dummy, female_d * dummyso = interaction of female director dummy and state ownership dummy, lta = logarithm of total assets, rg = revenue growth and roa = return on assets. table 8. the effect of female director and degree of state ownership on banks’ risk-taking variables npl car lr female_d 0.0012 (0.7712) -0.0147** (0.0116) 0.1158*** (0.0011) female_d * so -0.0007 (0.9496) 0.0287* (0.0906) -0.2603 (0.1015) so 0.0241 (0.2196) -0.0276 (0.1239) 0.2427*** (0.0036) lta -0.0078 (0.2329) 0.0229*** (0.0000) 0.0304 (0.4801) rg -0.0086 (0.3111) 0.0162* (0.0578) -0.1025 (0.5284) roa -0.0055 (0.4784) 0.0036 (0.5743) -0.0887* (0.0861) constant 0.1078* (0.0750) -0.0977** (0.0299) -0.1524 (0.7081) observation 40 40 40 r2 0.2546 0.3549 0.2745 note: ***, **, * denote significance at 1%, 5% and 10% significant levels respectively. npl = non-performing loans ratio, car = capital adequacy ratio, lr = liquidity ratio, female_d = female director dummy, so = state ownership percentage, female_d * so = interaction of female director dummy and state ownership percentage, lta = logarithm of total assets, rg = revenue growth and roa = return on assets. 4.7 robustness checks the panel regression model is reexamined using the generalized method of moments (gmm) estimation since the results presented earlier may subject to endogeneity and unobservable time-invariant individual effects. following roodman (2009), this paper employs the two-step system gmm. the robustness result is presented in table 9. the gmm result shows that most of the signs are consistent with the baseline result presented in table 2, only the level of significance differs. the robustness result strengthens the finding of hypothesis state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 87 1a. this is evidence by greater credit risk and lower level of capital. besides, the control variables become more significant after the inclusion of lag variables. the result highlights that the banks that are larger in size and the banks that are more capable at earning the revenue using assets, are exposed to lower risks. table 9. robustness check variables npl car lr (1) (2) (3) (4) (5) (6) dummyso 0.0669*** 0.0130 -0.0413*** -0.0847 0.3682*** -0.7731 (0.0000) (0.9112) (0.0006) (0.4103) (0.0020) (0.3867) dummyso* so 0.1925 0.5267 8.6280 (0.8311) (0.2373) (0.1410) so -0.2441 -0.4523 -9.8971* (0.8070) (0.1775) (0.0861) lta -0.0375*** -0.0177 0.0622*** 0.0613*** -0.0670 0.0704 (0.0023) (0.6614) (0.0001) (0.0021) (0.6179) (0.7541) rg -0.1048 -0.1206 0.1160 0.5014*** -0.1961 -1.2725 (0.4863) (0.4594) (0.4422) (0.0009) (0.6989) (0.4215) roa -0.0483** -0.0245 0.0341** 0.0096 0.3999*** 0.0503 (0.0114) (0.6869) (0.0364) (0.8045) (0.0022) (0.8697) constant 0.3460*** 0.2161 -0.4866*** -0.5948*** 0.1933 0.8273 (0.0041) (0.5591) (0.0086) (0.0084) (0.9020) (0.7642) ar(1) test statistic -0.5200* -0.7985** -0.2983* -1.4643* -0.6734* -1.5671* (0.0603) (0.0425) (0.0766) (0.0943) (0.0501) (0.0871) ar(2) test statistic -1.1620 -0.2008 -1.2582 -1.0004 -0.0872 -0.5830 (0.2452) (0.8409) (0.2083) (0.3171) (0.9305) (0.5599) hansen test of 2.4087 4.1272 2.3782 1.7791 1.5028 3.3094 overidentification (0.6610) (0.8455) (0.6666) (0.9871) (0.8262) (0.9135) note: ***, **, * denote significance at 1%, 5% and 10% significant levels respectively. npl = non-performing loans ratio, car = capital adequacy ratio, lr = liquidity ratio, dummyso = state ownership dummy, dummyso *so = interaction of state ownership dummy and state ownership percentage, so = state ownership percentage, lta = logarithm of total assets, rg = revenue growth and roa = return on assets. 88 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 5. conclusion and implications the objective of this study is to investigate how state ownership affects banks’ risk-taking behaviour with a sample of banks in malaysia. in summary, we find that state government incurred higher risk-taking behaviour in creating and lending out loans. this is consistent with the social lending view which shows that state-owned banks are more likely to focus on developing nations and lending out loans to project that comes with high social returns. the robustness result strengthens this finding, indicating that state-owned banks are associated with higher risk-taking behaviour. however, the banks’ risk-taking can be lower with more effective governance from the board of directors. also, there is evidence that a higher degree of state ownership has a more significant impact on the banks’ risk-taking behaviour. in facing poor earnings prospects in the current challenging economic environment, this paper shares important implications for malaysia’s banking industry. the government involvement through glics ownership in local commercial banks was seen as an important move to mitigate the increasing downside risks to malaysia’s economic development. this can be done through maintaining an optimal level of decision making. in relation to the significant impact of how excessive risk-taking behaviour by banks has an effect on the national economy in the previous financial crises, the malaysian banks should mitigate high risk-taking particularly in relative to loans portfolio. on the other hand, the bank management should remain focused on maintaining a healthy liquid position by building an efficient capital structure and liquidity management in banks. even the findings show that state shareholders concern more on banks’ solvency, thus, the importance of liquidity position placed in banks should not be neglected. the ease availability of liquid funds is crucial at all times. last but not least, the findings of this paper suggest that the supervision from the board of directors, somehow plays a crucial role in reducing the banks’ risk-taking behaviour. reviewing the board of directors’ structure can help to monitor the banks’ activities and tackle the flaw of corporate governance mechanisms which is very relevant in the context of malaysia where state-owned banks are one of the drivers of the economic development. 6. endnotes the data are extracted from malaysia banking industry report 1h2011 1 published by emerging markets direct. state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 89 shares ownership data are extracted from bursa malaysia banks’ 2015 2 annual report. 7. acknowledgments this work was supported by the usm fellowship provided by the institute of postgraduate studies, universiti sains malaysia. references abdul rahman, n. a., & rejab, a. f. (2013). the effect of risk taking on ownership structure and bank performance: a malaysia case. international review of business research papers, 9(6), 68-82. abdul razak, n. h., ahmad, r., & aliahmed, h. j. (2008). government ownership and performance: an analysis of listed companies in malaysia. corporate ownership and control, 6(2), 434-442. abdul razak, n. h. (2011). government ownership and corporate performance: a comparative analysis of malaysian and singaporean listed companies (doctoral thesis, university of malaya, kuala lumpur, malaysia). university of malaya research repository. retrieved from http:// studentsrepo.um.edu.my/id/eprint/1703 adams, r. b., & ferreira, d. (2009). women in the boardroom and their impact on governance and performance. journal of financial economics, 94(2), 291–309. anderson, r. c., mansi, s. a., & reeb, d. m. (2004). board characteristics, accounting report integrity, and the cost of debt. journal of accounting and economics, 37(3), 315-342. anginer, d., demirguc-kunt, a., huizinga, h., & ma, k. (2018). corporate governance of banks and financial stability. journal of financial economics, 130(2), 327-346. athanasoglou, p. p., brissimis, s. n., & delis, m. d. (2008). bank-specific, industry-specific and macroeconomic determinants of bank profitability. journal of international financial markets institutions and money institutions and money, 18(2), 121–136. 90 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 barr, r. s., seiford, l. m., & siems, t. f. (1994). forecasting bank failure: a non-parametric frontier estimation approach. recherches economiques de louvain/louvain economic review, 60(4), 417–429. berger, a. n., clarke, g. r., cull, r., klapper, l. f., & udell, g. f. (2005). corporate governance and bank performance: a joint analysis of the static, selection, and dynamic effects of domestic, foreign, and state ownership. journal of banking & finance, 29(8-9), 2179–2221. berger, a. n., hasan, i., & zhou, m. (2009). bank ownership and efficiency in china: what will happen in the world’s largest nation? journal of banking & finance, 33(1), 113–130. brandao-marques, l., correa, r., & sapriza, h. (2020). government support, regulation, and risk taking in the banking sector. journal of banking & finance, 112(march), article 105284. chin, y. w. (2015). changes of ownership and identities of malaysian banks: ethnicity, state and globalization (copenhagen discussion papers no. 51). retrieved from https://www.econstor.eu/bitstream/10419/208649/1/ cbs-cdp2015-51.pdf cornett, m. m., guo, l., khaksari, s., & tehranian, h. (2010). the impact of state ownership on performance differences in privately-owned versus state-owned banks: an international comparison. journal of financial intermediation, 19(1), 74–94. de haan, l., & van den end, j. w. (2013). bank liquidity, the maturity ladder, and regulation. journal of banking & finance, 37(10), 3930–3950. demirguc-kunt, a. (1989). deposit-institution failures: a review of empirical literature. economic review, 25(4), 2–18. dong, y., liu, z. t., shen, z., & sun, q. (2016). does state ownership really matter in determining access to bank loans? evidence from china's partial privatization. pacific-basin finance journal, 40(part a), 73-85. du, x., jian, w., & lai, s. (2017). do foreign directors mitigate earnings management? evidence from china. international journal of accounting, 52(2), 142–177. ezat, a., & el-masry, a. (2008). the impact of corporate governance on the timeliness of corporate internet reporting by egyptian listed companies. managerial finance, 34(12), 848–867. gillan, s. l., & starks, l. t. (2003). corporate governance, corporate ownership, and the role of institutional investors: a global perspective. journal of applied finance, 13(2), 4-22. state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 91 global financial development report. (2019/2020). bank regulation and supervision a decade after the global financial crisis. world bank. retrieved from https://openknowledge.worldbank.org/bitstream/handle/ 10986/28482/9781464811487.pdf gomez, e. t. (2005). the state, governance and corruption in malaysia. in nicholas tarling (ed.), corruption and good governance in asia (pp. 214-244). london: routledge. gomez, e. t., padmanabhan, t., kamaruddin, n., bhalla, s., & fisal, f. (2018). minister of finance incorporated: ownership and control of corporate malaysia. singapore: palgrave macmillan. gulamhussen, m. a., & santa, s. f. (2015). female directors in bank boardrooms and their influence on performance and risk-taking. global finance journal, 28, 10-23. haggarty, l., & shirley m. m. (1997). a new data base on state-owned enterprises. the world bank economic review, 11(3), 491-513. hamid, a. a. (2011). the corporate governance structures of glcs and nglcs and firm performance in malaysia. social and management research journal, 8(2), 1-28. hossain, m., jain, p. k., & mitra, s. (2013). state ownership and bank equity in the asia-pacific region. pacific-basin finance journal, 21(1), 914-931. iannota, g., nocera, g., & sironi, a. (2007). ownership structure, risk and performance in the european banking industry. journal of banking & finance, 31(7), 2127–2149. iannotta, g., nocera, g., & sironi, a. (2013). the impact of government ownership on bank risk. journal of financial intermediation, 22(2), 152176. imbierowicz, b., & rauch, c. (2014). the relationship between liquidity risk and credit risk in banks. journal of banking & finance, 40, 242–256. jensen, m. c., & meckling, w. h. (1976). theory of the firm: managerial behavior, agency cost and ownership structure. journal of financial economics, 3(4), 305–360. jiraporn, p., chatjuthamard, p., tong, s., & kim, y. s. (2015). does corporate governance influence corporate risk-taking? evidence from the institutional shareholders services (iss). finance research letters, 13, 105-112. karim, m. a., hassan, m. k., hassan, t., & mohamad, s. (2014). capital adequacy and lending and deposit behaviors of conventional and islamic banks. pacific-basin finance journal, 28, 58–75. 92 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 klein, a. (2002). audit committee, board of director characteristics, and earnings management. journal of accounting and economics, 33(3), 375–400. koudstaal, m., & wijnbergen, s. v. (2012). on risk, leverage and banks: do highly leveraged banks take on excessive risk? (ssrn working paper 2170008). retrieved from https://papers.tinbergen.nl/12022.pdf krugman, p. (2009). the return of depression economics and the crisis of 2008. new york: w. w. norton & company. laeven, l., & levine, r. (2009). bank governance, regulation and risk taking. journal of financial economics, 93(2), 259–275. lassoued, n., sassi, h., & ben rejeb attia, m. (2016). the impact of state and foreign ownership on banking risk: evidence from the mena countries. research in international business and finance, 36, 167–178. lee, a.-x., & hooy, c.-w. (2020). banks’ risk-taking and state ownership: evidence from asian emerging markets. malaysian journal of economic studies, 57(1), 59–80. liang, q., xu, p., & jiraporn, p. (2013). board characteristics and chinese bank performance. journal of banking & finance, 37(8), 2953-2968. liu, y., wei, z., & xie, f. (2014). do women directors improve firm performance in china? journal of corporate finance, 28, 169-184. naima, l., houda, s., & mouna, b. r. a. (2016). the impact of state and foreign ownership on banking risk: evidence from the mena countries. research in international business and finance, 36, 167–178. najid, n. a., & rahman, r. a. (2011). government ownership and performance of malaysian government-linked companies. international research journal of finance and economics, 61(1), 1450-2887. nguyen, b. d., & nielsen, k. m. (2010). the value of independent directors: evidence from sudden death. journal of financial economics, 98(3), 550–567. oecd. (2006). policy brief on corporate governance of banks in asia. paris: oecd publishing. oecd. (2013). oecd investment policy reviews: malaysia 2013. paris: oecd publishing. raheja, c. g. (2005). determinants of board size and composition: a theory of corporate boards. journal of financial and quantitative analysis, 40(2), 283–306. state ownership and risk-taking behaviour: evidence from malaysia's banking industry: 65-94 93 reinhart, c. m., & rogoff, k. s. (2011). from financial crash to debt crisis. american economic review, 101(5), 1676–1706. roodman, d. (2009). how to do xtabond2: an introduction to difference and system gmm in stata. stata journal, 9(1), 86–136. rosenstein, s., & wyatt, j. g. (1990). outside directors, board independence, and shareholder wealth. journal of financial economics, 26(2), 175–191. samaha, k., dahawy, k., hussainey, k., & stapleton, p. (2012). the extent of corporate governance disclosure and its determinants in a developing market: the case of egypt. advances in accounting, 28(1), 168–178. shaban, m., & james, g. a. (2018). the effects of ownership change on bank performance and risk exposure: evidence from indonesia. journal of banking & finance, 88, 483–497. shehzad, c. t., haan, j. d., & scholtens, b. (2010). the impact of bank ownership concentration on impaired loans and capital adequacy. journal of banking & finance, 34(2), 399–408. shleifer, a., & vishny, r. w. (1986). large shareholders and corporate control. journal of political economy, 94(3), 461–488. shleifer, a., & vishny, r. w. (1994). politicians and firms. quarterly journal of economics, 109(4), 995–1025. stiglitz, j. e. (1993). the role of the state in financial markets. the world bank economic review, 7(suppl_1), 19-52. taufil-mohd, k. n., md-rus, r., & musallam, s. r. m. (2013). the effect of ownership structure on firm performance in malaysia. international journal of finance and accounting, 2(2), 75–81. uddin, m. h. (2016). effect of government share ownership on corporate risk taking: case of the united arab emirates. research in international business and finance, 36, 322–339. vo, x. v. (2018). do firms with state ownership in transitional economies take more risk? evidence from vietnam. research in international business and finance, 46, 251–256. world bank (1993). the east asian miracle: economic growth and public policy. retrieved from http://www.worldbank.org/curated/ en/975081468244550798/pdf/multi-page.pdf zhu, j., ye, k., tucker, j. w., & chan, j. k. c. (2016). board hierarchy, independent directors, and firm value: evidence from china. journal of corporate finance, 41, 262–279. 94 the international journal of banking and finance, vol. 15, no 2, 2020 : 65-94 zhu, w., & yang, j. (2016). state ownership, cross-border acquisition, and risktaking: evidence from china’s banking industry. journal of banking & finance, 71, 133–153. the contagion from the 2007-09 us stock market crash 67 the international journal of banking and finance, volume 8 (number 4) 2011: pages 67-81 the contagion from the 2007-09 us stock market crash arnulfo m. castellanos, francisco s. vargas and luis g. rentería universidad de sonora, mexico ____________________________________________________________ abstract the global financial crisis that took place during the period 2007-09 had its most prominent manifestation in the general stock market crash. this could be studied from the perspective of financial contagion, using a mathematical tool known as wavelets. this paper aims to assess the impact of the us stock market crash on other stock markets all over the world. as an initial point the assumption that the former was the epicenter of the global financial crisis stands out. in order to determine the existence of differentiated impacts that show the presence of inertial factors in different stock exchange markets, a filtering technique is used on stock market indexes to assess such impacts. the data series are worked out on different time scales in order to identify short and long term effects. key words: contagion, global stock market crash, united states, wavelets jel classification: g01, g15, g12, f36. _____________________________________________ 1. introduction in the summer of 2007 the first global financial panic of the 21 st century was sparked: see timemap of events in ariff (2012). in january 2009, asset losses of financial institutions were estimated at $700 billion. the global stock market losses exceeded thirty billion dollars. the stock market index of the five hundred largest companies in the world in the s&p500 had fallen 39 percent in 2008: the nasdaq stocks fell by 42 percent while that of dow jones by35 percent. the financing capacity of global financial system had declined significantly in 2008 relative to 2007. the issuance of securities in 2008 for debt funding fell 37.7 percent compared to the previous year, indeed an unprecedented decline in the last twelve years. the stock issuance was dramatically reduced by 73 percent for the same period. this implied a reduction in bond issues amounting to $269.86 billion and 148.1 billion dollars decline in stock castellanos et al.: the contagion from the 2007-09 us stock market crash 68 financing. the aim of this study is to evaluate the impact of this us-origin stock market crash on other major stock exchanges in the world. the main hypothesis is that stock market crash in the us was the epicenter of the global stock market crash. the analysis has been conducted by a technique of filtering the data using wavelets. the paper is organized as follows. previous works that have been written about the stock market using wavelets are briefly summarized in the next section. the filtering process of stock series using a discrete transformation of wavelet coefficients is outlined in section 3. the results of regression and pearson correlation coefficients are summarized and inferences drawn in section 4. some concluding remarks are given in the final section. 2. previous studies using wavelets capobianco (2004) uses search algorithms with wavelet shaped dictionaries to break down the scale of the dynamics of stock returns. the wavelet analysis is used to identify intraday periodicities of both one and five minutes, in the timeline. in order to find side effects out of returns in the stock markets at different time scales using wavelet analysis, fernandez (2004) worked out with stock markets in north america and found that there were side effects to latin america, emerging asia, far east countries and pacific markets. she also identifies side effects from europe and latin america to north american markets. to study the relationship between the markets of south korea and us, lee (2004) worked with wavelets. he used a multi resolution technique at different scales. a strong evidence of side effects and price volatility from the stock markets of developed countries to those of developing nations is reported in that study. vuorenmaa (2004) studied the stock share volatility of nokia using wavelet multi resolution analysis to find out that wavelet variance and covariance revealed a considerable amount of stock market activity in intraday levels. moreover, applying a rule of local scale and long memory to volatility, he found that the variation in long memory was supported in the medium term (months). with high-frequency financial data and a markov tree model, gençay and whitcher (2005) used wavelets to establish a new stylized fact about volatility: the low volatility of longer maturities is more likely to be followed by short period low volatility and it would not be the international journal of banking and finance, vol. 8, iss. 4 [2011], art. 4 69 case for a high volatility of shorter time horizons. this phenomenon is called asymmetric vertical dependence. 3. what are wavelets? one way to introduce wavelets is from the fourier analysis, which is a process to analyze components in a function. such an analysis could represent a sound, an electrical signal, a light beam, and so on. by considering time dependent functions, as might be the case of a measuring device that connects one side of the power meter from one’s house to record voltage transients to find that varies with time. the fourier transform is a mathematical procedure that follows the shape:     deff i 2 )( 2 1 )(     (1) it is a change of one function, f, to be named the signal, whose variable  is the time, to another function, f, whose variable is the frequency of oscillations . for instance, in the case of a light signal, this procedure allows us to extract the frequencies (patterns) that make up the signal, and its importance. however, if the signal is localized, fourier transform begins loosing quality because it does not help much to locate in what time interval a signal was allocated, in such a way that tends to be blind to the details located at specific interval spaces. one way to address the solution of this problem is by modifying the fourier transform to get an expression like the following:     deftgtf ti 2 )()( 2 1 ),(     (2) in the previous equation a function that places the signal in a time interval around the instant t is added. such an addition was introduced by dennis gabor in the 1940s. an improvement developed nearly forty years later is due to morlet and grossman, who modified the gabor transformation adding up the frequency as a factor multiplying the time difference, which resulted as the following equation: castellanos et al.: the contagion from the 2007-09 us stock market crash 70       deftgtf ti 2 )()( 2 1 ),(     (3) these kinds of transformations are called wavelet transformations and allow for a great variety of them, depending on the g function chosen. the above approach is useful when the ability to observe an event at any given time is feasible. in addition, such a process permits to manage time as a continuous variable. in contrast, when the signals are collections of data taken at specific moments, whose modification is not convenient or when it cannot be changed, it is better to use discrete wavelet transforms. in these cases the time variable is a collection of moments that are usually equally spaced, as is the case of the data used in economics. for discrete wavelet transforms several propositions are put forward. the first one is due to the hungarian mathematician alfred haar, who in 1909 outlined a system of square waves to pass or bypass certain portions of a signal. the haar wavelet is defined as follows:         otherwise tif t _0 10__1 )( (4) in order to place the square at different time intervals the following algebraic expression is used: n kynwith kt n t kn 20_,...2,1_ ),2()( ,   (5) which provides the following set of formula: if n = 1 the only option is {-1 +2 t}, which provides the instant t = 0.5 and it occurs around the square. if n = 2 options are {-1+4t, -2+4t, 3+4t}, which provide the instants t = 0.25, t = 0.5, t = 0.75. if n = 3 options are {-1+8t, -2+8t, 3+8t, -4+8t, -5+8t, -6+8t, -7+8t}, the moments t = 0.125, t = 0.25, t = 0.375, t = 0.5, t = 0.625, t = 0.75, t = 0.875 are provided. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 4 71 the allocation of squares at different moments to accept or delete frequencies at different instants is performed by the simple method of multiplying by 1 to pass up, or by 0 to remove them out. the graphs below give a rough idea of how to carry out this procedure. to sum up, a set of simple functions that take one and zero as value, can be arranged to pass a section of frequencies and to eliminate others. for example, the following figure shows how a set of frequencies counted on the horizontal axis, with weights computed on the vertical axis, can be canceled to filter them, i.e. do not pass up. firstly, a low-frequency filter is presented in figure 1. the shaded area on the left indicates that lower frequencies are eliminated when multiplied by zero. figure 1: low frequency filter subsequently, a high frequency graph is included as shown in the shaded area on the right of figure 2. figure 2: high frequency filter a combination of both filters allows the researcher to pass a defined band of frequencies. therefore, given that the filters have different sets of moments, wherein one can be selected under certain frequency band, it is possible to have the most important frequencies in different points on time. castellanos et al.: the contagion from the 2007-09 us stock market crash 72 it is possible to find details of frequencies of oscillation in time, what is difficult with the fourier transform. to understand the importance of this development, it is worth posing the following question: is it feasible to have a computer program such that a musical signal is received by a microphone, so that once you send an electrical signal to a system you can allocate the frequency to identify musical notes issued? if one wants to pass it on to writing musical language, it is also necessary to determine how long the note lasted to be automatically written on the score. the wavelets are designed to solve that problem, because due to its definition, includes a location in a time interval. whichever signal, the filtered process can provide frequencies on specific time intervals in the same way as a musician that listen to or imagine a musical piece and writes the musical score down indicating which note (frequency), which tune (time location) and an intensity (strong or soft sounds) are introduced. however, it is not possible to achieve full precision simultaneously both in time and frequency. as a result of a theorem called uncertainty principle, it happens that as long as more precision on frequency is achieved less accuracy in time location is reached and the reverse is true. for this reason, one of the results of wavelets uses, which is called spectrogram, locate frequencies on instants when they occurred. it has a property, when many time periods are involved, the frequency knowledge decline. conversely, when the number of periods decreases, frequency knowledge accuracy increases. likewise, an analogous situation is faced when looking at a digitalized picture in a computer, in a zoom out, some details are fainted. 4. pearson parameter the pearson parameter is useful to compute as the correlation degree between two statistical variables. assuming n measures of a variable x and n measures of any other variable y, the question to answer is: what is the correlation between both? such a parameter is symbolized as r and it is defined as 2 1 2 1 1 )()( ))(( yyxx yyxx r n i i n i i i n i i        (6) where x and y are the averages. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 4 73 by definition r is a parameter that takes values in between -1 and 1, in such a way that those two extremes indicate total correlation between two variables. while the minus sign shows that if one of the variables increases the other decreases, a positive sign points out that when one of the variables increases, the same occurs with the other. absence of correlation corresponds to r = 0. however, there exist intermediate values which provide with a number for a visual image as follows: if each pair of values of the random variables (xi, yi) is matched with a dot in the cartesian plane, a scatter of n points will show up. depending on the value of the pearson parameter, the scatter will look as follows: for r=1, the dots resemble a positive oriented line. for r=-1, results in a negative oriented line. for r=0, a cloud of unorganized dots shows up as follows: for r=0.4, a quite organized scatter is observed meanwhile with an r = -0.4, the following image is perceived: castellanos et al.: the contagion from the 2007-09 us stock market crash 74 in this paper is shown that if the values of x variable are the main oscillation frequencies of the new york stock exchange (independent variable) and y represents the oscillation frequencies of any overseas stock exchange, such frequencies are randomly distributed, with a correlation coefficient close to zero that look like spherical distributions. on the contrary, when observation periods are longer than daily, for instance, weekly, bimonthly or monthly, the distributions appear much more organized. the pearson parameter takes values around 0.9. 5. data and results comparison of stock exchange indexes encompasses the period of march the 15 th , 2006 till march the 9 th , 2009. in order to synchronize physical time, the data for continents different from america was lagged one day with respect to new york stock exchange (nyse) values. the coefficients of discrete transform wavelets were obtained for each stock exchange. here a graphic example of different levels of such coefficients for the american case is presented. the bottom part of graph no. 3 shows the dow jones index evolution. then, eight levels of wavelets coefficients are displayed, identified as wn. once the discrete transform wavelet coefficients were obtained, the computations were carried out in order to calculate the pearson correlation coefficients, so as to measure the synchrony, if there was any. the applied method allows for a selection of periods to measure frequencies. measurement was performed in six periods, each of which is called level. the studied period roughly includes: each working day as the first level, every two working days, on the second level, every four working days, for the third level, every eight working days, on the fourth level (two weeks approximately), every sixteen working days, on the fifth level (one month approximately) and every thirty two working days, on the sixth level (two months approximately). international journal of banking and finance, vol. 8, iss. 4 [2011], art. 4 75 figure 3: discrete transform wavelets (wn) obtained from dow jones index (period: march the 15 th , 2006 to march the 9 th , 2009) 0 500 1000 1500 8 0 0 0 1 2 0 0 0 y t t 2 w1 t 2 w2 t 3 w3 t 3 w4 t 3 w5 t 3 w6 t 3 w7 t 3 w8 t 2 v8 the results were used to calculate the oscillation frequency correlation of each stock exchange to the nyse values. with this purpose a linear regression was run. the method used was to approximate a straight line with ordinary least squares. then the pearson parameter was calculated so that to search dispersion around the approximated line. based on the previous output, a qualitative concept called tune up, that shows the synchrony of the stock exchanges with the nyse was introduced. the following results were found: in the first level, calculated frequencies weakly resemble a straight line, which means that the pearson parameter, in absolute value, is greater than one tenth only for a limited number of cases, such as mexico, chile, peru, canada, germany, hong kong, czech republic and austria. the remaining countries show such a high dispersion that pearson parameters lies below one tenth. on the second level, correlations between frequencies are similar. as a result, the tune up for every two days between the overseas stock exchanges and nyse has not grown significantly. the tune up tends to grow for longer periods. for instance, in the third level there are 21 stock exchanges whose pearson parameter is over one tenth in absolute value. castellanos et al.: the contagion from the 2007-09 us stock market crash 76 table 1: regression coefficients (wnb) with the main stock exchanges and pearson correlation coefficients (wncp) for every level of discrete transform wavelet coefficient (wn), of world stock exchanges with respect to the dow jones index pai country code w1b w1cp w2b w2cp w3b w3cp w4b w4cp w5b w5cp w6b w6cp egypt ccsi 0.038 -0.026 0.034 0.031 0.022 0.037 -0.010 -0.016 0.102 0.236 0.020 0.076 argentine mval 0.012 0.062 0.032 0.162 0.016 0.065 -0.014 -0.050 -0.025 -0.102 0.113 0.494 mexico ipc 0.432 0.183 0.581 0.189 0.573 0.171 1.200 0.374 1.943 0.558 2.370 0.860 brazil vbpa 0.209 0.032 2.130 0.285 -0.544 0.067 -0.775 -0.108 0.200 0.029 2.174 0.358 chile ipsa 0.026 0.140 0.003 0.011 0.057 0.170 0.005 0.014 -0.019 -0.060 0.110 0.485 peru igvl 0.148 -0.106 0.186 0.107 -0.093 0.038 -0.363 -0.143 0.093 0.039 1.316 0.555 canada gtse 0.132 0.124 0.074 0.064 0.225 0.180 -0.150 -0.104 23.865 0.221 0.301 0.341 germany dax 0.058 0.107 0.035 0.057 -0.101 0.160 0.021 0.027 -0.312 -0.452 0.478 0.861 venezuela ibc 0.009 0.002 0.180 0.037 -0.430 0.083 0.230 0.033 -0.121 -0.026 1.767 0.351 japan n225 0.034 0.026 0.150 0.096 0.242 0.141 0.105 0.078 0.429 0.245 0.244 -0.173 indonesia jkse 0.004 -0.019 0.019 0.082 0.032 0.094 0.030 0.123 0.058 0.218 0.001 -0.005 hong kong h s i 0.254 -0.102 0.065 0.024 0.822 0.262 1.308 0.440 1.954 0.656 2.662 0.830 india bsn 0.009 0.006 0.097 0.047 -0.021 0.009 0.054 0.021 -0.892 -0.355 0.100 -0.050 taiwan tw 0.026 -0.042 0.009 0.012 0.028 0.029 0.147 0.218 0.001 0.001 0.179 -0.191 singapore sti 0.006 -0.025 0.019 0.077 -0.012 0.037 0.042 0.144 0.005 0.014 0.166 0.559 china sse 0.006 -0.013 0.002 0.004 -0.059 0.102 0.072 0.131 0.019 0.035 0.075 0.118 filipinas psei 0.000 0.000 0.008 0.026 0.042 0.106 0.007 0.019 0.036 0.102 0.143 -0.396 pakistan kse 0.007 -0.008 0.002 0.001 -0.073 0.049 0.028 0.023 -0.156 -0.096 0.471 -0.246 malaysia klse 0.001 0.020 0.003 0.032 -0.021 0.188 0.018 0.180 0.021 0.181 0.031 0.342 sri lanka cse 0.001 -0.011 0.019 0.113 0.020 0.104 0.018 0.082 -0.008 -0.034 0.080 0.354 south korea ks11 0.009 0.065 0.036 0.197 -0.024 0.138 -0.010 -0.045 -0.011 -0.059 0.068 0.418 denmark kfmx 0.091 -0.041 0.007 0.018 0.029 0.070 -0.014 -0.036 -0.059 -0.141 0.051 -0.265 greece ase 0.203 0.050 0.001 0.002 0.052 0.112 -0.006 -0.016 -0.018 -0.057 0.046 0.249 switzerland ssmi 0.025 0.043 0.147 0.204 0.025 0.034 0.444 0.567 0.591 0.773 0.550 0.909 slovakia sax 7.232 -0.037 6.890 0.028 40.323 0.154 66.493 0.247 78.385 0.229 9.879 0.046 russia rts 0.016 -0.083 0.006 0.024 0.058 0.184 0.043 0.131 0.060 0.184 0.134 0.613 czech rep. px 0.014 0.106 0.043 0.257 0.012 0.070 0.067 0.424 0.152 0.729 0.171 0.902 portugal ps20 0.002 -0.003 0.021 0.029 -0.019 0.023 0.127 0.140 -0.179 -0.202 0.071 -0.074 norway oseax 0.002 -0.038 0.002 0.028 0.014 0.203 -0.013 -0.223 0.016 0.238 0.019 0.357 sweden ompi 0.000 -0.008 0.002 0.048 0.012 0.357 0.000 0.001 0.001 0.018 0.014 0.457 n. zealand nz50 0.009 0.052 0.009 0.042 0.073 0.313 0.141 0.550 0.210 0.768 0.256 0.919 italy mtel 0.016 0.009 0.018 0.009 -0.641 0.278 0.162 0.065 -1.286 -0.508 1.731 0.815 spain ibex 0.051 0.049 0.032 0.030 0.030 0.024 -0.051 -0.048 -0.055 -0.071 0.225 -0.197 united kingdom ftse 0.001 -0.001 0.107 0.194 0.055 0.102 -0.111 -0.182 -0.204 -0.384 0.484 0.925 international journal of banking and finance, vol. 8, iss. 4 [2011], art. 4 77 belgium bfx 0.026 -0.096 0.074 0.219 0.113 0.324 0.105 0.280 0.064 0.160 0.055 0.151 austria atx 0.039 -0.105 0.000 0.001 0.006 0.012 0.174 0.369 0.226 0.379 0.407 0.883 netherlands aex 0.003 -0.095 0.006 0.152 -0.001 0.029 0.000 -0.003 -0.021 -0.352 0.030 0.608 france fchi 0.031 -0.073 0.100 0.209 -0.038 0.081 -0.018 -0.037 -0.174 -0.342 0.260 0.640 israel ta 0.003 0.038 0.006 0.075 0.009 0.088 0.005 0.045 -0.043 -0.418 0.058 0.572 australia aord 0.022 0.054 0.078 0.161 0.181 0.367 0.127 0.245 -0.198 -0.407 0.272 0.577 essentially, the number of countries is threefold as much as for the first level. moreover, the value of the pearson parameter tends to grow. there are stock exchanges that do not show any tune up in the first two levels. however, there is some tune up in the third level. the stock market is synchronized all over the world in an approximated period of two months. in effect, there are only five stock exchanges whose pearson parameter lies below one tenth in absolute value. in the opposite case, for instance, we have that the mexican stock exchange has 0.860, the czech republic 0.902, and the united kingdom 0.925. in the previous pages, it was found the average frequencies of oscillation of the world's stock market became alike. this is a clear trend towards homogenization of the frequencies. the phenomenon could be clearer if a frequency transformation were made, so that each economic frequency e were associated with a light frequency, v. the procedure is similar to the one used to change fahrenheit to celsius degrees and vice versa. the transformation is: v e+r (7) where a = 61199.2 is a number that is calculated as follows: calculate the difference between the maximum frequency (pattern 2.1) visible and low frequency visible (pattern 2.7). next, the difference between the maximum and minimum economic frequency is calculated. patterns are allocated based on the following table 2. the standard deviations of the regressions of various indices of stock exchanges in the world against the u.s. dow jones converted as variations in pattern range are shown. it is observed that the longer the time span, the larger the homogeneity degree of oscillations, which is clearly reflected in the predominance of one pattern. castellanos et al.: the contagion from the 2007-09 us stock market crash 70 table 2: allocation of patterns international journal of banking and finance, vol. 8, iss. 4 [2011], art. 4 map no. 1. map no. 3 map no. 5 map no. 2 map no. 4 map no. 6 6. concluding remarks the first financial crises of the 21st century that took place during 2007-09 disclosed several drawbacks. the global stock market underwent considerable losses. capital generation diminished to unprecedented levels. the main hypothesis of this paper is that us stock market was the epicenter of the world market crash. nyse movements on other stock markets is studied using the wavelets mathematical tool to assess the effects. frequencies of oscillation for the nyse as well as stock exchanges of 40 other countries were calculated and filtered. such frequencies were correlated between us and other stock markets. it was found that nyse movements exerted a great influence, during the first two working days, in a limited group of countries such as mexico, chile, peru, canada, germany, hong kong, czech republic and austria. after eight working days, it is observed that 21 stock exchanges (see map of the world) were significantly affected. castellanos et al.: the contagion from the 2007-09 us stock market crash 71 2 the most important concluding remark is that the stock market is synchronized all over the world in an approximated period of two months. author information: arnulfo m. castellanos, the submitting author, is a professor at the universidad de sonora in mexico: e-mail: castell@fisica.uson.mx. the two coauthors are staff members at the same university. they are respectively, francisco s. vargas (fvargas@guaymas.uson.mx) and luis g. rentería (lrenter@guaymas.uson.mx). international journal of banking and finance, vol. 8, iss. 4 [2011], art. 4 1 mailto:fvargas@guaymas.uson.mx mailto:lrenter@guaymas.uson.mx 3 references ariff, m., (2012). the origin of the global financial crisis: an alternative view, in m ariff, j farrar, and a khalid (eds) (2012). regulatory failure and global financial crisis. edward elgar, uk and usa. pp 15-35. arouri, mohamed el hedi; fredj jawadi y duc khuong nguyen. (2010). the dynamics of emerging stock markets. empirical assessments and implications. heidelberg: springer-verlag. capobianco, e., (2004). multiscale analysis of stock index return volatility. computational economics 23(3): 219–237. fernandez, v.. (2004). time-scale decomposition of price transmission in international markets. working paper 189, center for applied economics, university of chile, santiago, chile. gençay, r., and fan, y.. (2006). unit root and cointegration tests with wavelets. manuscript, simon fraser university, canada. gençay, r. f., and whitcher, b., (2001). an introduction to wavelets and otherfiltering methods in finance and economics. san diego, ca: academic press. gençay, r. f., and whitcher, b., (2003), asymmetry of information flow between volatilities across time scales. unpublished manuscript, simon fraser university, canada. gençay, r. s. f., and whitcher, b., (2005). multiscale systematic risk. journal of international money and finance 24(1): 55–70. lee, h., (2004). international transmission of stock market movements: a wavelet analysis. applied economics letters 11: 197–201. renaud, o., starck, j.l., and murtagh, f., (2003). prediction based on multiscale decomposition. international journal of wavelets, multiresolution and information processing 1(2): 217–232. vuorenmaa, t., (2004). a multiresolution analysis of stock market volatility using wavelet methodology. bank of finland, discussion paper. castellanos et al.: the contagion from the 2007-09 us stock market crash the time-varying nature of the overreaction effect: evidence from the uk 1 international journal of banking and finance, volume 8 (number 3), 2011: pages 1-36 the time-varying nature of the overreaction effect: evidence from the uk panagiotis andrikopoulos, arief daynes and paraskevas pagas de montfort university, university of portsmouth and university of portsmouth, united kingdom _____________________________________________ abstract previous studies on the overreaction effect in the uk show that prior losers consistently outperform prior winners in the period 1975 to 1990. this paper extends current knowledge by assessing the above phenomenon in the uk market for the period 1987 to 2007. in contrast to earlier research, we produce evidence of a weak presence of the overreaction effect for the latest test period. further, we show that, after adjusting for size, the overreaction effect almost disappears and any additional excess post-formation return to prior-losers is attributable to market cycles. this study implies that the presence of the overreaction effect in the uk stock market is time-varying and difficult to exploit in practice. keywords: overreaction, stock market efficiency, small-size effect, time-variation, behavioral finance jel classification: g14, g32 ____________________________________________ 1. introduction this study is on the overreaction hypothesis in the united kingdom (uk). one of the central hypotheses of behavioral finance is that stock prices systematically overreact. in their seminal study, de bondt and thaler (1985) show that stocks that have earned the highest positive abnormal returns during the pre-formation period tend to exhibit negative abnormal returns during the post-formation period and vice versa. moreover, it is further reported that stocks that have overreacted the most that is where the pre-formation period is longer and, hence, where the abnormal returns are more extreme tend to exhibit a correspondingly greater correction during the postformation period. the overreaction hypothesis is further corroborated in a follow up andrikopoulos et al.: overreaction effect 2 paper (de bondt and thaler, 1987), in which the authors conclude that the overreaction phenomenon persists even after controlling for size and risk. such findings pose significant questions on the validity of the efficient markets (emh) paradigm. if consistent mispricing phenomena such as those supported by de bondt and thaler (1985: 1987) do exist, market participants can devise profitable trading rules resulting in consistent above-average profits thereby violating emh‟s key assumption of investors‟ rationality. it poses questions regarding the extent to which the stochastic process of securities‟ market pricing may indeed represent a fair game. four main hypotheses have been advanced in the now larger and controversial literature on the above hypothesis. the first of these is the behavioral explanation given originally by de bondt and thaler (1985) namely that investors fail to act as ideal bayesian decision-makers, systematically overreacting to recent events such as companies‟ earnings announcements and under-reacting to prior information. analytically, in their study of stock price reversals in the united states (us) market for the period 1926 to 1982, de bondt and thaler (1985) report a difference in the cumulative average residuals between extreme prior losing and extreme prior winning securities of approximately 25 percent (24.6 percent, t-statistic of 2.20). these economically significant excess returns for prior losers are found to be persistent even after controlling for size effect and changes in the systematic risk element for prior losers (de bondt and thaler, 1987) indicating a consistency with the behavioral hypothesis of investor overreaction. this hypothesis was further corroborated by subsequent studies in the us and abroad (dissanaike, 1997; 2002; dreman and lufkin, 2000; nam, pyun and avard, 2001). the remaining three explanations come from within the efficient markets paradigm. the first of these is that the overreaction effect is encompassed by already existing anomalies and in particular the size and seasonal effects (zarowin, 1990; grinblatt and moskowitz, 1999). 1 by replicating the de bondt and thaler (1985) results with appropriate adjustments for risk and seasonal effects, zarowin (1990) shows that the mean reversion phenomenon is not due to investors‟ overreaction but 1 this is classified as an explanation within the efficient markets framework since the small firm and seasonal effects are often regarded as temporary anomalies within the wider emh research programme. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 3 rather due to size discrepancies between winners and losers. as the author concludes (zarowin, 1990 p.130) “…the winner vs. loser phenomenon found by de bondt and thaler appears to be another manifestation of the size phenomenon in finance”. this plausible relationship between the overreaction hypothesis and its size/seasonal explanation is also corroborated by clare and thomas (1995) in their study of the uk market. the second explanation focuses on the relationship between the reported overreaction premium and the time-varying risk characteristics for losing and winning portfolios (ball and kothari, 1989). in detail, as ball and kothari (1989) show, on an aggregate level 97.4 percent of the variation in pre-formation and postformation returns can be explained by analogous changes in the systematic risk component of such portfolios proxied by beta. this hypothesis was further corroborated by chen and sauer (1997). in their study on the overreaction effect in the us for the period 1926 to 1992, the authors show that the winner-loser relationship is vastly inconsistent over their selected four time-regimes (a) pre-war period, b) 1940s-1950s c) pre-energy-crisis d) post-energy-crisis) while the overall effect appears to be non-stationary and mostly related to alternative market cycles. as the authors conclude (chen and sauer, 1997, p.63) “…during economic downturns, losers go down faster and deeper than winners. on the other hand, losers go up faster than winners during economic upturns. during periods of economic stability, losers perform just as well as winners and there are little abnormal profits for the arbitrage portfolio”. 2 this non-stationary nature of the overreaction effect poses an important challenge to the applied aspect of any strategy aiming to exploit such an effect. an arbitrage portfolio consisting of buying past losers and shorting past winners is unlikely to be exploitable and, in particular, might even result in zero or negative abnormal returns for extended periods of years or even longer. in addition, in line with the efficient markets paradigm, the observation of an apparent anomaly in one time period might have little predictive power for either its persistence or reversal in a subsequent period as subsequent empirical evidence suggests that the market over 2 in chen and sauer‟s (1997) study, the arbitrage portfolio returns are defined as the return differential between a long position in the extreme loser portfolio and a short position in the extreme winner portfolio. andrikopoulos et al.: overreaction effect 4 extended periods under-reacts as often as it overreacts (fama, 1998). this is especially valid in the case of the overreaction effect due to its link with the size/calendar anomaly whose time-varying nature is documented in prior research (dimson and marsh, 1999; andrikopoulos et al., 2008). contrary to these studies, subsequent research on the overreaction effect in the us and abroad failed to corroborate the relationship between size/seasonal effects and time-varying risk with the winner-loser mean reversion hypothesis. in their study on the us market chopra et al. (1992) provide evidence of an economically significant overreaction effect of approximately 5 percent per year even after controlling for size and risk. in addition, as the authors documented, the effect appears to be asymmetric amongst larger and smaller companies with extreme winner and loser deciles suggesting that the underlying effect could be caused by the difference in the investment patterns between individual and institutional investors (chopra et al., 1992 p.262). these results were further corroborated by dissanaike (2002, p.152) who suggests that “…the size and winner-loser effects are not completely independent of each other, but there is no evidence to suggest that the size effect subsumes the winner-loser effect”. the third and last main explanation is that the overreaction effect is, at least to some extent, spurious, and is caused by bias in the data or in the methods used in computing returns. survivorship bias in the data may create a spurious overreaction effect by excluding bankrupt stocks, but may also understate a genuine overreaction effect by excluding takeover premiums on acquired companies not contained in the data set (dissanaike, 1997). in addition, computing returns by summing monthly abnormal returns causes an upward bias in the post-formation returns of past losers (conrad and kaul, 1993) while an apparent anomaly may also be created by data mining or methodology mining (fama, 1991). it is noted, however, that the criticism of survivorship bias has little force for most us studies, since the crsp returns data is generally regarded as being free from survivorship bias and any weaknesses and/or omissions from the crsp returns data appear to have little impact on the results or main conclusions. moreover, as boynton and oppenheimer (2006) show for the us market, even after correcting for both survivorship and micro-structure distortions from the bid-ask bounce biases, international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 5 past-losers outperform past-winners by an average annual return of 8.3 percent during the sample-period 1926 to 2002. however, what is most interesting in this most recent study on the us market is that contrarian strategies appear to have a neutral performance in more recent years (boynton and oppenheimer, 2006, pp.2626-2627) indicating that the overreaction effect may have gone into reverse. the rest of the paper deals with the uk scenario as this study is about overreaction in the uk. the next section is a brief review of the findings from the uk market. data and computation issues are described in section 3 and the findings are then presented and discussed in sections 4 and 5: the latter section is about adjusting the results for size effect. finally, this paper concludes in section 6 with a summary of the main findings and their empirical implications. 2. the overreaction effect in the uk a similar picture of the persistence of the overreaction effect has been also documented in the uk using out-of-sample data. using a data set covering the period 1955 to 1990, clare and thomas (1995) report an annual out-performance of 1.7 percent for portfolios consisting of prior losers compared to that of prior winners. according to statistical evidence reported in table 1 in this paper, the performance reported in clare and thomas‟ (1995) study is supportive of most prior findings for the us market, with a strong momentum for the short-term period following portfolio formation and a reversal afterwards (de bondt and thaler, 1985; 1987). nonetheless, this effect becomes marginal and statistically insignificant after controlling for size. hence, their conclusion is similar to that of zarowin (1990) that the overreaction effect is just another manifestation of the size effect. these results were corroborated by campbell and limmack (1997). using a dataset covering the period 1979 to 1990, the authors provide evidence of a strong momentum effect in the first twelve months following portfolio formation and subsequently an economically and statistically significant reversal in the fortunes of prior losers for a period of up to five years (campbell and limmack, 1997, p. 544). in addition, as their results suggest, the reversal documented was restricted only to the smallest „loser‟ andrikopoulos et al.: overreaction effect 6 companies indicating that this evident long-term overreaction is asymmetric and could be explained by the tax-loss selling hypothesis. these findings were further confirmed by george and hwang (2007) in their study of the hong-kong market. subsequent studies of the overreaction effect in the uk appear to provide an alternative picture. the studies of dissanaike (1997; 2002) are of particular value in that it is substantially free from the biases discussed earlier for the us studies. the research is based on a data set constructed by the author that appears to be free from survivorship and look-ahead biases. in addition, the buy and hold returns metric is used to avoid upward bias in the post-formation returns of past loser stocks, as documented in conrad and kaul (1993), while restricting the data set to large and liquid stocks within the ftse500 index minimised the impact induced by bid-ask spreads and infrequent trading. as the overreaction hypothesis has mostly been studied in the us context, dissanaike (1997) also addresses the problem of datasnooping bias by carrying out tests on out-of-sample uk data. table 1: prior findings on the overreaction effect in the uk (w-l%) notes: this table provides a summary of prior literature on the overreaction hypothesis in the uk. all results are based on equally weighted portfolios of prior „losers‟ and „winners‟. campbell and limmack‟s (1997) study uses as a sample the complete universe of the risk measurement service compiled by london business school. the study of clare and thomas (1995) is based on a random selection of 1000 securities in each calendar year and for all years under examination. w stands for „winners‟ and l stands for „losers‟. the coding system used for the description of the adopted methodologies is as follows: bhar-buy and hold abnormal returns; car-cumulative abnormal returns; q – based on quintiles; # – abnormal returns based on capm; ddeciles; ~ aggregate t-statistic not provided by the author; c – aggregate results based on a 48 month pre-formation period. *, **, *** indicate significance at the 10%, 5% and 1% level respectively. study date meth/logy sample period t+1 to t+12 (w-l%) t+1 to t+24 (w-l%) t+1 to t+36 (w-l%) t+1 to t+48 (w-l%) t+1 to t+60(w -l%) clare and thomas 1995 car q,# 1955-1990 +0.37 -1.68 ** -1.57 * -//-//car q,#,sa 1955-1990 +0.57 -1.13 *** -0.83 campbell and limmack 1997 car d,# 1979-1990 +10.07 *** -4.44 -9.35 * -//-//car d,sa -//+11.11 *** -0.69 +0.20 dissanaike 1997 bhar d 1979-1988 -0.05 ~,c -0.09 ~ -0.31 ~ -0.99 ~ international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 7 the main conclusion of this study is that the overreaction effect existed in the uk market for the period 1979 to 1988, thus validating prior evidence that the effect is both economically and statistically significant. in a follow-up paper the author also provided evidence to show that the overall overreaction effect, although not independent of the size effect, is not subsumed under the latter, results that are difficult to reconcile with the joint market efficiency and capm hypothesis (dissanaike, 2002). this paper reports the findings of a study to extend our current knowledge on the overreaction effect in the uk market by assessing chen and sauer‟s (1997) hypothesis that the overall effect is associated with alternative market cycles and more specifically that prior losers‟ abnormal performance is mostly triggered during periods of economic recovery. following this initial hypothesis, the study also addresses the relationship between the overreaction and size effects. as recent evidence suggests, the small size effect in the uk appears to be non-stationary in the post-1990 period and highly sensitive to the return measurement methodology adopted (dimson and marsh, 1999; andrikopoulos et al., 2008). hence, under all prior findings that the overreaction and small-size effects are interrelated, we would expect a similar reversal in the performance for the latter in the post-1990 period. compared to prior overreaction studies in the uk that were restricted by limited samples or used an ftse500 data set and covered periods only up to 1990, this study examines the presence of overreaction in the uk market using a bias-free dataset comprised of the entire universe of securities in the official listing of the london stock exchange for the entire period 1987 to 2007. as the data set used here is different from and was constructed independently of those used in all previous studies, it can provide further evidence on the effect in the uk limiting possible data snooping bias that may have been present in prior studies using almost identical data sets and covering similar periods. previewing the main results and conclusions, this paper finds weak evidence for the overreaction hypothesis for the uk market for the 1987 to 2007 period. considering that this paper and dissanaike‟s (1997; 2002) studies control for the main sources of bias discussed in the literature, the overall conclusion is that a genuine overreaction effect exist in the uk from 1975 to the late 1980s and andrikopoulos et al.: overreaction effect 8 weakened subsequently. this is consistent with the claim that the overreaction premium is compensation for time-varying risk and with the arguments of fama (1991; 1998), that many anomalies can be merely the result of data mining and that regularities observed in one period are equally likely to persist or reverse in subsequent periods. 3. data and return computation 3.1 the data compared to prior uk studies on the overreaction effect that used the london share price database as the main source of data for dataset selection (clare and thomas, 1995; campbell and limmack, 1997; dissanaike, 1997; 2002), our data set is obtained from the combination of two alternative sources. most of our out-of-sample data are obtained from uk equity data (uked). 3 this data set comprises of the entire universe of uk securities that have been fully listed on the london stock exchange at any time during the period july 1987 to march 2004, excluding investment companies and investment trusts. the data selected are free from survivorship bias and look-ahead bias. for each company, data are given for the entire period covered by the file, not merely for the period when the company was fully listed. the data files used in the present study are the total returns file and the market-listing file. uked‟s total returns file gives monthly returns for each stock, adjusted for dividends, capital changes, mergers and acquisitions and other events impacting upon the computation of monthly total returns and covers the twenty-two year period april 1982 to april 2004. the market listing file also covers the same period and contains for each month the end-of-month market listing and trading status for each stock. as the present study covers the period january 1987 to december 2007, all december 1986 to july 1987 and post-april 2004 market listings and returns data were obtained from the london share price database (lspd) currently maintained and updated by london business school. 3 a detailed discussion of the characteristics of the uked is presented in andrikopoulos et al. (2008). international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 9 to avoid any errors in the reconciliation of the data, all securities were matched manually using sedol numbers and official company names and the entire universe of stocks was further re-validated by a cross examination of all post-2004 securities against the lehmann communications company guide and pinsents company guide for accuracy and re-confirmation of their listing status. in total, for the period december 1986 to december 2006, when the last portfolio was formed, there were 3063 securities officially listed shares on london‟s main market. from this data set, 235 securities were excluded as they were either associated with non-ordinary securities and/or those shares that were found to be suspended, de-listed or traded under rule 535.2, rule 4.2(a) or rule 163(2) at each of the portfolio formation dates. figure 1: qxl ricardo stock price and bhars series notes: this figure illustrates closing prices and buy-and-hold returns of qxl ricardo plc for the period october 1999 to december 2007. buy-and-hold abnormal returns for qxl ricardo plc are estimated as                        1111 48 1 , 48 1 ,48, k t tftseall k t tqxlkqxl rrbhar where k refers to the holding period of k=48 months taking a long position on the 1 st of january 2004 and selling the stock on 31december 2007. all stock closing prices and returns for the company are obtained from the london share price database (lspd) maintained and updated by london business school. returns on the ftse all-share index are obtained from thomson‟s datastream international. the large decrease in the company‟s share price in december 2006 was caused by a 20-for-1 stock split. all returns are adjusted for dividend payments and capital changes. in addition, further examination of the properties of the dataset lead us to exclude one particular security, qxl ricardo plc. our rationale for such a decision 0 2000 4000 6000 8000 10000 12000 14000 o c t9 9 d e c -9 9 f e b -0 0 a p r0 0 ju n -0 0 a u g -0 0 o c t0 0 d e c -0 0 f e b -0 1 a p r0 1 ju n -0 1 a u g -0 1 o c t0 1 d e c -0 1 f e b -0 2 a p r0 2 ju n -0 2 a u g -0 2 o c t0 2 d e c -0 2 f e b -0 3 a p r0 3 ju n -0 3 a u g -0 3 o c t0 3 d e c -0 3 f e b -0 4 a p r0 4 ju n -0 4 a u g -0 4 o c t0 4 d e c -0 4 f e b -0 5 a p r0 5 ju n -0 5 a u g -0 5 o c t0 5 d e c -0 5 f e b -0 6 a p r0 6 ju n -0 6 a u g -0 6 o c t0 6 d e c -0 6 f e b -0 7 a p r0 7 ju n -0 7 a u g -0 7 o c t0 7 d e c -0 7 0% 5000% 10000% 15000% 20000% 25000% 30000% closing price bhar andrikopoulos et al.: overreaction effect 10 was the unique performance of this company mostly caused by the enormous amount of speculative trading associated with this stock during the period 2004-2005, mostly a result of ongoing speculation regarding the solvency of the business and a potential takeover bid fed by the continuous increase in the equity holdings in the business of two competing investment banks. as figure 1 illustrates, including this company in any of our stock portfolios would misrepresent its true return properties as a long position in this security starting on the 1 january 2004 and held up to the end of the examined period would have yielded a unique outlier return of 25,513 percent. our final sample comprises of 2,827 securities representing 92.30 percent of the entire universe of fully-listed securities. figure 2 illustrates the distribution of the selected sample over the period under examination. figure 2: sample selection and official listing coverage. notes: this figure summarises the distribution of the selected sample. from a total universe of 3063 securities officially listed on london‟s main market during the period december 1986 to december 2006, we exclude 235 securities as they were either associated with non-ordinary securities and/or those shares that were found to be suspended, de-listed or traded under rule 535.2, rule 4.2(a) or rule 163(2) during each of the portfolio formation dates. the final sample comprises of 2827 securities (after excluding qxl ricardo plc), representing 92.30% of the entire universe of fullylisted securities. the largest number of stocks included in a portfolio formation is reported for december 1986 with 1583 securities actively traded. nonetheless, the highest 1 ,3 5 4 1 ,3 9 8 1 ,4 0 7 1 ,3 7 8 1 ,3 4 9 1 ,3 3 1 1 ,3 0 6 1 ,3 1 0 1 ,3 9 5 1 ,5 2 2 1 ,5 8 3 1 ,5 7 3 1 ,5 1 3 1 ,3 6 5 1 ,2 2 3 1 ,1 7 3 1 ,0 7 0 9 7 0 9 1 5 8 1 8 7 7 0 9 4 .9 % 9 4 .9 %9 8 .1 % 9 7 .0 %9 6 .6 %9 6 .6 %9 2 .5 %9 6 .7 % 9 5 .9 %9 4 .3 % 9 3 .9 % 9 4 .9 % 9 1 .2 % 9 2 .5 % 9 5 .3 % 9 5 .4 % 9 4 .7 % 9 4 .4 % 9 3 .8 % 9 2 .0 % 9 1 .9 % 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 excluded selected sample international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 11 percentage of coverage is reported for december 2004 with 98.1 percent of all actively-traded officially-listed securities included in the study. what is most interesting from these statistics is the falling trend of ukregistered companies listed on london‟s main market in the post-1999 period. the inflation of the dot.com bubble in the middle of 1998 led to an increasing preference for companies in this newly established sector to become listed on aim rather than on the main market. nonetheless, following the bursting of the bubble, the trend did not reverse resulting in a number of uk companies remaining in the main market consistently falling from 1,578 securities listed in december 1998 to 811 in december 2006. 4 3.2 the buy-and-hold abnormal returns model because of the impact on the results produced from alternative return measurement methodologies and to allow comparison with prior research in the uk, this study uses two methods for computing returns. the first of these two, bhars, is the methodology recommended in conrad and kaul (1993) and adopted in the studies of dissanaike (1997; 2002). in addition, due to the reduced transaction costs associated with such an investment strategy, its use especially within a uk perspective is more realistic in practical terms. 5 analytically, at the end of each calendar year all securities that were actively traded in the official listing of the uk market were sorted on the basis of their performance for the period of k=12,24,…,60 months prior to portfolio formation, with each k-month period referring to an alternative holding time-interval. using the market listing file of uked, we identified all stocks that were fully listed and trading in the market at the end of each calendar year and for which uked reported a monthly return for each of the k-month periods. for example, for the k=48 months in december 1986, only those securities that reported a return for the entire period january 1983 to december 1986 were selected. the only exception is the k=60 months strategy for the december 1986 formation, where pre-formation 4 these numbers include only uk registered companies. since 2001, the london stock exchange‟s main market has attracted a large number of foreign companies, while uk registered companies have preferred to list on aim where listing criteria can be met more easily and an initial listing is more cost-effective. andrikopoulos et al.: overreaction effect 12 bhars are based on 57 observations, that is, starting in april 1982. the buy-andhold returns for each stock with monthly compounding were calculated as:    11 1 ,,    k t tiki rbhr (1) where, ki bhr , is the buy-and-hold return for security i for a holding period of k = 12, 24,…,60 months and, ti r , is the raw return on security i in month t. using the k=12,24,36,48 and 60 months pre-formation bhrs, all stocks were then sorted into equally-weighted deciles, with the number of stocks in each decile differing by up to one stock and ranked from highest to lowest k-month preformation return. decile one (d1) is therefore the ten percent of stocks with the highest k-month pre-formation bhrs, with decile ten (d10) containing the ten percent of stocks with the lowest k-month pre-formation performance. for each stock, post-formation buy-and-hold returns were computed using the returnestimation methodology in (1) for all alternative k=12, 24,…,60 months holding periods. post-formation returns for each of the ten decile-portfolios were then estimated as the simple average of the returns computed for each stock in each decile. 6 this is algebraically formulated as             n i k t tikp r n bhr 1 1 ,, 11 1 (2) where           11 1 , k t ti r is the k-months bhrs for security i, and ti r , is the raw return in security i at month t. monthly returns for all stocks that became de-listed during the post-formation 5 buy-and-hold post-formation decile returns equal the returns on a strategy of buying the equally weighted portfolio of all stocks in the decile on the formation date, reinvesting dividends in the stock paying the dividend and investing any proceeds from de-listed stocks equally across the remaining stocks in the portfolio. bhars‟ transaction costs are much lower than those of cars, not only because of much less frequent trading, but also because the long-term buy-and-hold strategy adopted in this study would allow the investor to minimise the price impact by building holdings in smaller stocks over a period of time. thus, if a large overreaction effect is found using the bhar methodology, these profits would probably be exploitable in the uk stock market. 6 these post-formation portfolio returns were computed under the assumption that equal-weighted portfolios are constructed on the portfolio-formation date and held throughout the post-formation periods. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 13 period were replaced by the average monthly return of the remaining stocks in the decile portfolio. the only exceptions were those stocks that became valueless. we classified a stock as valueless whenever there was a subsequent uk inland revenue pronouncement that the stock was deemed to have become valueless and we assigned the –1 return to the month in which the stock was suspended for the last time. 7 this procedure corresponds to a strategy of investing any proceeds from a dead stock equally across the remaining stocks in the decile-portfolios. overall, the study examined twenty-one k=12 non-overlapping formation periods starting from december 1986 and ending in december 2006, twenty k=24, nineteen k=36, eighteen k=48 and seventeen k=60 months overlapping portfolio formation periods, generating a total of 950 investment portfolios for the entire examined period 1january 1987 to 31 december 2007. finally, the assessment of post-formation momentum profits and the „winner-loser‟ effect is reported in terms of buy-and-hold abnormal returns. this is estimated as the k-month difference between the average kp bhr , of each momentum portfolio and the buy-and-hold returns of the ftse all-share index, or kftsekpkp bhrbhrbhar ,,,  (3) where kftse bhr , is the k-month buy-and-hold returns on the index calculated as           11 1 , k t tftse r . as prior uk studies on the overreaction effect have reported a relationship between „winner-loser‟ and „size‟ effects (clare and thomas, 1995; dissanaike, 2002), this study also addressed this issue by adopting a methodology similar to that introduced in lakonishok et al. (1994). analytically, at the end of each calendar year and for the entire sample period all securities included in our dataset were sorted on the basis of market capitalisation. all stocks were then assigned to decile portfolios with decile one (d1) consisting of the largest 10% of stocks and decile ten (d10) of the smallest. post-formation average buy-and-hold returns for all size portfolios were 7 the adoption of this methodology will help us avoid introducing survivorship and sample selection bias into our results. these problems were found to be present in conrad and kaul‟s (1993) methodology and are thoroughly investigated and discussed in loughran and ritter (1996). andrikopoulos et al.: overreaction effect 14 calculated as in (2). finally, the k-month size-adjusted sa kip bhar ,, for all portfolios was then estimated as the difference between the post-formation ki bhr , of each security and their corresponding matching size-portfolio size kp bhr , for k=12, 24, 36, 48 and 60 post-formation holding periods, or                     n i k t ti k t ti sa kip r n rbhr 1 1 , 1 ,,, 11 1 11 (4) moreover, average size-adjusted buy-and-hold abnormal returns, sa kip bhar ,, were estimated following a similar methodology to that described in (3). overall, if the „size-effect‟ is subsumed within the „winner-loser‟ effect as dissanaike (2002) suggested, the adoption of this methodology will allow the exclusion of common returns elements between the two and provide a better picture of the pure performance of the mean reversion effect in the uk during the 1990s. as regards the estimation of statistical significance, the t-statistics for the k=12 months holding period were computed using the conventional t-statistic methodology given as   st  (5) where  is the mean difference in the k-month average buy-and-hold abnormal returns between winners and losers,  l k w k bharbhar  ,  is equal to zero,  s is the standard error of the mean difference estimated as w and w is the sum of weights. however, as the portfolio formation periods for the k=24, 36, 48, and 60 months investment strategies overlap, the conventional t-statistic methodology would lead to overestimated t-values and possibly erroneous inferences. to solve this problem, statistical significance for the „winner-loser‟ return differences for the k=24, 36, 48, and 60 months was computed using the fama and macbeth (1973) procedure with certain adjustments for the relevant k-th order autocorrelation. hence, similar to prior studies (chopra et al., 1992; dissanaike, 2002), the t-statistic for the international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 15 kp abhar , difference between „winners‟ and „losers‟ for the k=60 months holding period is estimated as in (5) but we adjust  s for fourth-order autocorrelation as 4321 )4(2)3(2)2(2)1(2     nnnnn n s (6) where n = 17 (december 2002 being the last portfolio formation period ),  is the standard deviation of the time series of the average buy-and-hold returns‟ differential between winners and losers  l k w k bharbhar  and, t is the estimated nth order autocorrelation coefficient. a similar methodology is also adopted for the remaining k=24, 36 and 48 months holding periods. 3.3 the cumulated abnormal returns model the second methodology for calculating momentum abnormal returns is similar to the one adopted in de bondt and thaler (1985). cumulative monthly portfolio abnormal returns were obtained by taking the average abnormal returns for all stocks in the portfolio, which corresponds to an assumption of monthly rebalancing by forming an equally weighted portfolio at the end of every month. nonetheless, due to the presence of uk stamp duty of 0.5 percent which is payable on all stock purchases, realistically, this strategy is highly cost-inefficient. in addition, small capitalisation stocks in the uk market are in general much smaller and less liquid than in the us. hence, as past-losers in particular might contain a very high proportion of highly illiquid penny shares, any possible overreaction effect under car is unlikely to be exploitable in the uk because of transaction costs. in addition, in contrast to the buy-and-hold returns methodology, the monthly rebalancing process used in the car methodology will tend to put more weight on current performance leading to a different classification of companies with extreme recent performances into „winning‟/„losing‟ deciles compared to that obtained using the former method. nonetheless, as loughran and ritter (1996, p.1963) argue “…the buy-and-hold method provides a sharper distinction between portfolios when classifying firms. however, once the portfolios are selected, cars and buy-and-hold returns give rise to similar empirical conclusions”. andrikopoulos et al.: overreaction effect 16 although the buy-and-hold returns approach is the preferred method for the above methodological and empirical reasons, the use of cars will allow comparison with those results reported in earlier work in the us and uk markets (de bondt and thaler, 1985; 1987; clare and thomas, 1995; campbell and limmack, 1997; chen and sauer, 1997). in detail, for each calendar month prior to the portfolio formation, abnormal stock returns were calculated as the difference between the monthly raw return on each stock and the monthly return of the average of all stocks actively trading in the uk equity market or          n i tititi r n rarr 1 ,,, 1 (7) cumulative abnormal returns (car) are defined as the sum of the abnormal monthly returns notated as:    k i tiki arrcar 1 ,, (8) where ti arr , is the abnormal raw returns on each security i at time t, and k is the number of months prior to the portfolio formation date, k = 12, 24,…, 60 months. all equally weighted momentum deciles were constructed by sorting all actively-traded stocks at the portfolio formation date on the basis of prior k-months ki car , with decile one (d1) being the extreme prior-winners and decile ten (d10) the extreme prior losing stocks. similarly, the average abnormal post-formation decile monthly return is defined as                  )( 1 1 ,,, 1 )( 1 pu i n i tititp r n r pu aar (9) where         n i ti r n 1 , 1 is the average return on the market index for each calendar month t following portfolio formation, p is the decile portfolio formed at the end of december of each calendar year (p = d1, …, d10), u(p) is the number of stocks in p and where ti r , is the monthly return on stock i. finally, the cumulative abnormal post-formation decile return is the sum of the monthly average abnormal postinternational journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 17 formation decile returns or    k i tpkp aarcar 1 ,, (10) where tp aar , is the average abnormal returns on each momentum portfolio p at time t and k is the number of months after the portfolio formation date, k = 12,24,36,48 and 60. similarly to de bondt and thaler‟s (1985) methodology, our universe of stocks only includes all those with 60 months of unbroken pre-formation returns. the present study uses all stocks in the uked universe that have monthly returns throughout the pre-formation period, i.e. sixty continuous observations for the k=60 months investment horizon, forty-eight for the k=48 months one and so on. 4. results 4.1 results for the bhar methodology table 2 contains the averages of pre-formation and post-formation bhars for each decile, with the averages taken across all portfolio formation dates. table 2 reveals the extreme variations in pre-formation abnormal returns across past winner and past loser deciles and for all k-month holding periods. according to panel a, for the k=12 months test-period, the abnormal returns range from 129 percent for prior winners (d1) to a loss of 52.5 percent for prior losers (d10). as regards post-formation performances, the results suggest a consistent momentum effect with prior winners still outperforming prior losers. the difference in performance between (d1) and (d10) is 7.4 percent, a result which is only significant at the 15 percent level (t-value of 1.551). furthermore, these k=12 months post-formation returns increase almost uniformly from decile 1 to decile 10, while the same pattern is evident when individual formation dates are considered. a similar picture is reported in panel b which summarises the results from the k=24 months holding period. winners again appear to outperform losers up to the end of the buy-and-hold period with a total return differential of 2.4 percent. nonetheless, although these results are significant in economic terms these andrikopoulos et al.: overreaction effect 18 table 2: preand post-formation average kp bhar , on momentum portfolios for the period january 1987 to december 2007 panel a: average bhars for momentum deciles based on 12 months pre-formation performance d1 (w) d2 d3 …... d8 d9 d10 (l) diffd1-d10 n 126 126 126 126 126 126 k=[-12,0] 1.298 0.544 0.356 -0.132 -0.266 -0.525 1.822 *** k=[0,+12] 0.073 0.058 0.050 0.001 -0.005 -0.001 0.074 panel b: average bhars for momentum deciles based on 24 months pre-formation performance d1 (w) d2 d3 …... d8 d9 d10 (l) diffd1-d10 n 123 123 123 123 123 123 k=[-24,0] 2.473 0.965 0.638 -0.165 -0.346 -0.638 3.111 *** k=[0,+12] 0.045 0.048 0.054 0.023 0.031 0.037 0.008 k=[0,+24] 0.033 0.047 0.063 0.058 0.040 0.009 0.024 panel c: average bhars for momentum deciles based on 36 months pre-formation performance d1 (w) d2 d3 …... d8 d9 d10 (l) diffd1-d10 n 120 120 120 120 120 120 k=[-36,0] 3.624 1.369 0.892 -0.177 -0.394 -0.701 4.325 *** k=[0,+12] 0.017 0.050 0.029 0.028 0.041 0.040 -0.023 k=[0,+24] -0.012 0.079 0.039 0.075 0.092 0.070 -0.082 k=[0,+36] -0.066 0.053 0.047 0.062 0.123 0.075 -0.140 panel d: average bhars for momentum deciles based on 48 months pre-formation performance d1 (w) d2 d3 …... d8 d9 d10 (l) diffd1-d10 n 116 116 116 116 116 116 k=[-48,0] 5.098 1.914 1.252 -0.119 -0.370 -0.705 5.803 *** k=[0,+12] 0.007 0.040 0.035 0.061 0.073 0.046 -0.039 k=[0,+24] -0.017 0.041 0.031 0.101 0.127 0.058 -0.075 k=[0,+36] -0.054 0.038 0.022 0.107 0.132 0.077 -0.131 k=[0,+48] -0.091 0.023 -0.007 0.120 0.148 0.034 -0.125 panel e: average bhars for momentum deciles based on 60 months pre-formation performance d1 (w) d2 d3 …... d8 d9 d10 (l) diffd1-d10 n 113 113 113 113 113 113 k=[-60,0] 6.669 2.565 1.706 -0.014 -0.320 -0.691 7.360 *** k=[0,+12] -0.002 0.025 0.015 0.048 0.088 0.038 -0.040 k=[0,+24] -0.029 0.011 0.021 0.107 0.128 0.056 -0.085 k=[0,+36] -0.085 -0.014 0.018 0.129 0.152 0.031 -0.116 k=[0,+48] -0.109 -0.033 0.043 0.173 0.172 0.019 -0.129 k=[0,+60] -0.155 -0.085 -0.005 0.125 0.161 0.053 -0.208 * notes: this figure illustrates the difference in the buy-and-hold abnormal returns (bhars) between past winners and losers for all k=12,24,…60 months holding periods. all portfolios are calculated as follows: at the beginning of each calendar year for the period january 1987 to december 2007, our final dataset is sorted into deciles on the basis of their k-month pre-formation bhars performance. post-formation bhars for each decile-portfolio are estimated as the difference between the simple average of the returns computed for each stock in each decile and the bhrs for the ftse all-share index. all t-statistics are estimated using the fama and macbeth (1973) procedure and are adjusted for the relevant kth order autocorrelation. * indicates significance at the 10% level, ** indicates significance at the 5% level, *** indicates significance at the 1% level. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 19 differences are insignificant in statistical terms. these results confirm those findings reported in prior momentum literature (rouwenhorst, 1998; grinblatt and moskowitz, 1999) and they are consistent with de long et al.‟s (1990) argument that short-term momentum is not inconsistent with overreaction over 3 to 5-year horizons. for the remaining k=36, 48 and 60 months holding periods, according to panels c, d and e there is evidence of a „winner-loser‟ effect, as prior „losers‟ report a mean reversion in the post-formation period. the most immediate reaction on observing these post-formation returns is that they appear to be normal and unremarkable for both past winners and past losers and for all these three alternative post-formation holding periods. nonetheless, a closer examination reveals certain weak regularities in these post-formation returns figures. for example, according to all the panels c, d and e the post-formation returns for the extreme prior losers (d10) on average revert within the first year followed by a continuous and symmetrical increase from t+2 onwards (bhars‟ difference on k=12 of -2.27 percent for panel c, -3.93 percent for panel d and -4.05 for panel e). nonetheless, although these short-term reversals are significant in economic terms, they are not significant at all in statistical terms (unreported t-value of -0.401 for the k=36 months pre-formation strategy, t-values of -0.746 and -0.853 for the k=48 and k=60 months respectively). moreover, the reported out-performance of prior „losers‟ in these postformation periods appears not to be associated with a considerable improvement in their reported performance but rather appears to be associated with an asymmetric and detrimental deterioration in the market performance of the prior „winning‟ stocks. for example, according to panel e, in the k=60 buy-and-hold period there is an inconsistent performance for the prior „losers‟ portfolios by reporting an improvement for the first twenty-four months but then followed by a deterioration in the reported bhars in the k=36 to k=48 holding periods and a reversion afterwards. the reported difference in performance for k=60 months between winners and losers is -20.8 percent, significant at the 10 percent level (t-value of -1.952). as the results in table 2 indicate, there is no clear evidence that the uk market during the 1987 to 2007 period consistently overreacts in the way reported in andrikopoulos et al.: overreaction effect 20 previous overreaction studies (de bondt and thaler, 1985; dissanaike, 1997; 2002). for example, for the average bhrs for a k=48 months holding period dissanaike (1997, p. 32) reports excess abnormal returns of 98.9 percent for prior losers while the differences in bhars between the winner and loser portfolios are strongly significant in eight out of the ten formation periods. a possible explanation for such a large difference between our own results and the results of dissanaike (1997) might be the different dataset selected, with the former study restricted to the largest five hundred companies in the uk market. nonetheless, the more plausible explanation is that the effect varies over time. this time-variability nature of the effect is illustrated in figure 3. compared to most previous uk studies that cover periods up to 1990 (clare and thomas, 1995; campbell and limmack, 1997; dissanaike, 1997) and report an almost consistent appearance of the „winner-loser‟ effect, figure 3 shows prior-losing portfolios failing to outperform prior-winners in every year under examination. for example, in the k=48 months holding period, losers outperform winners in 10 out of 18 portfolio formation periods, with most of the aggregate bhars differential reported in the period following the dot.com bubble of 1998-2000. a clearer but equally inconsistent pattern is evident in the k=60 holding period, where the year-by-year results show extreme mean reversion for losers in the period 1990-1992 followed by a clear momentum for 1993 to 1996 and then a more extreme reversal in the post-1997 post-formation test-period. this latest reversal was caused by the bursting of the dot.com bubble and its effects on a large proportion of companies listed in this period and an increasing number of companies falling into bankruptcy/administration in the period 2000-2002. as the formation periods 1993 to 1996 report post-formation results from 1998 and up to 2001, the use of the buyand-hold methodology may have been affected by these failing stocks. on the contrary, the remainder of the prior-losing companies that survived the crash and the economic turmoil of the period soon recovered leading to the asymmetric post-formation performance reported for the december 1997 to december 2002 test-period. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 21 figure 3: average bhars‟ difference for “winner” and “loser” portfolios for all k-month holding periods (january 1987 to december 2007) 0 .1 1 0 0 .0 0 6 0 .1 3 3 0 .2 2 4 0 .3 1 8 0 .3 0 7 -0 .2 1 8 -0 .0 3 9 0 .1 7 6 0 .1 0 7 0 .1 4 8 0 .2 7 4 -0 .2 1 1 -0 .1 8 4 0 .0 3 0 0 .4 5 9 -0 .4 7 7 0 .0 0 2 0 .2 2 3 0 .2 1 5 -0 .0 4 4 -0.50 -0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 0.30 0.40 0.50 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 0 .0 9 4 0 .0 4 0 0 .0 6 5 0 .2 2 4 0 .3 5 3 0 .1 4 7 0 .4 5 7 0 .1 0 0 0 .3 7 5 0 .2 5 8 0 .2 4 1 0 .1 2 2 0 .4 5 7 0 .4 0 1 0 .7 9 2 0 .2 7 6 0 .4 7 5 0 .1 9 1 0 .4 8 8 0 .2 6 6 -0.90 -0.70 -0.50 -0.30 -0.10 0.10 0.30 0.50 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 -0 .3 0 0 -0 .0 9 5 -0 .2 0 1 -0 .0 4 5 0 .1 2 1 0 .0 6 2 -0 .4 2 1 0 .1 3 2 0 .1 0 1 0 .1 8 6 0 .0 7 8 0 .0 3 9 -0 .0 2 7 -0 .6 1 6 -1 .1 7 7 -0 .1 7 9 -0 .1 6 1 -0 .5 8 8 0 .4 2 4 -1.30 -1.10 -0.90 -0.70 -0.50 -0.30 -0.10 0.10 0.30 0.50 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 0 .1 5 9 0 .0 8 3 0 .2 4 1 0 .0 9 5 0 .1 6 6 0 .1 8 5 0 .3 0 0 0 .0 1 8 0 .0 3 4 0 .7 2 7 0 .0 3 6 0 .1 3 1 0 .3 1 1 0 .6 8 2 1 .3 0 6 0 .7 3 0 0 .0 1 9 0 .6 7 8 -1.50 -1.00 -0.50 0.00 0.50 1.00 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 -0 .1 5 3 0 .0 1 9 -0 .1 2 6 -0 .0 4 8 -0 .5 1 8 -0 .4 0 6 -0 .4 5 6 0 .1 5 2 0 .2 8 8 0 .3 0 6 0 .1 9 8 -0 .0 9 5 -0 .3 2 4 -0 .5 8 4 -0 .8 8 0 -0 .3 6 3 -0 .5 4 6 -1.00 -0.80 -0.60 -0.40 -0.20 0.00 0.20 0.40 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 lw k bhar  12 lw k bhar  48 lw k bhar  60 lw k bhar  24 lw k bhar  36 andrikopoulos et al.: overreaction effect 22 this hypothesis becomes even more plausible when examining table 3. this table reports the aggregate number of stocks in each decile and for each formation that became valueless during the post-formation holding period. clearly the valueless securities are heavily concentrated in the losing decile-portfolios with a maximum of 44.2 percent of all de-listed stocks occurring in the extreme past-loser deciles for the k=12 months period and a minimum of approximately 24.0 percent for k=60 months. table 3: distribution of valueless securities across decile portfolios and alternative k-month holding periods notes: this table reports the distribution of stocks becoming valueless during the post-formation period across deciles d1 (winners), d1,…, d10 (losers) and across all portfolio-formation periods. uked classifies a stock as valueless when there is a pronouncement by the uk inland revenue that the stock is deemed to be valueless. uked then assigns the –1 end-of-month return to the stock in that month when the stock is suspended for the last time. for the period after april 2004, all valueless securities were collected from the london share price database (lspd) master index file and cross-examined against the lehmann communications company guide and pinsents company guide. this pattern also indirectly reveals that the probability of a prior-losing portfolio being affected by a large number of bankruptcies is more likely to be reduced as the k-months holding period increases since most prior-losers at risk of liquidation will mostly go bust within the first thirty-six months subsequent to the formation of the portfolios. a way to validate this hypothesis is by examining the results from the cars methodology. as the use of cars assumes monthly rebalancing, the impact of valueless securities should be minimal as we should expect a stronger reversal in the post-formation performance of the „loser‟ portfolio and a clearer picture of a „winner-loser‟ effect. this is discussed in the next section. 4.2 results from the car methodology the average cars results are summarised in table 4. d1 (w) d2 d3 .... d8 d9 d10 (l) k=[0,12] 33 13 8 19 29 114 k=[0,24] 23 23 9 33 53 150 k=[0,36] 39 28 24 49 72 144 k=[0,48] 33 35 36 56 74 129 k=[0,60] 32 38 33 59 72 121 international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 23 table 4: preand post-formation acars on momentum decile-portfolios during the period january 1987 to december 2007 panel a: average cars for momentum deciles based on 12 months pre-formation performance d1 (w) d2 d3 ..... d8 d9 d10 (l) diffd1-d10 n 126 126 126 126 126 126 k=[-12,0] 0.839 -0.386 -0.236 0.214 0.359 -0.798 1.637 *** k=[0,+12] 0.014 -0.054 -0.016 0.002 0.036 -0.026 0.040 panel b: average cars for momentum deciles based on 24 months pre-formation performance d1 (w) d2 d3 ..... d8 d9 d10 (l) diffd1-d10 n 123 123 123 123 123 123 k=[-24,0] 1.260 0.538 0.322 -0.352 -0.566 -1.067 2.327 *** k=[0,+12] -0.031 -0.000 0.014 -0.022 -0.014 0.002 -0.033 k=[0,+24] -0.099 -0.052 -0.011 0.005 0.010 0.029 -0.128 panel c: average cars for momentum deciles based on 36 months pre-formation performance d1 (w) d2 d3 ..... d8 d9 d10 (l) diffd1-d10 n 120 120 120 120 120 120 k=[-36,0] 1.560 0.671 0.403 -0.423 -0.680 -1.323 2.882 *** k=[0,+12] -0.067 -0.011 -0.002 -0.006 0.007 0.043 -0.110 ** k=[0,+24] -0.164 -0.049 -0.020 0.009 0.059 0.074 -0.238 * k=[0,+36] -0.233 -0.080 -0.036 0.005 0.094 0.076 -0.310 ** panel d: average cars for momentum deciles based on 48 months pre-formation performance d1 (w) d2 d3 ..... d8 d9 d10 (l) diffd1-d10 n 116 116 117 117 116 116 k=[-48,0] 1.765 0.770 0.454 -0.475 -0.757 -1.451 3.216 *** k=[0,+12] -0.082 -0.018 -0.013 0.021 0.022 0.068 -0.150 ** k=[0,+24] -0.165 -0.055 -0.046 0.044 0.066 0.096 -0.261 ** k=[0,+36] -0.220 -0.084 -0.063 0.061 0.065 0.109 -0.329 *** k=[0,+48] -0.263 -0.112 -0.091 0.087 0.081 0.087 -0.350 *** panel e: average cars for momentum deciles based on 60 months pre-formation performance d1 (w) d2 d3 ..... d8 d9 d10 (l) diffd1-d10 n 113 113 113 112 113 113 k=[-60,0] 3.238 1.213 0.740 -0.411 -0.678 -1.237 4.475 *** k=[0,+12] -0.073 -0.023 -0.027 0.017 0.034 0.062 -0.134 *** k=[0,+24] -0.153 -0.063 -0.054 0.040 0.070 0.096 -0.249 *** k=[0,+36] -0.194 -0.110 -0.071 0.055 0.079 0.088 -0.282 *** k=[0,+48] -0.229 -0.124 -0.082 0.056 0.082 0.072 -0.301 *** k=[0,+60] -0.268 -0.169 -0.099 0.067 0.056 0.058 -0.326 *** notes: this table gives the average k-month pre-formation and k-month post-formation cumulative monthly abnormal return (car), with the average taken over all formations with the k-month preformation period. all t-statistics are estimated using the fama and macbeth (1973) procedure and are adjusted for the relevant k-th order autocorrelation. * indicates significance at the 10% level, ** indicates significance at the 5% level, *** indicates significance at the 1% level. for the methodological reasons discussed earlier, the average cumulative abnormal returns (car) results are of less interest than the bhar outcomes and will be andrikopoulos et al.: overreaction effect 24 discussed only briefly. these results largely corroborate the results reported in de bondt and thaler (1985). they also give evidence of a larger mean reversion for prior losers than those reported in prior uk research using the same return metric (clare and thomas, 1995; campbell and limmack, 1997) both in terms of direction and magnitude of the results as well as in their statistical significance. however, there are minor differences. the use of cars in this study has an asymmetric impact on the estimation of the pre-formation returns with more extreme pre-formation abnormal returns reported for prior losers (d10) than for prior winners (d1). as the de bondt and thaler (1985) study included only stocks with 84 months of unbroken pre-formation returns, it perhaps lead to the omission of some of the more extreme past losers and past winners. furthermore, in contrast to the buy-and-hold returns, the monthly rebalancing procedure in cars was also expected to lead to the overestimation of post-formation performance, especially in the k=48 and 60 month holding periods, as bankrupt stocks and their losses will be excluded/accumulated soon after they arise and they will not affect the performance at the end of the holding period. in detail, according to table 4, there is economically and statistically significant evidence of overreaction for the k=36, 48 and 60 months holding-periods. on average, the reversal in performance is found to be asymmetric between the two groups with d1 reporting a deterioration in performance for k=36, 48 and 60 months of approximately -179.3 -202.8 and -350.6 percent respectively while d10 reports an improvement in performance of +139.9, +153.8 and +129.5 percent over the same k-month periods. in terms of statistical significance, the k=48 and k=60 formation replications are significant at the 1% level, with t-values after adjusting for a thirdand fourth-order autocorrelation of -3.791 and -17.256 respectively. the results associated with the k =12 and k = 24 month periods confirm those reported earlier using the bhars methodology. this negative return differential between d1 and d10 is found to be evident throughout the entire period under examination. figure 4 summarises the k=60 months difference for individual portfolio formation dates. compared to the bhrs results, prior losers outperform prior international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 25 winners in 16 out of the 17 formation dates, while, as mentioned before, the return differences are found to be more extreme. figure 4: average cars difference between winners and losers for the t+60 months holding period january 1987 to december 2007 overall, the results for bhars and cars taken together partly contradict loughran and ritter‟s (1996, p.1963) findings that the use of bhars and cars in the post-formation periods lead to similar empirical conclusions. these results are found to be mostly supportive of the claim by conrad and kaul (1993) that the apparent overreaction effect reported in de bondt and thaler (1985) is at least in part due to bias in the method used in computing returns and, in particular, that a significant proportion of the positive post-formation cumulative abnormal returns on past loser portfolios is spurious. in any case, any overreaction effect under the latter methodology is unlikely to be exploitable in the uk because of transaction costs incurred in building and rebalancing monthly long positions in past loser stocks as discussed above. 5. decomposing the size effect from the „winner-loser‟ effect and assessing its relationship to alternative market cycles prior studies in the uk and us document a relationship between „winner-loser‟ and „small-size‟ effects (zarowin, 1990; campbell and limmack, 1997; dissanaike, -0 .2 1 4 -0 .3 8 7 -0 .2 1 6 -0 .4 9 5 -0 .4 8 9 -0 .4 9 2 -0 .5 1 3 -0 .0 5 3 -0 .3 9 5 -0 .2 2 5 0 .1 3 7 -0 .1 9 4 -0 .4 3 5 -0 .6 5 2 -0 .5 7 4 -0 .1 0 7 -0 .2 3 4 -0.70 -0.60 -0.50 -0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 1 9 8 6 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 w-l (t+60) lw k car  60 andrikopoulos et al.: overreaction effect 26 2002). the picture emerging from these studies is that the overreaction effect does not appear to be subsumed by the size effect even though both contrarian and small size investment strategies share common risk characteristics and common assets due to the association between past negative performance and current low market values. nonetheless, under the recent findings documented in dimson and marsh (1999) regarding the non-stationary nature of the „small size‟ premium in the uk in the 1990s, it is interesting to re-address this relationship. at the same time, by decomposing the excess returns reported earlier for the „loser‟ portfolios from their size-related component, we would be able to assess the true magnitude of the „winner-loser‟ effect in recent years. to do so, we adopted the methodology introduced in lakonishok et al. (1994) for estimating size-adjusted returns and combined it with the bhars methodology examined earlier. according to results reported in table 5, adjusting the bhars of all k-month portfolios for the size component has a large impact on post-formation performance and especially on the returns of past-loser portfolios. for the k=12 and k=24 holding periods in panels a and b, adjusting for size magnifies the previously reported momentum effect. for the remainder of the k-month holding periods, the size adjustment again has an asymmetric impact on the returns from all portfolios. although the reduction in the size component to the post-formation returns of prior winners has a marginal impact, prior losers‟ portfolio returns are severely affected in both magnitude and change of direction. the end-of-period bhars difference between winners and losers remains negative and follows a similar direction as before. however, these excess negative returns are drastically reduced. the sizeadjusted bhars’ difference between prior winners and losers for the k=36, 48 and 60 months are reported as -4.38, -3.50 and -11.30 percent, results which are statistically insignificant at all of the conventional acceptance levels. in terms of the impact of size component on these k-month holding periods, the worst affected is k=48 reported in panel e, where the losers‟ premium is reduced by almost 72 percent (71.95 percent). the relationship between these two effects is also illustrated in figure 5. adjusting post-formation returns for size will explain a large proportion of the international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 27 table 5: preand post-formation average size-adjusted average buy-and-hold returns ( sa klw bhar , ) on momentum deciles during the period 1987-2007. panel a: size-adjusted abhrs for momentum deciles based on 12 months pre-formation bhrs d1 (w) d2 d3 d8 d9 d10 (l) diffd1-d10 n 126 126 126 126 126 126 k=[-12,0] 1.084 0.334 0.164 -0.271 -0.394 -0.623 1.706 *** k=[0,+12] 0.042 0.029 0.018 -0.031 -0.041 -0.043 0.085 * panel b: size-adjusted abhrs for momentum deciles based on 24 months pre-formation bhrs d1 (w) d2 d3 d8 d9 d10 (l) diffd1-d10 n 123 123 123 123 123 123 k=[-24,0] 1.982 0.505 0.205 -0.481 -0.623 -0.856 2.838 *** k=[0,+12] 0.017 0.014 0.016 -0.022 -0.016 -0.016 0.034 k=[0,+24] -0.000 0.008 0.015 -0.004 -0.026 -0.068 0.068 panel c: size-adjusted abhrs for momentum deciles based on 36 months pre-formation bhrs d1 (w) d2 d3 d8 d9 d10 (l) diffd1-d10 n 120 120 120 120 120 120 k=[-36,0] 2.868 0.665 0.216 -0.666 -0.839 -1.047 3.915 *** k=[0,+12] -0.005 0.019 -0.000 -0.017 -0.006 -0.015 0.010 k=[0,+24] -0.041 0.032 -0.008 -0.002 0.006 -0.030 -0.012 k=[0,+36] -0.083 0.015 0.006 -0.024 0.027 -0.040 -0.044 panel d: size-adjusted abhrs for momentum deciles based on 48 months pre-formation bhrs d1 (w) d2 d3 d8 d9 d10 (l) diffd1-d10 n 116 116 116 116 116 116 k=[-48,0] 3.954 0.835 0.225 -0.902 -1.061 -1.256 5.210 *** k=[0,+12] -0.016 0.013 0.001 0.015 0.015 -0.016 0.000 k=[0,+24] -0.043 0.006 -0.019 0.031 0.036 -0.040 -0.004 k=[0,+36] -0.081 0.004 -0.029 0.027 0.024 -0.038 -0.043 k=[0,+48] -0.108 0.001 -0.048 0.037 0.044 -0.073 -0.035 panel e: size-adjusted abhrs for momentum deciles based on 60 months pre-formation bhrs d1 (w) d2 d3 d8 d9 d10 (l) diffd1-d10 n 113 113 113 113 113 113 k=[-60,0] 5.057 1.060 0.234 -1.125 -1.323 -1.499 6.555 *** k=[0,+12] -0.018 -0.002 -0.012 -0.002 0.033 -0.028 0.009 k=[0,+24] -0.050 -0.026 -0.018 0.031 0.041 -0.048 -0.002 k=[0,+36] -0.095 -0.041 -0.010 0.048 0.060 -0.077 -0.018 k=[0,+48] -0.123 -0.062 0.012 0.081 0.069 -0.091 -0.032 k=[0,+60] -0.153 -0.096 -0.018 0.048 0.073 -0.040 -0.113 notes: this figure illustrates the size-adjusted buy-and-hold abnormal returns ( sa klw bhar , ) between past winners and losers for all k=12,24,…,60 months holding periods. these returns are computed using the methodology introduced in lakonishok et al. (1994). all t-statistics are estimated using the fama and macbeth (1973) procedure and are adjusted for the relevant kth order autocorrelation. *indicates significance at the 10% level, **indicates significance at the 5% level, *** indicates significance at the 1% level. andrikopoulos et al.: overreaction effect 28 previously reported overreaction effect. in addition, as the figures for the k=36, 48 and 60 holding periods illustrate, the bhars‟ difference for the winner-loser portfolios after adjusting for the size effect is almost eliminated in the first twentyseven post-formation months for k=36 and k=48, while for the k=60 months‟ investment horizon prior-losers start to consistently outperform prior-winners only after month thirty-nine. overall, compared to prior uk studies, the results reported so far give evidence of a weak presence of a „winner-loser‟ effect during the 1990s. although the effect has not reversed completely, it appears to be considerably more inconsistent than in previous decades. this finding gives most support to the argument that the effect may be associated with time-varying risk or with market cycles as argued by chen and sauer (1997) in their study of the us market. to assess this possible explanation we compare the „winner-loser‟ return differential to that of the market premium for a similar time period. analytically, following chen and sauer‟s (1997) methodology we estimated the returns on an arbitrage portfolio consisting of taking a long position in the loser portfolio while shorting the winner portfolio for the entire period under examination. monthly excess returns on the arbitrage portfolio are estimated as the difference in the monthly average abnormal returns between the extreme loser and extreme winner portfolios or w t l t wl t aaraaraar   . the estimation of the market risk premium is calculated as the difference between the monthly raw returns of the ftse all-share index ( m r ) and the returns on one-month treasury gilts ( f r ). finally, to compare the two return series on an annual basis over the entire examined period 1987-2007, we compute annual risk premium as       12 1 ,,12, k t tftftsekrp rrcrr (11) where 12, krp crr are the annual cumulative returns, tftse r , is the monthly return on the ftse all-share index and tf r , is the monthly return on one-month treasury gilts. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 figure 5: average bhars and average size-adjusted bhars difference between „winners‟ and „losers‟ for all k=12,…,60 months holding periods. t+ 1 t+ 2 t+ 3 t+ 4 t+ 5 t+ 6 t+ 7 t+ 8 t+ 9 t+ 1 0 t+ 1 1 t+ 1 2 -0.02 0.00 0.02 0.04 0.06 0.08 0.10 series1 series2 t+ 1 t+ 2 t+ 3 t+ 4 t+ 5 t+ 6 t+ 7 t+ 8 t+ 9 t+ 1 0 t+ 1 1 t+ 1 2 t+ 1 3 t+ 1 4 t+ 1 5 t+ 1 6 t+ 1 7 t+ 1 8 t+ 1 9 t+ 2 0 t+ 2 1 t+ 2 2 t+ 2 3 t+ 2 4 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 t+ 1 t+ 3 t+ 5 t+ 7 t+ 9 t+ 1 1 t+ 1 3 t+ 1 5 t+ 1 7 t+ 1 9 t+ 2 1 t+ 2 3 t+ 2 5 t+ 2 7 t+ 2 9 t+ 3 1 t+ 3 3 t+ 3 5 -0.16 -0.14 -0.12 -0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 s t+ 1 t+ 3 t+ 5 t+ 7 t+ 9 t+ 1 1 t+ 1 3 t+ 1 5 t+ 1 7 t+ 1 9 t+ 2 1 t+ 2 3 t+ 2 5 t+ 2 7 t+ 2 9 t+ 3 1 t+ 3 3 t+ 3 5 t+ 3 7 t+ 3 9 t+ 4 1 t+ 4 3 t+ 4 5 t+ 4 7 -0.14 -0.12 -0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 s s t+ 1 t+ 3 t+ 5 t+ 7 t+ 9 t+ 1 1 t+ 1 3 t+ 1 5 t+ 1 7 t+ 1 9 t+ 2 1 t+ 2 3 t+ 2 5 t+ 2 7 t+ 2 9 t+ 3 1 t+ 3 3 t+ 3 5 t+ 3 7 t+ 3 9 t+ 4 1 t+ 4 3 t+ 4 5 t+ 4 7 t+ 4 9 t+ 5 1 t+ 5 3 t+ 5 5 t+ 5 7 t+ 5 9 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05 s s 12,  klw bhar sa klw bhar 12,  24,  klw bhar sa klw bhar 24,  36,  klw bhar sa klw bhar 36,  60,  klw bhar sa klw bhar 60,  48,  klw bhar sa klw bhar 48,  andrikopoulos et al.: overreaction effect in a similar manner, annual cumulative abnormal returns on the arbitrage portfolio are estimated as the sum of the monthly abnormal returns on the arbitrage portfolio averaged across the different portfolio formation periods so as to accommodate the effect of overlapping periods for the k=60 months post-formation performance, or               12 1 5 1 ,,,,12, 1k t h thwthlkwl arrarr h car (12) where thl arr ,, and thw arr ,, are the abnormal returns on the prior “loser” and “winner” portfolios for formation h and for month t and 12,  kwl car is the annual cumulative abnormal returns on the arbitrage portfolio. for example, the cumulative abnormal returns on the arbitrage portfolio for the year 1993 is estimated as the sum of the average abnormal monthly returns between losers and winners across all portfolio formations i.e. k=1,…,12 for december 1992, k=13,…,24 for the december 1991 and so on. figure 6 illustrates the relationship between the two series for the entire period 1987 to 2007. for most of the examined period, the relationship between the overreaction effect and the risk premium appears to be positive. apart from isolated periods, such as the years 1987-1989 and post 2001, for the majority of the 1990s, the average cars differential for prior losers is evidently related to the movements of the market risk premium. in the post dot.com era, the relationship has reversed dramatically and may be related to i) the continuous and periodic reduction in the uk‟s interest rates thus increasing further the market risk premium and/or ii) a possible change in the investment preferences of institutional and large private investors following the collapse of the internet bubble. overall, the „winner-loser‟ effect is more evident during periods of financial and/or economic recovery such as the years of 19871988, 1993-1994, 1997 to 2002 and post 2006. to validate the above argument and test the statistical significance of both the relationship between the “winner-loser” effect with the a) “size” effect and b) the market risk premium we consider the following two-factor regression model     tftmtt l t w t urrabmsaaarrarr  210 (13) international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 where  l t w t arrarr  is the difference in monthly abnormal returns between the “winner” and the “loser” portfolios, t bms is the monthly average abnormal returns between the largeand small-size portfolios and   ftmt rr  is the market risk premium. to accommodate the effect of heteroscedasticity and autocorrelation in the data, all coefficients are estimated using the newey-west (1987; 1994) procedure. figure 6: average annual cars for the arbitrage portfolio vs. the market risk premium notes: this figure illustrates the annual return differential between the arbitrage portfolio and the market risk premium. the returns on the arbitrage portfolio consist of taking a long position in the loser portfolio while shorting the winner portfolio for the entire period under examination. monthly excess returns on the arbitrage portfolio are estimated as the difference in the monthly average abnormal returns between the extreme loser and winner portfolios or w t l t wl t aaraaraar   . the estimation of the market risk premium is calculated as the difference between the monthly raw returns of the ftse all-share index ( m r ) and the returns on one-month treasury gilts ( fr ). annualised returns for the market risk premium are estimated as the cumulative raw returns on the monthly arrs; while, for the arbitrage portfolio, annualised cars are estimated as               12 1 5 1 ,,,,12, 1k t h thwthlkwl arrarr h car where thl arr ,, and thw arr ,, are the abnormal return on the prior “loser” and “winner” portfolios for formation h and for month t. according to evidence presented in table 6, both explanatory variables selected are different from zero and the coefficients have the predicted signs. nonetheless, only the size-effect variable (bms) can reliably explain the returns on the arbitrage portfolio. as the average coefficient suggests, every 1 percent increase in the excess returns on the large capitalisation portfolio will result in an increase of 0.49 percent in the arbitrage portfolio‟s returns (t-statistic of 4.77,4 significant at the -0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 0.30 0.40 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 rp arrsw-l12, krpcrr 12,  kwlcar andrikopoulos et al.: overreaction effect 10 percent level). the relationship between the arbitrage portfolio‟s returns and market risk premium suggests a negative relationship indicating that decreases in the market risk premium will lead to increases in the excess returns for prior winners compared to prior losers. however, the coefficient is very small ( 2 a = -0.1207) and not statistically significant (t-value of -0.9082, p.0.1920). table 6: ols regression of the arrs difference between “winner” and “loser” portfolios for the k=60 months post-formation ranking period with the arrs difference of largeand small-size portfolios and the market risk premium. intercept bms risk premium r 2 adjusted coefficient estimates 0.0001 0.4857 -0.1207 t-statistic # (0.5595) (4.7737) (-0.9082) 0.4340 f-statistic 29.4521 *** durbin-watson statistic 1.9623 notes: this table reports the results from the statistical examination of the relationship between the “winner-loser” effect with the a) “size” effect and b) the market risk premium. the regression estimation is formulated as     tftmtt l t w t urrabmsaaarrarr  210 where  l t w t arrarr  is the difference in monthly abnormal returns between the “winner” and the “loser” portfolios; t bms is the monthly average abnormal returns between the largeand small-size portfolios, and   ftmt rr  is the market risk premium. to accommodate the effect of heteroscedasticity and autocorrelation in the data, all regression coefficients are estimated using the newey-west (1994) procedure. the most plausible explanation for such small values on the residual might be the change in the relationship between the two variables in the post-2001 period discussed earlier. the intercept term ( 0 a coefficient) is very close to zero (0.000051) and it is not statistically significant. according to the adjusted r 2 results, the model explains only approximately 43% ( 2 r =0.4340) of the total variation in the endogenous variable. overall, these results confirm the link between “winner-loser” effect and „size-effect‟ by corroborating the findings of dissanaike (2002) that size can only partially explain the former return differential, while providing weak support to the arguments of chen and sauer (1997) that excess-returns for priorlosers are associated with market cycles. 6. conclusion dissanaike (1997, 2002) investigated the overreaction effect in the uk market for the period 1979 to 1988. these studies are substantially free from the main sources international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 of bias discussed in the literature on „winner-loser‟ effect and, in particular, from survivorship and look-ahead bias. the research presents convincing evidence that a statistically and economically significant overreaction effect existed in the uk market for the studied period. the overreaction effect was apparent for large liquid stocks in the ftse 500 index, the universe of stocks under investigation. this is remarkable in that it is often claimed that any market inefficiencies that do exist are more likely to be found amongst small, illiquid or neglected stocks. thus, restricting the studies to the large, liquid stocks that are followed by analysts would presumably have the effect of biasing the research towards finding no evidence of overreaction. the present study examines the overreaction hypothesis for the uk market for the later period january 1987 to december 2007. again, the research is largely free from the most commonly discussed sources of bias. the main difference between this and the previous uk studies, apart from the time period covered, is that the universe of stocks is far wider, covering all fully-listed uk stocks (excluding investment trusts and investment companies). in this case, the inclusion of many small, illiquid stocks should, if anything, bias the research towards finding evidence for the overreaction effect. the results suggest that evidence for the “winner-loser” effect, even if still present in the uk during the last decade, appears to be weaker and inconsistent. from all the alternative k-months holding periods under examination, only the k=60 appears to provide statistically significant results. taking these studies together provides strong evidence that an overreaction effect did indeed exist in the uk from 1975 to around the late 1980s, but weakened considerably thereafter. these results are therefore consistent with the us literature on long-term returns (chen and sauer, 1997), where it is shown that the overreaction effect in the us has been non-stationary during the time period 1926 to 1992. the accumulating evidence of the time-varying nature of the overreaction effect supports the efficient markets‟ explanations that the overreaction premium is compensation for time-varying risk and that it can partially be explained by the other popular anomaly in finance literature the „small-size‟ effect. from the perspective of the behavioral finance research paradigm, further corroboration of the time non-stationarity of the overreaction effect may require non ad hoc revisions to the basic overreaction hypothesis, i.e. revisions that are wellandrikopoulos et al.: overreaction effect motivated in terms of the underlying behavioral theories and not proposed merely to evade falsifying empirical evidence. author information: panagiotis andrikopoulos, the corresponding author, is a principal lecturer in the department of accounting and finance, leicester business school, de montfort university, uk. tel: +44(0)1162506331; email: pandrikopoulos@dmu.ac.uk. arief daynes is a senior lecturer in the department of accounting and finance, portsmouth business school, university of portsmouth, portsmouth, uk. tel: +44(0)2392844182; email:arief.daynes@port.ac.uk. paraskevas pagas is a senior lecturer in the department of accounting and finance, portsmouth business school, university of portsmouth, portsmouth, uk. tel: +44(0)2392844721; email:paraskevas.pagas@port.ac.uk. acknowledgements: the authors would like to thank david crowther, david latimer, peter scott and participants in the 2009 bafa and 2008 bafa-seag annual conferences for their valuable comments on earlier drafts of the paper. they are also grateful to the two anonymous referees for their constructive comments and suggestions that helped improve the paper. references andrikopoulos p., daynes, a., latimer, d., and pagas, p., (2008). size effect methodological issues and „risk-to-default‟: evidence from the uk stock market. the european journal of finance, 14: 299–314. ball, r., and kothari, s.p., (1989). non-stationary expected returns: implications for tests of market efficiency and serial correlations in returns. journal of financial economics, 25: 51-74. boynton, w., and oppenheimer, h.r., (2006). anomalies in stock market pricing: problems in return measurements. journal of business, 79: 2617-2631. campbell, k., and limmack, r.j., (1997). long-term overreaction in the uk stock market and size adjustments. applied financial economics, 7: 537-548. chopra, n., lakonishok, h.j., and ritter, j.r., (1992). measuring abnormal performance: do stocks overreact? journal of financial economics, 31: 235-268. chen, c.r., and sauer, d.a., (1997). is stock market overreaction persistent over international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 mailto:pandrikopoulos@dmu.ac.uk time? journal of business finance and accounting, 24: 51-66. clare, a., and thomas, s., (1995). the overreaction hypothesis and the uk stock market. journal of business finance and accounting, 22: 961-973. conrad, j., and kaul, g., (1993). long-term market overreaction or biases in computed returns? journal of finance, 48: 39-63. de bondt, w.f.m., and thaler, r.h., (1985). does the stock market overreact? journal of finance, 40: 793-805. ---------------(1987). further evidence on investor overreaction and stock market seasonality. journal of finance, 42: 557-582. de long, j.b., shleifer, a., summers, l., and waldmann, r., (1990). positive feedback investment strategies and destabilizing rational speculation. journal of finance, 45: 375-395. dimson, e., and marsh, p., (1999). murphy‟s law and market anomalies. journal of portfolio management, 26: 53-69. dissanaike, g., (1997). do stock market investors overreact? journal of business finance and accounting, 24: 27-48. ------------(2002). does the size effect explain the uk winner-loser effect? journal of business finance and accounting, 29: 139-54. dreman, d.n., and lufkin, e.a., (2000). investor overreaction: evidence that its basis is psychological. journal of psychology and financial markets, 1: 61-75. fama, e.f., and macbeth, j.d., (1973). risk, return, and equilibrium: empirical tests. journal of political economy, 71: 607-636. fama, e.f., (1991). efficient capital markets ii. journal of finance, 46: 1575-617. -----(1998). market efficiency, long-term returns and behavioural finance. journal of financial economics, 49: 283-306. george, t.j., and hwang c., (2007). long-term return reversals: overreaction and taxes? journal of finance, 62: 2865-2896. grinblatt, m., and moskowitz, t.j., (1999). do industries explain momentum? journal of finance, 54: 1249-1290. lakonishok, h.j., shleifer, a., and vishny, r., (1994). contrarian investment, extrapolation and risk. journal of finance, 49: 1541-1577. loughran, t., and ritter, j.r., (1996). long-term market overreaction: the effect of low-priced stocks. journal of finance, 51: 1959-1970. nam, k., pyun, c.s., and avard, s.l., (2001). asymmetric reverting behaviour of andrikopoulos et al.: overreaction effect short-horizon stock returns: an evidence of stock market overreaction. journal of banking and finance, 25: 807-824. rouwenhorst, k.g., (1998). international momentum strategies. journal of finance, 53: 267–84. zarowin, p., (1990). size, seasonality, and stock market overreaction. journal of financial and quantitative analysis, 25: 113-126. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 1 cetak a4 ijbf6(2)09.indd the international journal of banking and finance, 2008/09 vol. 6. number 2: 2009: 129-138 129 ijbf interest rate and foreign exchange risk exposures of australian banks: a note abul f. m. shamsuddin university of newcastle, australia abstract the abolition of most government controls over the australian fi nancial system in the 1980s, the advent of a fl exible exchange rate regime in 1983 and the globalisation of the fi nancial system in the 1990s have created new opportunities for australian banks but exposed them to new sources of risk. this study estimates systematic risk exposure of publicly listed australian banks with respect to market, interest rate and foreign exchange rate using a garch-inmean model. not surprisingly, the results suggest that nearly all banks exhibit varying degrees of market risk exposure. however, stock returns of large banks are highly sensitive to interest rate changes, while most small banks are almost immune to both interest and exchange rate changes. keywords: interest rate risk, foreign exchange rate risk, australian banks jel classifi cation: g12, g21 1. introduction the traditional focus of the literature on modelling bank stock returns is on the interest rate risk. this emphasis on interest rate exposure is primarily attributed to banks’ historical reliance on net interest income as the primary source of profi ts. however, the removal of most government controls over the australian banking system in the 1980s and the introduction of a fl exible exchange rate regime in 1983 have produced new opportunities for australian banks but exposed them to new risk factors.1 in this study, we estimate the sensitivity bank stock returns with respect to these risk factors. 1 australian banks were subject to three forms of regulatory controls, following the second world war: directing bank loans to specifi c sectors, regulating the quantity of bank loans or deposits and controlling interest rates for deposits and loans. by 1991 all of these quantitative controls were removed, which increased competition in the banking industry but also relaxed the upper limit on bank risk-taking. thus, in recent years regulatory authorities have increased their emphasis on prudential supervision to limit risk-exposure of fi nancial institutions and ultimately, to protect depositors’ funds and maintain the stability of the fi nancial system. ht tp :// ijb f.u um .e du .m y 130 interest rate and foreign exchange risk exposures of australian banks: a note: 129-138 the fundamental model of valuation suggests that a bank’s stock price depends on its discounted stream of expected cash fl ows, which in turn is partly determined by how effectively the bank manages its risk exposures. total risk tolerated by a bank’s shareholders results from the bank’s two unique intermediation functions. first, the bank performs the brokerage function, acting as an agent for households and businesses in providing fi nancial products and transaction services. second, the bank performs the traditional asset transformation function by issuing fi nancial claims that are more attractive to savers than the claims directly issued by corporations. in conducting these two core functions, the bank faces a range of risks, including market risk, interest rate risk and foreign exchange risk. these risk factors infl uence the bank’s expected stream of cash fl ows and ultimately, returns on bank stocks. thus, the sensitivity of bank stock returns with respect to these risk factors reveals market’s perception of risk exposure of banks. prior to the adoption of a fl oating exchange rate in december 1983, the australian dollar was overvalued. between january 1984 and december 2001 the australian dollar depreciated against the us dollar by about 44 percent. since 2002 the australian dollar has appreciated signifi cantly against the us dollar primarily due to its spectacular economic performance, primarily fueled by the global resources boom. the post-deregulation period also witnessed highly volatile interest rates, ranging from 18 percent to 4.25 percent. the high volatility in interest rates and foreign exchange rates was accompanied by an average return on australian bank stocks of over 10 percent per annum. these stylized facts seem to indicate that in general australian banks have been effective in managing their risk exposures in the post deregulation period. this paper investigates risk exposures of ten publicly listed australian banks over the post-deregulation period, from 1994 to 2007. more specifi cally, the sensitivity of bank stock returns with respect to market, interest rate and foreign exchange rate is evaluated, taking into account the potential trade-off between bank stock return and its forecast volatility. the remainder of this paper is organized as follows. the second section reviews the literature. the empirical model is presented in section 3 and the corresponding results are interpreted in section 4. the fi nal section presents some concluding remarks. 2. literature review from the vast empirical literature on risk exposures of commercial banks, this section puts forward a review some representative australian and international studies, which evaluated risk exposure based on the sensitivity of bank stock returns to interest and exchange rates and market risk factors. there have been some attempts in evaluating risk exposures of australian banks since the early 1990s. harper and scheit (1992) investigated the impact of fi nancial deregulation on the riskiness of three major australian banks: anz bank, national australia bank and westpac banking corporation. using monthly data from january 1974 to december 1989, they demonstrated that ht tp :// ijb f.u um .e du .m y interest rate and foreign exchange risk exposures of australian banks: a note: 129-138 131 fi nancial deregulation did not infl uence the systematic or market risk exposures of these banks or of the banking industry as a whole. they did not consider other sources of risks such as the interest rate or foreign exchange rate risks. daugaard and valentine (1993), however considered other sources of risks. they investigated the sensitivity of bank share prices to changes in variables such as interest rates, exchange rates, a market risk factor (dow jones us banking – money centre index) and bank profi ts. using quarterly data for the period, 1983 to 1991, they demonstrated that an appreciation of the australian dollar signifi cantly increased bank share prices, measured by the asx banking and fi nance price index. this study has some shortcomings. first, the scope of their study was limited to modeling the composite stock price index of the banking and fi nance industry rather than stock prices of individual banks or the price index for the banking industry alone. second, their sample includes a major structural break – the introduction of the fl oating exchange rate regime in december 1983. no attempt was made to investigate the impact of this structural change on the exchange rate sensitivity of bank share prices.2 third, the use of the dow jones us banking – money centre index as a proxy for the australian market portfolio is incompatible with the capital asset pricing model. iorio and faff (2000) examined the potential asymmetric response of industry stock index returns to the foreign exchange rate. they considered 24 australian industries, including the banking industry but ignored the interest rate risk factor.3 using monthly data for the period 1988 to 1996, they found that the foreign exchange risk was irrelevant in pricing bank stocks. in daily data, however, the foreign exchange risk was relevant only when daily appreciation of the australian dollar was at least 1 percent. thus, their results indicate that the banking industry as a whole manage the foreign exchange risk effectively, making expected cash fl ows insensitive to exchange rate fl uctuations at least in the mediumto long-term. in a subsequent study on bank risk exposures in ten markets within the asia-pacifi c region, yong and faff (2007) observed that the australian bank stock portfolio was exposed to market and long-term interest rate risks, but not foreign exchange risk. some us and canadian studies, however, incorporated market, interest rate and exchange rate risk factors in modeling bank stock returns. for example, choi, et al. (1992) used monthly data over the period january 1975 to december 1987 for the 48 largest us commercial banks. they observed that bank stock returns respond negatively to interest rates and the impact of the exchange rate depends on the banks’ net position in foreign currencies. their results confi rmed that a depreciation of foreign currencies negatively infl uenced bank stock returns 2 indeed, a later study (dennis and jeffrey, 2002) documented signifi cant structural changes in australian bank risk over the period 1981 to 1993. these changes coincided with the deregulatory initiative in the early 1980s, fl otation of the australian dollar and the implementation of basel accord. 3 however, in an earlier study, faff and howard (1999) estimated market and interest rate risk exposures of australian fi nancial sector companies at the portfolio level. ht tp :// ijb f.u um .e du .m y 132 interest rate and foreign exchange risk exposures of australian banks: a note: 129-138 prior to october 1979, when banks had a positive net position in major foreign currencies. after that date, banks’ foreign currency position changed from a positive net position to a negative net position and a depreciation of foreign currency increased bank stock returns. thus, the exchange rate was an important risk factor for us bank stock returns. a comparable study for the six largest canadian chartered banks was conducted by atindéhou and gueyie (2001). they observed that in general a depreciation of foreign currencies against the canadian dollar increased bank stock returns. this result is consistent with the negative foreign currency position of canadian banks over their sample period. a few studies examined risk exposures of banks using models from the garch family. the fi rst attempt in this area is a study by elyasiani and mansur (1998) on interest risk exposures of the us banks. they found that both interest rate and interest rate volatility infl uence mean as well as the conditional volatility of bank stock returns. ryan and worthington (2004) replicated their study in the context of australia, taking into account foreign exchange rate as an additional source of risk. however, these studies examined risk exposures at the industry or portfolio level, not at the individual bank level. our study is a novel attempt in evaluating risk exposures of individual australian banks within a garch framework. 3. empirical model a multi-index model of bank stock returns is used in this paper with return on the market portfolio, interest rate, foreign exchange rate and conditional volatility as indexes. conditional variance of bank stock returns is generated by a garch (1,1) process and a lagged interest volatility term. the latter explanatory factor was proposed by a us study on bank stock returns (elyasiani and mansur, 1998) to capture incremental information on macroeconomic volatility that is contained in the second moment of the distribution of interest rates. in this paper, the following garch(1,1)-in-mean model is used for empirical investigation: tjtjtttmtj heyrr ,,32,10, εγββββ +++∆++= (1) )( 11,2 2 1,10, −−− +++= ttjtjtj civhh λαεαα (2) ),0(~ ,, tjtj hnωε (3) in the mean equation (1), r j is the rate of return on bank stock or bank portfolio j, r m is the rate of return on the market portfolio; ∆y is the change in the long term interest rate; e is the rate of change in the value of the australian dollar (expressed in terms of usd); is the conditional standard deviation of bank stock returns; and ε j is a serially uncorrelated random error term. conditional jh ht tp :// ijb f.u um .e du .m y interest rate and foreign exchange risk exposures of australian banks: a note: 129-138 133 variance, h j,t, is determined by a constant or unconditional variance, α 0 , and the weighted sum of three components – last period’s news about volatility or squared innovation, , last period’s conditional variance, h j,t , and last period’s conditional interest rate volatility, civ t-1 . the conditional interest rate volatility was estimated from a garch(1, 1) model. this study considers ten fi nancial fi rms that are regarded as banks under the global industry classifi cation standard and currently listed on the australian stock exchange. these are: anz banking group ltd, bendigo bank limited, bank of queensland limited, commonwealth bank of australia, national australia bank limited, the rock building society limited, st. george bank limited, suncorpmetway limited, wide bay australia ltd and westpac banking corporation. the following null hypotheses are tested for the individual bank stocks as well as for the portfolio of bank stocks: h 1 : bank stock return is invariant to market conditions: β 1 = 0 under this hypothesis, banks are not exposed to systematic risk. h 2 : the conditional interest rate volatility has no infl uence on the conditional stock return volatility: λ = 0. in this case, the conditional stock return volatility follows a pure garch(1,1) process without any exogenous macroeconomic variable. h 3 : neither the change in interest rate nor the interest rate volatility infl uences the stock return generating process: β 2 = λ = 0 . under this hypothesis, the bank has no net exposure to interest rate risk. h 4 : bank stock return is insensitive to the change in the foreign exchange rate: β 3 = 0 . put differently, the bank has no net exposure to foreign exchange rate risk. h 5 : conditional volatility is not a relevant factor in pricing bank stocks: γ =0. under this null hypothesis, there is no trade-off between bank stock return and its conditional volatility. h 6 : volatility is time invariant: α1 = α2 = λ = 0. 2 1,tj 4. data and results this study uses weekly data from 26 september 1994 to 4 june 2007. all data are retrieved from datastream. weekly returns on individual bank stocks are calculated from the total return index (ri) series, created by datastream for individual equities. this series is suitable for calculating holding period total return as it captures both the price change and the dividend income components. the rate of return on an individual bank stock, r j is calculated as 100 times logarithm of return index relatives. the return on the bank stock ht tp :// ijb f.u um .e du .m y 134 interest rate and foreign exchange risk exposures of australian banks: a note: 129-138 portfolio is market capitalisation weighted returns on individual bank stocks. the datastream total stock market index for australia is used as a proxy for the market portfolio, and total return is calculated as 100 times logarithm of the relevant return index relatives we use the weekly change in the 10-year treasury bond yield as a proxy for interest rate change, ∆y. the rate of change in the value of the australian dollar against the us dollar (e) is measured as 100 times log exchange rate relatives. the results for the market capitalization weighted bank portfolio and individual banks are presented in table 1. the banking sector as a whole is exposed to market risk but the beta coeffi cient is less than unity and statistically signifi cant. the results for individual banks show that the four major banks are more sensitive to the change in market conditions than that of small banks. interest rate risk arises from mis-matching of maturity of a bank’s assets and liabilities. any change in the interest rate has the potential to infl uence a bank’s net interest income as well as market values of its assets and liabilities, which in turn is refl ected in bank stock price changes. the model used in this paper allows both the interest rate and the conditional volatility of the interest rate to infl uence bank stock risk and return. for the banking sector as a whole, the mean equation shows that one percentage point increase in the interest rate leads to 1.68 percentage point decrease in return. interestingly, the interest sensitivity coeffi cient is negative for all banks but statistically signifi cant for only fi ve large banks: national australia bank limited, commonwealth bank of australia, anz banking group, westpac banking corporation and st. george bank. as noted earlier, we estimate the conditional interest rate volatility (civ) using a garch (1, 1) model without any exogenous variables and then use the lagged civ variable in the second moment equation for bank stock returns. it is argued that the interest rate volatility conveys incremental information regarding macro economic conditions and may infl uence the volatility of bank stock returns. this is indeed the case for eight out of ten banks considered in this study. the coeffi cient of the civ variable is negative and highly signifi cant for the bank portfolio. 4 transaction exposure measures the variability of the bank’s (contractually specifi ed) future cash transactions resulting from exchange rate fl uctuations. this exposure arises from a bank’s foreign currency denominated payables and receivables. the second dimension of foreign exchange risk is translation exposure, which arises from conversion of foreign currency denominated assets and liabilities into domestic currency. although this exposure leads to unrealized gains and losses, it might affect a bank's expected cash fl ows and hence, its share price. finally, operating exposure measures the long-term impact of exchange rate changes on the bank’s profi tability. for example, exchange rate fl uctuations may infl uence the bank’s cash fl ows by changing the cost of a bank’s inputs and revenue earned from its fi nancial products. unlike translation and transaction exposure, operating exposure is diffi cult to quantify and can not be eliminated using market based hedging instruments. ht tp :// ijb f.u um .e du .m y interest rate and foreign exchange risk exposures of australian banks: a note: 129-138 135 next we turn to foreign exchange risk, which has three dimensions: transaction exposure, translation exposure and operating exposure. these underlying exposures determine the degree of sensitivity of bank stock returns with respect to changes in foreign exchange rates. the exchange rate parameter is positive and highly signifi cant for the bank portfolio, indicating that an appreciation of the australian dollar exerts positive infl uence on bank stock returns. at the individual bank level, this empirical assertion only holds for anz bank and bank of queensland. the results suggest that stock returns for most banks in australia are not sensitive to the change in the exchange rate. perhaps, these banks adequately mange their foreign exchange risk exposure. the garch-in-mean model also allows us to examine the potential tradeoff between bank stock returns and its conditional volatility, which has been approximated by the parameter, γ. this coeffi cient is statistically insignifi cant at the 5 percent level in all but the model for anz bank. we observe a positive tradeoff between risk and return for anz bank, which implies that the conditional volatility is priced. however, for the banking sector as a whole the conditional volatility is irrelevant in pricing bank stocks. finally, a wald test suggests that the volatility of bank stock return is time variant. that is, the null hypothesis of α 1 = α 2 = λ = 0 can be rejected at the 5 percent level of signifi cance. 5. conclusion this study contributes to the australian literature on bank risk in two important dimensions. first, this paper focuses on risk exposure of individual banks in the post-deregulation period, when australian banks were exposed to new opportunities as well as increased risk. second, this study allows both the level and the volatility of the interest rate to infl uence the distribution of individual bank stock returns. in general, the results suggest that large banks are highly sensitive to interest rate changes. most small banks are immune to both interest and foreign exchange rate changes. we observe time variation in the volatility of bank stock returns but no signifi cant trade-off between return and the conditional volatility for bank stocks. further research can be directed to examine factors driving cross-sectional variation in risk exposure of australian banks. author statement: abul shamsuddin is a professor at the newcastle graduate school of business, the university of newcastle, callaghan, nsw 2308, australia. tel: +61-2-4921 8971; fax: +61-2-4921 7398, e-mail: abul. shamsuddin@newcastle.edu.au. the author acknowledges fi nancial support for this study from a new staff grant. an earlier version of the paper was presented at the fourth international conference on banking and finance 2008, langkawi island, malaysia. the author acknowledges with thanks the editorial review comments as responded in the revised version of this paper. ht tp :// ijb f.u um .e du .m y 136 interest rate and foreign exchange risk exposures of australian banks: a note: 129-138 t ab le 1 : m ax im um l ik el ih oo d e st im at es o f th e g a r c h ( 1, 1) -i nm ea n m od el ; e st im at io n p er io d: 2 6 s ep te m be r 19 94 t o 4 ju ne 2 00 7 b an k t ot al a ss et s (a ve ra ge , 19 94 -2 00 7, m il li on a u d ) β 1 β 2 β 3 γ α 1 α 2 λ r -s qu ar ed a ll b an k s 0. 88 24 01 -1 .6 84 30 7 9. 06 90 08 0. 24 39 57 0. 01 16 84 0. 97 24 28 -9 .3 11 9 0. 66 53 52 p -v al ue 0. 00 00 0. 00 00 0. 00 00 0. 21 78 0. 24 34 0. 00 00 0. 00 81 l ar ge b an k s n a b 32 0, 99 6 0. 83 17 93 -2 .0 34 95 9 4. 74 51 25 0. 15 40 25 0. 02 91 03 0. 92 65 61 -4 1. 80 05 5 0. 33 95 01 0. 00 00 0. 00 01 0. 37 97 0. 38 78 0. 07 75 0. 00 00 0. 08 58 c b a 21 8, 66 1 0. 80 20 81 -2 .2 54 51 1 -3 .4 70 62 8 0. 07 38 13 0. 02 03 17 0. 95 79 55 -3 0. 73 62 5 0. 28 70 85 0. 00 00 0. 00 00 0. 50 81 0. 51 40 0. 15 66 0. 00 00 0. 00 00 a n z 19 7, 65 4 0. 96 48 41 -2 .4 53 53 7 11 .7 08 37 0. 38 19 10 0. 09 48 90 0. 48 01 80 -4 5. 68 70 3 0. 40 63 33 0. 00 00 0. 00 00 0. 03 53 0. 02 03 0. 01 50 0. 04 80 0. 04 45 w b c 18 8, 01 2 0. 89 67 76 -2 .6 77 08 3 3. 16 37 40 0. 15 47 49 -0 .0 16 84 3 1. 00 60 18 -1 2. 99 53 3 0. 36 38 87 0. 00 00 0. 00 00 0. 60 00 0. 41 45 0. 00 03 0. 00 00 0. 00 30 s g b 55 ,9 50 0. 64 37 08 -1 .2 14 43 8 -7 .6 32 79 3 0. 08 71 85 0. 07 29 25 0. 69 33 76 -2 4. 76 07 9 0. 17 63 96 0. 00 00 0. 03 56 0. 20 53 0. 59 32 0. 03 05 0. 00 00 0. 03 13 t j t j t t t m t j h e y r r , , 3 2 , 1 0 , ; ) ( 1 1 , 2 2 1 , 1 0 , t t j t j t j c iv h h t a b le c o n ti n u es o n t h e p a g e ht tp :// ijb f.u um .e du .m y interest rate and foreign exchange risk exposures of australian banks: a note: 129-138 137 t ab le 1 : ( co n ti n u ed ) s m al l b an k s s u n 31 ,8 45 0. 69 46 67 -1 .1 49 38 9 -1 0. 44 03 3 0. 20 36 62 0. 14 79 75 0. 53 50 95 -3 1. 93 09 1 0. 18 38 04 0. 00 00 0. 07 18 0. 10 23 0. 32 82 0. 00 12 0. 00 00 0. 00 03 b e n 7, 21 0 0. 56 69 49 -0 .5 69 03 7 2. 75 36 94 0. 01 60 43 0. 06 55 13 0. 82 08 57 22 .5 06 89 0. 13 18 89 0. 00 00 0. 29 94 0. 65 17 0. 62 89 0. 13 12 0. 00 00 0. 24 67 b o q 6, 57 7 0. 40 39 94 -0 .1 78 07 3 15 .1 13 08 -0 .0 75 33 6 0. 24 34 69 0. 34 34 43 -1 0. 93 62 7 0. 09 78 28 0. 00 00 0. 77 47 0. 01 82 0. 75 29 0. 00 00 0. 00 45 0. 27 73 w b b 80 0 0. 06 27 66 -0 .0 23 52 1 3. 99 72 96 -0 .2 95 18 8 0. 10 96 30 0. 65 71 85 -1 2. 46 96 5 0. 01 11 90 0. 22 22 0. 97 08 0. 57 29 0. 27 30 0. 00 09 0. 00 00 0. 02 36 r o k 48 5 0. 17 83 33 -0 .8 23 57 8 -0 .2 64 70 9 -0 .1 37 87 6 0. 09 83 58 0. 82 44 57 -5 9. 02 86 9 0. 02 53 93 0. 00 03 0. 12 56 0. 96 65 0. 51 48 0. 00 00 0. 00 00 0. 00 17 n ot es : a n z = a n z b an ki ng g ro up l td , b e n = b en di go b an k l im it ed , b o q = b an k of q ue en sl an d l im it ed , c b a = c om m on w ea lt h b an k of a us tr al ia , n a b = n at io na l a us tr al ia b an k l im it ed , r o k = t he r oc k b ui ld in g s oc ie ty l im it ed , s g b = s t. g eo rg e b an k l im it ed , s u n = s un co rp m et w ay l im it ed , w b b = w id e b ay a us tr al ia l td , w b c = w es tp ac b an ki ng c or po ra ti on . ht tp :// ijb f.u um .e du .m y 138 interest rate and foreign exchange risk exposures of australian banks: a note: 129-138 references akhigbe, a., and whyte, a.m. (2003). changes in market assessments of bank risk following the riegle-neal act of 1994. journal of banking and finance, 27, 87-102. atindéhou, r., and gueyie, j. (2001). canadian chartered banks’ stock returns and exchange rate risk. management decision, 39 (4), 285-295. choi, j.j., elyasiani, e., and kopecky, k.j. (1992). the sensitivity of bank stock returns to market, interest and exchange rate risks. journal of banking and finance, 16, 983-1004. dennis, s.a., and jeffrey, a. (2002). structural changes in australian bank risk. journal of international financial markets, institutions and money, 12, 1-17. daugaard, d., and valentine, t. (1993). bank share prices and profi tability. school of finance and economics. working paper series (no 31). university of technology sydney. elyasiani, e., and mansur, i. (1998). sensitivity of the bank stock returns distribution to changes in the level and volatility of interest rates: a garch-m model. journal of banking and finance, 22, 535–563 faff, r.w., and howard, p.f. (1999). interest rate risk of australian fi nancial sector companies in a period of regulatory change. pacifi c-basin finance journal, 7, 83-101. harper, i.r., and scheit, t. (1992). the effects of fi nancial market deregulation on bank risk and profi tability. australian economic papers, 260-271. iorio, a.d., and faff, r. (2000). an analysis of asymmetry in foreign currency exposure of the australian equities market. journal of multinational financial management, 10, 133-159. loudon, g. (1993). the foreign exchange operating exposure of australian stocks. accounting and finance, 32, 19-32. merrett, d.t. (2002). the internationalization of australian banks. journal of international financial markets, institutions and money, 12, 377-397. ryan, s., and worthington, a. (2004). market, interest rate and foreign exchange rate risk in australian banking: a garch-m approach. international journal of applied business and economic research, 2(2), 81-103. wetmore j.l., and brick, j.r. (1994). commercial bank risk: market, interest rate, and foreign exchange. journal of financial research, 17, 585-96. yong, h., and faff, r. (2007). asia-pacifi c banks risk exposures: pre and post the asian fi nancial crisis. applied financial economics, 18(6), 431-449ht tp :// ijb f.u um .e du .m y 79 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 how to cite this article: toh, j. n., & khaw, k. l. -h. (2021). fossil fuel price, carbon dioxide emission, and renewable energy capacity: evidence from asian developing countries. international journal of banking and finance, 16(1), 79-96. fossil fuel price, carbon dioxide emission, and renewable energy capacity: evidence from asian developing countries 1toh jia ni & 2karren lee-hwei khaw department of finance and banking faculty of business and accountancy university of malaya 2corresponding author: karren@um.edu.my received: 3/11/2020 revised: 11/12/2020 accepted: 14/12/2020 published: 30/1/2021 abstract this paper examined the impact of fossil fuel price and carbon dioxide (co2) emission on renewable energy, using a sample of 14 asian developing countries from the years 2000 to 2018. fossil fuel prices, mainly those of crude oil and coal, are positively related to renewable energy capacity. co2 emission is also a positive driver, indicating the significance of environmental concern. the results were consistent for both the upper-middle-income and lower-middle-income countries. between fossil fuels and co2 emission, the positive impact of co2 emission outweighed that of fossil fuels. from a policy perspective, this paper concurs the need to shift huge subsidies away from fossil fuels to renewable energy and to enforce a heavy tax on co2 emission for a sustainable environment. keywords: renewable energy, fossil fuel, co2 emission, asian developing countries. jel classification: p28, q30, q42 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance https://doi.org/10.32890/ijbf2021.16.1.5 80 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 introduction this paper examines the impact of fossil fuel price and carbon dioxide (co2) emission on renewable energy capacity from the context of asian developing countries. asia has great potential for renewable energy resources, such as terrestrial wind and solar power in china and vietnam, hydro and solar power in malaysia, and geothermal energy in indonesia and the philippines. nonetheless, the transition from non-renewable to renewable energy is slow. asia accounted for 54 percent of the new renewable energy capacity in 2019; however, this capacity was mainly in china (international renewable energy agency, 2019). excluding china, the potential in other asian developing countries, many of which are lower-middle-income countries, is relatively untapped (kariuki, 2018). one of the barriers to switching to renewable energy sources is cost disadvantage. despite the decreasing price of renewable technologies, the production of renewable energy remains expensive and unaffordable for many developing countries as the renewable technologies have to be imported (kariuki, 2018). therefore, fossil fuels remain the dominant energy sources, although they are unsustainable given the rapid population growth, increasing energy demands, and volatile fossil fuel prices. hypothesis 1 examines the relationship between fossil fuel prices (oil, coal, and natural gas) and renewable energy. it is hypothesized that fossil fuel prices are positively related to renewable energy capacity on the basis of cost disadvantage. higher fuel prices increase the burden on countries to continue supplying affordable energy to individuals and businesses. globally, fossil fuel subsidies increased from $287 billion in 2016 to $438 billion in 2018 with increasing fuel prices. in 2019, the fuel subsidies decreased by $120 billion, largely due to decreasing fuel prices (international energy agency, 2020). instead of increasing the fuel subsidies, it would be more sustainable to subsidize renewable investments so that renewable energy would be cost competitive as compared to non-renewable fossil fuels (carley, 2009). this study also determines whether environmental concern motivates the adoption of renewable energy among asian developing countries. the major concern is the greenhouse gas emissions from the burning 81 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 of fossil fuels. although there is a decreasing trend of co2 emissions, the effort is mainly driven by advanced economies (international energy agency, 2020). excluding these advanced economies, about 80 percent of the 400 metric megaton increase in co2 emissions in 2019 were due to asian countries. evidently, six of the asian developing countries (bangladesh, myanmar, nepal, the philippines, thailand, and vietnam) are among the ten countries that most affected by climate change in the last two decades (eckstein et al., 2019). therefore, hypothesis 2 posits that co2 emission should be positively related to renewable energy capacity because it is critical to mitigate the adverse effect of climate change on achieving sustainability (rustemoglu & andres, 2016). this study contributes toward the growing literature on renewable energy in the following ways. first, it provides empirical evidence from the perspectives of asian developing countries, where the motivations to adopt renewable energy among the countries, whether upper-middle-income or lower-middle-income countries, are significantly driven by fossil fuel prices and co2 emission. the results agree with the common view that fossil fuel prices (e.g. apergis & payne, 2014; bird et al., 2005; chang et al., 2009) and co2 emission (e.g. aguirre & ibikunle, 2014; omri & nguyen, 2014; sadorsky, 2009a; salim & rafiq, 2012) are positively related to renewable energy. second, this study concurs that renewable energy can be a substitute for crude oil and coal (apergis & payne, 2014) and as a complementary energy source for natural gas because the latter is relatively clean as compared to crude oil and coal (sadorsky, 2009a; omri & nguyen, 2014). third, environmental concerns are found to outweigh cost concerns. this paper argues that this is potentially due to the global pressure to decrease co2 emissions that are currently rising. the rest of the paper is organized as follows: section 2 details the methods used, while section 3 is on the data, followed by section 4 that discusses the results. section 5 concludes the paper. methodology this study examines the hypotheses by using the multivariate panel data regression model (gozgor eta al., 2020; przychodzen & przychodzen, 2020). in line with gozgor et al. (2020) and sisodia and soares (2014), the random effects are controlled. the choice is further 82 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 validated using the hausman test. the empirical model is specified below: (1) data a sample of 14 asian developing countries were used in this study, namely bangladesh, bhutan, cambodia, china, india, indonesia, laos, malaysia, mongolia, myanmar, the philippines, sri lanka, thailand, and vietnam. these countries were selected based on the availability of data. in total, there is a balanced panel dataset of 266 annual observations, spanning from 2000 through 2018. the dependent variable, renewable energy capacity, was collected from the international renewable energy agency (irena) website. figure 1 presents the average annual renewable energy capacity for the sample. figure 1. renewable energy capacity of asian developing countries for the years 2000 to 2018. 3 methodology this study examines the hypotheses by using the multivariate panel data regression model (gozgor eta al., 2020; przychodzen & przychodzen, 2020). in line with gozgor et al. (2020) and sisodia and soares (2014), the random effects are controlled. the choice is further validated using the hausman test. the empirical model is specified below: 𝐿𝐿𝐿𝐿𝑅𝑅𝑅𝑅𝐿𝐿𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝐿𝐿𝑅𝑅𝑒𝑒𝑒𝑒𝑒𝑒𝑖𝑖,𝑡𝑡 = 𝛼𝛼 + 𝛽𝛽1𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑅𝑅 𝑓𝑓𝑓𝑓𝑅𝑅𝑅𝑅 𝑝𝑝𝑒𝑒𝐿𝐿𝑝𝑝𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿2𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐺𝐺𝐺𝐺𝐺𝐺 𝑒𝑒𝑒𝑒𝐿𝐿𝑅𝑅𝑔𝑔ℎ𝑖𝑖,𝑡𝑡, + 𝛽𝛽4𝐺𝐺𝐿𝐿𝑝𝑝𝑓𝑓𝑅𝑅𝑅𝑅𝑔𝑔𝐿𝐿𝐿𝐿𝐿𝐿 𝑒𝑒𝑒𝑒𝐿𝐿𝑅𝑅𝑔𝑔ℎ𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝑈𝑈𝐿𝐿𝑅𝑅𝑈𝑈𝑝𝑝𝑅𝑅𝐿𝐿𝑒𝑒𝑈𝑈𝑅𝑅𝐿𝐿𝑔𝑔𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐿𝐿ℎ𝐿𝐿𝐿𝐿𝑅𝑅𝑖𝑖 + 𝛽𝛽7𝑈𝑈𝑝𝑝𝑝𝑝𝑅𝑅𝑒𝑒 𝑈𝑈𝐿𝐿𝑚𝑚𝑚𝑚𝑅𝑅𝑅𝑅𝑖𝑖 + 𝑓𝑓𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (1) data a sample of 14 asian developing countries were used in this study, namely bangladesh, bhutan, cambodia, china, india, indonesia, laos, malaysia, mongolia, myanmar, the philippines, sri lanka, thailand, and vietnam. these countries were selected based on the availability of data. in total, there is a balanced panel dataset of 266 annual observations, spanning from 2000 through 2018. the dependent variable, renewable energy capacity, was collected from the international renewable energy agency (irena) website. figure 1 presents the average annual renewable energy capacity for the sample. over the years, the renewable energy capacity of these developing countries had increased from 8,728 megawatts in 2000 to 62,883 megawatts in 2018. this significant rise was mainly driven by china’s renewable energy policies. excluding china, the annual average renewable energy capacity showed a slower increase, from 3,565 megawatts in 2000 to 14,194 megawatts in 2018. 4 figure 1. renewable energy capacity of asian developing countries for the years 2000 to 2018. moving on, relatively close per capita values occurred even when china was excluded from the sample, as shown in figure 2. this implied an underutilization of renewable energy sources among these developing nations. 0 10000 20000 30000 40000 50000 60000 70000 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 re ne w ab le e ne rg y ca pa ci ty (m eg aw at ts ) yearasian developing countries excluding china 83 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 over the years, the renewable energy capacity of these developing countries had increased from 8,728 megawatts in 2000 to 62,883 megawatts in 2018. this significant rise was mainly driven by china’s renewable energy policies. excluding china, the annual average renewable energy capacity showed a slower increase, from 3,565 megawatts in 2000 to 14,194 megawatts in 2018. moving on, relatively close per capita values occurred even when china was excluded from the sample, as shown in figure 2. this implied an underutilization of renewable energy sources among these developing nations. figure 2. renewable energy capacity per capita of asian developing countries for the years 2000 to 2018. the independent variables, fossil fuel price and co2 emission, were collected from the world bank and global carbon atlas websites, respectively. the prices of crude oil, coal, and natural gas were then recorded. the total co2 emissions arising from the use of fossil fuels were measured per capita. both the independent and dependent variables appeared in natural logarithmic form. additionally, this study controlled gdp growth, population growth, and unemployment rate, while china and the upper-middle-income countries were included using dummy variables to account for any potential bias that might result from the significant differences among the countries. the description of each variable is summarized in table 1. 5 figure 2. renewable energy capacity per capita of asian developing countries for the years 2000 to 2018. the independent variables, fossil fuel price and co2 emission, were collected from the world bank and global carbon atlas websites, respectively. the prices of crude oil, coal, and natural gas were then recorded. the total co2 emissions arising from the use of fossil fuels were measured per capita. both the independent and dependent variables appeared in natural logarithmic form. additionally, this study controlled gdp growth, population growth, and unemployment rate, while china and the upper-middle-income countries were included using dummy variables to account for any potential bias that might result from the significant differences among the countries. the description of each variable is summarized in table 1. 0 1 2 3 4 5 6 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 re ne w ab le e ne rg y ca pa ci ty p er c ap ita year asian developing countries excluding china 84 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 table 1 description of the observed variables variables description source lnrenewable energy the natural logarithm form of the maximum generating capacity of power installations that use renewable sources to generate electricity, measured in megawatts. irena lncrude oil the natural logarithm form of the equally weighted average price of brent, dubai, and west texas intermediate crude oil in us$ per barrel. world bank lncoal the natural logarithm form of the coal price in us$ per metric ton. world bank lnnatural gas the natural logarithm form of the natural gas price in us$ per million btu. world bank lnco2 the natural logarithm form of the carbon dioxide (co2) emissions from the use of fossil fuels, measured in million tons of co2. global carbon atlas lnco2 per capita the total carbon dioxide (co2) emissions divided by the population of a country. the value is also in the natural logarithm form. global carbon atlas gdp growth the annual percentage growth rate of gross domestic product (gdp) at market prices based on constant local currency. world bank population growth the annual percentage growth rate of individuals in a population. world bank unemployment the percentage of labor force that is without work but is available for and seeking employment. world bank china a dummy variable that takes the value of 1 if the observed country is china and 0 otherwise. upper middle a dummy variable that takes the value of 1 if the observed country is an uppermiddle-income country and 0 otherwise. world bank 85 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 results table 2 summarizes the descriptive statistics of the variables for sample 1 and sample 2, which excludes china. panel a presents the non-transformed values of the key variables, while panel b shows the natural logarithmic values of the variables. for sample 1, the average renewable energy capacity was at 25,911 megawatts, reaching a maximum capacity of 695,831 megawatts. in sample 2, the average renewable energy capacity was only 7,183 megawatts, with a maximum capacity of 117,955 megawatts. however, the average renewable energy per capita was about 196 watts, with a maximum value of 2,238.76 watts for both samples. the average prices of crude oil, coal, and natural gas were us$62.84 per barrel, us$69.22 per metric ton, and us$4.68 per million btu, respectively. in terms of greenhouse gas emissions, for sample 1, the average co2 emission was 733 million tons, or 2.24 tons per capita, while for sample 2, the average co2 emission was 220.34 million tons, or 2 tons per capita. in terms of gdp growth, population growth, and unemployment rate, the average values were approximately similar for both samples. table 3 presents the mean values of the country-specific variables. among the 14 developing countries, china, malaysia, sri lanka, and thailand are upper middle-income countries. out of these four, china reported the highest renewable energy capacity of 269,375 megawatts or 198.82 per capita. malaysia reported the highest co2 emissions per capita of 7.14, followed by china at 5.52 and thailand at 3.62. on the other hand, india led the lower-middle-income countries with the highest renewable energy capacity of 55,618 megawatts; nevertheless the per capita value was low at 44.63. in contrast, bhutan’s per capital value was 1.611.23. india also had relatively higher co2 emissions. it ranked fourth after mongolia, indonesia, and vietnam in terms of co2 emissions per capita. table 4 presents the pearson correlation matrix for the observed variables. although lncrude oil and lncoal were highly correlated with a coefficient of 0.88, this was not a problem because these two variables were used in different models. overall, the correlation matrix suggested that thee regression model did not suffer from serious multicollinearity problems, which was also confirmed using the variance inflation factor (vif) test (mean vif value = 1.43). 86 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 ta bl e 2 su m m ar y st at is tic s of th e o bs er ve d va ri ab le s o ve r th e ye ar s of 2 00 0 to 2 01 8 sa m pl e 1 sa m pl e 2 (e xc lu di ng c hi na ) m ea n st d. d ev . m in m ax m ea n st d. d ev . m in m ax p an el a : r en ew ab le e ne rg y (m eg aw at ts ) 25 .,9 11 .3 10 0 86 ,2 36 .4 90 0 10 .1 00 0 69 5, 83 1. 10 00 7, 18 3. 33 90 16 ,1 43 .6 60 0 10 .1 00 0 11 7, 95 5. 00 00 r en ew ab le e ne rg y pe r c ap ita 19 6. 81 44 45 3. 56 73 0. 82 22 2, 23 8. 76 30 19 6. 66 01 46 9. 23 5 0. 82 22 2 2, 23 8. 76 30 c ru de o il pr ic e (u s$ ) 62 .8 42 6 27 .6 24 2 24 .3 50 0 10 5. 01 00 62 .8 42 6 27 .6 28 2 24 .3 50 0 10 5. 01 00 c oa l p ri ce (u s$ ) 69 .2 22 6 30 .8 56 8 25 .3 10 0 12 7. 10 00 69 .2 22 6 30 .8 61 3 25 .3 10 0 12 7. 10 00 n at ur al g as p ri ce (u s$ ) 4. 67 68 1. 91 55 2. 49 00 8. 92 00 4. 67 68 1. 91 58 2. 49 00 8. 92 00 c o 2 e m is si on s (m il) 73 3. 71 38 2, 01 2. 39 40 0. 29 05 10 ,0 64 .6 90 0 22 0. 33 68 46 4. 41 83 0. 29 05 2, 65 4. 10 10 c o 2 e m is si on s pe r c ap ita 2. 24 88 2. 55 57 0. 16 25 15 .1 15 2 1. 99 69 2. 43 53 0. 16 25 15 .1 15 2 g d p gr ow th (% ) 6. 59 57 2. 78 20 (1 .5 45 4) 17 .9 25 8 6. 69 74 2. 85 59 (1 .5 45 4) 17 .9 25 8 po pu la tio n gr ow th (% ) 1. 24 19 0. 47 03 0. 12 90 2. 32 46 1. 29 46 0. 44 58 0. 12 90 2. 32 46 u ne m pl oy m en t r at e (% ) 3. 36 31 2. 04 16 0. 39 30 8. 76 00 3. 28 29 2. 09 55 0. 39 30 8. 76 00 p an el b l nr en ew ab le e ne rg y 7. 73 12 2. 34 42 2. 31 25 13 .4 52 9 7. 38 38 2. 04 51 2. 31 25 11 .6 78 1 l nr en ew ab le e ne rg y pe r c ap ita 3. 92 91 1. 73 26 -0 .1 95 7 7. 71 37 3. 84 25 1. 75 81 -0 .1 95 7 7. 71 37 l nc ru de o il 4. 03 11 0. 48 73 3. 19 25 4. 65 41 4. 03 11 0. 48 74 3. 19 25 4. 65 41 l nc oa l 4. 12 23 0. 50 42 3. 23 12 4. 84 50 4. 12 23 0. 50 43 3. 23 12 4. 84 50 l nn at ur al g as 1. 46 77 0. 37 95 0. 91 23 2. 18 83 1. 46 77 0. 37 95 0. 91 23 2. 18 83 l nc o 2 em is si on 4. 00 31 2. 50 29 (1 .2 36 2) 9. 21 68 3. 63 05 2. 18 78 (1 .2 36 2) 7. 88 39 l nc o 2 em is si on p er c ap ita 0. 20 10 1. 13 67 (1 .8 17 2) 2. 71 57 0. 08 92 1. 09 86 (1 .8 17 2) 2. 71 57 o bs er va tio ns 26 6 24 7 n um be r o f c ou nt ri es 14 13 87 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 ta bl e 3 m ea n va lu es o f k ey v ar ia bl es b y c ou nt ry r en ew ab le en er gy (m eg aw at ts ) r en ew ab le en er gy p er ca pi ta c o 2 e m is si on s (m il) c o 2 e m is si on s pe r c ap ita g d p gr ow th (% ) po pu la tio n gr ow th (% ) u ne m pl oy m en t r at e (% ) u pp er -m id dl e in co m e c hi na 26 9, 37 4. 90 00 19 8. 82 03 7, 40 7. 61 40 5. 52 32 5. 27 32 0. 55 76 4. 40 56 m al ay si a 4, 23 3. 17 90 14 9. 61 14 19 8. 73 04 7. 14 21 5. 26 56 1. 73 81 3. 31 05 sr i l an ka 1, 52 4. 54 20 75 .3 01 8 14 .8 48 8 0. 73 08 5. 34 81 0. 78 62 5. 81 51 t ha ila nd 5, 04 0. 07 40 74 .7 62 1 24 2. 03 41 3. 61 86 4. 05 61 0. 57 03 1. 13 33 lo w er -m id dl e in co m e b an gl ad es h 29 0. 24 74 1. 97 39 51 .9 90 3 0. 35 08 6. 05 02 1. 33 57 4. 09 32 b hu ta n 1, 12 0. 06 30 1, 61 1. 23 00 0. 62 77 0. 90 26 7. 20 61 1. 40 29 2. 67 73 c am bo di a 35 3. 23 16 22 .7 10 7 4. 99 44 0. 33 96 9. 13 63 1. 64 57 0. 88 42 in di a 55 ,6 18 .0 20 0 44 .6 27 5 1, 68 2. 70 90 1. 36 40 6. 63 07 1. 39 31 5. 56 23 in do ne si a 6, 73 2. 58 90 27 .8 29 0 41 9. 25 53 1. 73 46 7. 79 06 1. 31 18 5. 82 77 l ao p d r 2, 18 9. 01 60 33 6. 27 62 5. 11 35 0. 77 11 7. 16 88 1. 57 49 1. 05 13 m on go lia 55 .8 73 7 19 .0 65 7 17 .9 92 8 6. 38 94 9. 93 45 1. 51 73 6. 00 52 m ya nm ar 1, 73 7. 61 10 33 .6 66 6 14 .4 76 9 0. 28 42 6. 93 54 0. 79 47 0. 87 05 ph ili pp in es 4, 96 3. 86 30 53 .4 98 5 88 .0 37 8 0. 94 14 5. 09 43 1. 76 08 3. 44 38 v ie tn am 9, 52 5. 10 00 10 6. 02 79 12 3. 56 83 1. 39 09 6. 44 98 0. 99 81 2. 00 37 88 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 ta bl e 4 c or re la tio n m at ri x ln r e l nr e pc l nc ru de o il l nc oa l l nn at ur al ga s l nc o 2 l nc o 2 pe r ca pi ta g d p gr ow th po pu la tio n gr ow th u ne m pl oy m en t l nr en ew ab le en er gy (l nr e ) 1. 00 00 l nr en ew ab le en er gy p er c ap ita (l nr e pc ) 0. 52 21 ** * 1. 00 0 l nc ru de o il 0. 13 50 ** 0. 15 88 ** * 1. 00 00 l nc oa l 0. 16 89 ** * 0. 19 95 ** * 0. 87 72 ** * 1. 00 00 l nn at ur al g as -0 .1 59 1* ** -0 .1 95 3* ** 0. 07 47 -0 .0 96 4 1. 00 00 l nc o 2 0. 75 26 ** * -0 .0 45 8 0. 08 06 0. 10 42 * -0 .0 87 3 1. 00 00 l nc o 2 p er ca pi ta 0. 39 05 ** * 0. 34 66 ** * 0. 14 11 * 0. 18 51 ** * -0 .1 61 9* ** 0. 58 00 ** * 1. 00 00 g d p gr ow th -0 .3 81 3* ** -0 .2 21 1* ** 0. 06 77 -0 .0 08 1 0. 21 07 ** * -0 .2 70 8* ** -0 .1 47 0* * 1. 00 00 po pu la tio n gr ow th -0 .3 75 5* ** -0 .1 33 5* * -0 .1 92 8* ** -0 .2 14 3* ** 0. 05 01 -0 .3 11 8* ** -0 .1 15 6* 0. 14 55 ** 1. 00 00 u ne m pl oy m en t (u ne m p) 0. 10 71 * -0 .1 48 5* * -0 .1 06 1* -0 .1 13 1* 0. 11 58 * 0. 34 25 ** * 0. 30 69 ** * 0. 08 29 0. 02 17 1. 00 00 n ot es . * , * *, a nd * ** im pl y th e re je ct io n of th e nu ll hy po th es is a t 1 0% , 5 % , a nd 1 % s ig ni fic an ce le ve ls , r es pe ct iv el y. 89 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 table 5 reports the estimates of the regression analysis. based on models 1 and 2, crude oil and coal prices were positively related to renewable energy capacity. this positive relationship implied that as the prices of crude oil and coal increased, there was a higher demand for renewable energy to substitute fossil fuels (apergis & payne, 2014; bird et al., 2005; chang et al., 2009), seemingly driven by cost concerns. it is argued that when fossil fuel prices are high, additional fuel subsidies have to be incurred to continue supplying affordable energy, especially for lower-middle-income countries. nonetheless, this is deemed as unsustainable in the long run. instead, the emphasis should be on allocating subsidies to renewable energy sources to improve the cost competitiveness of renewable energy as compared to non-renewable fossil fuels (carley, 2009). gradually, the lower cost of renewable energy would replace fossil fuels like crude oil and coal, which have rising costs due to increased fossil fuel consumption and high risk of resource scarcity (kaberger, 2018). this is specifically true for asian developing countries with rapidly growing populations that depend on fossil fuels. these statistics support hypothesis 1. on the other hand, natural gas prices were negatively related to renewable energy capacity. a potential explanation for this inverse relationship is that natural gas is a relatively clean fossil fuel as compared to crude oil and coal. this suggests that renewable energy could be more of a complementary energy source (sadorsky, 2009a; omri & nguyen, 2014). this study found consistent evidence in support of hypothesis 2 that co2 emission was positively related to renewable energy capacity, which was in line with existing studies (aguirre & ibikunle, 2014; omri & nguyen, 2014; sadorsky 2009a; salim & rafiq, 2012). a consistent positive relationship was reported when co2 emissions per capita were used as proxy for co2 emissions (refer to model 4). therefore, the higher the co2 emissions, the more urgent the call becomes for a country to take drastic initiatives to promote and increase the renewable energy capacity. between fossil fuels and co2 emission, the effect of co2 emission on renewable energy outweighed that of fossil fuels. for the control variables, only gdp growth and population growth were significant. the negative coefficients of gdp growth suggested that higher gdp growth led to lower renewable energy capacity. this discovery contradicted with existing research, such as that of przychodzen and przychodzen (2020), with evidence from transitional economies in central and eastern europe, caucasus, and central asia. 90 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 ta bl e 5 f os si l f ue l p ri ce , c ar bo n d io xi de e m is si on a nd r en ew ab le e ne rg y sa m pl e 1 sa m pl e 2 (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) (9 ) l nc ru de o il 0. 16 98 ** 0. 19 45 ** 0. 18 63 ** (0 .0 31 3) (0 .0 15 6) (0 .0 24 7) l nc oa l 0. 19 86 ** 0. 22 65 ** * 0. 20 14 ** (0 .0 18 5) (0 .0 08 6) (0 .0 23 2) l nn at ur al g as -0 .3 35 9* ** -0 .3 45 9* ** -0 .3 12 1* ** (0 .0 00 5) (0 .0 00 5) (0 .0 02 4) l nc o 2 em is si on 1. 13 96 ** * 1. 10 25 ** * 1. 09 59 ** * 1. 10 96 ** * 1. 07 64 ** * 1. 07 13 ** * (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) l nc o 2 pe r c ap ita 1. 30 63 ** * 1. 25 52 ** * 1. 25 42 ** * (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) g d p gr ow th -0 .0 34 2* * -0 .0 31 5* * -0 .0 20 0 -0 .0 37 8* ** -0 .0 34 6* * -0 .0 22 8 -0 .0 34 2* * -0 .0 31 3* * -0 .0 21 5 (0 .0 15 1) (0 .0 23 9) (0 .1 54 1) (0 .0 08 5) (0 .0 15 0) (0 .1 13 3) (0 .0 17 2) (0 .0 28 7) (0 .1 38 2) po pu la tio n gr ow th -0 .2 29 1 -0 .1 91 6 -0 .3 01 4* * -0 .3 33 5* * -0 .2 87 5* -0 .4 19 6* ** -0 .2 22 3 -0 .1 94 0 -0 .3 07 1* * (0 .1 23 2) (0 .2 10 1) (0 .0 28 0) (0 .0 27 9) (0 .0 66 7) (0 .0 02 6) (0 .1 43 4) (0 .2 15 7) (0 .0 29 1) (c on tin ue d) 91 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 sa m pl e 1 sa m pl e 2 (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) (9 ) u ne m pl oy m en t 4. 80 14 5. 07 26 5. 49 86 5. 01 49 5. 28 00 5. 28 57 4. 61 76 4. 64 50 4. 48 29 (0 .2 93 0) (0 .2 66 2) (0 .2 15 1) (0 .2 85 8) (0 .2 60 2) (0 .2 47 5) (0 .3 32 0) (0 .3 29 1) (0 .3 31 3) c hi na -0 .8 81 1 -0 .7 04 4 -0 .7 36 2 3. 15 74 3. 21 07 3. 14 42 (0 .6 33 3) (0 .7 03 2) (0 .6 89 3) (0 .1 17 7) (0 .1 11 5) (0 .1 19 0) u pp er m id dl e -0 .6 10 7 -0 .5 51 5 -0 .5 55 7 -0 .9 00 3 -0 .8 19 5 -0 .8 35 7 -0 .5 75 4 -0 .5 22 7 -0 .5 32 4 (0 .5 56 3) (0 .5 94 9) (0 .5 91 5) (0 .4 35 8) (0 .4 78 4) (0 .4 69 1) (0 .5 79 3) (0 .6 14 3) (0 .6 07 2) c on st an t 3. 07 09 ** * 2. 98 22 ** * 4. 37 02 ** * 7. 21 12 ** * 6. 95 80 ** * 8. 49 49 ** * 3. 10 28 ** * 3. 07 42 ** * 4. 46 96 ** * (0 .0 00 0) (0 .0 00 1) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 1) (0 .0 00 1) (0 .0 00 0) r -s qu ar ed 0. 54 45 0. 54 87 0. 54 90 0. 29 69 0. 30 06 0. 30 17 0. 43 09 0. 43 14 0. 43 22 w al d c hi 2 (p -v al ue ) 0. 00 00 0. 00 00 0. 00 00 0. 00 00 0. 00 00 0. 00 00 0. 00 00 0. 00 00 0. 00 00 o bs er va tio ns 26 6 26 6 26 6 26 6 26 6 26 6 24 7 24 7 24 7 n um be r o f co un tr ie s 14 14 14 14 14 14 13 13 13 n ot es . * , * *, a nd * ** i m pl y th e re je ct io n of t he n ul l hy po th es is a t 10 % , 5 % , a nd 1 % s ig ni fic an ce l ev el s, r es pe ct iv el y. f ig ur es i n pa re nt he se s ar e pva lu es . 92 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 their argument was consistent with apergis and payne (2014), fan and hao (2020), gan and smith (2011), marques et al. (2010), and sadorsky (2009a), whereby high income countries are more likely to invest in renewable energy because these countries have the capacity to afford the cost of renewable energy technologies. it is argued that the conflicting results were mainly due to the different development goals of lower-middle-income countries from those of high-income countries. for example, from table 3, even though the lower-middleincome countries such as cambodia (9.13% ) and mongolia (9.93%) reported the highest mean gdp growth, their mean renewable energy capacities were among the lowest in the sample. models 3 through 6 indicated that population growth was related to lower renewable energy capacity, which agreed with the work by aguirre and ibikunle (2014). high population growth translates to high energy demand. renewable energy is a new concept for many developing countries, and thus the supply may not be sufficient or consistent enough to meet the demands of an entire country’s population. to provide sufficient energy supply, these countries must depend on fossil fuels, which are more affordable, instead of renewable energy. the analysis was then repeated using sample 2, which excluded china. the results are reported in models 7 through 9. the analysis was also repeated using the natural logarithm of renewable energy capacity per capita as the dependent variable. consistently, the results provided evidence in support of hypotheses 1 and 2. although the results are not included here for brevity, they are available upon request. table 6 reports the results of the sub-sample analysis to control any potential bias due to differences in the countries’ size and income levels. the regression analysis was repeated using the lower-middleincome countries, upper middle-income countries, and upper middleincome countries excluding china. consistently, results from these three sub-samples were in line with those reported in table 5. therefore, it can be concluded that the capacity of renewable energy increases when the fossil fuels become more expensive, especially crude oil and coal. in simple terms, cost disadvantage is a significant concern. co2 emissions are another significant reason for these countries to increase renewable energy capacity, which would also mean increasing investment in renewable energy. 93 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 ta bl e 6 su bs am pl e a na ly si s l ow er -m id dl e in co m e u pp er -m id dl e in co m e u pp er -m id dl e in co m e (e xc lu di ng c hi na ) (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) (9 ) l nc ru de o il 0. 24 95 ** 0. 26 59 ** * 0. 18 23 ** (0 .0 13 3) (0 .0 03 1) (0 .0 34 6) l nc oa l 0. 26 64 ** 0. 36 42 ** * 0. 25 86 ** * (0 .0 14 5) (0 .0 00 0) (0 .0 00 8) l nn at ur al g as -0 .3 37 7* * -0 .5 51 4* ** -0 .4 28 8* ** (0 .0 12 0) (0 .0 00 0) (0 .0 00 0) l nc o 2 em is si on 1. 10 45 ** * 1. 06 51 ** * 1. 08 47 ** * 0. 62 77 ** * 0. 61 06 ** * 0. 63 10 ** * 0. 47 47 ** * 0. 48 26 ** * 0. 45 34 ** * (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) g d p gr ow th -0 .0 31 1* -0 .0 26 8 -0 .0 16 6 -0 .0 24 7 -0 .0 25 4 -0 .0 01 2 -0 .0 25 4 -0 .0 24 6 -0 .0 15 4 (0 .0 87 8) (0 .1 40 5) (0 .3 69 9) (0 .2 64 1) (0 .2 08 6) (0 .9 48 3) (0 .1 62 3) (0 .1 46 8) (0 .3 13 8) po pu la tio n gr ow th -0 .1 25 4 -0 .0 80 4 -0 .2 43 8 -0 .3 44 3* ** -0 .3 08 2* ** -0 .3 70 8* ** -0 .2 24 1* * -0 .2 16 6* ** -0 .2 10 7* ** (0 .5 24 8) (0 .6 93 6) (0 .1 94 6) (0 .0 00 8) (0 .0 01 2) (0 .0 00 0) (0 .0 11 5) (0 .0 09 7) (0 .0 05 5) u ne m pl oy m en t 4. 69 01 4. 55 71 5. 26 48 12 .9 87 2* ** 12 .7 59 7* ** 12 .8 19 6* ** 4. 22 99 5. 19 82 2. 91 50 (0 .5 05 6) (0 .5 17 6) (0 .4 57 4) (0 .0 02 5) (0 .0 01 4) (0 .0 00 5) (0 .3 22 8) (0 .1 94 0) (0 .4 04 1) c hi na 1. 22 10 ** * 1. 31 56 ** * 1. 18 65 ** * (0 .0 02 1) (0 .0 00 4) (0 .0 00 5) c on st an t 2. 70 73 ** * 2. 65 88 ** * 4. 31 45 ** * 4. 15 29 ** * 3. 77 38 ** * 5. 93 69 ** * 5. 35 23 ** * 4. 94 09 ** * 6. 79 40 ** * (0 .0 03 3) (0 .0 04 4) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) (0 .0 00 0) r -s qu ar ed 0. 41 33 0. 41 29 0. 41 06 0. 97 11 0. 97 50 0. 97 85 0. 80 96 0. 83 03 0. 86 06 w al d c hi 2 (p -v al ue ) 0. 00 00 0. 00 00 0. 00 00 0. 00 00 0. 00 00 0. 00 00 0. 00 00 0. 00 00 0. 00 00 o bs er va tio ns 19 0 19 0 19 0 76 76 76 57 57 57 n um be r o f c ou nt ri es 10 10 10 4 4 4 3 3 3 n ot es . * , * *, a nd * ** i m pl y th e re je ct io n of t he n ul l hy po th es is a t 10 % , 5 % , a nd 1 % s ig ni fic an ce l ev el s, r es pe ct iv el y. f ig ur es i n pa re nt he se s ar e pva lu es . 94 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 concluding remarks even though asia has great potential for renewable energy resources, the potential is relatively untapped, particularly among developing countries that are highly dependent on fossil fuels for energy. this mainly results from cost disadvantage. in addition, co2 emissions are high in these countries. this study shows that fossil fuel prices and co2 emission provide significant motivation for the adoption of renewable energy, which can substitute crude oil or coal but not for natural gas, for both the upper-middle-income and lower-middleincome countries. the findings further supported the call to reduce fossil fuel dependency and co2 emissions. policymakers should revisit the existing fossil fuel subsidies regulations and respond to the global call to remove fossil fuel subsidies progressively so that fossil fuels are no longer a cheaper energy alternative. instead, the subsidies should be allocated to finance investment in renewable energy development to increase the cost competitiveness of renewable energy sources. lower cost translates to better investment returns and could attract renewables investment from the private sector. however, a lack of legislation would discourage investment in renewable energy from the private sector. this paper proposes that governments and policymakers consider stakeholders’ interests and shareholders’ protections in their decisions to attract not only local investments from the private sector but also foreign investment. next, this study found that high co2 emission led to high renewable energy capacity. it is argued that these developing countries are under pressure to decrease their high co2 emissions due to environmental concerns. however, the transition to renewable energy is still slow. therefore, the second call for legislative change lies in the importance of developing policies and regulations that restrict greenhouse gas emissions to motivate the adoption of renewable energy to mitigate the adverse consequences of climate change. note that due to data limitations, the study only included 14 asian developing countries. thus, it is recommended for future studies to consider more asian countries in the sample of examination. future studies can also conduct a comparison study between developing and developed countries and/or examine renewable energy by types such as wind, hydro, geothermal, biomass, and others. 95 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 acknowledgment this research received no specific grant from any funding agency. references aguirre, m., & ibikunle, g. (2014). determinants of renewable energy growth: a global sample analysis. energy policy, 69, 374–384. apergis, n., & payne, j. e. (2014). renewable energy, output, co2 emission, and fossil fuel prices in central america: evidence from a nonlinear panel smooth transition vector error correction model. energy economics, 42, 226–232. bird, l., bolinger, m., gagliano, t., wiser, r., brown, m., & parsons, b. (2005). policies and market factors driving wind power development in the united states. energy policy, 33(11), 1397–1407. carley, s. (2009). state renewable energy electricity policies: an empirical evaluation of effectiveness. energy policy, 37(8), 3071–3081. chang, t. h., huang, c. m., & lee, m. c. (2009). threshold effect of the economic growth rate on the renewable energy development from a change in energy price: evidence from oecd countries. energy policy, 37, 5796–5802. gan, j. b., & smith, c. t. (2011). drivers for renewable energy: a comparison among oecd countries. biomass and bioenergy, 35, 4497–4503. eckstein, d., künzel, v., schafer, l., & winges, m. (2019, december 4). global climate risk index 2020. who suffers most from extreme weather events? weather-related loss events in 2018 and 1999 to 2018. https://germanwatch.org/sites/germanwatch. org/files/20-2-01e%20global%20climate%20risk%20 index%202020_10.pdf fan, w., & hao, y. (2020). an empirical research on the relationship amongst renewable energy consumption, economic growth and foreign direct investment in china. renewable energy, 146, 598–609. gozgor, g., mahalik, m. k., demir, e., & padhan, h. (2020). the impact of economic globalization on renewable energy in the oecd countries. energy policy, 139, 111365. 96 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 79–96 international energy agency. (2020). energy subsidies: tracking the impact of fossil-fuel subsidies. https://www.iea.org/topics/ energy-subsidies international renewable energy agency. (2020, march 31). renewable capacity statistics 2020. https://www.irena.org/ publications/2020/mar/renewable-capacity-statistics-2020 kariuki, d. (2018, january 25). barriers to renewable energy technologies development. energy today. https://doi.10.1515/ energytoday-2018-2302 kaberger, t. (2018). progress of renewable electricity replacing fossil fuels. global energy interconnection, 1, 48–52. marques, a. c., fuinhas, j. a., & manso, j. r. p. (2010). motivations driving renewable energy in european countries: a panel data approach. energy policy, 38, 6877–6885. omri, a., & nguyen, d. c. (2014). on the determinants of renewable energy consumption: international evidence. energy, 72, 554–560. przychodzen, w., & przychodzen, j. (2020). determinants of renewable energy production in transition economies: a panel data approach. energy, 191, 116583. rustemoglu, h., & andres, a. r. (2016). determinants of co2 emissions in brazil and russia between 1992 and 2011: a decomposition analysis. environmental science and policy, 58, 95–106. sadorsky, p. (2009a). renewable energy consumption, co2 emissions and oil prices in the g7 countries. energy economics, 31, 456–462. sadorsky, p. (2009b). renewable energy consumption and income in emerging economies. energy policy, 37(10), 4021–4028. salim, r. a., & rafiq, s. (2012). why do some emerging economies proactively accelerate the adoption of renewable energy? energy economics, 34, 1051–1057. sisodia, g. s., & soares, i. (2014). panel data analysis for renewable energy investment determinants in europe. applied economics letters, 22, 397–401. international journal of banking and finance (ijbf) is a refereed journal that publishes two issues every year in january and july by the uum press, universiti utara malaysia. ijbf is indexed by the malaysian citation index (myjurnal), american economic association (econlit), british library, asian digital library (adl) and index corpernicus international. it is an open access journal in which articles on this site are available in full-text and free of charge to our web visitors. this website does not require any personal information about its visitors to read, download, copy, distribute, print, search, or link to the full texts of these articles. ijbf welcomes international scholars and practitioners in banking, finance and insurance to disseminate their theoretical, empirical and applied research through anonymously doubleblind peer review process. the journal provides a global forum wherein academicians and practitioners share ideas, elucidate practices and enhance the existing theoretical and practical positions in the following niche areas: • behavioural economics • macroeconomics • corporate finance • monetary policy • financial institutions and markets • financial economics • financial services, • portfolio and security analysis • insurance, • real estate, ethical finance • international economics and trade • money and banking • international finance • accounting and financial reporting authors are required to strictly adhere to the journal manuscript format. the format is available at http://e-journal.uum.edu.my/index.php/ijbf/manuscript-guideline. all manuscripts should be submitted through the ijbf editorial system by the corresponding author. submission can be made at http://e-journal.uum.edu.my/index.php/ijbf/guidelineauthors authors are responsible to ensure that any manuscript submitted to this journal should be original, which means it should not have been published before in its current or similar form. authors have to make sure that the manuscript submitted has not been submitted simultaneously for reproduce copyrighted material from other sources. if any substantial element of the article has been previously published, the author has to declare this to the journal editor upon submission. please note that the journal editor may check on the originality of submissions received. in ensuring that the articles published meet acceptable quality standards, the manuscripts will undergo a double-blind peer review process. authors will be informed about the acceptance or rejection of manuscript by the managing editor. please refer to http://www.ijbf.uum.edu.my/index.php for further information regarding ijbf. call for papers the international journal of banking and finance (ijbf) 109 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: wenhao, m., jaidi, j., & mohidin, r. (2022). board independence, corporate social responsibility and firm performance: evidence from china. international journal of banking and finance, 17(1), 109 130. https://doi.org/10.32890/ ijbf2022.17.1.5 board independence, corporate social responsibility and firm performance: evidence from china 1miao wenhao, 2junainah jaidi & 3rosle mohidin faculty of business, economics and accountancy universiti malaysia sabah, malaysia 2corresponding author: nenjaidi@ums.edu.my received: 8/12/2020 revised: 8/4/2021 accepted: 16/4/2021 published: 2/12/2021 abstract the purpose of this paper was to investigate the relationship between board independence and the firm performance of chinese firms listed in the shanghai stock exchange, under the moderating role of corporate social responsibility (csr). a total of 860 firm-year observations over a period of ten years, that is from 2010 to 2019 was collected. the panel data regression technique was employed to analyze the data and determine the relationship between board independence and the firm performance of the chinese firms under investigation. after a robustness check, the empirical results showed that the level of the csr moderated (reduced) the positive relationship between board independence and firm performance. therefore, the results seemed to imply that although the csr has been seen as a useful business strategy, the level of the csr in china still needed to be improved. 110 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 in order to improve firm performance through practicing the csr, the chinese government and enterprises should be encouraged to continuously improve the level of the csr. keywords: board independence, firm performance, csr, panel data regression, china stock market. jel classification: g3. introduction the world market in the 21st century has already undergone profound and complex changes due to the accelerated development of economic globalization. today’s enterprises are facing great challenges and even cutthroat market competition as a result of the unstable and uncertain business environment in this new century (taouab, 2019). it seems that high and stable business performance is the key driving force for the firm’s long-term sustainability in today’s dynamic business environment. the rapid growth in the capital market in asia, especially in china, has emphatically underscored the contribution of corporate governance (cg) performs in influencing future firm’s performance. corporate governance is considered an important element in improving a firm’s performance. li et al. (2012) highlighted that the topic of corporate governance has become a central discussion among academicians after the 1997 asian financial crisis as a result of frequent outbreaks of disgraceful incidents such as the infamous enron and worldcom scandals. as a result of these corporate scandals, corporate governance has become increasingly important for firms to establish reasonable and good practices of corporate governance mechanisms. according to al-matari et al. (2014), good corporate governance played a momentous role in preventing the likelihood of a financial crisis, attract more investment for enterprises to maximize its capital, as well as to solidify the pillars of the firm, which could boost firm’s performance. it is undeniable that independent directors have a significant role in corporate governance codes and guidance (merendino & melville, 2019). jensen and meckling, (1976) stated that directors had to be independent as they represent the interests of shareholders, and their main function was to monitor the behaviour of managers to protect the shareholders’ interests. in response to 111 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 the widespread concern of corporate governance issues, most of the governance reforms implemented by various countries have mainly concentrated on the board of directors’ effectiveness in performing their responsibility as board directors. it has been suggested that there is an urgent need to raise the number of independent directors in the firm so that the board could be more effectively play a supervisory role (gaur et al., 2015). according to report published by the organization for economic co-operation & development (oecd) in 2004, good corporate governance should be encompassed of an active board of directors, chairperson and chief executive officer (ceo) should be held by two different persons, outside directors should constitute the majority members of the board, and there should be a two-tier board which to be the supervisory board ( tier-one) and the other being the board of directors as tier-two. in terms of the role of independent directors in china, one of the key provisions of the corporate governance code has been the requirement that the composition of boards should consist of outside directors or independent directors. in august 2001, the china securities regulatory commission (csrc) had published a guidelines that highlighted on the requirement of independent directors sitting on the company’s board in listed companies in china. it is clearly stated in the guidelines that all listed companies in china are required to have at least one-third of their board members as independent directors by 2003. as a consequence, both theory and practice emphasized the importance of board independence. kao et al. (2019) argued that independent directors in china could indeed reduce agency problems, as these independent directors could be regarded as a bridge between managers and stakeholders. however, dian (2014) claimed that there was no conclusive research finding on the connection between firm performance and board independence in the unique setting of the chinese listed firms. the reason behind that is whether the so-called good corporate governance practices depend on whether those practices fit into the institutional environment. empirical findings from the west might not necessarily have a positive impact on the firm performance of chinese companies because of the different institutional environments and culture, such as the role of the independent directors. in western countries, the agency cost between internal managers and external shareholders could be effectively mitigated by board independence. however, in the case of chinese companies, controlling-shareholder expropriation, which has become an increasingly common phenomenon, meant 112 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 that independent directors were usually not independent to large or controlling shareholders, and even considered as having a low status and weak power. under these circumstances, the independent directors could not play an effective supervisory role in a company, and therefore, will not be able to achieve the independent directors’ goal of helping to enhance firm performance. therefore, due to the institutional differences between advanced and developing countries, this paper will fill the research gap by examine the association between board independence and firm performance, especially in the case of growing economies like china. given the foregoing description of the background situation, the present study has, therefore, sought to examine the relationship between board independence and firm performance in the context of listed firms in china can be moderated by the corporate social responsibility (csr) disclosure. the idea of csr, which was initiated from the west, has created a mixed reaction in china market. as a consequence of the imposition of the chinese firm law in 2005, which required that enterprises had to undertake social responsibilities while engaging in business activities. yang et al. (2019) has highlighted that since the 20th century, the field of csr has begun to develop in value creation, especially the business value it could bring, including attracting, retaining and motivating valuable employees; reducing costs by saving energy and reducing a firm’s unnecessary inputs; promoting corporate innovation; improving firm brand image by developing new products and providing services to help deal with social problems. yang et al. (2019) further pointed out another important reason why the csr issue has received much attention; it is basically because csr could promote sustainable corporate development. this view was in line with mohr et al.(2001), who argued that csr is seen as a commitment by companies to maximize and minimize their longterm positive and negative impact on society. similarly, numerous past studies support that csr was related to a firm’s sustainable development (cheng et al., 2016a; feng et al., 2017; liu and zhang, 2016) and there was the consensus that firm performance would be improved along with the enhancement of corporate competitiveness as a result of implementing the csr (famiyeh, 2017). therefore, awareness of the critical role of social responsibility incorporates sustainable development has been increasing over the years, and 113 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 firms wanting to undertake more social responsibility has become an indispensable part of the corporate management system when ensuring a firm’s sustainable development. however, kao et al. (2018) believed that most empirical studies on csr have been concentrated on developed economies, such as the united states and europe, where the agency cost between managers and shareholders was low, so it was widely reported that csr has had a positive impact on firm performance. companies in emerging and developed markets were seen to practice organizational and behavioural differences as compared to firms from developed countries (fan et al., 2011). in particular, the study by yang et al. (2019) indicated that developing countries had incomplete csr frameworks, which has made it interesting to see how csr activities could affect firm performance in emerging markets such as china. the foregoing discussions of some past studies have shown that both board independence and csr did have a significant influence on firm performance. hence, it has become clear the relationship between board independence and firm performance can be moderated by the level of csr (either enhance or reduce). however, there remains a paucity of evidence on the role the csr disclosure as a moderator of the relationship between board independence and firm performance. therefore, this paper will fill this research gap through further investigating the contribution of csr in moderating the relationship between board independence and firm performance of listed firms in china. literature review theoretical discussion in this paper, the agency theory was employed to investigate the association between board independence and firm performance under the unique chinese market environment. according to hu, tam & tan (2010), chinese enterprise reform and corporate governance reform in china, were largely influenced by the agency theory. in china itself, china securities regulatory commission (csrc) is the authority given the responsibility to introduce the guidelines with regards to monitoring the role of independent directors, in addition, 114 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 chinese listed companies were required to have at least one-third of their board members as independent directors by 2003. referring to jensen and meckling (1976), the principal (shareholder) and the agent (manager) are agency relationships. under this agency relationship, the principal grants certain power to the agent, such as giving the agent decision-making power, and the agent can carry out certain services on behalf of the principal. however, the interests of the two parties are not always in alignment under this agency relationship. therefore, agency theorists of corporate governance have held the idea that under any given situation, it would be impossible for managers to maximize shareholders’ interests unless there was in place a proper governance structure (used to monitor costs) to protect shareholders’ interests. therefore, an appropriate corporate governance structure could guarantee that the managers would be able to act to maximize the interests of shareholders, instead of their interests. to protect the interests of shareholders, monitoring the managers’ behaviours has been considered as the primary function of the boards. therefore, the directors in the boards must be independent. in this paper, an independent director was defined as an outside director who had no business relationship with the company, except for having one seat on the board (shao, 2019). in terms of the csr-firm performance relationship, there is still an ongoing debate as to whether engaging the csr hurts or improves performance. in this paper, two quite different theories were used to explain the csr-firm performance relationship. firstly, based on the classic agency theory, the csr engagement was considered as a relationship between principal-agent or in other words relationship between shareholders and managers. according to this perspective, as was best exemplified by friedman (1970), it was believed that csr could be seen as a selfish act by corporate management to improve their public image at the expense of shareholders’ interests. barnea and rubin (2010) further argued that under the agency relationship, managers were interested in overinvesting in csr since managers could build a good reputation as social citizens through participating in csr, but it might damage the interests of shareholders and hurt firm performance. furthermore, barnea and rubin (2010) highlighted that the negative impact of the csr on firm performance was due to a large number of agency costs in the process of implementing the csr, and the high costs outweighed the benefit, which would then lower firm 115 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 performance and put the firm at an economic disadvantage. li et al. (2015) noted that these additional costs associated with implementing the csr might include substantial charitable contributions, extensive investments in projects to promote community development, and the cost of establishing environmental protection programs. contrary to the over-investment hypothesis based on the agency theory, the stakeholder theory (freeman & mcvea, 1984) has extended the agency problem to a multilateral relationship amongst all stakeholders, such as employees, stockholders, customers, community and other related stakeholders. under the perspective of the stakeholder theory, managers need to pay attention to all the groups that affect and are affected by the activities of the business, rather than simply serving the shareholders’ interests. the stakeholder theory highlighted the point that firms and society are interdependent. therefore, the implementation of csr enables firms to increase value. empirically, the study by jo and harjoto (2011) confirmed that managers’ active participation in csr indeed could eliminate conflicts among various stakeholders, and thus, enhance firm performance. consistent with the research results of jo and harjoto (2011), li et al. (2015) who based their study on the assumptions of the agency theory, studied whether ceos with considerable power were more likely to invest in more social responsibility to improve firm performance. the results showed that the implementation of the csr was value enhancement rather than value destruction, which supported the conflict resolution hypothesis based on the stakeholder theory, yet did not support the over-investment hypothesis based on the agency theory. in this paper, csr has been regarded as a broad concept, which referred to the responsibility that firms had to perform for its impact on society (cheng et al., 2016a). literature review and hypothesis development to date, there has been no consensus on how board independence could influence firm performance. fuzi fuzi et al. (2016) stated that independent directors are persons entrusted by shareholders to represent the organization. thus, independent directors should act in the interest of shareholders. the results showed that there was a mixed relationship between the proportion of independent directors and firm performance. although the company had one of the largest numbers of independent directors, this did not guarantee the improvement 116 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 of company performance. therefore, to bring positive value to shareholders, independent directors should be effectively supervised. numerous empirical studies of agency theory ( gaur et al., 2015; assenga et al., 2018; kao et al., 2019) have found that there was a significant positive relationship between board independence and firm performance. in particular, donadelli et al. (2014) documented the proportion of independent directors was positively correlated with firm performance the study employed data from eight countries, namely the usa, canada, france, germany, uk and australia, italy and japan. their findings were consistent with previous literature that independent directors were perceived to be better able to supervise managers and ceos (fama & jensen, 1983). kao et al. (2019) pointed out that the supervisory value of independent directors was more significant in a market with weak corporate governance mechanisms. therefore, the findings seemly to imply that the policy to reform the corporate governance of the independent director system was successful and that it had to be implemented by newly listed companies. liu et al. (2015) provided comprehensive evidence on the relationship between board independence and firm performance in chinese firms. the study found that independent directors had a significant positive impact on the firm performance of chinese enterprises. conversely, there have been some other studies that showed that there was a negative correlation or even no correlation between board independence and firm performance (allam, 2018; merendino & melville, 2019; rashid, 2018). merendino and melville (2019) attributed these negative findings to the adherence of italian independent directors to soft and hard laws on corporate governance, which would increase firm’s costs and thus, negatively impacted firm performance. another reason might be that the ceos could use some strategies to offset the power of independent directors, and as such the independent directors would be unable to play an effective role in the company. in addition, fan et al. (2020) has pointed out that the fundamental reason behind the value loss caused by the proportion of independent directors might be that enterprises were more willing to substitute the existing non-independent directors with new independent directors. new independent and non-independent directors might have the same qualifications, but engagement and responsibility toward shareholder maximization may differ. therefore, the short supply of qualified independent directors might explain the negative valuation effect. 117 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 considering the literature review of previous studies in the importance of understanding the role of board directors and shareholder wealth maximization, therefore, this paper hypothesis is: h1: there is a positive relationship between board independence and firm performance in the capital market in china. empirical studies on firm csr and firm performance by several scholars have generally shown mixed results. friedman’s (1970) classic argument was that csr was an agency problem and claimed that it had a negative impact on firm performance because it had imposed costs directly or implicitly. kao et. al (2018) also believed that the relationship between csr and firm performance was negative because opportunistic managers might pursue their private interests at the expense of the interests of shareholders and stakeholders. especially when a company’s performance has been poor, managers might engage in more social programs, such as csr activities, to cover up their disappointing performance results. in addition, brammer, brooks, and pavelin (2006) also found that the firm performance is affected natively by csr. in other different study, nelling and webb (2009) found that contradict finding where there was no significant connection between firm performance and csr. in supporting the effect of csr on firm performance, malik (2015) had documented the a positive relationship between csr disclosure and firm performance. the study of li et al. (2015), jo and harjoto (2011) showed that companies engaging in more csr activities could indeed improve firm performance as the csr activities could resolve conflicts between managers and stakeholders by reducing information asymmetry and agency costs. firms with a higher level of csr would bring additional social capital to firms, attract more investors for them and therefore, improve their firm performance. studies by yeh et al. (2019) and el ghoul et al. (2011) also disclosed that the csr activities as a useful business strategy, might increase the satisfaction of all stakeholders, strengthen the company brand image, and therefore, increase firm performance and lower the cost of capital. in the chinese context, many scholars, for example, chen and wang (2011), jo and harjoto (2011), famiyeh (2017); and rahman and fang (2019), have revealed that the more chinese firms invested 118 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 in the csr, the better its firm performance was. therefore, according to these previous studies, both board independence and csr could influence firm performance. the findings seemed to suggest that it would be reasonable to assume that the relationship between corporate governance and firm performance is moderated by csr the. in light of this consensus, the research reported hypothesis is: h2: the relationship between board independence and firm performance in the capital market in china is moderated by csr. methodology data and sample the data consists of 860 firm-year observations spanning from 2010 to 2019. panel data regression technique was used in order the determine the relationship between variables used. the sse 180 index is known as the core index among all the a-shares listed on the shanghai stock exchange. the sse 180 index is represented by the industry’s leading enterprises. this index acts as a market barometer of the overall performance and operation of the shanghai securities market. hence, this study used the individual stock composition represented by the index to represent the sample data. data for this study was collected from the accounting research database (csmar) and the china stock market. if there was missing data, a reference was made to the annual report. all the data collected from the database were analyzed through the e-views and stata software. model design this paper used a panel data analysis. both the fixed effect model and the random effect model are used in the analysis. hausman test was performed to detect the problem of endogeneity and the appropriateness of either the fixed effect model or the random effect model. the ordinary least squares (ols) or random effect could obtain a more efficient estimation will be employed if all the independent variables were exogenous. furthermore, the lagrange multiplier (lm) test was used to test whether the ols or the random effect model was more appropriate for this paper’s data set. according to the results, the econometric model was specified as follows: 119 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 (1) (i =1,…,n; t=1,…t) where, fpit represents the roa to measure firm performance for firm i at time t. in this paper, the roa has been defined as the ratio of consolidated net earnings to average assets. the econometric model of this paper has included three key variables, which were the bi, csr and interaction beta bi*csr respectively. if the interaction coefficient beta is significant, it would indicate that the existence of moderation effect. in this paper, board independence (bi)is computed the the percentage of the total number of independent directors sitting in the board. the csr was a moderator variable. following the study of yang et al. (2019), the csr was obtained via the score rating of a-shares listed in the firms’ csr reports, which could be found at the following url, http://www.hexun.com/. the quality of the csr disclosure is based on the csr score. according to yang et al. (2019) hexun evaluation system was more widely used in the performance evaluation of csr activities in chinese firms.. in addition, ceo duality (cd), ceo compensation(cc), state ownership (so), debt (debt) and board size (bs), and were employed as control variables in this paper, as these variables could also have an impact on firm performance (jiang, fuxiu & kim, 2015; kao et al., 2019). board size (bs) was obtained by calculating the total number of directors representing the board. dummy variable equal to 1 was used to represent ceo duality (cd) if the ceo acted as the chairman of the company’s board, and 0 otherwise (kao et al., 2019). on the other hand, ceo compensation (cc) is calculated as an average of top 3 executives’ salaries in the firm, excluding allowance received by them. state ownership (so) embodies the ratio of stateowned shares in proportion to the total number of shares (hu et. al., 2010). debt on the other hand was calculated by the total debt over the total assets to measure the firm’s capital structure. in addition to that, fsit represented the firm size, which was the control variable that was included to ensure the robustness of the analysis. the methodology of this study is adopted by kao et al. (2019). they also 120 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 used year and industrial dummy variables in the model able to capture the regulation effect. empirical analysis descriptive and correlation analysis table 1 shows the descriptive analysis variables used in this study. the range of firm performance (roa) is in between -11.9635 (minimum value)) and 47.7017 (maximum value), respectively with an average mean of 6.9889. this indicated that there are significant differences among these sample firms, and some firms demonstrated poor performance. in addition, the average proportion of independent directors (board independence) was 38.85 percent, and the median was 36.36 percent, which was in line with the provisions on the composition of the board of directors of listed companies in china (as of 2003, one-third of the board members must be independent). however, the minimum of bi was 12.50 percent, implying that a small number of companies in the sample did not have enough independent directors on their boards. in terms of csr, the quality of csr disclosure varied widely (the score of csr disclosure ranged from 0 to 85.77). table 1 the descriptive statistics of the main variables variables mean median maximum minimum std. dev. roa (%) 6.9889 5.6762 47.7017 -11.9635 6.1267 csr 37.4673 28.9400 85.7700 0.0000 20.8460 bi 0.3885 0.3636 0.8000 0.1250 0.0781 bs 9.5488 9.0000 17.0000 5.0000 1.9079 cd 0.1267 0.0000 1.0000 0.0000 0.3302 so 0.0544 0.0000 0.7682 0.0000 0.1406 incc 14.8625 14.7930 17.7457 11.8241 0.7664 debt 0.5127 0.5168 0.8858 6.1700 0.1842 infs 24.3930 24.3337 28.6364 19.7325 1.6891 observations 860 table 2 shows the variance inflation factor (vif) analysis for each variable. the results seemed to imply that the regression models 121 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 applied to test the research hypotheses were free from the problems of multi-collinearity effect, as all the vif values were less than 10 (ali et al., 2020). in addition to that, table 3 provides the correlation matrix among all key variables in the regression analysis. the correlation coefficient between all independent variables was -0.6090 to 0.2434, implying that there was no multicollinearity problem. based on the study by shao (2019), multi-collinearity problem might exist if the correlation between independent variables higher than the absolute value of 0.7. this is consistent with the results of the vif analysis in the present study. table 2 inflation factor statistics variables vif tolerance bi 1.31 0.7617 csr 1.04 0.9629 bs 1.27 0.7809 cd 1.06 0.9447 so 1.06 0.9456 cc 1.15 0.8725 debt 1.28 0.7809 fs 1.48 0.6766 mean vif 1.21 regression results the hausman-test (prob.=1.000) showed that the random effect model is appropriate for the data set. however, if one or more independent variables in the model correlated with the random disturbance term, the panel regression results might suffer from the problem of endogeneity. if all the independent variables were exogenous, the ordinary least squares (ols) or random effect (re) could obtain a more efficient estimation. to check the problem of endogeneity, the hausman test was carried out and the results x2 =11.08, prob=0.9953) showed that there were no endogeneity problems in the research model. furthermore, to test whether the ols and re were appropriate for the data set used in this research, the lagrange multiplier (lm) test was also conducted. the results x2 = 461.18, prob=0.000) strongly confirmed that the random effect 122 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 ta bl e 3 th e c or re la tio n m at ri x w ith p ro ba bi lit y of m ai n va ri ab le s in th e r eg re ss io n a na ly si s c or re la tio ns 1 r o a 2 b i 3 c sr 4 b s 5 c d 6 so 7 c c 8 d e b t 9 fs 1 r o a 1. 00 00 2 b i -0 .1 47 9* ** (0 .0 00 0) 1. 00 00 3 c sr 0. 18 84 ** * (0 .0 00 0) 0. 00 28 (0 .9 33 7) 1. 00 00 4 b s 0. 01 07 (0 .7 53 4) -0 .3 69 9* ** (0 .0 00 0) 0. 13 69 ** * (0 .0 00 1) 1. 00 00 5 c d 0. 09 82 ** (0 .0 04 0) 0. 00 72 (0 .8 32 8) -0 .0 69 4* * (0 .0 42 0) -0 .1 15 9* ** (0 .0 00 7) 1. 00 00 6 so -0 .0 23 1 (0 .4 97 8) 0. 05 70 * (0 .0 94 6) 0. 07 01 ** (0 .0 39 9) 0. 09 49 ** * (0 .0 05 3) -0 .0 82 0* * (0 .0 16 2) 1. 00 00 7 c c 0. 18 58 ** * (0 .0 00 0) -0 .0 26 3 (0 .4 41 0) 0. 07 87 ** (0 .0 21 0) 0. 05 69 * (0 .0 95 3) 0. 10 07 ** * (0 .0 03 1) -0 .1 61 7* ** (0 .0 00 0) 1. 00 00 8 d e b t -0 .6 09 0* ** (0 .0 00 0) 0. 21 16 ** * (0 .0 00 0) 0. 03 41 (0 .3 17 1) 0. 03 37 (0 .3 23 5) -0 .0 51 5 (0 .1 31 5) 0. 04 00 (0 .2 41 0) 0. 00 31 (0 .9 26 8) 1. 00 00 9 fs -0 .2 48 3* ** (0 .0 00 0) 0. 23 54 ** * (0 .0 00 0) 0. 10 72 ** * (0 .0 01 7) 0. 12 00 ** * (0 .0 00 4) -0 .1 42 2* ** (0 .0 00 0) 0. 04 71 (0 .1 67 7) 0. 24 34 ** * (0 .0 00 0) 0. 44 24 56 ** * (0 .0 00 0) 1. 00 00 n ot e. * ** , * *, * re pr es en t s ig ni fic an ce a t t he 0 .0 1 le ve l, 0. 05 le ve l, 0. 1 le ve l r es pe ct iv el y. 123 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 was valid. after controlling the endogeneity problems, the roe was used to replace the roa to measure the robustness test for firm performance. the specific results are as shown in table 4, which displays the influence of csr on the relationship between board independence and firm performance. table 4 the influence of the csr on the relationship between board independence and firm performance (with probability) dependent variable (dv): roa random effect bi 9.6891** (0.029) csr 0.1446***(0.000) bi*csr -0.2726***(0.001) bs 0.2930***(0.009) cd 1.0069* (0.052) so 0.2465 (0.824) cc 1.5085*** (0.000) debt -19.7961*** (0.000) fs 0.6775*** (0.004) c (constant term) -26.2766*** (0.000) industry yes year yes n r2 860 0.328 f-statistic 15.6296***(0.000) hausman-test for choosing model p=1.000 hausman-test for endogeneity problem x2 =11.08, prob=0.9953 lm test x2 = 461.18, prob=0.000 note. ***, **, * represent significance at the 0.01 level, 0.05 level, 0.1 level respectively. as shown in table 4, the coefficient of board independence was positive and statistically significant for the roa at the level of 1 percent, implying that board independence was positively related to firm performance. it is worth noting that the regression coefficient of the interaction term (bi*csr) was negatively related to firm performance at the significance level of 1 percent, indicating that 124 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 the csr could negatively moderate the positive effect of board independence on firm performance. in other words, csr has reduced the relationship between board independence and firm performance. in addition, there was a positive relationship between board size and firm performance; both ceo duality and ceo compensation could improve firm performance, and debt was negatively related to firm performance. discussion and conclusion the research reported here has studied the relationship between board independence and corporate performance under the moderating effect of csr. to test the relationship among these three variables, the study has adopted the panel data regression method to analyze the data from the sse 180 index of the capital market in china. the empirical results showed that board independence has had a positive impact on firm performance. more importantly, the results showed that csr could reduce the positive relationship between board independence and firm performance. the findings of the present research have made it possible to draw the following conclusions. firstly, it was found that board independence could improve firm performance, a standpoint that was consistent with the agency theory. according to the agency theory, outside directors would carry out their responsibility to monitor top management because they had the incentive to develop their reputation in decision control (fama & jensen, 1983). they were believed to bring more diversity and greater objectivity to decision-making, and consequently were better representatives of shareholders’ interests (mutlu et al., 2018). therefore, the possibility of collusion and expropriation of shareholder wealth by top management might be lowered with a greater proportion of outside directors on the board, which further reduce the agency costs and improve firm performance. more importantly, in terms of the moderating role of the csr, the interaction term bi*csr has a negative impact on firm performance, which means that the csr reduces the positive relationship between board independence and firm performance. according to sial et al. (2018), one might be the possible reason is that the development of chinese firms was at a stage of excitement ed a feeling for quick success and quick profits. with the rapid development of china’s economy, most firms practice disclosure of csr only to meet the requirements 125 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 of the government. the boundary of csr has not been captured by chinese firms. as a result, a lot of firms have been indecisive about what csr actions to undertake and what csr activities to report. inevitably, the managers or shareholders may sometimes overinvest in the csr or cover up a firm’s improper behaviours in pursuit of their interests, engaging in themselves in negative actions which will maximizing the shareholder wealth. because of china’s relevant laws, regulations and policies which are less compromise to encourage corporate social responsibility, the csr did not effectively optimize the corporate governance structure, which in turn, could not improve firm performance. although the number of csr disclosure has been increasing, the quality of the csr was low and many firms published information that lacked thirdparty certification (cheng et al., 2016b). therefore, managers should carefully consider csr and invest in csr according to a firm’s ability when they make csr decisions. the study recommended that to strengthen the relationship between board independence and firm performance through csr, there are alternatives implications for regulators and managers. first, law enforcement through national legislation and power execution by the government need to be reviewed and strengthened to enlighten the whole society. firms need to emphasize the role of csr as part of the business philosophy and social responsibility as a kind of good business strategy. meanwhile, under the supervision of stakeholders and society, firms should create a good corporate social image and improve core competitiveness on top of that, relevant departments should construct a suitable csr evaluation system and implement the csr standard certification to make china’s csr management in line with the relevant international standards. the study reported here has been able to contribute to the field in studies. firstly, the research found that board independence can increase firm performance in china capital market increasing the proportion of independent directors and put forward clear requirements for the selection and appointment of independent directors. secondly, this study can be used as a business strategy in improving firm performance and maximizing shareholder’s wealth. therefore, improvise a firm’s csr strategy might improve firm business visibility and sustainability. the results also have important implications for managers involved in making csr decisions, and they should not overinvest in resources and energy that can undermine the profitability of their businesses. 126 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 it also recommended that further study should be conducted by including a larger sample size that is categorized by industries sectors or even share type available in shanghai stock exchange. this study only used a-share listing on the shanghai stock exchange, only 86 representative listed large companies were selected as they represent the sse 180 index it is also recommended that future studies on csr need to be conducted using various methods and other related variables such as csr disclosure levels, csr ranking scores and others which help to provide greater robustness to the study findings. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. references al-matari, e. m., al-swidi, a. k., & fadzil, f. h. b. (2014). the measurements of firm performance’s dimensions. asian journal of finance & accounting, 6(1), 24. https://doi.org/10.5296/ ajfa.v6i1.4761 ali, r., sial, m. s., brugni, t. v., hwang, j., khuong, n. v., & khanh, t. h. t. (2020). does csr moderate the relationship between corporate governance and chinese firm’s financial performance? evidence from the shanghai stock exchange (sse) firms. sustainability (switzerland), 12(1). https://doi. org/10.3390/su12010149 allam, b. s. (2018). the impact of board characteristics and ownership identity on agency costs and firm performance: uk evidence. corporate governance (bingley), 18(6), 1147–1176. https:// doi.org/10.1108/cg-09-2016-0184 assenga, m. p., aly, d., & hussainey, k. (2018). the impact of board characteristics on the financial performance of tanzanian firms. corporate governance (bingley), 18(6), 1089–1106. https:// doi.org/10.1108/cg-09-2016-0174 barnea, a., & rubin, a. (2010). corporate social responsibility as a conflict between shareholders. journal of business ethics, 97(1), 71–86. https://doi.org/10.1007/s10551-010-0496-z 127 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 brammer, s., brooks, c., & pavelin, s. (2006). corporate social performance and stock returns: uk evidence from disaggregate measures. financial management, 35(3), 97–116. https://doi. org/10.1111/j.1755-053x.2006.tb00149.x burinwattana, s. (2018). the mediating effect of csr reporting on the relationship of board structure and financial performance: a case of the stock exchange of thailand. srinakharinwirot business journal, 9(2), 72–81. chen, h., & wang, x. (2011). corporate social responsibility and corporate financial performance in china: an empirical research from chinese firms. corporate governance, 11(4), 361–370. https://doi.org/10.1108/14720701111159217 cheng, s., lin, k. z., & wong, w. (2016a). corporate social responsibility reporting and firm performance: evidence from china. journal of management and governance, 20(3), 503– 523. https://doi.org/10.1007/s10997-015-9309-1 cheng, s., lin, k. z., & wong, w. (2016b). corporate social responsibility reporting and firm performance: evidence from china. journal of management and governance, 20(3), 503– 523. https://doi.org/10.1007/s10997-015-9309-1 dian, y. (2014). corporate governance and firm performance: a sociological analysis based on chinese experience. social sciences in china, 35(1), 44–67. https://doi.org/10.1080/0252 9203.2013.875662 donadelli, m., fasan, m., & magnanelli, b. s. (2014). the agency problem, financial performance and corruption: country, industry and firm level perspectives. industry and firm level perspectives, 11(3–4), 259–272. https://doi.org/10.1111/ emre.12038 el ghoul, s., guedhami, o., kwok, c. c. y., & mishra, d. r. (2011). does corporate social responsibility affect the cost of capital? journal of banking and finance, 35(9), 2388–2406. https:// doi.org/10.1016/j.jbankfin.2011.02.007 fama, e. f., & jensen, m. c. (1983). seperate of ownership and control. journal of law & economics, 26(2), 301–325. famiyeh, s. (2017). corporate social responsibility and firm’s performance: empirical evidence. social responsibility journal, 13(2), 390–406. https://doi.org/10.1108/srj-042016-0049 fan, j. p. h., wei, k. c. j., & xu, x. (2011). corporate finance and governance in emerging markets: a selective review and an agenda for future research. journal of corporate finance, 17(2), 207–214. https://doi.org/10.1016/j.jcorpfin.2010.12.001 128 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 fan, y., jiang, y., kao, m. f., & liu, f. h. (2020). board independence and firm value: a quasi-natural experiment using taiwanese data. journal of empirical finance, 57(april), 71–88. https:// doi.org/10.1016/j.jempfin.2020.04.001 feng, m., wang, x., & kreuze, j. g. (2017). corporate social responsibility and firm financial performance: comparison analyses across industries and csr categories. american journal of business, 32(3/4), 106–133. https://doi.org/10.1108/ ajb-05-2016-0015 freeman, e., & mcvea, j. (1984). a stakeholder approach to strategic management. working paper no. 01-02. darden business school, 01, 1–32. https://doi.org/10.2139/ssrn.263511 friedman, m. (1970). the social responsibility of business is to increase its profits. new york times magazine, 32-33,122,124,126. fuzi, s. f. s., halim, s. a. a., & julizaerma, m. k. (2016). board independence and firm performance. procedia economics and finance, 37(16), 460–465. https://doi.org/10.1016/s22125671(16)30152-6 gaur, s. s., bathula, h., & singh, d. (2015). ownership concentration, board characteristics and firm performance. management decision, 53(5), 911–931. https://doi.org/10.1108/md-082014-0519 hu, h. w., tam, o. k., & tan, m. g. s. (2010). internal governance mechanisms and firm performance in china. asia pacific journal of management, 27(4), 727–749. https://doi. org/10.1007/s10490-009-9135-6 jensen, m. c., & meckling, w. h. (1976). theory of the firm: managerial behavior, agency costs and ownership structure. journal of financial economics, 3(december 2000), 305–360. https://doi.org/10.1016/0304-405x(76)90026-x jiang, fuxiu, & kim, k. a. (2015). corporate governance in china: a modern perspective. journal of corporate finance, 32, 190– 216. https://doi.org/10.1016/j.jcorpfin.2014.10.010 jo, h., & harjoto, m. a. (2011). corporate governance and firm value: the impact of corporate social responsibility. journal of business ethics, 103(3), 351–383. https://doi.org/10.1007/ s10551-011-0869-y kao, e. h., yeh, c. c., wang, l. h., & fung, h. g. (2018). the relationship between csr and performance: evidence in china. pacific basin finance journal, 51(april), 155–170. https://doi.org/10.1016/j.pacfin.2018.04.006 129 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 kao, m. f., hodgkinson, l., & jaafar, a. (2019). ownership structure, board of directors and firm performance: evidence from taiwan. corporate governance (bingley), 19(1), 189–216. https://doi. org/10.1108/cg-04-2018-0144 li, w.,xu, y., niu, j., & qiu, a. (2012). a survey of corporate governance: international trends and china’s mode. nankai business review international, 3(1), 4–30. https://doi. org/10.1108/20408741211201890 li, z. f., li, tao, & minor, d. (2015). ceo power, corporate social responsibility, and firm value: a test of agency theory. ssrn electronic journal, february 2018. https://doi.org/10.2139/ ssrn.2612733 liu, x., & zhang, c. (2016). corporate governance, social responsibility information disclosure, and enterprise value in china. journal of cleaner production. https://doi.org/10.1016/j. jclepro.2016.09.102 liu, y., miletkov, m. k., wei, z., & yang, t. (2015). board independence and firm performance in china. journal of corporate finance, 30, 223–244. https://doi.org/10.1016/j. jcorpfin.2014.12.004 malik, m. (2015). value-enhancing capabilities of csr: a brief review of contemporary literature. journal of business ethics, 127(2), 419–438. https://doi.org/10.1007/s10551-014-2051-9 merendino, a., & melville, r. (2019). the board of directors and firm performance: empirical evidence from listed companies. corporate governance (bingley), 19(3), 508–551. https://doi. org/10.1108/cg-06-2018-0211 mohr, l. a., webb, d. j., & harris, k. . (2001). do consumers expect companies to be socially responsible? the impact of corporate social responsibility on buying behavior. j. consum. aff, 35, 45–72. https://doi.org/10.5267/j.uscm.2019.1.005 mutlu, c. c., van essen, m., peng, m. w., saleh, s. f., & duran, p. (2018). corporate governance in china: a meta-analysis. journal of management studies, 55(6), 943–979. https://doi. org/10.1111/joms.12331 nelling, e., & webb, e. (2009). corporate social responsibility and financial performance: the “virtuous circle” revisited. review of quantitative finance and accounting, 32(2), 197–209. https://doi.org/10.1007/s11156-008-0090-y 130 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 109–130 rahman, m. j., & fang, y. (2019). the relationship between corporate social responsibility and firm performance in china. risk governance and control: financial markets and institutions, 9(4), 41–48. https://doi.org/10.22495/rgcv9i4p4 rashid, a. (2018). board independence and firm performance: evidence from bangladesh. future business journal, 4(1), 34–49. https://doi.org/10.1016/j.fbj.2017.11.003 shao, l. (2019). dynamic study of corporate governance structure and firm performance in china: evidence from 2001-2015. chinese management studies, 13(2), 299–317. https://doi.org/10.1108/ cms-08-2017-0217 sial, m. s., zheng, c., khuong, n. v., & usman, m. (2018). does firm performance influence corporate social responsibility reporting of chinese listed companies ? sustainability, 10, 1–12. https:// doi.org/10.3390/su10072217 taouab, o. (2019). firm performance : definition and measurement models. 15(1), 93–106. https://doi.org/10.19044/esj.2019. v15n1p93 yang, m., bento, p., & akbar, a. (2019). does csr influence firm performance indicators? evidence from chinese pharmaceutical enterprises. sustainability (switzerland), 11(20), 1–18. https:// doi.org/10.3390/su11205656 yeh, c. c., lin, f., wang, t. s., & wu, c. m. (2019). does corporate social responsibility affect cost of capital in china? asia pacific management review, xxxx. https://doi.org/10.1016/j. apmrv.2019.04.001 customer adoption of banking technology in private banks of india 73 the international journal of banking and finance, volume 8 (number 3), 2011: pages 73-88 customer adoption of banking technology in private banks of india bindiya tater, manish tanwar, and krishna murari suresh gyan vihar university-jaipur, rampuria jain college and mits university, india ___________________________________________________________ abstract this paper explores the perception of indian customers towards the use of technologies with respect to such factors as convenience, privacy, security, ease of use, real time accessibility, and accurate record of varied transaction that enable customer’s adoption of banking technology. other factors such as slow transfer speed, technical failure, frauds and unawareness among customers that make hindrance in adoption, are also tested. the results show that demographic variables such as gender, age, qualification and income play a positive role in adoption of banking technology. all the banks are using information technology as a strategic vehicle to stay competitive against other players. there is no significant difference between adoption rates of banking technologies by the customers of different private banks. the paper also shows that banking technology helps in increasing customer satisfaction, customer loyalty, improvised growth, and performance of the banks. keywords: private banks, technical failures, atm, branch, internet and mobile banking, kruskal wallis test, chi-square test jel classification: g 21 ___________________________________________________________ 1. introduction the banking industry in india is steadily expanding. the liberalization of the economy has created a competitive culture that has taken the service industry and particularly the banking industry by storm. the banking sector has been the backbone of every emerging country. it implements and brings about economic reforms. any change in this sector through technology adoption has a sweeping impact on a country’s growth. the development in information collection, storage, processing and transmission technologies have influenced all aspects of the banking activity. tater et al.: adoption of banking technology 74 information technology is a medium that has revolutionized banking and everyday bank operations at the click of a button thus enabling sophisticated product development, better market infrastructure, implementation of reliable techniques for control of risks and reaching geographically distant and diversified markets (marion 2008). banking technology offers benefits to both banks and customers. pikkarainen et al. (2004) mention two fundamental reasons underlying banking technology development and penetration. first, the bank gets significant cost savings in their operation through e-banking services. it has been proven that online banking channel is the cheapest delivery channel for banking products once established. secondly, the banks have reduced their branch networks and downsized the number of service staff which paved the way to self-service channels. customers too enjoy self-service, freedom from time and place constraint and reduced stress of queuing in banking hall. it was indicated that electronic banking services delivery are the cheapest and the most profitable and wealthiest delivery channel for banking products. this paper is organized as follows. section 2 attempts a quick review of the literature on this topic. in section 3, we describe the methodology adopted for this empirical work. the results are presented thereafter in sections 3 and 4 with the latter section providing statistical tes results. the paper ends in section 6 as conclusion. 2. literature review most studies found to-date focuses on the individual banking technologies and the customer satisfaction level used in the case of banks. for example mobarek (2007) in his study in botswana region pointed out that the delivery channels are lacking in meeting the demands of the customer by not making them aware of e-banking and up-to-date technology. the hypotheses were tested, which showed that there is a relationship between age group, occupation and some aspects of e-banking. another study by islam, sheel and biswas (2007) investigate the satisfaction levels of hsbc atm cardholders (both staff and non-staff) with respect to various aspects (for instance, promptness of card delivery, the performance of hsbc atm, and the service quality of atm personnel) of using hsbc atm and their opinions on various other related issues such as positive and inconvenient features of hsbc atm, recommendation to improve the service quality. the findings provide significant results related to use and worth of holding the atm. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 4 75 further, the findings helped the atm section to identify their positive and negative features and the customer recommendation. wan et al. (2005) study the customers’ adoption of banking channels in hong kong. they covered four major banking channels namely atm, branch banking, telephone banking and internet banking. the study segmented the customers on demographic variables and psychological beliefs about the positive attributes processed by the channels. the psychological factors were ease of use, transaction security, transaction accuracy, speediness, convenience, time utility, provision of different personal services, social desirability, usefulness, economic benefits, and user involvement. another study investigated inhibitors and enablers of internet banking in oman, comparing this to the situation in australia. data were collected from interviews with bank managers in each country, based on a consideration of each bank manager’s perceptions of four factors that might affect their decisions to adopt internet technologies: relative advantage, organizational performance, customer/organizational relationship and ease of use. it is hoped that the results will be useful in seeing why omani banks have been slow to adopt internet technology and helps to encourage them to make the change. there are several competitive advantages associated with the adoption of technology in the banking sector, including the creation of entry barriers, enhancement of productivity, and increased revenue generation from new services (fitzsimmons and fitzsimmons, 1997). delivery methods have become an increasingly important technique to retain customers in today’s dynamic banking environment since customers can make withdrawals, deposits and access balances at their own convenience (tanzi, 1997). the importance of security and privacy for the acceptance of online banking has been noted in many banking studies (sathye, 1999; hamlet and strube, 2000; tan and teo, 2000; polatoglu and ekin, 2001; black et al., 2002; giglio, 2002; and howcroft et al., 2002). roboff and charles (1998) found that people have a weak understanding of online banking security risks although they are aware of the risks. furthermore, they found that consumers often trust that their bank is more concerned about privacy issues and will protect them. finally, they argue that although consumers' confidence in their bank is strong, their confidence in technology is weak. tater et al.: adoption of banking technology 76 3. research methodology 3.1 objective of the study this paper focuses on exploring the four major factors that influence the adoption of information technology in private banks of india. these factors are: the variables that enable and inhibit customer’s adoption for banking technology; influence of demographic variables on banking technology adoption; and level of customer satisfaction with banking technology. the paper also examines the variables associated with atm banking, branch banking, internet banking and mobile banking. 3.2 hypotheses of the study eight null hypotheses (starting with h01, h02 …) and alternate hypotheses (starting with ha1, ha2) are formulated. tested results of these hypotheses are presented in the section 5 of this study. h01: usefulness of banking services is same for all the banks. ha1: h01 is not true. h02: there is a significant difference between characteristics of atm banking services provided by different private banks to the customers. ha2: h02 is not true. h03: there is a significant difference between branch banking services and customer satisfaction. ha3: h03 is not true. h04: there is a significant difference between internet banking services and customer satisfaction. ha4: h04 is not true. h05: there is a significant difference between mobile banking services and customer satisfaction. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 4 77 ha5: h05 is not true. h06: there is a significant difference between adoption of information technology and customer satisfaction. ha6: h06 is not true. h07: there is a significant difference between customer adoption and benefits of it banking services for all four private banks. ha7: h07 is not true. h08: there is a significant difference between unwillingness to use e-channel and customer rejection for all four banks. ha8: h08 is not true. 3.3 sample and tools for data analysis this study is conducted using respondents involved with four private sector banks (icici, hdfc, axis, and indusind) in india from bikaner to jaipur regions of rajasthan. descriptive research methodology is used to accomplish the objectives using random and convenience sampling techniques. a questionnaire was designed consisting of the following four parts. part 1: demographic profile of the respondents, part 2: customer’s account profile of the respondents, part 3: personal characteristics of customers with different banking services, and part 4: characteristic and usefulness of different banking services. a five point likert scale is used to elicit responses to the questionnaire. the data are collected from 500 bank customers (icici bank-154, hdfc bank88, axis bank77, indusind bank84) through structured questionnaire method out of which 403 responses are obtained. the data collected are tabulated and analyzed for the purpose of giving precise and concise information. descriptive frequency statistical tool is adopted for interpretation and hypothesis tater et al.: adoption of banking technology 78 testing is done using kruskal wallis and chi-square tests as appropriate for non-parametric statistics applied to survey data analysis. 4. empirical results tables 1 through 4 show the demographic descriptive statistics of the respondents from the selected banks. gender as a personal variable was found to have a significant role in customer’s banking technology adoption. from the table 1, it is found that icici, hdfc, axis and indusind bank’s male respondents are five times more as users as compared to female respondents. females are still comparatively lacking banking habits. table 1: demographic profile of the respondents demographic variables categories icici (%) hdfc (%) axis (%) indusind (%) gender male female 122 (79) 32 (21) 74 (84) 14 (16) 65 (84) 12 (16) 68 (81) 16 (19) age 18 yrs-30 yrs 30 yrs-45 yrs 45 yrs+ 42 (27) 76 (49) 36 (23) 42 (48) 31 (35) 15 (17) 45 (58) 21 (27) 11 (14) 31 (37) 36 (43) 17 (20) qualification under graduate graduate post graduate 28 (18) 53 (34) 73 (47) 14 (16) 28 (32) 46 (52) 21 (27) 24 (31) 32 (42) 21 (25) 24 (29) 38 (46) monthly income (rs.) < 10000 10001-20000 20001-35000 35001 & above 38 (25) 58 (38) 34 (22) 24 (16) 21 (24) 25 (28) 28 (32) 16 (10) 18 (23) 29 (38) 21 (27) 10 (13) 11 (14) 30 (36) 37 (44) 6 (7) note: figures in parenthesis denotes % to column total. source: primary data from questionnaire and interviews. qualification is the factor that makes the customer aware of the banking technology and also helps them in easy adoption. among the four banks, about 27 per cent of axis bank respondents belong to the qualification category of undergraduate: this could proxy for income effect since educated persons earning more money are likely to use banking. thirty-four per cent of icici, 32 per cent of hdfc, 31 percent of axis and 29 per cent of indusind bank customers belong to the qualification category of graduate while 52 per cent of hdfc, 47 per cent of icici, 46 per cent of indusind and 42 per cent of axis bank customers belong to the qualification category of post-graduate. the survey reflects that about 25 per cent of icici bank respondents, 24 per cent of hdfc bank respondents, 23 per cent of axis bank respondents and 14 per cent indusind bank respondents belong to the monthly income category of less than indian rupees=rs 10,000.00; 38 per cent of icici and axis bank respondents belong to the monthly income international journal of banking and finance, vol. 8, iss. 3 [2011], art. 4 79 category of rs10,000-20,000; 44 per cent of indusind, 32 per cent of hdfc, 27 per cent of axis and 22 per cent icici bank respondents belong to the monthly income category of rs 20,001-35,000; and 16 per cent of icici and hdfc bank, 13 per cent of axis and 7 per cent indusind bank respondents belong to the monthly income category of rs 35,001 and above. table 2 represents the statistics of the respondents about the form of association with the banks. from the table it was found that 56 per cnt of respondents having current account and 44 per cent of respondents having saving account belongs to icici bank, 18 per cent of respondents having current account and 82 per cent of respondents have saving account belongs to hdfc bank, 38 per cent of respondents having current account and 63 per cent of respondents have saving account belongs to axis bank, 32 per cent of respondents having current account and 68 per cent of respondents have saving account belongs to indusind bank. table 2: association of the respondents with bank customer’s account variables categories icici (%) hdfc (%) axis (%) indusind (%) type of account current saving 87 (56) 67 (44) 16 (18) 72 (82) 29 (38) 48 (63) 27 (32) 57 (68) time of association with banks <2yr 2 – 4 yrs 4 yrs & above 52 (34) 64 (42) 38 (25) 38 (43) 27 (31) 23 (26) 27 (35) 37 (48) 13 (17) 31 (37) 36 (42) 17 (20) frequency of visit to banks everyday 2-3 times per week 1-2 times per month 39 (25) 46 (30) 69(45) 31 (35) 34 (39) 23 (26) 24 (31) 19 (25) 34 (44) 31 (37) 26 (31) 27 (32) access of electronic banking services atm banking branch banking internet banking mobile banking 56 (36) 43 (28) 31 (20) 24 (16) 32 (36) 30 (34) 14 (16) 12 (14) 32 (42) 24 (31) 13 (17) 8 (10) 35 (42) 20 (24) 18 (21) 11 (13) note: figures in parenthesis denote % to column total. source: primary data from questionnaire and interviews. the length of association represents the state of connection of customer and bank. the result shows that most of respondents have 2-4 years of association with their banks. people have long periods of relationship with public bank since private banks are new. so, the length of association is less and these banks need to generate more publicity among customers about their products and services. the longer the duration of holding an account with the banks, the more trust is developed which adds to customers’ satisfaction. from the numbers in table 2, it is found that about 45 per cent of icici bank respondents visit the banks 1-2 times per month, 39 tater et al.: adoption of banking technology 80 per cent of hdfc bank respondents visit 2-3 times per week while 44 per cent of indusind bank respondents visit 1-2 times per month. the survey reflects that for icici, 36 per cent of respondents prefer atm banking, 28 per cent prefer branch banking, 20 per cent prefer internet banking and 16 per cent prefer mobile banking. similarly for hdfc: 36 per cent prefer atm banking, 34 percent branch banking, 16 per cent internet banking and 14 per cent mobile banking. in the axis bank, 42 per cent of respondents prefer atm banking, 31 per cent branch banking, 17 per cent internet banking and 10 per cent mobile banking. about 42 per cent of respondents prefer atm banking, 24 per cent branch banking, 21 per cent internet banking and 13 per cent mobile banking in the case of indusind bank. the result shows that atm and branch banking remain as the two popular banking services. table 3 shows the customer adoption of technology on the basis of demographic features. table 3: age and banking technology adoption icici (%) hdfc (%) axis (%) indusind (%) variables 18-30 30-45 45+ 18-30 30-45 45+ 18-30 30-45 45+ 18-30 30-45 45+ atm banking 18 (32.1) 28 (50.0) 10 (45.5) 15 (46.9) 12 (37.5) 5 (15.6) 13 (37.1) 14 (40.0) 8 (22.9) 13 (37.1) 14 (40) 8 (22.9) branch banking 13 (30.2) 21 (48.8) 9 (20.9) 13 (43.3) 14 (46.7) 3 (10.0) 6 (30.0) 10 (50.0) 4 (20.0) 6 (30) 10 (50) 4 (20) mobile banking 6 (19.4) 17 (54.8) 8 (25.8) 7 (50.0) 4 (28.6) 3 (21.4) 8 (44.4) 8 (44.4) 2 (11.1) 8 (44.4) 8 (44.4) 2 (11.1) internet banking 5 (20.8) 10 (41.7) 9 (37.5) 7 (58.3) 1 (8.3) 4 (33.3) 4 (36.4) 4 (36.4) 3 (27.3) 4 (36.4) 4 (36.4) 3 (27.3) note: figures in parenthesis denote % to column total. source: primary data from questionnaire and interviews. banks in india have adapted and continue to adopt different technologies. customers were asked to indicate the various technologies their banks have adopted. this was queried to know if the customers are aware of the technologies provided by their banks. four common technologies were selected: atm, branch banking, internet banking, and mobile banking. from the tables, it is found that both male and female customers of icici belonging to the 30-45 years age category are seen to favor atm banking, branch banking, internet banking and mobile international journal of banking and finance, vol. 8, iss. 3 [2011], art. 4 81 banking. it shows younger customers belonging to the 30-45 years are tech-savvy and find these services comfortable, to be friendly and easy to use. customers from different educational background were asked to indicate the various technologies they like to adopt. four common technologies selected to present the variables include atm, branch banking, internet banking, and mobile banking. the variables are labeled in four banks and the result is presented in table 4. from the figures shown in the table, it is found that both male and female post-graduate customers of icici bank favor atm banking, branch banking, internet banking and mobile banking. the study shows that higher qualification is associated with bringing attention towards new tech-banking services and qualification is a factor found to be relevant. it shows that majority of post-graduate and graduate customers favor atm banking followed by mobile banking, branch banking and internet banking. customers with post-graduate and graduate qualifications are mostly adopters of it banking services table 4: educational qualification and banking technology adoption icici (%) hdfc (%) axis (%) indusind (%) variables ug g pg ug g pg ug g pg ug g pg atm banking 6 (10.7) 19 (33.9) 31 (55.4) 4 (13) 9 (28) 19 (59) 6 (19) 12 (38) 14 (44) 7 (20) 11 (31) 17 (49) branch banking 8 (18.6) 16 (37.2) 19 (44.1) 5 (17) 10 (33) 15 (50) 8 (33) 7 (29) 9 (38) 4 (22) 6 (33) 8 (44) mobile banking 7 (22.5) 12 (38.7) 12 (38.7) 2 (14) 5 (36) 7 (50) 4 (31) 3 (23) 6 (46) 3 (17) 6 (33) 9 (50) internet banking 7 (29.1) 6 (25) 11 (45.8) 3 (25) 4 (33) 5 (42) 3 (38) 2 (25) 3 (38) 4 (36) 3 (27) 4 (36) note: figures in parenthesis denote % to column total; ug-under graduate, g-graduate and pgpostgraduate. source: primary data from questionnaire and interviews. 5. testing of hypotheses 5.1 kruskal wallis test this is a distribution-free non-parametric test used to compare three or more independent groups of sampled data statistics. seven respondents from each of the four private banks (icici, hdfc, axis and indusind bank) were randomly sampled to determine whether customer’s ranking of different banking services were the same. tater et al.: adoption of banking technology 82 h01: usefulness of banking services is the same for all the banks. ha1: h01 is not true. if null hypothesis is not true then chi-square value tends to get large and the hypothesis is rejected; vice versa otherwise. the ranks are from high value to low value of banking services. table 5: frequency usage of banking services by respondents customer’s usage variables categories icici (%) hdfc (%) axis (%) indusind (%) frequency of usage once in a day thrice in a week once in a week 42 (27) 73 (47) 39 (25) 31 (35) 32 (32) 29 (33) 25 (32) 38 (49) 14 (18) 28 (34) 34 (41) 21 (25) note: figures in parenthesis denote % to column total. source: primary data from questionnaire and interviews. the numbers in table 5 show that the most frequent usage of banking services are: 35 per cent once in a day, 49 per cent of axis respondent were found to use banking services thrice in a week, followed by icici bank (47 per cent), indusind bank (41 per cent) and hdfc bank (32 per cent). table 6: kruskal wallis test to determine usefulness of banking services s. no. banking services degree of freedom (4-1=3) kruskal wallis test value result 1 cash withdrawal 3 -63 reject h01, χ 2 <7.815 at 5% level of significance 2 tax, insurance, utility bill payments 3 -28.0 reject h01, χ 2 <7.815 at 5% level of significance 3 account balance 3 -55.1 reject h01, χ 2 <7.815 at 5% level of significance 4 fund transfer 3 -48.7 reject h01, χ 2 <7.815 at 5% level of significance 5 investment information and online loan related information 3 -58.9 reject h01, χ 2 <7.815 at 5% level of significance 6 download online forms 3 -47.3 reject h01, χ 2 <7.815 at 5% level of significance 7 e-ticketing 3 -28.0 reject h01, χ 2 <7.815 at 5% level of significance source: primary data from questionnaire and interviews. compare this to 33 per cent of hdfc respondents are found to use banking services once a week, followed by hdfc bank (33 per cent), icici and indusind bank (25 per cent each). international journal of banking and finance, vol. 8, iss. 3 [2011], art. 4 83 table 6 shows that chi-square value is less than kruskal wallis test value for all the variables. so, the null hypothesis (h01) is rejected in favor of alternate hypothesis (ha1). this means there is a significance difference in the usefulness of banking services among the four private sector banks in india. 5.2 chi-square test the chi-square values are used to test the significance of association between two attributes. the data gathered are rated on 5-point likert scale. the results so obtained are tabulated based on their specific factors. the banking services were classified into four major categories: atm banking, branch banking, internet banking and mobile banking with their variable factors. the results are given the in the table 7. in order to test the hypothesis based on the above mentioned four banking services, we use chi-square values to find the significance of the responses attributed by the respondents. the summarized results of the chi-square test values at 5 per cent level of significance are given in table 8. h02: there is a significant difference between characteristics of atm banking services provided by different private banks to the customers. ha1: h02 is not true. the test results are shown in table 8. it shows that at 5 per cent level of significance, the calculated value (1.956) is less than the tabulated value (21.026). so, the null hypothesis is rejected. this reveals that there is no significant difference between atm banking services provided by different private bank’s adoption of it banking services by customers. h03: there is a significant difference between branch banking services and customer satisfaction. ha3: h03 is not true table 8 shows that at 5 percent significance level, the calculated value (1.000326) is less than the tabulated value (16.919). so, the null hypothesis (h03) is rejected. this reveals that there is no significant difference between branch banking services provided by different private banks to the adoption of it banking services by customers. tater et al.: adoption of banking technology 84 table 7: customer adoption of banking services factor atm banking icici hdfc axis indusind f1 convenient location 81.42 76.87 76.87 76.57 f2 ease of use 78.57 81.25 75.62 76.00 f3 24*7 environment of operation 83.92 82.50 75.00 76.00 f4 variety of transactions at atm network 85.71 78.75 72.50 69.14 f5 accurate records of all transaction 88.57 77.5 71.25 68.00 branch banking f1 friendliness of bank personnel 80.9302 78 80 77 f2 bank branch’s reputation 82.7907 80 78.5167 75 f3 time taken to process the transaction 84.1860 80.6666 86.3157 78 f4 working hours 85.1162 81.3333 73.1100 73 internet banking f1 page set up/menu flow 82.58065 81.42857 81.33333 71.11111 f2 speed of page loading 79.35484 81.42857 77.33333 68.88889 f3 easy of use/navigation 83.22581 72.85714 79.33333 67.77778 f4 convenient hours of operation 76.12903 80 79.33333 72.22222 f5 variety of transaction 83.87097 74.28571 81.33333 76.66667 f6 real time access to information 80.64516 72.85714 84 77.77778 f7 accurate records of all transaction 80 80 73.33333 81.11111 f8 support service(customer feedback/complaint management services) 80.64516 77.14286 80.66667 73.33333 mobile banking f1 call answering time 83.33333 80 81.36364 70.90909 f2 flawless/correct options 83.33333 71.66667 77.67442 67.27273 f3 understanding and replying queries quickly 82.5 76.66667 73.02326 69.09091 f4 communication skill/positive approach 82.5 75 72.27273 76.36364 f5 educate customers how to use options of different delivery channel 81.66667 75 73.48837 76.36364 h04: there is a significant difference between internet banking services provided by different private banks to the adoption of it banking services by customers. ha4: h04 is not true international journal of banking and finance, vol. 8, iss. 3 [2011], art. 4 85 as shown in the table 8, at 5 percent significance level, the calculated value (3.959) is less than the tabulated value (32.671). so, the null hypothesis (h04) is rejected. this reveals that there is no significant difference between internet banking services provided by different private banks to the adoption of it banking services by customers. h05: there is a significant difference between mobile banking services provided by different private banks to the adoption of it banking services by customers. ha5: ha5 is not true as shown in the table 8, at 5 percent significance level, the calculated value (1.774) is less than the tabulated value (21.026). so, the null hypothesis is rejected. this reveals that there is no significant difference between mobile banking services provided by different private banks to the adoption of it banking services by customers. table 8: testing of hypotheses (χ 2 test) level of significance number of rows number of columns degrees of freedom p-value calculated value tabulated value result 0.05 5 4 12 0.9994 1.956361 21.026 reject h02 0.05 4 4 9 0.9994 1.000326 16.919 reject h03 0.05 8 4 21 0.9999 3.959468 32.671 reject h04 0.05 5 4 12 0.9996 1.774393 21.026 reject h05 table 9: adoption of information technology and customer satisfaction (χ 2 test) level of significance number of rows number of columns degrees of freedom pvalue calculated value tabulated value result 0.05 13 4 36 1 5.4870 79.083 reject h06 h06: there is a significant difference between adoption of information technology and customer satisfaction. ha6: h06 is not true. tater et al.: adoption of banking technology 86 table 9 shows the result of the chi-square test used at 5 percent significance level. the result reveals that the calculated value (5.4870) is less than the tabulated value (79.083). so, the null hypothesis (h06) is rejected. this reveals that there is no significant difference between adoption of information technology and customer satisfaction. h07: there is a significant difference between benefits of it banking services and customer satisfaction as compared in different banks. ha7: h07 is not true factors responsible for adoption of it banking services among customers include it provide faster, easier, reliable and continuous access to information, second, it reduces cost in accessing and using the banking services; easy to learn (increased automation of process); fourth, better management of services (easy to follow up requests/complaints). table 10 shows that as per the chi-square test used at 5 percent significance level, the result reveals that the calculated value (0.462) is less than the tabulated value (21.026). so, the null hypothesis is rejected indicating that there is no significant difference between adoption of information technology and customer satisfaction as compared in different banks. table 10: benefits of it services for respondents (χ 2 test) level of significance number of rows number of columns degrees of freedom pvalue calculated value tabulated value result 0.05 5 4 12 0.9999 0.461701 21.026 reject h07 h08: there is a significant difference between unwillingness to use e-channel and customer dissatisfaction. ha8: h08 is not true. the it banking services has many benefits but there are many factors that inhibit adoption of it banking services. these are: lack of awareness/knowledge about internet technology and accessibility of service, increase in fraud due to inefficient safety and security features, slow transfer speed, and delay in transmission due to machine breakdown / machine complexity/technical failure. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 4 87 table 11: respondents’ unwillingness to use e-channel for commercial purpose (χ 2 test) level of significance number of rows number of columns degrees of freedom pvalue calculated value tabulated value result 0.05 4 4 9 0.9972 1.487 16.919 reject h08 table 11 shows that, at 5 per cent significance level, the calculated value (1.487) is less than the tabulated value (16.919). so, the null hypothesis (h08) is rejected. this reveals that there is no significant difference between unwillingness to use e-channel and customer dissatisfaction. 6. conclusion customers of private sector banks agree that there exist relationship between factors such as age, gender, income, qualification and adoption of banking technology by customers. young generation belonging to a category of 30-45 years finds the services comfortable, friendly and easy to use. customers with post-graduate and graduate qualifications are found to be mostly adaptors of it banking services it is reflected from the survey that atm banking remains the most popular banking service among customers after branch banking, mobile banking and internet banking respectively as they provide convenience, privacy, security, ease of use, real time accessibility, and accurate record of various transaction. kruskal wallis test applied to the data collected ensures that customer’s usage of different banking services is same for all the banks. from these, it is clear that there is no significant difference between atm banking, branch banking, mobile banking and internet banking services provided by different private banks to the customers. there is a relation between benefits of banking services and increasing banking technology adoption. customer’s unwillingness to use e-channel for commercial purpose decreases banking technology adoption. author information: submitting author is bindiya tater, research scholar, suresh gyan vihar university, jaipur, rajasthan-india, email id: bindiya.tater@gmail.com, contact no: 919602026899. manish tanwar is assistant professor, rampuria jain college, bikaner (raj) india, email id:imanishtanwar@gmail.com, contact no: 91-941-309-2282 and krishna murari tater et al.: adoption of banking technology 88 is corresponding author and assistant professor at mits university, laxmangarh, sikar, raj, india, email: krishnamurari_1980@rediffmail.com, contact no.-91-9983733735 references ajzen, i. and fishbein, m., (1980). understanding attitudes and predicting social behavior (3 rd ed.). englewood cliffs, prentice-hall, london. al-sabbagh, i., and molla, a., (2004). adoption and use of internet banking in the sultanate ofoman: an exploratory study. journal of internet banking and commerce, 9: 1-7. davis, f., (1986). a technology acceptance model for empirically testing new enduserinformation systems: theory and results. mit. ph.d. thesis, boston. fishbein, m. and ajzen, i., (1975). belief, attitude, intention, and behavior: an introduction to theory and research. addison-wesley, boston. hain, d., tootell, h. and alcock c., (2002). understanding attitudes towards internet banking. seventh annual collector conference on electronic commerce, melbourne, australia, heikki, k., minna, m. and taplo, p., (2002). factors underlying attitude formation towards online banking in finland. the international journal of bank marketing, 20: 261-272. kalfan, a., al-refaei, y., al-hajry, m., (2006). factors influencing the adoption of internet banking in oman: a descriptive case study analysis. international journal of financial services management, 1: 155-172. marlin, s., (2005). banking for the 21st century, bank systems and technology, july 1. perumal, v., and shanmugam b., (2004), internet banking: boon or bane, journal of internet banking and commerce, 9: 1-6, ramsay, j., and smith, m., (1999). managing consumer channel usage in the australian banking sector, managerial auditing journal, 14: 32-33, sohail, s. m., and shanmugam, b., (2002). electronic banking and customer preferences in malaysia: an empirical investigation, fourth conference on information & computer science, king fahd university of petroleum & minerals, saudi arabia, 355-364 tan, m., and teo, t., (2000). factors influencing the adoption of internet banking. journal of the association for information systems 1: 1-42. wan, w.n., luk, c.l., and chow c.w.c., (2005). consumers' adoption of banking channels in hong kong. international journal of bank marketing, 23: 255-272. wang, y., wang, y., lin, h. and tang, t., (2003). determinants of user acceptance of internet banking: an empirical study. international journal of service industry management 14: 501519. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 4 is islamic banking capable of meeting corporate social responsibility? ijbf is islamic banking capable of meeting corporate social responsibility? sarath delpachitra flinders university, australia ________________________________________________________________ abstract this paper provides an overview of the islamic banking system and its product ranges, and examines their ability to meet corporate social responsibility obligations while maintaining efficiency and managing risk. the growing literature on islamic banking products, including their derivatives indicates that islamic banking is popular worldwide and has global attention. similarly, the diversity of banking options is growing in middle eastern countries with some banks offering separate windows for conventional and islamic banking options to customers. due to the unique nature of islamic banking, which is constrained not only by the standard operating regulations of the respective countries, but also by islamic law, the sharī’ah, there are significant differences in the philosophy behind its finance transactions, the nature of transactions, goals and the obligations between the two types of institutions. whether these differences lead to changes in risk preferences, lending decision-making processes, and the ultimate efficiency of operations are some of the key empirical questions yet to be fully answered. the paper introduces a theory based on classical utility theory, and then extends it to show its relevance to islamic banking. in particular this paper shows that islamic banking may not only improve operating performance, but also increase profitability for all parties concerned. key words: islamic banking; corporate social responsibility; sharīʿah jel classification: g 34 ________________________________________________________________ 1. introduction islamic banking is growing fast, but the increase in the size of islamic banking ib is questionable (abdallah & delpachitra, 2008). over the past decade, islamic banking has substantially broadened its publicity worldwide by taking overall leadership in many islamic communities. for instance, total islamic assets worldwide expanded from us$137 billion in 1996 to approximately us$900 billion in 2010, a linear annual growth rate of 40%: in 2013, the size is reported to be us$1,700 billion. despite this progress, the total volume of international delpachitra: is islamic banking capable of meeting corporate social responsibi 50 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 transactions is still very small, as would be the case of a niche banking with just about 50 year history. however, the growing credibility of the islamic principledriven, interest-free financial intermediary based on either fee-based or profitshared pricing of financial products in the international financial markets and the positive evidence of its adaptations worldwide in particular, should receive the attention of regulators, researchers and investors. the growth of credibility is considered the driving force behind creating a viable alternative, at least where the customers demand such products as being more in line with human ethics, to the conventional banking system. on closer inspection, both the conventional and ib systems have similarities and differences. both systems focus on the classical wisdom of shareholder or corporate wealth maximization. however, islamic banking attempts to achieve these objectives through a set of islamic laws (sharīʿah) that basically prevents managing the financial institution (and the market) through the margins of interest income and interest cost. basically it promotes principal– agent relationships in financial transactions in which profits or returns reflecting pure opportunity costs are divided among stakeholders fairly and equitably. in that sense this paper assumes that the islamic banks should play a role similar to that of a venture capitalist (vc). on the other hand, as sairelly (2007, p 279) states, the ethical credentials of islamic financial institutions make them attractive to not only muslims, but also to a wide spectrum of ethically-conscious consumers who desire a socially just financial system. by extending this proposition to the basic role of the financial system, both saving and borrowing units expect that the islamic financial intermediation ensures that returns are distributed (among investors, entrepreneurs and other stakeholders) in a socially responsible manner. based on the guidance from the holy text¾the qurãn¾the rules for islamic economic and financial practices, for individual entrepreneurs and investors, and institutional investors should not only be concerned with what kind of business transactions they should be involved with, but also with how such businesses should be funded, because islamic transactions should not only be sharīʿah-compliant (ḥalāl) but should also be free of exploitation of the weak, free of interest (ribā) and exclude prohibited transactions such as those involving gharar, etc. sairally (2007) pinpoints a number of criticisms of the practice of islamic finance. these criticisms cover the heavy bias towards ‘mark-up financing’ contracts (murābaḥah) in profit-and-loss-sharing arrangements, and ‘trustee finance’ contracts (murābaḥah): it is also claimed that islamic banking is mirroring the practices of conventional banking. these criticisms were further supported by abdallah and delpachitra (2008) suggesting that ib uses identical strategies in risk preference and risk profiling to those of conventional banking. despite these criticisms, it can be shown that ib provides the best solution to the issue of ethically and socially responsible banking. in this paper, we try to establish that. essentially this paper shows that if the principles of sharīʿah are applied, ib participation in the financial system can improve the operating performance of firms, and as ib shares a higher percentage of profits with international journal of banking and finance, vol. 10, iss. 2 [2013], art. 6 is islamic banking capable of meeting corporate social responsibility?: 49-66 51 entrepreneurs, the investee firms’ operating performance is improved. this provides more incentives for ib to act in a socially responsible manner. this paper is organized as follows. section 2 provides a brief overview of ib transactions relevant to this paper. section 3 introduces the concept of corporate social responsibility (csr) and its relevance to ib. section 4 provides a detailed description of the quantitative framework for classical interpretation of principal-agent relationships using utility functions. the last section summarizes and concludes this paper. 2. islamic banking transactions islamic banks are required to comply with islamic ethics and in order to do so every islamic bank must have a board of sharīʿah scholars, or a sharīʿah supervisory board (ssb) to review the juristic correctness of the bank’s transactions (archer & karim, 2002). furthermore, as sarker (1999) points out, the primary objectives of islamic finance cover broad-based economic wellbeing, social and economic justice, and equitable distribution of income and wealth. in order to achieve these objectives, different islamic financial instruments and their derivative products have been introduced by islamic institutions. for instance, mark-up pricing (murābaḥah), ultimately creates instruments to finance economic transactions. consequently, murābaḥah is ‘a common instrument used for short-term financing based on the conventional concepts of purchase finance’ (dhumale & sapcanin, 2003, p. 34). the seller reports to the buyer the cost of acquiring or producing a good, and then a profit margin is negotiated between the two parties. ‘typical asset-backed security in the conventional system is a claim against a pool of assets; islamic instruments are claims against individual assets’ (anwar, 2003, p. 63). a distinct feature of such financial securities is that they resemble conventional debt securities characterised by a predetermined pay-off, with the difference being that islamic instruments and their derivatives are collateralized against a real asset or economic activity. in contrast in ‘trustee finance’ (murābaḥah) contracts, a principal with capital can develop a partnership with an agent (or entrepreneur) who has expertise in deploying capital in real economic activities with an agreement to share the profits (hasan, 2002). losses are borne by the capital owner only, as the other party does not hold any capital. further, though the capital owner is exposed to loss or risk, he is not entitled to participate in the management of the funds; this is exclusively left to the other partner. however this form of funding is no more than at best 25% of the total funding by banks given its huge still unresolvable asymmetric information problem. furthermore, in ‘equity participation’ (mushārakah), which is a sharelike arrangement, the capital owner enters into a partnership by contributing equity together with the entrepreneur(s). their contributions need not be equal, and contributions may be in the form of physical or monetary assets. intangible capital, such as labour, management, skill and goodwill can form part of the delpachitra: is islamic banking capable of meeting corporate social responsibi 52 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 capital (lewis & algaoud, 2001). profits can be shared in any pre-agreed ratio, but losses must be borne strictly in proportion to equity participation. this contract is suitable for long-term project financing. some scholars consider this to be the purest form of islamic finance since according to them it is closest to the spirit of sharīʿah, does not have exploitative provisions, and is a fairly equal contract for both parties. mushārakah financing is closer to a traditional equity stake with rights of control. in between these three commonly used instruments and their derivatives of mushārakah, murābaḥah and murābaḥah, there are other collateralized securities, such as ijārah (similar to a lease or a lease purchase arrangement), kafālah and amānah contracts. these instruments can be classified exclusively into transactional and intermediation contracts in an islamic financial system. transactional contracts govern retail sector transactions that include exchange, trade and other financing activities. the intermediation contracts not only govern indirect financial instruments, but also facilitate the efficiency and transparency of the execution of transactional contracts (el-hawary, grais & iqbal, 2004). in each of these cases, ib ensures that transactions comply with not only prudential standards, but also with the rules of sharīʿah, to ensure a fair deal for both contracting parties unlike in interest-only contract, which is onesided. from a historical point of view profit-and-risk-shared contracting was the dominant form of funding economic activities until the birth of fractional banking some 250 years ago. the old practice was replaced slowly with interestonly no-risk-sharing lending as a one-sided cheaper form of lending while banks used deposits to create more money on each dollar of deposits to make more interest-based no-risk-shared lending. 3. csr and islamic banking islamic finance, given its ethical characteristics some of which has been highlighted in the previous section, is often associated with ethical investment, which itself is linked to socially responsible investing. the restrictions on investment choices such as prohibition of investment in armaments sector, pornography, etc. are important examples. the concepts in question are similar. an islamic bank provides financial services according to religious precepts in the same way as ethical finance also focuses on moral values (guyot, 2009). at a contemporary time when capitalism is growing blindly as evidenced by financial scandals such as those of enron and later the maddof scam, the notion of ethics has jumped for attention and companies are reflecting more and more on ethics. many organizations are now displaying their ethical and moral concerns in the media. the aim is to reassure stakeholders (i.e., any party or person who maintains links, or indirectly are linked to a company such as suppliers, customers, shareholders, etc.), that all their interests matter to the company, not just the profits. this paper considers ethics as an integral part of international journal of banking and finance, vol. 10, iss. 2 [2013], art. 6 is islamic banking capable of meeting corporate social responsibility?: 49-66 53 csr. howard r. bowen is often seen as the founder of csr after publication of his book titled social responsibilities of the businessman (bowen, 1953) in which he attempts to convince business leaders to avoid regulatory constraints that reduce their freedom of action. according to the liberal vision of north america, ‘businessmen should therefore incorporate the public interest in their decisions’ (acquires & aggeri, 2008). since this contribution from bowen, the definition of csr was gradually reformulated. that facilitated further research on csr. according to carroll (1979), csr refers to ‘the expectations of society towards the company in the economic, legal, ethical and discretionary plans’. in this context of csr, socially responsible investment (sri) has been developed as a form of ethical finance. we talk about sri when ‘the primary purpose of the investment is profitable but additional constraints are imposed’ (winant, 2008). however, we should be cautious about the conceptual vagueness that still exists between various forms of investment despite avowed commitment to csr. unless each product is designed with csr concerns and there is a body of regulations to certify that as a fact in the design of financial products, statement such as what we quoted holds no binding commitment. socially responsible investments are selected on ‘ethical characteristics, assessed by the rating agencies’ (winant, 2008) thus, islamic finance can be considered a form of socially responsible finance in that it chooses legal and lawful financial assets, but both have their difference in religious character. 4. islamic banking: a classical view in general business, entities are growing larger and more complicated than ever with an increasing number of business models. in many cases ownership and management are separated. as a result, the agents or the management tend to put their personal interests as the top priority (lin, 2004). they initiate decisions to maximize their benefits, instead of the principals’ benefits, let alone stakeholders’ benefits. due to the principals’ limited capabilities to process information, agency issues are created. as noted before, ib differs from conventional financial institutions and investors because ib not only provides funds and services to the investee firms, but also supervises the investee firms based upon their investment plans and their compliance to a set of religiously-oriented restriction. the ibs’ ultimate goal is to ensure that the investee firms make sound investment and development decisions under strict supervision, thereby not only improving the investee firms’ operating performance and values based upon their experience as well as their knowledge and technical capabilities, while ensuring that they comply with the principles of sharīʿah, which aims to advance human welfare in ethics-based financial transactions. in many cases information asymmetry becomes a serious issue due to high risk and the limited knowledge of the technical processes. therefore, it is an important task to analyze models based upon the agency theory. delpachitra: is islamic banking capable of meeting corporate social responsibi 54 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 based upon the classical model inference by holmstrom’s (1979) assumption regarding the agent’s utility function, it is assumed that both principal and agents are fully aware that they both rationally pursue the maximization of utility. therefore, the agents strive to increase the firms’ outputs; and then principals and agents distribute the outputs according to the compensation system. in other words, agents work hard in order to improve the firms’ value, thereby increasing their wealth, and thus the agents’ utility is increased. however, the firms’ outputs are determined in accordance with the agents’ contributions and other reasons beyond the agents’ control. the factors beyond the agents’ control are called uncontrollable risk. as far as agency relationship is concerned, wealth is the only element that determines the agents’ utility function. however, the agents’ utility functions are determined according to their wealth and their contributions to the investee firm. therefore, contributions result in a negative utility and the negative utility increases as contributions increase. agency relationship research often places equal emphasis on both supervision and incentive plans (shleifer & vishny, 1997). however, more recent research shows that supervision and incentive supplement each other (lin, 2004). ib provides incentive compensation to inspire the entrepreneurs to improve the investee firms’ operating performance. the entrepreneurs receive bonuses and dividends, in addition to their fixed salaries. the ibs design the incentive mechanism to ensure that the clients’ interests are satisfied through the entrepreneurs’ decisions and behavior. the control mechanism is designed to minimize the agency-related issues and harmonize the relationships between principals and agents. ibs differ from conventional fund providers because they not only provide funds, but are also required to take part in the administration or decision-making processes of the investee firms in order to ensure proper compliance with the principles of islam. if the conventional agency theory by amit, glosten and muller (1990) is interpreted in the islamic sense, ibs supervise the agents as long as information asymmetry exists between ibs and investee firms. holmstrom (1979) states that agents’ compensation serves as an incentive mechanism. however, the supervision mechanism was not included in the model. a firm’s production would also be affected by supervision. therefore, following holmstrom (1979) the principal’s participation is included as the supervision mechanism for agents in the model used in this paper. holmstrom (1979) incorporates the neo-classical utility theory into his model and assumed that human beings pursue maximization of personal interests. the basic model between principals and agents of holmstrom stresses the value of information. but it overlooks the contributions made by technology and knowledge to production. the utility function is a useful model to describe the behavior of principal and agents. ib behavior can be analyzed in accordance with agency theory. in considering model specification and the robustness of the inferences, this paper assumes that an agency relationship exists between islamic investors (or ib) and entrepreneurs (investee firms). it thus attempts to solve the limitations of the basic model between principals and agents to create a well-defined model of the agency relationship between ib and investee firms. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 6 is islamic banking capable of meeting corporate social responsibility?: 49-66 55 agency theory is incorporated into the islamic banking model with emphasis on the saving units (su) (the investors) contributions to, and influence on the borrowing units (bu) (the entrepreneurs) when information asymmetry exists between the su and bu. islamic investors and entrepreneurs are assumed to be risk-averters as in standard theory and, as far as the agency relationship is concerned, neither party may disturb the other party. this is further complemented by their islamic faith, that is, the relationships are sharīʿah-compliant as to be continually supervised. therefore, the sus and bus attempt to maximize their personal utility and abide by the contracts at the same time. entrepreneurs act as agents for islamic investors and are therefore required to work in the best interests of the investors. in the light of information asymmetry, islamic investors need to launch incentive plans to inspire the entrepreneurs to work for the best interest of the islamic investors, thereby minimizing the conflicts of interest between the entrepreneurs and islamic investors. in other words, the investee firms’ operating performance changes the entrepreneurs’ compensation positively, thus the islamic investors and entrepreneurs share a common interest. operating performance determines the profit level. in the case of islamic finance, an incentive plan alone cannot solve the conflicts of interest resulting from attitudes towards risk. in this case, the supervision mechanism can reduce conflicts of interest and check for any deviation from islamic principles. as stated earlier, ibs provide not only funds, but also extra services to the investee firms commensurate with sharīʿah. therefore, a key decision variable¾ib participation¾is included in the model to determine the contributions made by ibs to investee firms. an information asymmetry between the islamic investors and entrepreneurs is assumed; and ib participation serves as a supervisory mechanism.as far as the agency relationship between an ib and entrepreneurs is concerned, it is assumed that the wealth possessed by the ib, or the difference between the firm’s profits and the entrepreneur’s income, determines the ib’s utility function. the entrepreneurs’ utility function is determined by the firms’ profits and the investee firms’ operational achievement. further, it is assumed that operational advancement creates negative utility for the entrepreneurs. this paper adopts the theoretical framework of holmstrom’s (1979) to explain its relevance to ib and financial transactions. holstrom’s model is extended to include the critical contributions made by firms as well as the principals’ contributions to operating performance. the model makes the following assumptions: • firms increase their productivity through operational advancement. therefore, the investee firms’ operating capabilities determine the ib’s outputs. let be the investee firm’s operating capability, a be the set of relevant technologies and r be the set of real numbers. the investee firms are free to choose either to request ibs or research institutes for technological transfer or cultivate specialists to upgrade their firm’s operational capabilities. facing information asymmetry, 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and delpachitra: is islamic banking capable of meeting corporate social responsibi 56 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 entrepreneurs are likely to determine whether to initiate the decisions related to operational improvement based on their personal interests. thus, their decisions affect the firms’ profits; • let q represent the ib’s profits. the profits are determined in accordance with a number of factors beyond the entrepreneurs’ control, in addition to . these factors are represented by a random variable ; knowns a random state of nature. it is assumed to represent the external environmental factors affecting the ib’s profits. it is assumed that the entrepreneurs face after is selected. in other words, the profits of the ib q are determined by the investee firms’ operational capabilities , together with the random variable ; that is. it is assumed that the entrepreneurs only adopt technologies that increase profits. further, it is assumed that the marginal profits created via the investee firms’ operational capabilities decrease. thus is continuously differentiable up to the second order partial derivative with respect to and (1) • ibs may use a number of methods to provide extra services to investee firms and thus help them increase profits. the services include technical transfer, supervision and consultation, and help financial administration. ibs provide these services to increase the firms’ profits. let represent ib participation. this paper attempts to find out whether due to ib participation, a, help the firms to increase their profits; • investors tend to support entrepreneurs to minimize agency-related issues between investors and entrepreneurs, and thus prevent entrepreneurs from putting their personal interests first and thereby hurting the firms’ interests. as far as income compensation is concerned, entrepreneurs are entitled to a fixed amount of remuneration and variable compensation – bonuses and dividends. the amount of variable compensation is determined in accordance with the profits of the investee firms. it is assumed that the entrepreneur’s compensation increases as the firms’ profit increases. let represent the proportion of profits shared with the entrepreneurs. an ib helps firms to increase profits through participation and, therefore, the firms’ profits are determined by the ib’s participation. hence represents the profits of firms with ib’s participation and also represents the entrepreneur’s variable compensation. let represent the fixed amount of compensation for entrepreneurs. then the incentive plan for entrepreneurs is represented by the linear structure as follows: . set and rearrange (2) 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4)     kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()( )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and international journal of banking and finance, vol. 10, iss. 2 [2013], art. 6 u (ce, )  m (ce)  k ( ) is islamic banking capable of meeting corporate social responsibility?: 49-66 57 and (3) then c e represents the entrepreneurs’ compensation and c v represents the compensation for the islamic investors. as the ibs increase their participation, islamic investors are more likely to launch a reward system to encourage entrepreneurs to improve the operating performance of the firms, thereby solving the agency-related issues between islamic investors and entrepreneurs and improving the firms’ operating performance; thus creating a win-win situation. if the whole process is sharīʿah compliant, islamic investors are likely to share a higher percentage of profits with entrepreneurs as ibs increase their participation in investee firms; • consistent with holmstrom (1979), this paper assumes that the von neumann-morgenstern utility function (abbreviated as vn-m utility function) represents the utility functions for the principles and agents. let (4) be the utility function for islamic investors and (5) be the utility function for entrepreneurs, where represents the utility of the profits earned by the entrepreneurs from the firms and represents the negative utility created by the investee firms’ technologies for the entrepreneurs. v, m and k are assumed to be continuously differentiable up to their second-order derivatives. in addition, it is assumed that islamic investors and entrepreneurs are risk averse and their utility increases and marginal utility decreases as their wealth increases, that is, and (6) function ( )gk represents the negative utility created by the investee firms’ technologies for the entrepreneurs. the negative effect is assumed to increase as the investee firms enhance their operational capabilities for the entrepreneurs and the marginal negative utility increases as the investee firms enhance their operational capabilities, i.e., 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 60 rα q    q   ,qq    ,q    0 ,     q ,   0 , 2 2     q (1) 0a  1,0p     ,1 qa     ,1 qa 0f    pqaf  ,1 pqafce ),()1(:  (2) fpqacv  )1)(,()1(  (3) ec vc  fpqavcv v  )1)(,()1()(  (4) )()(),(  kcmcu ee      kpqafm  ),()1( (5) )( ecm  k   0)('  v v v dc cdv cv ,   0)('' 2 2  v v v dc cvd cv and 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) delpachitra: is islamic banking capable of meeting corporate social responsibi 58 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 (7) 4.1 model inference and proposition the contractual behaviors between islamic investors and entrepreneurs are analyzed in the context of the assumptions made at the beginning. both parties understand that the other intends to maximize utility rationally. therefore, the contractual behavior is expressed in the optimization and implications equation as follows: (8) subject to (9) and (10) an optimum strategy or optimal solution of the optimization problem (8)–(10) is assumed to exist for the investor. according to equation (8), islamic investors select the most suitable contracts to maximize their expected utility. equation (9) represents the entrepreneurs’ considerations regarding opportunity costs, where ū represents the reserved utility level, which assumes entrepreneurs’ expected utility to stay above a certain level. in reality, ū is determined in accordance with the market situation. if the utility created by the agency contract is lower than the utility determined by the market, entrepreneurs are unlikely to accept the agency contract. equation (10) relates to the decisions that maximize the entrepreneurs’ expected utility under the incentive plan . islamic investors are not familiar with the operational capabilities of the investment objects when they enter into a contract. the asymmetry of the information between islamic investors and entrepreneurs is characterized by the dependence on g in their utilities. entrepreneurs are likely to initiate the decisions related to the operational advancement based upon their personal interest and religious beliefs, and thus maximize their expected utility. the necessary condition of the constraint (10) is given by its first order condition with respect tog . taking the utility function for entrepreneurs we differentiate this with respect to g to obtain the necessarycondition (11) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) international journal of banking and finance, vol. 10, iss. 2 [2013], art. 6 is islamic banking capable of meeting corporate social responsibility?: 49-66 59 taking the expression for entrepreneurs’ compensation given by and substituting this in the utility function for entrepreneurs we have from (11) that (12) assume the probability distribution of q is continuous. it then follows from leibnitz’s rule and (4.1.12) that (13) to simplify the discussion, assume necessary conditions for the implicit function theorem to hold so that can be solved implicitly from equation (1.13) in terms of other variables. therefore can be determined as a function of (a, p, f), that is, to obtain the necessary conditions for the optimization, the lagrangian multiplier is introduced (based on the necessary condition (13) of (10) and (8)−(9)), (14) where l and m are the corresponding lagrangian multipliers. under the assumption of the existence of the optimal solution of the optimization problem, the kuhn-tucker theorem (see pp. 740–741 in chapter 19 in taha, 1992) is used to obtain the following necessary conditions for the optimal solution (a*, p*, f*; g*)1 (15) (16) (17) therefore, if (a*, p*, f*, g*) is an optimal solution of the optimization problem, it should satisfy the necessary conditions (15)–(17). based on these assumptions, g*= g* (a, p, f) solves the equation (17). after substituting g*= g(a, p, f) into (1.15) and (1.16), the constrained optimization problem is treated as finding a solution in variable (a, p, f). the optimal solution is then substituted, say (a*, p*, f*) , back to g* to obtain g*= g*(a*, p*, f*) . 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) g g 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 61   0)('  e e e dc cdm cm ,   0)('' 2 2  e e e dc cmd cm (6)  k   0)('     d dk k ,   0)('' 2 2     d kd k (7) fpa max ,,  )( vcvε (8)  ),(  ecuε ū (9)  ),(maxarg     ea cue (10) ū   . ),( ecu ,    0 ),(     ecue (11) pqafce ),()1(      kcmcu ee ),( ,        0 ),1(      kpqafme (12)    0)()1(            k q pa dc cdm e e e (13)   ),,( fpa  .          ),()(:),,;,,(   ev cuhcvfpal ee +                )()1)((    k q paceme , (14) 62   );,,( **** fpa 1 ,0   a l ,0   p l ,0   f l (15)   ,),(   hcu e e ,0   0),(         hcu e  e (16)   0)()1(            k q pa dc cdm e e e (17) ),,,( **** fpa ),,(** fpa  ),,(* fpa  ),,( fpa . ),,( *** fpa , * ),,( ***** fpa  . p  .  ,0   p  (18)   )()1(:);,,(    k q pa dc cdm fpaw e e          e (19) );,,( fpaw =0 (20)  0        p w p w   (21) 1there is no condition on  after equation (1.17) since the constraint (1.13) is an equality, unlike the inequality constrain in equation (1.16) 62   );,,( **** fpa 1 ,0   a l ,0   p l ,0   f l (15)   ,),(   hcu e e ,0   0),(         hcu e  e (16)   0)()1(            k q pa dc cdm e e e (17) ),,,( **** fpa ),,(** fpa  ),,(* fpa  ),,( fpa . ),,( *** fpa , * ),,( ***** fpa  . p  .  ,0   p  (18)   )()1(:);,,(    k q pa dc cdm fpaw e e          e (19) );,,( fpaw =0 (20)  0        p w p w   (21) 1there is no condition on  after equation (1.17) since the constraint (1.13) is an equality, unlike the inequality constrain in equation (1.16) 62   );,,( **** fpa 1 ,0   a l ,0   p l ,0   f l (15)   ,),(   hcu e e ,0   0),(         hcu e  e (16)   0)()1(            k q pa dc cdm e e e (17) ),,,( **** fpa ),,(** fpa  ),,(* fpa  ),,( fpa . ),,( *** fpa , * ),,( ***** fpa  . p  .  ,0   p  (18)   )()1(:);,,(    k q pa dc cdm fpaw e e          e (19) );,,( fpaw =0 (20)  0        p w p w   (21) 1there is no condition on  after equation (1.17) since the constraint (1.13) is an equality, unlike the inequality constrain in equation (1.16) 1 there is no condition on m after equation (1.17) since the constraint (1.13) is an equality, unlike the inequality constrain in equation (1.16) delpachitra: is islamic banking capable of meeting corporate social responsibi 60 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 the following discussion examines how the profit-sharing ratio designed by an ib for entrepreneurs and the investee firms’ operational capabilities are related. intuitively, the profit-sharing ratio p designed by the ib for the entrepreneurs would encourage the entrepreneurs to improve the investee firm’s operational capabilities g. in other words, the entrepreneurs are likely to increase the investee’s operational capabilities g as more profits are allocated to the entrepreneurs. mathematically, this suggests that (18) the sufficient conditions under which (1.18) holds are now derived. denote (19) it then follows from the necessary condition (17) that (20) equation (20) is differentiated with respect to g to obtain (21) based on the assumption that the utility function m is twice continuously differentiable, leibnitz’s rule is used to obtain from (19) that (22) and (23) note that it then follows from (22) and (23) that (1.24) 62   );,,( **** fpa 1 ,0   a l ,0   p l ,0   f l (15)   ,),(   hcu e e ,0   0),(         hcu e  e (16)   0)()1(            k q pa dc cdm e e e (17) ),,,( **** fpa ),,(** fpa  ),,(* fpa  ),,( fpa . ),,( *** fpa , * ),,( ***** fpa  . p  .  ,0   p  (18)   )()1(:);,,(    k q pa dc cdm fpaw e e          e (19) );,,( fpaw =0 (20)  0        p w p w   (21) 1there is no condition on  after equation (1.17) since the constraint (1.13) is an equality, unlike the inequality constrain in equation (1.16) 62   );,,( **** fpa 1 ,0   a l ,0   p l ,0   f l (15)   ,),(   hcu e e ,0   0),(         hcu e  e (16)   0)()1(            k q pa dc cdm e e e (17) ),,,( **** fpa ),,(** fpa  ),,(* fpa  ),,( fpa . ),,( *** fpa , * ),,( ***** fpa  . p  .  ,0   p  (18)   )()1(:);,,(    k q pa dc cdm fpaw e e          e (19) );,,( fpaw =0 (20)  0        p w p w   (21) 1there is no condition on  after equation (1.17) since the constraint (1.13) is an equality, unlike the inequality constrain in equation (1.16) 62   );,,( **** fpa 1 ,0   a l ,0   p l ,0   f l (15)   ,),(   hcu e e ,0   0),(         hcu e  e (16)   0)()1(            k q pa dc cdm e e e (17) ),,,( **** fpa ),,(** fpa  ),,(* fpa  ),,( fpa . ),,( *** fpa , * ),,( ***** fpa  . p  .  ,0   p  (18)   )()1(:);,,(    k q pa dc cdm fpaw e e          e (19) );,,( fpaw =0 (20)  0        p w p w   (21) 1there is no condition on  after equation (1.17) since the constraint (1.13) is an equality, unlike the inequality constrain in equation (1.16) 62   );,,( **** fpa 1 ,0   a l ,0   p l ,0   f l (15)   ,),(   hcu e e ,0   0),(         hcu e  e (16)   0)()1(            k q pa dc cdm e e e (17) ),,,( **** fpa ),,(** fpa  ),,(* fpa  ),,( fpa . ),,( *** fpa , * ),,( ***** fpa  . p  .  ,0   p  (18)   )()1(:);,,(    k q pa dc cdm fpaw e e          e (19) );,,( fpaw =0 (20)  0        p w p w   (21) 1there is no condition on  after equation (1.17) since the constraint (1.13) is an equality, unlike the inequality constrain in equation (1.16) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) international journal of banking and finance, vol. 10, iss. 2 [2013], art. 6 is islamic banking capable of meeting corporate social responsibility?: 49-66 61 and (25) the sign of in (1.25) is now examined. from the assumption that and, it can be seen from (25) that (26) noting that , equation (24) can be rewritten as (27) entrepreneurs’ compensation includes a fixed amount of remuneration and variable remuneration (1 + a)q(g, q)p. therefore, f is considered the entrepreneurs’ fixed wealth and the variable remuneration (1 + a)q(g, q)p is considered a risky asset. the entrepreneurs’ risks increase as the risk of the risky asset increases. it is assumed that the entrepreneurs avoid risks and the relative risk aversion coefficient of the utility function m of entrepreneurs is smaller than one (see huang & litzenberger (1988) for the definition and related discussion), (28) note that and hence this implies that (29) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 63                       q a dc cdmq pa p c dc cmd p w e ee e e )1()1(2 2 e (22) and        '')1()1( 2 2 2 2 k q pa dc cdmq pa c dc cmdw e ee e e                     e (23) ),()1( qa p ce    ,   ) , ()1(         q pa ce                       q a dc cdmq pqa dc cmd p w e e e e )1(,)1( 2 2 2 e (1.24) and              222 2 2 )()1(  q pa dc cmdw e e e      '')1( 2 2 k q pa dc cdm e e e             (25)  w (1.25) 0,0'',0,0 2 2      q mpa 0'' k 0    w (26) 0'm                          q cm cmpqa cma p w e e e ' "),()1(1 1')1(e (27) pqa ),()1(  . pqa ),()1(    1 )(' "   e e e cm cm c (28) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)   0 )(' )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29) )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)   0 )(' )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)      ecm cmpqa (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) delpachitra: is islamic banking capable of meeting corporate social responsibi 62 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 note that m' > 0 and 1 + a + > 0, we have from (1.29) that (30) following from the assumption , we obtain from (27) and (30) that (31) this, together with (26) and (21), leads to (18). it is, therefore shown that (18) holds under condition (28). by summarizing the analysis, proposition 1 is obtained. proposition 1: assume then in other words, if the relative risk aversion coefficient of the utility function m of the entrepreneurs is strictly less than one, then as the ib distributes a higher percentage of profits to entrepreneurs, the investee firms’ operational capabilities are improved. similarly, the investees’ operational capabilities are expected to increase as the ib increases its participation. mathematically, this corresponds to (32) equation (1.32) holds under condition (1.28). equation (1.20) is differentiated with respect to a to obtain (33) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)   0 )(' )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)   0 )(' )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)   0 )(' )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)   0 )(' )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)   0 )(' )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)   0 )(' )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)   0 )(' )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) international journal of banking and finance, vol. 10, iss. 2 [2013], art. 6 is islamic banking capable of meeting corporate social responsibility?: 49-66 63 based on the assumption that the utility function m is twice continuously differentiable, leibnitz’s rule is used to obtain from (19) that (34) note that it then follows from (1.34) that (35) we have shown that from the assumption that, condition (1.28) and equation (1.35), we have it then follows from (33) that (32) holds. this leads to proposition 2. proposition 2. assume then, in other words, if the relative risk aversion coefficient of the utility function m of the entrepreneurs is strictly less than one, then as the ibs increase their participation, the investee firms’ operational capabilities are improved. 64 0,0",0'  fmm ecpqafpqa  ),()1(),()1(  1 )( )( )(' )(),()1(      e ee e e cm cmc cm cmpqa  0 )( )(),()1(1 1     e e cm cmpqa  (29)   0 )(' )("),()1(1 1')1(         e e e cm cmpqa cma  (30) 0    q       0 ' "),()1(1 1')1(                     q cm cmpqa cma p w e e ee (31)   1 )(' "   e ee cm cmc . then 0   p  0   a  (32) 0        a w a w   (33)                       q p dc cdmq pa a c dc cmd a w e ee e e )1(2 2 e (34) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q          0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) delpachitra: is islamic banking capable of meeting corporate social responsibi 64 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 the ultimate purpose of the model is to find out whether the changes in ibs’ participation and the changes of the entrepreneurs’ profit-sharing ratio affect the investee firms’ operating performance. in this model, profit q(g,q) represents the investee firms’ operating performance. therefore, the following are investigated: and it follows from q = q(g,q) and that g = g (a, p, f) (36) it then follows from , (18) and (32) that under the same assumption as (28) (37) this leads to the following proposition 3. proposition 3. assume then in other words, if the relative risk aversion coefficient of the utility function m of the entrepreneurs is strictly less than one, then the final propositions are: 1 : as ibs increase their participation, the investee firms’ operating performance is improved; 2 : as ibs share a higher percentage of profits with entrepreneurs, the investee firms’ operating performance is improved. 5. summary and conclusions information asymmetry is a fundamental problem in agency-related issues between the ib and entrepreneurs. ibs provide funds, extra services and 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 65 ),( pq a ce    .                 q pcm q qpacm a w ee )('),()1)((' 2e =                 )(' ),()1)((" 1)(' e e e cm pqacmq pcm   e (35) 0    w . 0,0,0'      q pm , 0   a w . 1 )( )(    e ee cm cmc 0   a  ),( q p q   ),(  a q   ),(  . ),( qq  ),,( fpa  p q p q           ),(),( , a q a q           ),(),( (36) 0 ),(     q 0 ),(    p q  , 0 ),(    a q  (37) 66 1 )( )(    e ee cm cmc 0 ),(    p q  , 0 ),(    a q  . 66 1 )( )(    e ee cm cmc 0 ),(    p q  , 0 ),(    a q  . international journal of banking and finance, vol. 10, iss. 2 [2013], art. 6 is islamic banking capable of meeting corporate social responsibility?: 49-66 65 supervisory measures to the investee firms to overcome both agency-related and religious-compliance-related issues. ibs also design the incentive plan to inspire the agents, thereby improving the investee firms’ operating performance through their participation and incentive system. typically, agency theories stress the value of information and advocate that the incentive system encourages the agents to increase the firms’ profits. conventional agency theory stresses the value of information and, as a result, overlooks the importance of technology, learning for the firms with respect to creating value, and more importantly the role of religion. as stated in proposition 1, the investee firms’ operating performance is improved as ibs increase their participation. the ibs’ supervision mechanism reduces the information asymmetry and therefore, affects the investee firms’ operating performance positively. the influence imposed by the incentive plan on the investee firms’ operating performance is also examined. proposition 2 states that the investee firms’ operating performance is improved as the ib shares a higher percentage of profits to the entrepreneurs. incentive plans launched by ibs affect investee firms’ operating performance positively. over all, if the objective of islamic banking is to promote the welfare of the society then, by carrying out financial services as designed by ib principles, ibs will be able to ensure that their activities are socially responsible. author information: the author wishes to record his gratitude to the following for their research assistance on this study: tieh shang jin and jamila alaktif. sarath delpachitra is a professor in the flinders business school, flinders university, australia. he may be contacted at: email: sarath.delpachitra@ flinders.edu.au; phone: 618 82013893 references abdallah, a., & delpachitra, s. b. (2008). is islamic banking a successful alternative to conventional banking? a comparison of tow banking. in proceedings on the symposium of islamic banking and finance: global perspective on ethics and financial practices, 20 december 2008. melbourne: national centre of excellence for islamic studies, university of melbourne. acquier, a. & aggeri, f. (2008). une généalogie de la pensée managériale sur la rse. revue française de gestion, 34(180), pp. 131-157. amit, r., glosten, l., & muller, e. (1990). does venture capital foster the most promising entrepreneurial firms? california management review, 32, pp.102-111. anwar, m. (2003). islamicity of banking and modes of islamic banking. arab law quarterly, 18(1), pp. 62-80. archer, s., & abdel karim, r. a. (2002). agency theory, corporate governance and the accounting regulation of islamic banks. research in accounting regulation, 1(1), pp. 97–114. delpachitra: is islamic banking capable of meeting corporate social responsibi 66 the international journal of banking and finance, vol. 10. number 2, august 2013: 1-24 bowen, h. (1953). social responsibilities of the businessman. new york: harper and brothers. caroll, a. (1979). a three-dimensional conceptual model of corporate social performance. academy of management review, 4(4), pp. 497-505. dhumale, r & sapcanin, a. (2003). an application of islamic banking principles to microfinance. technical note. a study by the regional bureau for arab states, undp, in cooperation with the middle east and north africa region, world bank. el-hawary, d., grais, w., & iqbal, z. (2004). regulating islamic financial institutions: the nature of the regulated. world bank policy research working paper, no. 3227, march. guyot, a. (2009). le coût de l’investissement islamique : l’application des critères dow jones islamic au marché d’actions euronext paris. euromediterranean economics finance review, special issue for ifc 5, 4(3), third quarter, pp. 127-140. hasan, z. (2002). murābaʿah as a mode of finance in islamic banking: theory, practices and problems. the middle east business and economic review, 14(2), pp. 41-53. holmstrom, b. (1979). moral hazard and observability. bell journal of economics, 10, pp. 74-91. huang, c., & litzenberger, r. h. (1988). foundations for financial economics. upper saddle river, nj: prentice hall. lewis, m. & algaoud, l. m. (2001). islamic banking. cheltenham, uk: edward elgar. lin, s. h. (2004). research on the uncertainty of venture capital funds performance. working paper, national taiwan university. sairally, s. (2007). evaluating the ‘social responsibility’ of islamic finance: learning from the experience of socially responsible investment funds. in munawar iqbal, salman syed ali and dadang muljawan, advances in islamic economics and finance vol. 1. jeddah: islamic research and training institute. sarker, a. m. (1999). islamic business contracts, agency problem and the theory of islamic firm. international journal of islamic financial services, 1(2), pp. 15-49. shleifer a. & vishny, r. w. (1997). a survey of corporate governance. the journal of finance, 52, pp. 737-783. taha, h. a. (1992). operation research: an introduction. new york: maxwell macmillan. winant, p. (2008). la finance éthique. regards croisés sur l’économie, 1(3), pp. 84-85. international journal of banking and finance, vol. 10, iss. 2 [2013], art. 6 how to cite this article: badru, b. o., & davies, n. o., & abdulkadir, r. i. (2020). board size determinants: evidence from nigeria. international journal of banking and finance, 15(1), 89-103. https://doi.org/10.32890/ijbf2020.15.1.9933 board size determinants: evidence from nigeria 1bazeet olayemi badru school of economics, finance and banking, universiti utara malaysia 2nordiana osagie davies waziri umaru federal polytechnic, birnin kebbi 3rihanat idowu abdulkadir faculty of management sciences, university of ilorin 1corresponding author: badru@uum.edu.my; bazeetolayemi@ gmail.com 2osagiedavies@gmail.com 3riolaq29@yahoo.com a r t i c l e i n f o article history: received 21 december 2019 revised 8 january 2020 accepted 13 january 2020 published 31 march 2020 keywords: board size; corporate governance; ceo ownership; ownership concentration; nigeria jel classification: g3; g34 a b s t r a c t this paper seeks to investigate the determinants of board size for nigerian companies. to accomplish the aim of the study, a panel data set of public listed companies in nigeria from 2005 to 2015 was employed. the results showed that the most common board size of nigerian companies ranged from four to 18 members. specifically, the findings indicated that board size was a function of company and industry characteristics. a significant and positive association was found between company size and board size, while ceo ownership and ownership concentration were negative. the results lend support to theoretical arguments that a company’s board structure is determined by the scope of company operations and monitoring costs associated with the company. since company-specific characteristics the international journal of banking and finance, vol. 15. number 1, 2020: 89-103 89 90 the international journal of banking and finance, vol. 15, no 1, 2020 : 89-103 1. introduction in today’s corporate environment, emphasis on improving corporate governance is mainly centred on the board of directors (bods). the bods is the most important corporate governance mechanism in companies and has a wide range of responsibilities, which include monitoring company operations, advising top management and making strategic decisions that may affect company financial performance and value as well as sustainability. however, for the bods to carry out its responsibilities effectively, understanding the environment in which the company operates is vital. one key factor that affects board effectiveness and which reflects the ability of a company to link the company with the environment and to secure critical resources is the size of the board (nguyen, rahman, tong, & zhao, 2015; pfeffer, 1972). board size, which is considered as the number of directors in the corporate boardroom and this number (small or large), is crucial in enabling the bods to discharge its responsibilities. for instance, scholars arguing from the agency theory perspective have suggested that a small board can monitor more effectively, while a large board can be ineffective and slow in decisionmaking because of coordination and communication problems leading to the deterioration of company value (jensen, 1993; lipton & lorsch, 1992). however, resource dependence theory scholars have opined that large boards provide directors with the chance to specialise, which in turn, can result in better advice on the company’s management (pfeffer, 1972; zahra & pearce, 1989). in fact, companies which are highly in need of advice benefit from having a large board (coles, daniel, & naveen, 2008). these two opposing arguments suggest that board size depends on the costs and benefits of board functions in terms of monitoring and advising. thus, the question is what are the drivers of a company’s board size? some scholars mentioned that the scope and complexities of a company’s operations as well as specific business and information environment, are the drivers of a determine board size, the impact of board size on corporate outcomes may differ based on these characteristics. therefore, it would be helpful if future studies could consider the interactive effect of company characteristics when investigating the impact of board size on corporate outcomes. board size determinants: evidence from nigeria: 89-103 91 company’s board structure, including its size (coles et al., 2008; lehn, patro, & zhao, 2009; raheja, 2005). however, empirical evidence in this respect is scarce with most being investigated in advanced countries (the united states and the united kingdom), with little attention given to emerging and frontier markets, like nigeria. therefore, this study examines the determinants of board size of companies in the nigerian capital market. investigating this issue is particularly important, considering the contextual differences in business practices and environments of frontier markets. similarly, there are divergent views and mixed findings on the impact of board size on its effectiveness, which in turn, influences corporate outcomes, such as performance and company value. there is also a concern with regards to the channel that board size influence on corporate outcomes passes through. in addition, capital market regulators in most countries require companies to have a reasonable number of board directors. specifically, in nigeria, section 4.2 of the code of corporate governance for public companies states that membership of the board should not be less than five. this means that companies have the discretion to decide on the number of directors that can sit on the board. similarly, in nigeria, which is one of the most important financial markets, corporate boards have been considered inefficient (adegbite, amaeshi, & nakajima, 2013). hence, it is important to examine what factors affect board size, as this would enable future researchers to identify how board size can influence company effectiveness and corporate outcomes. the remainder of the paper is structured as follows. section 2 discusses the literature review, while section 3 describes the research method used. section 4 reports the results and finally, section 5 presents a discussion followed by the conclusion. 2. literature review and hypotheses development the bods is defined as a group of individuals who represent the shareholders and other stakeholders of the company to carry out certain functions1, whereby the size of the board plays a vital role (golden & zajac, 2001). therefore, understanding the role that the size of the board plays in a company is important. there are different and opposing theoretical views on the exact role of board size. scholarly literature from the agency theory point of view has indicated that small boards are more effective in monitoring than large size boards. some of the reasons to this effect are that small boards have fewer opportunities for freeriding, are more cohesive and more productive, and these assist the directors in their monitoring function of the company as well as their ability to be involved in strategic decision-making. however, increase in board size makes coordination and communication difficult among directors, leading to internal conflicts (coles et al., 2008; 92 the international journal of banking and finance, vol. 15, no 1, 2020 : 89-103 firstenberg & malkiel, 1994; jensen, 1993; lipton & lorsch, 1992). this also creates difficulties for directors to organize board meetings and reach agreement due to their numbers and operational complexities (guest, 2009; jensen 1993). it also increases agency problems in the boardroom and inhibits the board’s ability to initiate strategic changes in the company. a typical example is goodstein, gautam, and boeker’s (1994) study which showed that boards with a large number of directors made slower and less-efficient decisions. such boards held back corporate restructuring decisions and were slow to react in difficult situations. large boards were also more likely to be controlled by the chief executive officer (ceo) instead of the other way around. it is based on this line of argument that shareholders generally consider small boards and pressure companies to reduce board size due to problems inherent in large boards. for example, jensen (1993) argued that increase in board size beyond seven or eight led to a negative relationship between board size and company value. likewise, in the case of the nigerian banking sector, uwuigbe and fakile (2012) showed that banks with boards less than 13 directors were more viable than banks with more than 13 directors. in addition, companies with large boards have been found to record lower profits than those with small boards. similarly, certain empirical literature has shown that a large board is detrimental to the company (cheng, 2008; guest, 2009; kumar & singh, 2013; o’connell & cramer, 2010; nguyen et al., 2015; yermack, 1996). in contrast to the agency theory perspective, the resource dependence theory is of the view that companies are better off with large boards. the theory posits that there is a tendency that a board with a large board size would have directors with diverse educational qualifications and industry experience, which would result in greater monitoring. it would also enhance the company’s access to more resources and expertise (goodstein et al., 1994). these resources include access to markets, new and better technologies and raw materials. for instance, booth and deli (1996) stated that large boards enabled companies to access a wide range of expertise needed to overcome environmental uncertainties. musteen, datta and kemmerer (2010) reported that companies with large boards had a better reputation compared to small boards. coles et al. (2008) mentioned that companies requiring more advice derived greater value by having a large board. cheng (2008) suggested that large boards were necessary for some types of companies and under certain situations. therefore, it can be concluded that large boards would provide management with high quality advice and strategic decisions that can create value for the company (dalton, daily, johnson, & ellstrand, 1999; pfeffer, 1972; zahra & pearce, 1989). these conflicting arguments therefore, indicate that each company has an optimal board size that can be explained with the company’s characteristics, in particular the scope and complexity of its operations, specific business board size determinants: evidence from nigeria: 89-103 93 and information environment and negotiation between the company’s ceo and outside board members (boone, field, karpoff, & raheja, 2007). these characteristics are what theoretical papers refer to as scope of operations and monitoring cost hypotheses (adams & ferreira, 2007; coles et al., 2008; fama & jensen, 1983; harris & raviv, 2007; lehn et al., 2009). under the scope of operations hypothesis, several studies are of the view that monitoring and providing strategic directions may be a challenging task for complex companies. this is because such companies exist in a complex business environment that would need external human capital for effective decisionmaking (booth & deli, 1996; boone et al., 2007; cicero et al., 2013; pfeffer, 1972). according to fama and jensen (1983), the organization of a company was a reflection of the scope and complexity of its production processes. therefore, large or more complex processes would lead to large and more hierarchical companies, thereby making the job of ratifying and monitoring senior managers’ decisions more complex. hence, the need for a large board becomes important. in addition, the ability of the board to carry out its resource dependence functions, such as provision of information and expertise and creation of channels of communication with constituents of importance for the company are dependent on the number of directors on the board (carter, d’souza, simkins, & simpson, 2010). indeed, boone et al. (2007) reported that companies with operations that were more complex may likely require large boards with a greater proportion of independent directors. accordingly, the requirement of the board to increase the number of independent directors would in turn, increase board size (coles et al., 2008; linck, netter, & yang, 2008). several studies have suggested that large companies tend to be more involved in diverse activities (e.g., merger and acquisition activities and using more sophisticated and financial marketing techniques). such companies therefore need more directors for the new tasks, including succession planning, compensation and auditing, as well as more information compared to small companies (bhagat & black, 1999; agrawal & knoeber, 1996; lehn et al., 2009; pearce & zahra, 1992). in another related study, nguyen et al. (2015) found that the sensitivity of ceo compensation to company size was more prevalent in australian companies with a large board. this indicated that a large board would exhibit lower operating performance and higher operating costs. in a similar vein, guest (2009) reported that the negative effect of board size on company performance was more pronounced in large companies that tended to have large boards. consistent with this notion, a number of studies have shown that board size is positively associated with company size. for example, germain, galy, and lee (2014) found that companies’ operational level was a significant determinant of the board size in the malaysian stock market. other studies have also shown that company size is positively associated with board size, which 94 the international journal of banking and finance, vol. 15, no 1, 2020 : 89-103 means larger and more complicated companies would need more directors’ expertise and external resources (boone et al., 2007; lehn et al., 2009; linck et al., 2008; min, 2018; yermack, 1996). therefore, in line with the scope of operations hypothesis, this study hypothesizes that: h1 : company size is positively associated with board size. in contrast to the positive effect of a company’s scope of operations, a considerable number of studies pointed out that free-riding problems caused by having larger boards may result in less monitoring services offered by board members, which makes the board less effective (adams & ferreira, 2007; harris & raviv, 2007; raheja, 2005). the net benefits of additional monitoring increase with managers’ opportunities to consume private benefits, but decrease with the costs of monitoring. therefore, it can be suggested that board size emerges from the trade-off between firm-specific benefits and costs of increased monitoring, which is determined by the environment that a company operates in. for example, linck et al. (2007) demonstrated that companies facing greater information asymmetry tended to have small and less independent boards because of the high costs of monitoring. an important example is a high growth company, which is often characterized as high in information asymmetry, may have a small board because of the high costs of monitoring (smith & watts, 1992; gaver & gaver, 1993). consistent with the monitoring cost hypothesis, germain et al. (2014) documented that board size was correlated with the monitoring cost hypothesis, such as market-to-book ratio, free cash flow and market concentration. lehn et al. (2009) found that board size was negatively associated with growth opportunities. min (2018), linck et al. (2008) and boone et al. (2007) demonstrated that board size was negatively related to the company’s costs of monitoring, such as ceo ownership and research and development (r&d) expenditure. this means that companies with high managerial ownership, high growth opportunities and high r&d expenditures are associated with small boards. kyereboah-coleman and biekpe (2007) also reported that ceo tenure, as a measure of monitoring costs, is negatively associated with board size. this implies that when the costs of monitoring are high, the board should be small. therefore, the next hypothesis is that: h2 : ceo ownership and ownership concentration are negatively associated with board size. 3. research design and data the unit of analysis in this study was the company. this study considered 80 listed companies on the nigerian stock exchange. based on data availability, the board size determinants: evidence from nigeria: 89-103 95 sample data was collected from the period, 2005 to 2015. the board structure and each company’s financial data were extracted manually from annual reports of the sampled companies and the factbook released by the nigerian stock exchange. in line with prior studies, financial companies were excluded due to their financial ratios characteristics (guest, 2008; germain et al., 2014). the dependent variable in this study was board size, while the independent variables were company size, age and ceo ownership. board size was measured as the natural logarithm of the total number of directors on the board as indicated in the annual reports of the companies (coles et al., 2008; germain et al., 2014). company size was measured as the natural logarithm of the total assets of the company in each year of analysis, while company age was computed as the natural logarithm of the company age from the time the company was established. these two variables were proxies for the scope of operations hypothesis. to capture the monitoring cost hypothesis, this study used ceo ownership, which was identified as the percentage of company shareholdings owned by the company’s ceo. other control variables considered in this study were ownership concentration, industry classification of the company and time dummies. another important variable that could determine board size was the company’s ownership structure (ning, davidson, & zhong, 2007), because a company with a concentrated ownership in the hands of a few directors tended to have a small board (denis & sarin, 1999). in addition, the industry in which a company operates may significantly affect its board size because of the variability associated with the costs and benefits of large boards across different industries (ning et al., 2007). since the sample contained data across companies and over time, this study employed the multivariate panel regression technique. this allowed the study to estimate the effects that were simply not detectable in pure crosssectional or time-series data (germain et al., 2014; ahmed sheikh et al., 2013). where represents board size for the ith company at time t, measured as the natural logarithm of number of directors on the board; captures company size for the ith company at time t and it is measured as the natural logarithm of total assets of the company; represents the company age for the ith company at time t, which is measured as the natural logarithm of the company’s age from year of establishment; is the company ceo’s ownership for ith company at time t, which is the proportion of shares owned by the ceo; represents ownership concentration for the ith company at time t, which is the percentage of shares controlled by block holders which is at least 10% of the total share of the company; 𝐿𝑁𝐵𝑆𝐼𝑍𝐸�� = 𝛽� + 𝛽� 𝐿𝑁𝑇𝐴�� + 𝛽� lncage�� + 𝛽� 𝐶𝐸𝑂𝑊𝑁�� + 𝛽� 𝑂𝑊𝐶𝑂𝑁�� + 𝛽� 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝛽� 𝑌𝐸𝐴𝑅 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝜀�� where 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 represents board size for the for ith company at time t, measured as the natural logarithm of number of directors on the board; 𝐿𝑁𝑇𝐴�� captures company size for the ith company at time t and it is measured as the natural logarithm of total assets of the company; lncage�� represents the company age for the ith company at time t, which is measured as the natural logarithm of the company’s age from year of establishment; 𝐶𝐸𝑂𝑊𝑁�� is the company ceo’s ownership for ith company at time t, which is the proportion of shares owned by the ceo; 𝑂𝑊𝐶𝑂𝑁�� represents ownership concentration for the ith company at time t, which is the percentage of shares controlled by block holders is at least 10% of total share of the company; 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 represents industry dummies; year represents year dummies; and 𝜀�� is the random error for the ith company at time t. 3.1 empirical results 3.1.1 descriptive statistics and correlation analysis the descriptive statistics of the variables considered in the study are displayed in table 1. based on the figures shown, the average board size is nine, the minimum is four and the maximum is 18. in addition, the average age of companies is 38 years and the maximum is 92 years, which means that a company is as old as 92 years. the average percentage of ceo ownership is 4% and a ceo owns as much as 63.65% of the company shares. there is also a high degree of ownership concentration among nigerian companies. on average, ownership concentration is 53.83% and is as much as 95%. the average total asset of nigerian companies is 29.360 million naira and the maximum value is 950 million naira. in addition, the correlation results shown in table 2 reveal that company age (lncage) and company size (lnta), are statistically significant and positively correlated to board size (lnbsize). 𝐿𝑁𝐵𝑆𝐼𝑍𝐸�� = 𝛽� + 𝛽� 𝐿𝑁𝑇𝐴�� + 𝛽� lncage�� + 𝛽� 𝐶𝐸𝑂𝑊𝑁�� + 𝛽� 𝑂𝑊𝐶𝑂𝑁�� + 𝛽� 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝛽� 𝑌𝐸𝐴𝑅 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝜀�� where 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 represents board size for the for ith company at time t, measured as the natural logarithm of number of directors on the board; 𝐿𝑁𝑇𝐴�� captures company size for the ith company at time t and it is measured as the natural logarithm of total assets of the company; lncage�� represents the company age for the ith company at time t, which is measured as the natural logarithm of the company’s age from year of establishment; 𝐶𝐸𝑂𝑊𝑁�� is the company ceo’s ownership for ith company at time t, which is the proportion of shares owned by the ceo; 𝑂𝑊𝐶𝑂𝑁�� represents ownership concentration for the ith company at time t, which is the percentage of shares controlled by block holders is at least 10% of total share of the company; 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 represents industry dummies; year represents year dummies; and 𝜀�� is the random error for the ith company at time t. 3.1 empirical results 3.1.1 descriptive statistics and correlation analysis the descriptive statistics of the variables considered in the study are displayed in table 1. based on the figures shown, the average board size is nine, the minimum is four and the maximum is 18. in addition, the average age of companies is 38 years and the maximum is 92 years, which means that a company is as old as 92 years. the average percentage of ceo ownership is 4% and a ceo owns as much as 63.65% of the company shares. there is also a high degree of ownership concentration among nigerian companies. on average, ownership concentration is 53.83% and is as much as 95%. the average total asset of nigerian companies is 29.360 million naira and the maximum value is 950 million naira. in addition, the correlation results shown in table 2 reveal that company age (lncage) and company size (lnta), are statistically significant and positively correlated to board size (lnbsize). 𝐿𝑁𝐵𝑆𝐼𝑍𝐸�� = 𝛽� + 𝛽� 𝐿𝑁𝑇𝐴�� + 𝛽� lncage�� + 𝛽� 𝐶𝐸𝑂𝑊𝑁�� + 𝛽� 𝑂𝑊𝐶𝑂𝑁�� + 𝛽� 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝛽� 𝑌𝐸𝐴𝑅 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝜀�� where 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 represents board size for the for ith company at time t, measured as the natural logarithm of number of directors on the board; 𝐿𝑁𝑇𝐴�� captures company size for the ith company at time t and it is measured as the natural logarithm of total assets of the company; lncage�� represents the company age for the ith company at time t, which is measured as the natural logarithm of the company’s age from year of establishment; 𝐶𝐸𝑂𝑊𝑁�� is the company ceo’s ownership for ith company at time t, which is the proportion of shares owned by the ceo; 𝑂𝑊𝐶𝑂𝑁�� represents ownership concentration for the ith company at time t, which is the percentage of shares controlled by block holders is at least 10% of total share of the company; 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 represents industry dummies; year represents year dummies; and 𝜀�� is the random error for the ith company at time t. 3.1 empirical results 3.1.1 descriptive statistics and correlation analysis the descriptive statistics of the variables considered in the study are displayed in table 1. based on the figures shown, the average board size is nine, the minimum is four and the maximum is 18. in addition, the average age of companies is 38 years and the maximum is 92 years, which means that a company is as old as 92 years. the average percentage of ceo ownership is 4% and a ceo owns as much as 63.65% of the company shares. there is also a high degree of ownership concentration among nigerian companies. on average, ownership concentration is 53.83% and is as much as 95%. the average total asset of nigerian companies is 29.360 million naira and the maximum value is 950 million naira. in addition, the correlation results shown in table 2 reveal that company age (lncage) and company size (lnta), are statistically significant and positively correlated to board size (lnbsize). 𝐿𝑁𝐵𝑆𝐼𝑍𝐸�� = 𝛽� + 𝛽� 𝐿𝑁𝑇𝐴�� + 𝛽� lncage�� + 𝛽� 𝐶𝐸𝑂𝑊𝑁�� + 𝛽� 𝑂𝑊𝐶𝑂𝑁�� + 𝛽� 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝛽� 𝑌𝐸𝐴𝑅 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝜀�� where 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 represents board size for the for ith company at time t, measured as the natural logarithm of number of directors on the board; 𝐿𝑁𝑇𝐴�� captures company size for the ith company at time t and it is measured as the natural logarithm of total assets of the company; lncage�� represents the company age for the ith company at time t, which is measured as the natural logarithm of the company’s age from year of establishment; 𝐶𝐸𝑂𝑊𝑁�� is the company ceo’s ownership for ith company at time t, which is the proportion of shares owned by the ceo; 𝑂𝑊𝐶𝑂𝑁�� represents ownership concentration for the ith company at time t, which is the percentage of shares controlled by block holders is at least 10% of total share of the company; 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 represents industry dummies; year represents year dummies; and 𝜀�� is the random error for the ith company at time t. 3.1 empirical results 3.1.1 descriptive statistics and correlation analysis the descriptive statistics of the variables considered in the study are displayed in table 1. based on the figures shown, the average board size is nine, the minimum is four and the maximum is 18. in addition, the average age of companies is 38 years and the maximum is 92 years, which means that a company is as old as 92 years. the average percentage of ceo ownership is 4% and a ceo owns as much as 63.65% of the company shares. there is also a high degree of ownership concentration among nigerian companies. on average, ownership concentration is 53.83% and is as much as 95%. the average total asset of nigerian companies is 29.360 million naira and the maximum value is 950 million naira. in addition, the correlation results shown in table 2 reveal that company age (lncage) and company size (lnta), are statistically significant and positively correlated to board size (lnbsize). 𝐿𝑁𝐵𝑆𝐼𝑍𝐸�� = 𝛽� + 𝛽� 𝐿𝑁𝑇𝐴�� + 𝛽� lncage�� + 𝛽� 𝐶𝐸𝑂𝑊𝑁�� + 𝛽� 𝑂𝑊𝐶𝑂𝑁�� + 𝛽� 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝛽� 𝑌𝐸𝐴𝑅 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝜀�� where 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 represents board size for the for ith company at time t, measured as the natural logarithm of number of directors on the board; 𝐿𝑁𝑇𝐴�� captures company size for the ith company at time t and it is measured as the natural logarithm of total assets of the company; lncage�� represents the company age for the ith company at time t, which is measured as the natural logarithm of the company’s age from year of establishment; 𝐶𝐸𝑂𝑊𝑁�� is the company ceo’s ownership for ith company at time t, which is the proportion of shares owned by the ceo; 𝑂𝑊𝐶𝑂𝑁�� represents ownership concentration for the ith company at time t, which is the percentage of shares controlled by block holders is at least 10% of total share of the company; 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 represents industry dummies; year represents year dummies; and 𝜀�� is the random error for the ith company at time t. 3.1 empirical results 3.1.1 descriptive statistics and correlation analysis the descriptive statistics of the variables considered in the study are displayed in table 1. based on the figures shown, the average board size is nine, the minimum is four and the maximum is 18. in addition, the average age of companies is 38 years and the maximum is 92 years, which means that a company is as old as 92 years. the average percentage of ceo ownership is 4% and a ceo owns as much as 63.65% of the company shares. there is also a high degree of ownership concentration among nigerian companies. on average, ownership concentration is 53.83% and is as much as 95%. the average total asset of nigerian companies is 29.360 million naira and the maximum value is 950 million naira. in addition, the correlation results shown in table 2 reveal that company age (lncage) and company size (lnta), are statistically significant and positively correlated to board size (lnbsize). 𝐿𝑁𝐵𝑆𝐼𝑍𝐸�� = 𝛽� + 𝛽� 𝐿𝑁𝑇𝐴�� + 𝛽� lncage�� + 𝛽� 𝐶𝐸𝑂𝑊𝑁�� + 𝛽� 𝑂𝑊𝐶𝑂𝑁�� + 𝛽� 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝛽� 𝑌𝐸𝐴𝑅 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝜀�� where 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 represents board size for the for ith company at time t, measured as the natural logarithm of number of directors on the board; 𝐿𝑁𝑇𝐴�� captures company size for the ith company at time t and it is measured as the natural logarithm of total assets of the company; lncage�� represents the company age for the ith company at time t, which is measured as the natural logarithm of the company’s age from year of establishment; 𝐶𝐸𝑂𝑊𝑁�� is the company ceo’s ownership for ith company at time t, which is the proportion of shares owned by the ceo; 𝑂𝑊𝐶𝑂𝑁�� represents ownership concentration for the ith company at time t, which is the percentage of shares controlled by block holders is at least 10% of total share of the company; 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 represents industry dummies; year represents year dummies; and 𝜀�� is the random error for the ith company at time t. 3.1 empirical results 3.1.1 descriptive statistics and correlation analysis the descriptive statistics of the variables considered in the study are displayed in table 1. based on the figures shown, the average board size is nine, the minimum is four and the maximum is 18. in addition, the average age of companies is 38 years and the maximum is 92 years, which means that a company is as old as 92 years. the average percentage of ceo ownership is 4% and a ceo owns as much as 63.65% of the company shares. there is also a high degree of ownership concentration among nigerian companies. on average, ownership concentration is 53.83% and is as much as 95%. the average total asset of nigerian companies is 29.360 million naira and the maximum value is 950 million naira. in addition, the correlation results shown in table 2 reveal that company age (lncage) and company size (lnta), are statistically significant and positively correlated to board size (lnbsize). 96 the international journal of banking and finance, vol. 15, no 1, 2020 : 89-103 represents industry dummies; year represents year dummies; and is the random error for the ith company at time t. 3.1 empirical results 3.1.1 descriptive statistics and correlation analysis the descriptive statistics of the variables considered in the study are displayed in table 1. based on the figures shown, the average board size was nine, the minimum was four and the maximum was 18. in addition, the average age of companies was 38 years and the maximum was 92 years, which meant that a company could be as old as 92 years. the average percentage of ceo ownership was 4% and a ceo owned as much as 63.65% of the company shares. there was also a high degree of ownership concentration among nigerian companies. table 1. descriptive statistics variable mean min max skewness kurtosis board size (bsize) 9.08 4.00 18.00 0.37 2.50 lnbsize 2.16 1.39 2.89 -0.17 2.27 company age (cage) 39.54 1.00 92 0.05 2.72 lncage 3.50 0.00 4.52 -1.67 6.54 total assets in millions naira (ta) 29360.44 65.31 950000 6.09 59.84 lnta 15.76 11.12 20.67 -0.08 2.60 ceo ownership (ceown) 4.50 0.00 63.65 2.67 9.95 ownership concentration (owcon) 53.83 0.00 95.00 -0.54 2.44 on average, ownership concentration was 53.83% and was as much as 95%. the average total asset of nigerian companies was 29.360 million naira and the maximum value was 950 million naira. in addition, the correlation results shown in table 2 revealed that company age (lncage) and company size (lnta) were statistically significant and positively correlated to board size (lnbsize). is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. captures the company size 𝐿𝑁𝐵𝑆𝐼𝑍𝐸�� = 𝛽� + 𝛽� 𝐿𝑁𝑇𝐴�� + 𝛽� lncage�� + 𝛽� 𝐶𝐸𝑂𝑊𝑁�� + 𝛽� 𝑂𝑊𝐶𝑂𝑁�� + 𝛽� 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝛽� 𝑌𝐸𝐴𝑅 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝜀�� where 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 represents board size for the for ith company at time t, measured as the natural logarithm of number of directors on the board; 𝐿𝑁𝑇𝐴�� captures company size for the ith company at time t and it is measured as the natural logarithm of total assets of the company; lncage�� represents the company age for the ith company at time t, which is measured as the natural logarithm of the company’s age from year of establishment; 𝐶𝐸𝑂𝑊𝑁�� is the company ceo’s ownership for ith company at time t, which is the proportion of shares owned by the ceo; 𝑂𝑊𝐶𝑂𝑁�� represents ownership concentration for the ith company at time t, which is the percentage of shares controlled by block holders is at least 10% of total share of the company; 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 represents industry dummies; year represents year dummies; and 𝜀�� is the random error for the ith company at time t. 3.1 empirical results 3.1.1 descriptive statistics and correlation analysis the descriptive statistics of the variables considered in the study are displayed in table 1. based on the figures shown, the average board size is nine, the minimum is four and the maximum is 18. in addition, the average age of companies is 38 years and the maximum is 92 years, which means that a company is as old as 92 years. the average percentage of ceo ownership is 4% and a ceo owns as much as 63.65% of the company shares. there is also a high degree of ownership concentration among nigerian companies. on average, ownership concentration is 53.83% and is as much as 95%. the average total asset of nigerian companies is 29.360 million naira and the maximum value is 950 million naira. in addition, the correlation results shown in table 2 reveal that company age (lncage) and company size (lnta), are statistically significant and positively correlated to board size (lnbsize). 𝐿𝑁𝐵𝑆𝐼𝑍𝐸�� = 𝛽� + 𝛽� 𝐿𝑁𝑇𝐴�� + 𝛽� lncage�� + 𝛽� 𝐶𝐸𝑂𝑊𝑁�� + 𝛽� 𝑂𝑊𝐶𝑂𝑁�� + 𝛽� 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝛽� 𝑌𝐸𝐴𝑅 𝐷𝑈𝑀𝑀𝐼𝐸𝑆�� + 𝜀�� where 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 represents board size for the for ith company at time t, measured as the natural logarithm of number of directors on the board; 𝐿𝑁𝑇𝐴�� captures company size for the ith company at time t and it is measured as the natural logarithm of total assets of the company; lncage�� represents the company age for the ith company at time t, which is measured as the natural logarithm of the company’s age from year of establishment; 𝐶𝐸𝑂𝑊𝑁�� is the company ceo’s ownership for ith company at time t, which is the proportion of shares owned by the ceo; 𝑂𝑊𝐶𝑂𝑁�� represents ownership concentration for the ith company at time t, which is the percentage of shares controlled by block holders is at least 10% of total share of the company; 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 represents industry dummies; year represents year dummies; and 𝜀�� is the random error for the ith company at time t. 3.1 empirical results 3.1.1 descriptive statistics and correlation analysis the descriptive statistics of the variables considered in the study are displayed in table 1. based on the figures shown, the average board size is nine, the minimum is four and the maximum is 18. in addition, the average age of companies is 38 years and the maximum is 92 years, which means that a company is as old as 92 years. the average percentage of ceo ownership is 4% and a ceo owns as much as 63.65% of the company shares. there is also a high degree of ownership concentration among nigerian companies. on average, ownership concentration is 53.83% and is as much as 95%. the average total asset of nigerian companies is 29.360 million naira and the maximum value is 950 million naira. in addition, the correlation results shown in table 2 reveal that company age (lncage) and company size (lnta), are statistically significant and positively correlated to board size (lnbsize). table 1: descriptive statistics variables mean min max skewness kurtosis board size (bsize) 9.08 4.00 18.00 0.37 2.50 lnbsize 2.16 1.39 2.89 -0.17 2.27 company age (cage) 39.54 1.00 92 0.05 2.72 lncage 3.50 0.00 4.52 -1.67 6.54 total assets in millions naira (ta) 29360.44 65.31 950000 6.09 59.84 lnta 15.76 11.12 20.67 -0.08 2.60 ceo ownership (ceown) 4.50 0.00 63.65 2.67 9.95 ownership concentration (owcon) 53.83 0.00 95.00 -0.54 2.44 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. 𝐿𝑁𝑇𝐴�� captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. however, ceo ownership (ceown) is statistically significant and negatively correlated to lnbsize. the results in table 3 also provide that the correlation terms among the variables are small, hence indicating no concerns with multicollinearity among variables, further evidenced in the variance inflation factors vifs results presented in column 6 of table 3, where all the vif values reported are below the critical level of 10. table 2: correlation results variables lnbsize lncage lnta ceown owncon lnbsize 1.000 lncage 0.081*** 1.000 lnta 0.531*** 0.157*** 1.000 ceown -0.322*** -0.195*** -0.303*** 1.000 owcon 0.019 -0.020** 0.113*** -0.046 1.000 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. lnta captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. table 1: descriptive statistics variables mean min max skewness kurtosis board size (bsize) 9.08 4.00 18.00 0.37 2.50 lnbsize 2.16 1.39 2.89 -0.17 2.27 company age (cage) 39.54 1.00 92 0.05 2.72 lncage 3.50 0.00 4.52 -1.67 6.54 total assets in millions naira (ta) 29360.44 65.31 950000 6.09 59.84 lnta 15.76 11.12 20.67 -0.08 2.60 ceo ownership (ceown) 4.50 0.00 63.65 2.67 9.95 ownership concentration (owcon) 53.83 0.00 95.00 -0.54 2.44 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. 𝐿𝑁𝑇𝐴�� captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. however, ceo ownership (ceown) is statistically significant and negatively correlated to lnbsize. the results in table 3 also provide that the correlation terms among the variables are small, hence indicating no concerns with multicollinearity among variables, further evidenced in the variance inflation factors vifs results presented in column 6 of table 3, where all the vif values reported are below the critical level of 10. table 2: correlation results variables lnbsize lncage lnta ceown owncon lnbsize 1.000 lncage 0.081*** 1.000 lnta 0.531*** 0.157*** 1.000 ceown -0.322*** -0.195*** -0.303*** 1.000 owcon 0.019 -0.020** 0.113*** -0.046 1.000 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. lnta captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. board size determinants: evidence from nigeria: 89-103 97 and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. represents ownership concentration and it is the percentage of shares controlled by block holders which is at least 10% of the total shares of the company. however, ceo ownership (ceown) was statistically significant and negatively correlated to lnbsize. the results in table 3 also showed that the correlation terms among the variables were small, hence indicating no concerns with multicollinearity among variables, further evidenced in the variance inflation factors (vifs) results presented in column 6 of table 3, where all the vif values reported were below the critical level of 10. table 2. correlation results variable lnbsize lncage lnta ceown owncon lnbsize 1.000 lncage 0.081*** 1.000 lnta 0.531*** 0.157*** 1.000 ceown -0.322*** -0.195*** -0.303*** 1.000 owcon 0.019 -0.020** 0.113*** -0.046 1.000 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. lnta captures the company size and is measured as the natural logarithm of the total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. represents ownership concentration and it is the percentage of shares controlled by block holders which is at least 10% of the total shares of the company. 3.1.2 regression results although panel data regression assisted in controlling for heterogeneity of crosssectional units (hsiao, 2007), the independence of the year-to-year companylevel observations was a concern because board structure was relatively persistent (boone et al., 2007; guest, 2008; hermalin & weisbach, 1988). therefore, the empirical results reported in table 2 were based on the robust white huber standard errors regression technique, which allowed observations to be clustered at company level and controlled for serial correlation and heteroskedasticity table 1: descriptive statistics variables mean min max skewness kurtosis board size (bsize) 9.08 4.00 18.00 0.37 2.50 lnbsize 2.16 1.39 2.89 -0.17 2.27 company age (cage) 39.54 1.00 92 0.05 2.72 lncage 3.50 0.00 4.52 -1.67 6.54 total assets in millions naira (ta) 29360.44 65.31 950000 6.09 59.84 lnta 15.76 11.12 20.67 -0.08 2.60 ceo ownership (ceown) 4.50 0.00 63.65 2.67 9.95 ownership concentration (owcon) 53.83 0.00 95.00 -0.54 2.44 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. 𝐿𝑁𝑇𝐴�� captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. however, ceo ownership (ceown) is statistically significant and negatively correlated to lnbsize. the results in table 3 also provide that the correlation terms among the variables are small, hence indicating no concerns with multicollinearity among variables, further evidenced in the variance inflation factors vifs results presented in column 6 of table 3, where all the vif values reported are below the critical level of 10. table 2: correlation results variables lnbsize lncage lnta ceown owncon lnbsize 1.000 lncage 0.081*** 1.000 lnta 0.531*** 0.157*** 1.000 ceown -0.322*** -0.195*** -0.303*** 1.000 owcon 0.019 -0.020** 0.113*** -0.046 1.000 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. lnta captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. table 1: descriptive statistics variables mean min max skewness kurtosis board size (bsize) 9.08 4.00 18.00 0.37 2.50 lnbsize 2.16 1.39 2.89 -0.17 2.27 company age (cage) 39.54 1.00 92 0.05 2.72 lncage 3.50 0.00 4.52 -1.67 6.54 total assets in millions naira (ta) 29360.44 65.31 950000 6.09 59.84 lnta 15.76 11.12 20.67 -0.08 2.60 ceo ownership (ceown) 4.50 0.00 63.65 2.67 9.95 ownership concentration (owcon) 53.83 0.00 95.00 -0.54 2.44 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. 𝐿𝑁𝑇𝐴�� captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. however, ceo ownership (ceown) is statistically significant and negatively correlated to lnbsize. the results in table 3 also provide that the correlation terms among the variables are small, hence indicating no concerns with multicollinearity among variables, further evidenced in the variance inflation factors vifs results presented in column 6 of table 3, where all the vif values reported are below the critical level of 10. table 2: correlation results variables lnbsize lncage lnta ceown owncon lnbsize 1.000 lncage 0.081*** 1.000 lnta 0.531*** 0.157*** 1.000 ceown -0.322*** -0.195*** -0.303*** 1.000 owcon 0.019 -0.020** 0.113*** -0.046 1.000 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. lnta captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. table 1: descriptive statistics variables mean min max skewness kurtosis board size (bsize) 9.08 4.00 18.00 0.37 2.50 lnbsize 2.16 1.39 2.89 -0.17 2.27 company age (cage) 39.54 1.00 92 0.05 2.72 lncage 3.50 0.00 4.52 -1.67 6.54 total assets in millions naira (ta) 29360.44 65.31 950000 6.09 59.84 lnta 15.76 11.12 20.67 -0.08 2.60 ceo ownership (ceown) 4.50 0.00 63.65 2.67 9.95 ownership concentration (owcon) 53.83 0.00 95.00 -0.54 2.44 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. 𝐿𝑁𝑇𝐴�� captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. however, ceo ownership (ceown) is statistically significant and negatively correlated to lnbsize. the results in table 3 also provide that the correlation terms among the variables are small, hence indicating no concerns with multicollinearity among variables, further evidenced in the variance inflation factors vifs results presented in column 6 of table 3, where all the vif values reported are below the critical level of 10. table 2: correlation results variables lnbsize lncage lnta ceown owncon lnbsize 1.000 lncage 0.081*** 1.000 lnta 0.531*** 0.157*** 1.000 ceown -0.322*** -0.195*** -0.303*** 1.000 owcon 0.019 -0.020** 0.113*** -0.046 1.000 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. lnta captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. 98 the international journal of banking and finance, vol. 15, no 1, 2020 : 89-103 problems that occurred mostly when using panel regression (cameron & miller, 2015). this was especially in situations where the panel regression results chose the fixed-effects model. on this note, the main result discussed in this study is the robust regression results displayed in column 5 of table 3. table 3. regression results variable pooled ols randomeffects model fixed-effects model huber white standard error vif lncage -0.019 -0.012 -0.000 -0.023* 1.06 lnta 0.081*** 0.061*** 0.044*** 0.097*** 1.13 ceown -0.004*** -0.001 -0.000 -0.003*** 1.13 owcon -0.061* 0.111** -0.158** -0.094** 1.02 industry dummies yes agriculture 0.194*** consumer 0.165*** health 0.198*** industrial products 0.138*** natural resources 0.161*** oil and gas 0.199** year dummies yes (2005 & 2006) r-squared 33% 41% breusch and pagan lagrangian multiplier test 0.000 hausman test 0.002 wald test 0.000 board size determinants: evidence from nigeria: 89-103 99 based on the results, a positive association was found between company characteristics and board size. specifically, company size was significantly and positively associated with board size at the 1% significance level. however, ceo ownership and ownership concentration had a significant and negative impact on board size; while ceo ownership had a 1% level of significance and ownership concentration which was significant at the 5% level. these results highlighted that a company’s board size was determined by the company’s characteristics. in addition, board size was found to be dependent on the type of industry that the company belonged to. for instance, a significantly positive influence was found in agriculture, consumer, health, industrial products, natural resources and oil and gas industries. similarly, there was also evidence of year effects in 2005 and 2006, which indicated that possible amendments to corporate governance regulations had impacted the number of directors in the company. is the natural logarithm of the number of directorson a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. lnta captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. represents ownership concentration and it is the percentage of shares controlled by block holders. 4. discussion and conclusion several scholars in the area of corporate governance have claimed that board size is one of the important internal governance mechanisms that have direct influence on board effectiveness. this mechanism reduces conflict of interest between managers saddled with the responsibility of controlling company resources and owners with equity holdings that do not amount to monitoring costs (lipton & lorsch, 1992; pfeffer, 1972). it therefore suggests that the number of directors on a company’s board should be appropriate. corporate governance studies have suggested that company-specific features may be important determinants of board size. thus, this study employed proxies related to scope of operations hypothesis and monitoring cost hypothesis to investigate the determinants of company board size in nigeria. consistent with prior empirical literature (e.g., boone et al., 2007; germain et al., 2014; lehn et al., 2009; linck et al., 2008), the regression results of this study has revealed that company size is positively associated with board size, which simply implies that the complexity of a company’s operations is a significant determinant of a company’s board size. thus, board size grows in response to a company’s complexity. similarly, in line with the prediction of the monitoring cost hypothesis that monitoring costs would have a negative impact on board size, this study has found that proxies table 1: descriptive statistics variables mean min max skewness kurtosis board size (bsize) 9.08 4.00 18.00 0.37 2.50 lnbsize 2.16 1.39 2.89 -0.17 2.27 company age (cage) 39.54 1.00 92 0.05 2.72 lncage 3.50 0.00 4.52 -1.67 6.54 total assets in millions naira (ta) 29360.44 65.31 950000 6.09 59.84 lnta 15.76 11.12 20.67 -0.08 2.60 ceo ownership (ceown) 4.50 0.00 63.65 2.67 9.95 ownership concentration (owcon) 53.83 0.00 95.00 -0.54 2.44 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. 𝐿𝑁𝑇𝐴�� captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. however, ceo ownership (ceown) is statistically significant and negatively correlated to lnbsize. the results in table 3 also provide that the correlation terms among the variables are small, hence indicating no concerns with multicollinearity among variables, further evidenced in the variance inflation factors vifs results presented in column 6 of table 3, where all the vif values reported are below the critical level of 10. table 2: correlation results variables lnbsize lncage lnta ceown owncon lnbsize 1.000 lncage 0.081*** 1.000 lnta 0.531*** 0.157*** 1.000 ceown -0.322*** -0.195*** -0.303*** 1.000 owcon 0.019 -0.020** 0.113*** -0.046 1.000 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. lnta captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. table 1: descriptive statistics variables mean min max skewness kurtosis board size (bsize) 9.08 4.00 18.00 0.37 2.50 lnbsize 2.16 1.39 2.89 -0.17 2.27 company age (cage) 39.54 1.00 92 0.05 2.72 lncage 3.50 0.00 4.52 -1.67 6.54 total assets in millions naira (ta) 29360.44 65.31 950000 6.09 59.84 lnta 15.76 11.12 20.67 -0.08 2.60 ceo ownership (ceown) 4.50 0.00 63.65 2.67 9.95 ownership concentration (owcon) 53.83 0.00 95.00 -0.54 2.44 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. 𝐿𝑁𝑇𝐴�� captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. however, ceo ownership (ceown) is statistically significant and negatively correlated to lnbsize. the results in table 3 also provide that the correlation terms among the variables are small, hence indicating no concerns with multicollinearity among variables, further evidenced in the variance inflation factors vifs results presented in column 6 of table 3, where all the vif values reported are below the critical level of 10. table 2: correlation results variables lnbsize lncage lnta ceown owncon lnbsize 1.000 lncage 0.081*** 1.000 lnta 0.531*** 0.157*** 1.000 ceown -0.322*** -0.195*** -0.303*** 1.000 owcon 0.019 -0.020** 0.113*** -0.046 1.000 𝐿𝑁𝐵𝑆𝐼𝑍𝐸 is the natural logarithm of the number of directors on a company’s board. lncage is measured as the natural logarithm of the company’s age from year of establishment. lnta captures the company size and is measured as the natural logarithm of total assets of the company. ceown is the company ceo’s ownership, which is the proportion of shares owned by the ceo. 𝑂𝑊𝐶𝑂𝑁 represents ownership concentration and it is the percentage of shares controlled by block holders is at least 10% of total share of the company. 100 the international journal of banking and finance, vol. 15, no 1, 2020 : 89-103 for monitoring costs, such as ceo ownership and ownership concentration, are negatively associated with board size. this implies that board structure is based on the monitoring requirements of the company’s business activities and increases in private benefits. all these findings are consistent with assertions from prior studies (boone et al., 2007; coles et al., 2008; cicero et al., 2013; linck et al., 2008) that board size is a function of monitoring costs combined with increase in benefits. in short, the findings provide at least some support for the theoretical prediction of the scope of operations hypothesis and the monitoring cost hypothesis in explaining board size. hence, it is recommended that studies examining the influence of board size on corporate outcomes could consider the interacting effects of company characteristics before regressing it against corporate outcomes, like performance, value and financial reporting quality. in addition, a number of proxies have been proposed by previous scholars for scope of operations and monitoring costs hypotheses. therefore, future studies can apply leverage and diversity in the context of scope of operations hypothesis and growth opportunities, such as market-to-book ratio and r&d expenditure in the context of the monitoring cost hypothesis. this would allow for a comprehensive conclusion on board size determinants in nigeria. the overall implication of these results is that regulators and managers should ensure that corporate boards commensurate with the size of the company, as this would enable the boards to act efficiently and to avoid boards which are persistently too large. end note 1 some of these functions include giving expert advice and guidance to the ceo and managers in strategy formulation and implementation. other functions include monitoring and disciplining of ineffective management teams, and providing access to critical information and resources that are needed for company survival (adams & ferreira, 2007; fama & jensen, 1983; hillman & dalziel, 2003). it is also considered as an instrument for dealing with external interdependence and uncertainty caused by its exchange of resources with the external environment (lynall, golden, & hillman, 2003; pfeffer, 1972). references adams, r. b., & ferreira, d. (2007). a theory of friendly boards. journal of finance, 62(1), 217-250. adegbite, e., amaeshi, k., & nakajima, c. (2013). multiple influences on corporate governance practice in nigeria: agents, strategies and implications. international business review, 22(3), 524-538. board size determinants: evidence from nigeria: 89-103 101 agrawal, a., & knoeber, c. r. (1996). firm performance and mechanisms to control agency problems between managers and shareholders. journal of financial and quantitative analysis, 31(3), 377-397. ahmed sheikh, n., wang, z., & khan, s. (2013). the impact of internal attributes of corporate governance on firm performance: evidence from pakistan. international journal of commerce and management, 23(1), 38-55. bhagat, s., & black, b. (1999). the uncertain relationship between board composition and firm performance. the business lawyer, 921-963. booth, j. r., & deli, d. n. (1996). factors affecting the number of outside directorships held by ceos. journal of financial economics, 40(1), 81104. boone, a. l., field, l. c., karpoff, j. m., & raheja, c. g. (2007). the determinants of corporate board size and composition: an empirical analysis. journal of financial economics, 85(1), 66-101. cameron, a. c., & miller, d. l. (2015). a practitioner’s guide to cluster-robust inference. journal of human resources, 50(2), 317-372. carter, d. a., d’souza, f., simkins, b. j., & simpson, w. g. (2010). the gender and ethnic diversity of us boards and board committees and firm financial performance. corporate governance: an international review, 18(5), 396-414. cheng, s. (2008). board size and the variability of corporate performance. journal of financial economics, 87(1), 157-176. cicero, d., wintoki, m. b., & yang, t. (2013). how do public companies adjust their board structures?. journal of corporate finance, 23, 108-127. coles, j. l., daniel, n. d., & naveen, l. (2008). boards: does one size fit all? journal of financial economics, 87(2), 329-356. dalton, d. r., daily, c. m., johnson, j. l., & ellstrand, a. e. (1999). number of directors and financial performance: a meta-analysis. academy of management journal, 42(6), 674-686. denis, d. j., & sarin, a. (1999). ownership and board structures in publicly traded corporations. journal of financial economics, 52(2), 187-223. fama, e. f., & jensen, m. c. (1983). separation of ownership and control. journal of law and economics, 26(2), 301-325. firstenberg, p. b., & malkiel, b. g. (1994). the twenty-first century boardroom: who will be in charge? mit sloan management review, 36(1), 27. gaver, j. j., & gaver, k. m. (1993). additional evidence on the association between the investment opportunity set and corporate financing, dividend, and compensation policies. journal of accounting and economics, 16(1-3), 125-160. germain, l., galy, n., & lee, w. (2014). corporate governance reform in malaysia: board size, independence and monitoring. journal of economics and business, 75, 126-162. 102 the international journal of banking and finance, vol. 15, no 1, 2020 : 89-103 golden, b. r., & zajac, e. j. (2001). when will boards influence strategy? inclination× power= strategic change. strategic management journal, 22(12), 1087-1111. goodstein, j., gautam, k., & boeker, w. (1994). the effects of board size and diversity on strategic change. strategic management journal, 15(3), 241-250. guest, p. m. (2009). the impact of board size on firm performance: evidence from the uk. the european journal of finance, 15(4), 385-404. guest, p. m. (2008). the determinants of board size and composition: evidence from the uk. journal of corporate finance, 14(1), 51-72. harris, m., & raviv, a. (2006). a theory of board control and size. the review of financial studies, 21(4), 1797-1832. hermalin, b. e., & weisbach, m. s. (1988). the determinants of board composition. the rand journal of economics, 19(4), 589-606. hillman, a. j., & dalziel, t. (2003). boards of directors and firm performance: integrating agency and resource dependence perspectives. academy of management review, 28(3), 383-396. hsiao, c. (2007). panel data analysis—advantages and challenges. test, 16(1), 1-22. jensen, m. c. (1993). the modern industrial revolution, exit, and the failure of internal control systems. journal of finance, 48(3), 831-880. kumar, n., & singh, j. p. (2013). effect of board size and promoter ownership on firm value: some empirical findings from india. corporate governance: the international journal of business in society, 13(1), 88-98. kyereboah-coleman, a., & biekpe, n. (2007). on the determinants of board size and its composition: additional evidence from ghana. journal of accounting & organizational change, 3(1), 6877. lehn, k. m., patro, s., & zhao, m. (2009). determinants of the size and composition of us corporate boards: 1935–2000. financial management, 38(4), 747-780. linck, j. s., netter, j. m., & yang, t. (2008). the determinants of board structure. journal of financial economics, 87(2), 308-328. lipton, m., & lorsch, j. w. (1992). a modest proposal for improved corporate governance. the business lawyer, 48(1), 59-77. lynall, m. d., golden, b. r., & hillman, a. j. (2003). board composition from adolescence to maturity: a multitheoretic view. academy of management review, 28(3), 416-431. min, b. s. (2018). determinants of board size in an emerging market. journal of international financial management & accounting, 29(1), 5-29. musteen, m., datta, d. k., & kemmerer, b. (2010). corporate reputation: do board characteristics matter? british journal of management, 21(2), 498510. board size determinants: evidence from nigeria: 89-103 103 nguyen, p., rahman, n., tong, a., & zhao, r. (2016). board size and firm value: evidence from australia. journal of management & governance, 20(4), 851-873. ning, y., davidson iii, w. n., & zhong, k. (2007). the variability of board size determinants: an empirical analysis. journal of applied finance, 17(2), 46-61. o’connell, v., & cramer, n. (2010). the relationship between firm performance and board characteristics in ireland. european management journal, 28(5), 387-399. pearce, j. a., & zahra, s. a. (1992). board composition from a strategic contingency perspective. journal of management studies, 29(4), 411438. pfeffer, j. (1972). size and composition of corporate boards of directors: the organization and its environment. administrative science quarterly, 17(2), 218-228. raheja, c. g. (2005). determinants of board size and composition: a theory of corporate boards. journal of financial and quantitative analysis, 40(2), 283-306. smith jr, c. w., & watts, r. l. (1992). the investment opportunity set and corporate financing, dividend, and compensation policies. journal of financial economics, 32(3), 263-292. uwuigbe, o. r., & fakile, a. s. (2012). the effects of board size on financial performance of banks: a study of listed banks in nigeria. international journal of economics and finance, 4(2), 260-267. yermack, d. (1996). higher market valuation of companies with a small board of directors. journal of financial economics, 40(2), 185-211. zahra, s. a., & pearce, j. a. (1989). boards of directors and corporate financial performance: a review and integrative model. journal of management, 15(2), 291-334. 23 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: ttaffi al-, l. h. a., abdul-jabar, h., & bin-nashwan, s. a. (2021). the role of income tax system structure in tax non-compliance behaviour among smes in yemen. international journal of banking and finance, 16(2), 23–49. https://doi. org/10.32890/ijbf2021.16.2.2 the role of income tax system structure in tax non-compliance behaviour among smes in yemen 1lutfi hassen ali al-ttaffi, 2hijattulah abdul-jabbar & 3saeed awadh bin-nashwan 1college of administrative science seiyun university, yemen 2tunku puteri intan safinaz school of accountancy universiti utara malaysia 3islamic business school, universiti utara malaysia 1corresponding author: alttaffilutfi@yahoo.com received: 2/11/2020 revised: 6/2/2021 accepted: 6/2/2021 published: 15/6/2021 abstract tax is the main source of government revenue. however, a number of countries worldwide are increasingly besieged by challenges regarding compliance levels with the rules of tax systems. thus, this paper aims to enhance an understanding of tax non-compliance behaviour by investigating the effect of the income tax system structure on yemeni taxpayers’ behaviour. the study focuses on income tax compliance behaviour of owner-managers of small and medium enterprises (smes), as the yemeni economy relies heavily on this sector. the sme sector represents 99.6 percent of business in yemen. 24 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 based on a quantitative approach using a self-administered survey instrument, a total of 330 valid questionnaires were collected and the feedback provided analyzed. the results demonstrate that sme taxpayers exhibited a high level of tax non-compliance. furthermore, the multiple regression analysis shows that the tax rate had a positive and significant influence on tax non-compliance behaviour, but the tax penalties rate did not. these results can be especially relevant to policymakers and practitioners of tax systems structures, particularly in a developing country such as yemen. keywords: tax non-compliance behaviour, tax system structure, smes, yemen. jel classification: m410. introduction tax non-compliance has become a serious impediment that affects the global economy and there has been widespread growing interest among tax researchers (ross & mcgee, 2012). all economies in the world, whether developed or developing economies were affected by this phenomenon (hindriks et al., 2008). mas’ud et al. (2014) argued that levels of tax non-compliance in developing communities were much greater than developed ones. previous research has vigorously investigated the issue of tax non-compliance, covering various countries in europe, asia, north america, south america, new zealand and australia (gaventa & mcgee, 2010; mcgee et al., 2009a; mcgee et al., 2009b; mcgee, 2007; mcgee, 2006). although the middle east region was certainly not immune to such a perplexing challenge of tax non-compliance (central organization for control and audit coca, 2018) – the middle east has become a vital region that contributes significantly to the growth of the global economy (carapico, 1998) – studies into this issue were very rare (aljaaidi et al., 2011). thus, the present research has sought to scrutinize tax noncompliance in yemen, one of the struggling economies of the middle east. yemen was considered as among the least developed countries that has long been plagued by high levels of administrative, economic and financial corruption. one of its most prominent financial and 25 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 economic barriers was the issue of tax non-compliance. the country recorded heavy losses over the last few decades (coca, 2018). despite numerous attempts made by the tax authorities to reform the tax system and mitigate tax non-compliance levels, the puzzle of tax non-compliance remains unresolved. over the 2004-2014 period, it has been observed that tax non-compliance had a steady increase, exceeding usd164 million in 2004, and increased to usd2 billion in 2009. in 2012, it reached usd2.5 billion, usd3 billion in 2013, and usd 4 billion in 2014 (mpic, 2009; coca, 2018). al-saadi, who was the minister of planning and international cooperation of yemen, (2014) stated that the issue of low collection of tax in yemen led to a serious financial problem. furthermore, in terms of the gdp, when comparing yemen with other countries, it seems that the average tax revenue proportion of yemen amounted to only 6.68 percent to gdp for the period between 2008 and 2013, compared to 14.3 percent and 16 percent of gdp in countries that have similar economic conditions, such as egypt and jordan respectively, and greater than 20 percent of their gdps as tax in morocco, tunisia and lebanon (world bank, 2018). overall, the low contribution of tax revenue in gdp might largely be due to the low tax compliance behaviour among yemeni taxpayers. the low compliance was a result of the fact that tax authorities did not adequately investigate the determinants of taxpayer’s non-compliance behaviours (obaid et al., 2020). from the 1960s until to date, small and medium enterprises (smes) have been recognized as playing a significant role in promoting rapid economic growth, development and stability for many economies (pashev, 2008). in yemen, the economy has been relying heavily on the smes sector; for the smes represented 99.6 percent of business in the country1 (mpic, 2009; central organization of statistic cos, 2019). furthermore, hanefah et al. (2002) has pointed out that several governments see the smes as the vehicle for greater economic growth. mwangi (2014) and fararah (2014) suggested concentrating on smes because this sector often constituted a very large segment of the economy of countries. in many developing countries, the phenomenon of tax non-compliance was attributed to many reasons, the most important of which was the tax system structure (twum, 2014). therefore, the present study was an attempt to explore tax non-compliance among smes in yemen and 26 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 the influence of the tax system structure on the behaviour of taxpayers. this study could contribute to the extant literature, particularly for the middle east region as this issue has rarely been examined. literature review tax non-compliance although tax revenue was deemed to be one of the main financial sources of yemen’s national economy, the reported problem of tax non-compliance has been very rarely investigated (bin-nashwan et al., 2020). the dearth of research has not helped to provide the possible explanations for why yemenis did not comply with tax laws, as well as investigated the determinants shaping their tax-paying behaviour. a descriptive study by al-doais (2008) on the role of the accounting information system (ais) in mitigating tax evasion has failed to provide a logical explanation for the real reasons behind tax non-compliance in yemen. aljaaidi et al. (2011) explored yemenis’ perceptions of tax evasion, specifically to determine whether they considered it a crime or not. the researchers revealed that the people perceived tax evasion as not a crime. meanwhile, gubran (2009) conducted an analytical research of tax non-compliance reality in yemen. he suggested that the problem of the government being unfair was the key reason of non-compliance among taxpayers. this has recently been supported in studies by helhel and ahmed (2014) and bin-nashwan et al. (2020) who argued that tax non-compliance could be traced to a wide range of determinants, such as the structure of the tax system, efficiency of tax administration, economic and political situation, and individual characteristics of taxpayers. however, in the study by helhel and ahmed’s (2014) the scope was very limited, as data was obtained only from one city of yemen (sana’a). nevertheless the study argued that the general public’s weak allegiance towards the yemeni regime was somewhat similar nationwide, so was the prevalent negative attitude towards the government. while the problem of tax non-compliance has been an issue in the middle east region, al-ttaffi and abduljabbar (2018) suggested that the problem of tax non-compliance has not been explored sufficiently. this was also despite the fact that the 27 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 region was one of the fastest-growing markets and has a vital role to play in the global economy. tax system structure several researchers, such as richardson (2013), borrego et al. (2013), palil and mustapha (2011), and palil (2010), have indicated that the significance of fischer et al.’s (1992) model in explaining compliance behaviour of taxpayers. mas’ud et al. (2014) also contended that fischer’s model could provide a good insight into the role of penalties and tax rate in taxpayers’ non-compliance decisions under the tax system structures. in developing countries, taxpayers might decide to evade tax after assessing two things: the tax rate offered and the level of tax penalties (umar et al., 2012). therefore, the present research sought to examine the effect of the tax rate and penalties on tax noncompliance in a developing country such as yemen. tax rate the influence of the tax rate on tax non-compliance has attracted the attention of tax researchers. this has led to many studies conducted to discover any tangible evidence of the impact of tax rates on noncompliance. many studies have found that tax non-compliance was positively related to the tax rate, whereas others did not find any relationship between the tax rate and tax non-compliance. in general, it could however, be assumed that the higher the tax rate, the more prevalent non-compliance behaviour among taxpayers (mwangi, 2014). several researchers have argued that when tax rates were high, it might lead to many taxpayers evading taxes (hai & see, 2011). spicer and becker (1980) indicated that the behaviour of non-compliance was reported to be at high levels among those taxpayers who became aware that the payable tax rates were greater than the average paid by others. rationally, taxpayers might perceive that the tax rate that they had to pay had outweighed the expected overpayment through noncompliance. recent studies had confirmed this association between the tax rate and tax non-compliance. for example, nor azrina et al. (2014) had found that the relationship was positively significant. however, other studies concluded that the high tax rate did not 28 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 necessarily increase tax non-compliance (porcano, 1988; trivedi et al., 2004; mcgee & an, 2007; modugu et al., 2012). in general, it seems from the literature review that some studies have claimed that the relationship between the tax rate and tax non-compliance was positive, while other studies argued that the relationship was negative. furthermore, some others have suggested that the relationship was neither positive nor negative. one study has reinforced this conclusion of conflicting findings (richardson, 2006). serén and panadés (2013) have recommended that since the literature on the effect of the tax rate and tax non-compliance was not definitive due to the mixed findings by various studies, the issue still requires further investigation. thus, the present study was one such attempt to contribute to the growing body of knowledge in the issue at hand and has sough to investigate the relationship between the tax rate and tax non-compliance behaviour in a developing country. tax penalties according to the deterrence theory proposed by becker (1968), because of the risk of being punished, people would not commit crimes. therefore, to resolve the tax non-compliance issue, there must be put in place an effective punishment structure. sanders et al. (2008) argued that deterrence theory has provided a logical explanation of the relationship between penalties and tax non-compliance. in the issue of tax non-compliance, when it is expected that taxpayers are able to evade tax, tax authorities must resort to stricter punishments to help ensure better levels of compliance with tax rules, as well as to mitigate non-compliance. therefore, non-compliance behaviour has let to counter measures such as punishment, audit, and the probability of detection (cherry, 2001). blank’s (2014) study indicated that tax penalties were basically a response to the different levels of tax noncompliance. he further provided explanations of tax penalties by indicating that penalties were divided into two types, namelymonetary tax penalties and collateral tax penalties. monetary tax penalties were civil penalties, which required taxpayers to pay additional money to the tax authority. governments also sometimes threatened taxpayers with criminal tax penalties, but they were rarely imposed. generally, monetary tax penalties were either percentage tax penalties, which consisted of a portion of the taxpayers’ underpayment of tax, or flat 29 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 tax penalties, which consisted of a stated amount that taxpayers had to pay for every instance of a particular offence. the other type of penalty was collateral tax penalties, which was defined as additional penalties that occurred outside of the tax system. this penalty imposed on citizens who failed to pay tax, deprived them from enjoying any benefits provided by the government. according to witte and woodbury (1985) and twum (2014), penalties could play a negative and significant role in tax non-compliance behaviours. sanders et al. (2008) examined the influence of sanctions on tax non-compliance and found it to be negative and significant. when there were stricter tax penalties imposed by tax authorities, taxpayers would be motivated to well comply with regulations and laws (doran, 2009). however, some previous studies had documented a positive relationship between punishments and non-compliance (crane & nourzad, 1986). azrina et al. (2014) argued that the risk of getting punished did not affect the behaviours of taxpayers to comply with paying a tax. in short, they might even decide to evade tax when there was the possibility of potential punishment. other researches had reported no effect of penalties on non-compliance decisions of taxpayers (kuria et al., 2013; kamdar, 1997; pommerehne & weckhannemann, 1996). blank (2014) described collateral tax sanctions as a choice that could foster voluntary compliance of taxpayers more effectively, rather than the risk of monetary sanctions. added to the above studies, morris (2010) argued that if penalties were not an effective way to deter non-compliance, then other option had to be proposed. research model and hypotheses the present work was underpinned by the deterrence theory of becker (1968). it was employed to explain the relationships between the constructs under study. thus, the hypotheses in this studyhave been developed on the basis of becker’smain theoretical framework, as well as the related literature reviewed earlier. furthermore, previous studies have suggested that public governance, service quality, and system structure have much to offer in understanding taxpayers noncompliance behaviour, especially in developing communities such as those found in yemen (egwaikhide, 2010; manaf et al., 2005; al30 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 ttaffi, 2009). therefore, these variables have been incorporated in the model of the present study and is expressed in the equation below: tax non-compliance behaviour = β0 β1 tax penalties + β2 tax rate – β3 tax service quality – β4 public governance quality + e the present study is a pioneer one as it was aimed at studying the role of the tax system structure in taxpayers’ decisions towards noncompliance behaviour. this is especially important in the context of the middle east where empirical studies on such a focus are relatively scarce. thus, in line with the literature reviewed earlier, it was assumed that taxpayers’ non-compliance behaviour would be low when tax penalties were more strict, as well as when tax rates were favourable.the hypotheses in the study were as follows: h1: tax non-compliance behaviour among smes is positively related to the tax rate. h2: tax non-compliance behaviour among smes is negatively related to penalties. methodology data collection procedures a survey questionnaire was employed in the present study because it was considered the most appropriate research design and method to collect primary data to obtain beliefs, personal and social facts, and attitude (kerlinger & lee, 2000). the population of the current study was the small and medium enterprises (smes) in yemen. the most updated number of smes in yemen was recorded as totalling 45,483 (national information system nis, 2018; ministry of industry & trade mit, 2018; cos, 2019). a minimum sample size of 381 was required for a population of 50,000, according krejeie and morgan (1970). however, israel (2009) argued that in order to obtain adequate data, the researchers should increase the sample size to compensate for likely non-response. particularly, israel (2009) recommended that researchers ought to increase the sample size by at least 30 percent to compensate for likely non-response. consequently, the sample size of the current study was increased to 500 to overcome the probable non-response rate. 31 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 for the purpose of representativeness, stratified sampling was utilised because a stratified sample was a more efficient sample that could be taken on the basis of simple random sampling. another reason for taking a stratified sample was the assurance that the sample accurately reflected the population based on the criterion or criteria used for stratification (zikmund, 2003). according to mit (2018), 78.4 percent of smes in yemen were located in seven cities, namely sana’a, aden, taiz, mukalla, seiyun, ibb and hodeidah. therefore, the data was collected only from these cities, while 20.6 percent of yemeni smes were distributed among nearly 20 cities. accordingly, the remaining cities were ignored as there was no detailed information available for each city. the period of data collection for each single city was between two to three weeks. however, the overall period of collecting datwas two months specificcaly april and may 2016. (april & may 2016). efforts made by the research assistants to distribute 500 questionnaires yielded a total of 339 returned questionnaires, obtaining a response rate of 68 percent. although the data of the current study were collected during the period of crisis in yemen, the response rate was considered high compared to previous studies in yemen (e.g., 57% in al-ttaffi et al., 2011; 53% in helhel & ahmed, 2014). this high response rate may be attributed to the fact that the researcher had used seven assistants with at least two follow-ups, which could have motivated the respondents to complete and return the questionnaires. a total of 11 questionnaires were discarded because they were not completed by the targeted owner or manager of smes; instead, they were completed by some of the employees or workers. in addition, two questionnaires had to be excluded because more than 50 percent of their questions had not been completed by the respondents (hair et al., 2010). besides, four questionnaires were discarded based on the outlier result. thus, 330 questionnaires were considered usable for analysis at a response rate of 66 percent. according to sekaran (2003), a response rate of 30 percent was sufficient for survey analysis. thus, this study has met this criterion. measurement of variables for the purpose of this study, taxpayers’ non-compliance was measured by an indirect hypothetical scenario (see appendix). this 32 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 measurement was utilized in the studies of falsetta, (2020), pham et al. (2020), al-ttaffi et al. (2020), al-ttaffi and abdul-jabbar (2020) and al-ttaffi and abdul-jabbar (2018). in this measurement, the participants were required to read the scenario, and then, answer four questions related to the components of tax non-compliance. these questions gave respondents the opportunity to express their view on (1) reporting non-compliance, which means not providing inaccurate information on the taxable income, (2) deduction noncompliance, which means claiming overstatement of deduction, (3) filing non-compliance, which means failure to submit a tax return on time voluntarily, and (4) payment non-compliance, which means non-compliance to pay taxes on time. the components of tax noncompliance were adopted from long and swingen (1991) and brown and mazur (2003). three options were given initially to respondents for each scenario, interpreted as compliance (coded 1), partial noncompliance (coded 2) and fully non-compliance (coded 3). the value of 1 was interpreted as compliance, meanwhile, the values of 2 and 3 were transformed and interpreted as non-compliance, which meant that the final responses for each scenario was categorized into compliance and non-compliance. the overall tax non-compliance was calculated by the overall mean values for the combined four scenarios. if the mean was equal to 1, it was interpreted as fully compliant, whereas if the mean was equal to 2, it was interpreted as fully non-compliant, while if the mean was between 1 and 2 it was interpreted as partly compliant. the tax rate and penalties are the variables under a tax system structure. the tax rate was measured with five items. respondents were asked to identify the degree of agreement or disagreement with questions regarding the tax rate. the mean score was used for the analysis. a high score of “5” (strongly agree) suggested that the tax rate was perceived to be fair, while a low score of “1” (strongly disagree) meant that the tax was less fair. the measurements of the five items were combined to obtain a total scale of the tax rate. a mean score of 3 and higher showed a strong belief of tax rate fairness, while a mean score lower than 3 was a signal of low perception about the fairness of the tax rate. penalties were measured with four items to show the level atd which taxpayers agreed or disagreed with the questions regarding penalties. a high score of “5” (strongly agree) was an indication that the penalties had a high impact on tax non-compliance, while a low score of “1” (strongly disagree) was described as a low level effect 33 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 of tax penalties. the measurement of the four items were combined to obtain a total scale of the penalties. a mean score of 3 and higher indicated a high perception of penalty, and a mean score lower than 3 was a signal of low perception about penalties. similarly, tax service quality and public governance quality were measured by the 5-point likert scale as well. tax service quality was measured using 18 items. respondents were asked to indicate their disagreement or agreement to statements relating to the quality of tax service. taken together, the instrument comprising the 18 items was a measure of the unit of analysis – the tax service quality. a high score indicated a high perception of the tax service quality, while a low score was an indication of the poor perception of the tax service quality. public governance quality was measured with 17 items. the high score of “5” (strongly agree) indicated that the respondents believed that the quality of public governance was high, whereas a low score of “1” (strongly disagree) meant low quality. research findings respondent profile the descriptive analysis of the respondent s’ profile showed that more than half (52%) of the sample was aged 30-50 years old, with 38 percent aged more than 50 years and the remaining 10 percent aged less than 30 years old. a large majority of the participants (94%) were male and married (85%). in terms of education, 44 percent of the sme owners and managers held a secondary school certificate and diploma, while 34 percent of the respondents were holders of a bachelor’s degree, and the remaining 22 percent of them had up to primary school education only. next, from the selected sample of smes, about 86 percent had a business experience of five or more years. as for the business sector, retail trade represented 42 percent of the sample, followed by 28 percent in wholesale trade, 16 percent in each of the following sectors in livestock, industry, farming and others, and service sector represented 13 percent. tax non-compliance behaviour tax non-compliance was computed in four situations, as follows: reporting non-compliance, deduction non-compliance, filing non34 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 compliance and payment non-compliance. also, tax non-compliance was described through its overall value. table 1 presents the level of the four situations of tax compliance and tax non-compliance of the respondents. table 1 situations of tax compliance and non-compliance behaviour tax non-compliance situations compliance non-compliance n % n % reporting non-compliance 175 53% 155 47% deduction non-compliance 175 53% 155 47% filing non-compliance 108 33% 222 67% payment non-compliance 108 33% 222 67% table 1 shows that 53 percent of the respondents complied by reporting their taxable income and in claiming the right deduction, while 47 percent of the sample of smes did not. concerning the level of filing and payment compliance, it represented 33 percent of the respondents, whilst 67 percent of the respondents did not comply with filing and paying their due tax. with regard to the general levels of tax non-compliance, three categories were employed to assess the perceptions of the participants: full compliance, partial compliance and full non-compliance. a respondent who can comply with the four components (see table 1) was considered as a compliant respondent, while a fully noncompliant respondent was the one who could not fulfil all the required four components, and partly compliant taxpayers were those who complied with some components and failed to comply with others. the results of the overall levels of non-compliance and compliance are as shown in table 2. from table 2, it can be seen that the three levels of compliance and non-compliance indicated that 47 percent of the smes were fully noncompliant, and 20 percent were partly compliant, while 33 percent were fully compliant. thus, it can be concluded that the smes in 35 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 yemen showed a high degree of tax non-compliance. generally, the results of table1 and table 2 supported the findings in previous studies and suggested that the issue of tax non-compliance was still a critical problem in yemen. table 2 results of overall tax compliance and non-compliance compliance non-compliance partial compliance full noncompliance n % n % n % overall tax compliance and non-compliance 108 33% 67 20% 155 47% from table 2, it can be seen that the three levels of compliance and non-compliance indicated that 47 percent of the smes were fully noncompliant, and 20 percent were partly compliant, while 33 percent were fully compliant. thus, it can be concluded that the smes in yemen showed a high degree of tax non-compliance. generally, the results of table1 and table 2 supported the findings in previous studies and suggested that the issue of tax non-compliance was still a critical problem in yemen. tax system structure the tax system structure was explained in the current study through two variables, namely the tax rate and penalties. the next subsections will present the descriptive statistics of these two variables. the tax rate, as one of the variables of the tax system structure in yemen, was measured through the use of five items (tss1 to tss5). table 3 shows a summary of the mean values, standard deviation values, minimum values and maximum values of the five-likert scale for the items. table 3 shows that the mean of the items was at 2.34 as minimum mean value, with a standard deviation of 1.031 and 2.59 as maximum value, with standard deviation 0.967. regarding the overall value of 36 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 the tax rate, the result was 2.44. as the values were interpreted as lower than 3.0, this indicated that the respondents believed that the tax rate structure was not fair. table 3 descriptive statistics of the tax rate codes items mean stdd min. max. tss1 tss2 tss3 tss4 tss5 the tax rate on higher-income should be more than the tax rate on lower-income. the high-income taxpayer should pay proportionately more than a low-income tax. the tax rate should not be the same for all taxpayers. it is fair to be tax noncompliant if the tax rates are too high. it is unfair to be tax noncompliant if the tax rate is fair. 2.34 2.39 2.41 2.59 2.53 1.031 1.041 1.169 0.967 0.986 1 1 1 1 1 5 5 5 5 5 overall 2.44 0.951 penalties were the second dimension among the components of the tax system structure, which have been used in the current study. data of this variable were collected through four items (tss6 to tss9). table 4 shows a summary of the mean values, standard deviation values, minimum values and maximum values of the five-likert scale for the items. table 4 reveals that the mean of the items was 2.41 as the minimum mean value, with a standard deviation of 1.124 and 2.60 as the maximum value, with standard deviation 1.006, and the overall value of the tax rate was 2.47. this result would seem to suggest that the respondents believed that penalties were less strict on tax noncompliance behaviour in yemen because the mean values were lower than 3.0. 37 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 table 4 descriptive statistics of penalties codes items mean stdd min. max. tss6 tss7 tss8 tss9 i think the person who is caught for tax noncompliance will be forced to pay the tax he owes with interest. i think that the taxpayer who is caught for tax non compliance will be forced to pay a substantial fine and pay the tax he owes with interest. i think the person who is caught for tax non compliance will be taken to court and pay the tax he owes with interest. i think the person who is caught for tax non compliance will be taken to court, pay a substantial fine and pay the tax he owes with interest. 2.41 2.49 2.60 2.45 1.124 1.095 1.006 1.127 1 1 1 1 5 5 5 5 overall 2.47 1.011 multiple regression results to test the hypothesized relationship between the tax system structure and taxpayers’ non-compliance decision, a multiple regression test was performed. the results of the multiple regression analysis is as displayed in table 5. the analysis revealed that the model was statistically significant as explained by the f values of 5.981 (p=0.000). this result meant that the proposed model could provide an understanding of the important determinants influencing tax non-compliance among yemeni smes. moreover, the analysis revealed that the adjusted r2 was 0.069, which meant that the tax system structure could explain on 6.9 percent of the 38 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 table 5 multiple regression results dependent variable: tax non-compliance variables coefficients sig. tax rate 0.210*** 0.009 tax penalties 0.040 0.505 tax service quality 0.009 0.906 public governance quality -0.150*** 0.006 r2 0.069 f-value 5.981*** adjusted r2 0.057 no of observations 330 note. *** indicate significance at 1% confidence level. tax compliance behaviour significantly. on the weight of the contribution of each independent variable, the regression pointed out that the tax rate (β = 0.210; p=0.009) was positively related to the tax non-compliance behaviour. meanwhile, the relationship between penalties with tax non-compliance was not significant (β = 0.040; p=0.505). also, the regression showed that public governance quality negatively affected the behaviour of taxpayers, whereas service quality had no significant influence on the behaviour of taxpayers among the smes in yemen. discussion of the results the descriptive statistics of the current study revealed that taxpayers in yemen believed that the tax rate structure was moderately not fair. the multiple regression results disclosed that the association between the tax rate and taxpayers’ non-compliance was positively significant. this result could be seen as consistent with postulates of the theory of deterrence, and the hypothesis of the current study. alm (1999) explained that it was natural human behavior to always act rationally by comparing the costs and benefits of any activity. hamm (1995) 39 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 suggested that according to the deterrence theory, a high tax rate could lead to a decrease in tax non-compliance. mwangi (2014) described the puzzle of tax non-compliance could be linked to a high degree of the tax rate. in the literature review carried out in this study, most of the earlier empirical studies were found to support the results of the present study, such as the studies by clotfelter (1983), feinstein (1991), christian et al. (1993), joulfaian and rider (1998), ali et al. (2001), lin and yang (2001), mcgee and ho (2006), mcgee and lingle (2006), and mcgee and rossi (2006). hai and see (2011) found that the high tax rate led to high tax non-compliance, and park and hyun (2003) found that increasing tax rates encouraged non-compliance behaviour. spicer and becker (1980) explained that taxpayers, who were aware that their tax rate was higher than the average recorded a high level of tax non-compliance. from a multi-national perspective, derwent (2000) examined behaviours of taxpayers from five countries, i.e., gambia, kenya, nigeria, south africa and usa, and found that the tax rate was a serious challenge facing tax authorities, in terms of its effect on tax compliance. arguably, several recent studies have confirmed the positive association between the tax rate and tax non-compliance, such as those by ser (2013), nor azrina et al. (2014) and levin and widell (2014). the current study has shown support for findings in previous literature and provided robust empirical evidence that the high tax rate significantly influenced the behaviour of taxpayers and could lead to an increased level of tax non-compliance. in the context of the second independent variable of this study, the descriptive statistics on the data of the current study revealed that the respondents believed that penalties were moderately strict on tax non-compliance behaviour in yemen. although the deterrence theory of becker (1968) has proposed that the fear of punishment would cause people not to commit crimes, the multiple regression results in the current study seemed to suggest that the relationship between tax penalties and tax non-compliance behaviour was not significant. this result might be explained by the ongoing war and the political instability in the country, which led to the weakness of governmental power and reduced accountability (al-rabaee, 2014). as a result, the government was not in a position to enforce rules and as a consequence, the penalties did not affect the behaviour of taxpayers in such troubled times. more specifically, the literature review has shown that the 40 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 results of the relationship between penalties and tax non-compliance were mixed, there were some studies which found that the relationship was negative (witte & woodbury, 1985; cherry, 2001; sanders et al., 2008; doran, 2009; twum, 2014), while other studies concluded that the relationship was positive (crane & nourzad, 1986; morris, 2010 and azrina et al., 2014). the results of the present study seemed to be more in line with the studies of kamdar (1997), pommerehne and weck-hannemann (1996), and kuria et al. (2013), which concluded that the relationship was not significant. rettig (2011) explained that the tax authorities should enhance voluntary compliance rather than force taxpayers to pay their tax. this suggestion has been proposed to the yemeni tax authority as well, because the tax penalties seemed to have no significant effect on the behaviour of taxpayers. conclusion and limitations in conclusion, the current study has presented a model of tax noncompliance behaviour that included two factors from an economic perspective. specifically, the study investigated the relationship between the tax rate and tax penalties for non-compliant behaviour. the findings of the study revealed that the tax rate was significantly associated with tax non-compliance behaviour. however, tax penalties had no significant contribution to the model of tax noncompliance. it is worth highlighting that the current study was the first study that investigated the influence of tax system dimensions among smes in yemen. thus, the study has contributed to theory and practice in the field, as it has helped to explain how tax compliance could be enhanced. specifically, the present study has incorporated the tax rate, tax penalties, tax service quality and public governance quality in one single model. as such the study could contribute to the tax compliance literature by incorporating these four variables in the model and test the influence of these variables in the behaviour of taxpayers. it is recommended that the yemeni government deal with the tax non-compliance phenomenon in an effective manner by imposing a reasonable and fair tax rate. also, while the penalties have not been relevant in reducing the level of tax non-compliance, the yemeni tax authority must enhance voluntary compliance rather than continue to force taxpayers to pay their tax. 41 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 however, this study, like any other research work, has a number of limitations that must be recognized. nevertheless, these limitations may present an opportunity for future studies to consider. the current study relied on the self-reporting of taxpayers, where tax non-compliance was measured by way of a hypothetical scenario. this method of measurement had depended on the judgement of respondents which might vary from one person to another. thus, the general conclusion from the findings of the current study must be understood within the particular context of the present study, granted its share of limitations which require further study. another important limitation is the concern about the political situation of the country during the period in which the study was conducted.yemen suffers from a crisis of violence under an unstable political situation, which could have had an effect on the psychological and behavioural state of the citizen taxpayers in the country. the respondents taking part in the study might have lost confidence in the government. therefore, the answers probably did not reflect their actual behaviour accurately. in sum, further studies should be carried out when the country is in a better political situation. finally, the literature review has showed conflicting results in studies carried out thus far, especially about the relationship between the tax rate and penalties with tax non-compliance. in this regard, kirchler et al. (2007) explained that the stark contrast in the research findings was an indication that the relationship might have been moderated by certain variables. thus, further research is recommended to scrutinize the moderation effect between tax non-compliance and its determinants. acknowledgment this research received no specific grant from any funding agency. end notes 1 the yemeni business law has defined the sme on the basis of two criteria; the first criterion is the number of employees and workers, and the second 42 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 criterion is the capital of the enterprise. an enterprise employing from 1 to 4 workers and has a capital of less than yr1.5 million [about 6,700 usd], is considered a small enterprise, meanwhile the medium enterprise is defined as the enterprise, which has from 5 to 10 employees and a capital between yr1.5 million to yr20 million [about usd93,000]. references al-doais, f. a. (2008). the role of accounting information system in reducing tax evasion n the republic of yemen (unpublished master's thesis). university of al-albait, jordan. ali, m. m., cecil, h. w., & knoblett, j. a. (2001). the effects of tax rates and enforcement policies on taxpayer compliance: a study of self-employed taxpayers. atlantic economic journal, 29(2), 186-202. alm, j. (1999). tax compliance and administration. public administration and public policy, 7(2), 741-768. al-ttaffi, l. h. a. (2009). determinants of tax evasion: an empirical evidence from the republic of yemen (unpublished master's thesis). universiti utara malaysia. al-ttaffi al-, l. h. a., & abdul-jabbar, h. (2018). geopolitical differences and tax non-compliance among yemeni smes. journal of business management and accounting, 8(1), 31-45 al-ttaffi al-, l. h. a., & abdul-jabbar, h. (2020). geopolitical differences and tax non-compliance among yemeni smes. journal of business management and accounting, 8(1), 31-45. al-ttaffi al-, l. h. a., & abdul-jabbar, h. (2020). public governance quality and income tax non-compliance among small and medium enterprises in yemen. asia-pacific management accounting journal, 15(2), 1-24. al-ttaffi al-,, l. h. a., bin-nashwan, s. a., & amrah, m. (2020). the influence of tax knowledge on tax compliance behaviour: a case of yemeni individual taxpayers. journal of business management and accounting, 10(2), 15-30. azrina mohd yusof, n., ming ling, l., & bee wah, y. (2014). tax non-compliance among smes in malaysia: tax audit evidence. journal of applied accounting research, 15(2), 215234. 43 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 becker, g. s. (1968). crime and punishment: an economic approach. journal of political economy, 76(2), 169-217. bin-nashwan, s. a., hamedi al-, a. m., marimuthu, m., & harethi al-, (2020). study on system fairness dimensions and tax compliance in the middle east context. problems and perspectives in management, 18(1), 181-191. blank, j. (2014). collateral compliance. university of pennsylvania law review, 162(4), 720-800. borrego, a. c., da mota lopes, c. m., & ferreira, c. m. s. (2013). tax non-compliance in an international perspective: a literature review. portuguese journal of accounting and management, 103(1), 9-41. brown, r. e., & mazur, m. j. (2003). irs’s comprehensive approach to compliance measurement. national tax association symposium. new york: national tax association. central organization of control & audit coca. (2018). annual report. http://www.arabosai.org central organization of statistics cos. (2019). small and medium enterprises smes. http://www.cosit.gov.iq/ar/ cherry, t. (2001). financial penalties as an alternative criminal sanction: evidence from panel data. atlantic economic journal, 29(4), 96-114. christian, c. w., gupta, s., & lin, s. (1993). determinants of tax preparer usage: evidence from panel data. national tax journal, 46(4), 487-503. clotfelter, c. t. (1983). tax evasion and tax rates: an analysis of individual returns. the review of economics and statistics, 65(3), 363-373. crane, s. e., & nourzad, f. (1986). tax evasion: an empirical analysis. the economics and statistics, 68(2), 217-228. derwent, j. (2000). dismantling the barriers: a pan-european survey on the use of patents and patent information by small and medium-sized enterprises. houghton. doran, m. (2009). tax penalties and tax compliance. harvard journal of legislation, 46(1), 111-161. falsetta, d. (2020). discussion of trust and compliance effects of taxpayer identity theft: a moderated mediation analysis.the journal of the american taxation association, 1(42), 79-81. egwaikhide, f. (2010). taxation and state-building in a democratic system. nigerian taxation, 11(1), 35-57. 44 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 fararah, f. s. (2014). islamic microfinance system and customer satisfaction: a study on smes sector in yemen (unpublished doctoral dissertation). college university of insaniah, kedah, malaysia. feinstein, j. s. (1991). an econometric analysis of income tax evasion and its detection. journal of economics, 22(1), 14-35. fischer, c. m., wartick, m., & mark, m. (1992). detection probability and tax compliance: a review of the literature. journal of accounting literature, 11(2), 1-46. freire-serén, m. j., & panadés, j. (2013). do higher tax rates encourage/discourage tax compliance? modern economy, 4(2) 809-817. doi. 10.4236/me.2013.412086. gaventa, j., & mcgee, r. (2010). citizen action and national policy reform: making change happen. zed books. hai, o. t., & see, l. m. (2011). intention of tax non-complianceexamine the gaps. international journal of business and social science, 2(7), 79-83. hair, j. f., black, w. c., babin, b. j., & anderson, r. e. (2010). multivariate data analysis (7th ed.). prentice hall. hamm, j. l. (1995). income level and tax rate as determinants of taxpayer compliance: an experimental examination (unpublished doctoral dissertation). texas technical university, usa. hanefah, m., ariff, m., & kasipillai, j. (2002). compliance costs of small and medium enterprises. journal of australian taxation, 4(1), 73-97. helhel, y., & ahmed, y. (2014). factors affecting tax attitudes and tax compliance: a survey study in yemen. european journal of business and management, 6(22), 48-58. hindriks, j., peralta, s., & weber, s. (2008). competing in taxes and investment under fiscal equalization. journal of public economics, 92(12), 2392-2402. israel, g. d. (2009). sampling. university of florida. jaaidi al-, k. s., manaf, n. a. a., & karlinsky, s. s. (2011). tax evasion as a crime: a survey of perception in yemen. international journal of business and management, 6(9), 190-201. joulfaian, d., & rider, m. (1998). differential taxation and tax evasion by small business. national tax journal, 51(4), 676-687. kamdar, n. (1997). corporate income tax compliance: a time series analysis. atlantic economic journal, 25(1), 37-49. 45 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 kerlinger, f. n., & lee, h. b. (2000). foundations of behavioural research (3rd ed.). holt, rinehart & winston. kirchler, e., muelbacher, s., kastlunger, b., & wahl, i. (2007). why pay taxes? a review of tax compliance decisions. working paper 07-03, georgia state university, altanta. krejcie, r. v., & morgan, d. w. (1970). determining sample size for research activities. education psychological meas. kuria, k., ngumi, p., & rugami, j. (2013). factors affecting rental income tax compliance among landlords in kilifi municipality in kenya. prime journal of business administration and management, 3(5), 997-1008. levin, j., & widell, l. m. (2014). tax evasion in kenya and tanzania: evidence from missing imports. economic modelling, 39, 151162. lin, w. z., & yang, c. c. (2001). a dynamic portfolio choice model of tax evasion: comparative statics of tax rates and its implication for economic growth. journal of economic dynamics and control, 25(11), 1827-1840. long, s. b., & swingen, j. a. (1991). taxpayer compliance: setting new agendas for research. law and society review, 25(3), 637684. manaf, n. a. a., hasseldine, j., & hodges, r. (2005). the determinants of malaysian land taxpayers’ compliance attitudes. e-journal of tax research, 3(2), 206-221. mas’ud, a., aliyu, a. a., gambo, e. m. j., al-qudah, a. a., & al sharari, n. (2014). tax rate and tax compliance in africa. european journal of accounting auditing and finance research, 2(3), 22-30. mcgee, r. w. (2006). three views on the ethics of tax evasion. journal of business ethics, 67(1), 15-35. mcgee, r. w. (2007). ethics and tax evasion in asia. icfai journal of public finance, 5(2), 21-33. mcgee, r. w., & and rossi, m. j. (2006). the ethics of tax evasion: a survey of law and business students in argentina. sixth annual international business research conference, co-sponsored by the coggin college of business, university of north florida and the school of management, warsaw university, february 10-11. mcgee, r. w., & ho, s. (2006). the ethics of tax evasion: a survey of accounting, business and economics students in hong kong. 46 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 published in the proceedings of the international academy of business and public administration disciplines (iabpad), winter conference, orlando, florida, january 3-6. mcgee, r. w., & lingle, c. (2006). the ethics of tax evasion: a survey of guatemalan opinion. presented at the sixteenth international atlantic economic conference, new york, october 6-9. mcgee, r. w., petrides, y., & ross, a. m. (2009a). ethics and tax evasion: a survey of mexican opinionopinion. in the ethics of tax evasion (387-403). springer new york. mcgee, r. w., serkan, b., benk, adriana m. ros., s & harun kılıçaslan. (2009). an empirical study of ethical opinion in germany. journal of accounting, ethics & public policy, 10(2), 243-259. mcgee, r., & an, y. (2007). an empirical study of tax evasion ethics in china. working paper, andreas school of business, florida university. ministry of industry & trade mit. (2018). small and medium enterprises smes. http://www.moit.gov.ye ministry of planning & international cooperation mpic. (2009). annual report. http://www.yemen.gov.ye modugu, p. k., eragbhe, e., & izedonmi, f. (2012). government accountability and voluntary tax compliance in nigeria. research journal of finance and accounting, 3(5), 69-76. morris, d. (2010). tax penalties and deterrence: determining effectiveness and taxpayer perception. the cpa journal, 43(3), 28-33. mwangi, p. c. n. (2014). factors influencing tax compliance among small and medium enterprises in nairobi’s industrial area, kenya (unpublished doctoral dissertation). university of nairobi. national information system nis. (2018). small and medium enterprises smes. http://www.yemen-nic.info/study_guide / index.php nor azrina mohd yusof, n., ming ling, l., & bee wah, y. (2014). tax non-compliance among smes in malaysia: tax audit evidence. journal of applied accounting research, 15(2), 215234. obaid, m. m., ibrahim, i., & udin, n. m. (2020). an investigation of the determinants of tax compliance among yemeni manufacturing smes using the fisher model. international journal of psychosocial rehabilitation, 24(4), 1809-1824. 47 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 palil, m. r. (2010). tax knowledge and tax compliance determinants in self-assessment system in malaysia (unpublished doctoral dissertation). university of birmingham. palil, m. r., & mustapha, a. f. (2011). factors affecting tax compliance behaviour in self-assessment system. african journal of business management, 5(33), 12864-12872. park, c. g., & hyun, j. k. (2003). examining the determinants of tax compliance by experimental data: a case of korea. journal of policy modelling, 25(8), 673-684. pashev, k. v. (2008). taxation and compliance of small business in transition economies: lessons from bulgaria. taxation and public finance in transition and developing economies, springer us. pham, t. m. l., le, t. t., truong, t. h. l., & tran, m. d. (2020). determinants influencing tax compliance: the case of vietnam. the journal of asian finance, economics, and business, 7(2), 65-73. pommerehne, w. w., & weck-hannemann, h. (1996). tax rates, tax administration and income tax evasion in switzerland. public choice, 88(1-2), 161-170. porcano, t. m. (1988). correlates of tax evasion. journal of economic psychology, 9(1), 47-67. rabaee al-, f. (2014). failure of the unity between the republics of yemen: a scientific approach under unity theories. madar centre for studies and research. philistine. rettig, c. p. (2011). enhancing voluntary compliance through the administration of civil tax penalties. journal of tax practice and procedure, 13(11), 17-20. richardson, g. (2006). determinants of tax evasion: a cross country investigation. journal of international of accounting, auditing and taxation, 15(2), 150-169. richardson, m. (2013). determinants of social security compliance in a small island developing states: an analysis analysis of anguilla. central bank of barbados, research and economic analysis department. cbb working paper no. wp/13/12. ross, a. m., & mcgee, r. w. (2012). education level and ethical attitude toward tax evasion: a six-country study. journal of legal, ethical and regulatory issues, 15(2), 93-138. saadi, al-, m. (2014). international cooperation: vision and mission. al-thawra gazette (issued daily by the yemeni prime minister). issue of 17 july 2014. sana’a. 48 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 sanders, d. l., reckers, p. m., & iyer, g. s. (2008). influence of accountability and penalty awareness on tax compliance. journal of the american taxation association, 30(2), 1-20. sekaran, u. (2003). research methods for business: a skill building approach (4th ed.). john wiley and sons. ser, p. c. (2013). determinants of tax non-compliance in malaysia (unpublished master's thesis). universiti tunku abdul rahman, petaling jaya, malaysia. spicer, m. w., & becker, l. a. (1980). fiscal inequity and tax evasion: an experimental approach. national tax journal, 33(2), 171175. trivedi, v. u., shehata, m., & mestelman, s. (2004). attitudes, incentives, and tax compliance. department of economics, mcmaster university. twum, b. e. (2014). gift tax compliance in ghana: an empirical study. journal of finance and accounting, 2(1), 1-7. umar, m. a., kasim, r., & martin, d. j. (2012). an overview of property tax collection as a tool for a sustainable local government reform in malaysia. international conference on technology management, business and entrepreneurship, 1819 december, melaka, malaysia. witte, a. d., & woodbury, d. f. (1985). the effect of tax laws and tax administration on tax compliance: the case of the us individual income tax. national tax journal, 38(1), 1-13. world bank. (2018). browse by country or indicator data bank microdata data catalo. https://data.worldbank.org/indicator/ sh.sta.diab.zs?view=chart zikmund, w. g. (2003). business research methods. south-western. 49 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 23–49 appendix the hypothetical scenario form measuring tax non-compliance read the following scenario and kindly indicate your opinion (by way of a circle): suppose that the annual sales income of ahmed’s enterprise amounted yr130.000, and the total operating expenses amounted yr30.000. the amount which has to be reported for tax purpose is yr100.000. however, ahmed is almost certain that the tax office will not audit him and would not know if 40% of the taxable income was not reported. additionally, he paid yr10.000 to repair his personal car. in preparing the tax return, he is thinking about claiming the costs of repair as if the car was used in his business. legally, such a claim is not allowable, but he is almost certain that he will not be audited and that the tax authority would not be able to detect the deduction. the income tax law provides that tax return has to be filed with tax authority within 120 days from commencement of year (january) of assessment, and tax should be paid within 60 days from the date of assessment notice. 1) indicate the taxable income that ahmed should include in his report: (a) yr100.000 (b) yr60.000 (c) other amount, please specify _________________ tnc1 2) how much should ahmed deduct as business expenditure? (a) yr 30.000 (b) 40.000 (c) other amount, please specify _________________ tnc2 3) which of the dates below should ahmed file his income tax? (a) on 30th april or before (b) after 30th april (c) other date, please specify _______________ tnc3 4) how many days after receiving assessment notice should ahmed pay his income tax: (a) 60 days or less (b) more than 60 days (c) other, please specify _________________ tnc4 decision to pay dividens and ownership structures in nigeria : 27-44 27 decision to pay dividends and ownership structures in nigeria adamu idris adamu* department of accounting, federal university dutsin-ma, nigeria tissa, universiti utara malaysia, malaysia rokiah ishak nor laili hassan tissa, universiti utara malaysia, malaysia *corresponding author: adamuidris48@gmail.com ________________________________________________________________ a r t i c l e i n f o _____________________________ article history: received 12 march 2019 revised 5 august 2019 accepted 13 august 2019 published 6 january 2020 ____________________ keywords: decision to pay dividends, ownership structures, emerging market. jel codes: ag32, g350, g390 a b s t r a c t _________________________________ empirical evidence on how ownership structures influence decision to pay dividends remain unclear in the dividend policy literature. this paper is set to investigate the association between ownership structures and decision to pay dividends. the sample firms of this study consist of nonfinancial firms listed on the nigerian stock exchange for the period 2011 to 2015 with 270 firm-year observations and logit regression models used to examine the relationship. the study revealed strong evidence that institutional investors were positively related to the decision to pay dividends. however, managerial shareholding was found to have an inverse effect on firms’ probability to pay dividends. additional analysis was carried out only on dividend payers and the results were also consistent with the hypothesis. despite this, the managerial investors were somehow weak when the sample was reduced to dividend payers. the international journal of banking and finance, vol. 14, 2018-2019 : 27-44 27 28 the international journal of banking and finance, vol. 14, 2018-2019: 27-44 cumulatively, the results are robust and show support of the agency theory and hence, imply that institutional investors in nigeria have preference for dividend payers. 1. introduction dividend policy has been a topical issue over the years and remains a subject of vital concern in modern finance (baker & weigand, 2015). additionally, almalkawi, rafferty, and pillai (2010) noted that dividend policy has become the top agenda item of managers in the modern corporate world and has emerged as a contending topic in the field of accounting and finance. karpavičius (2014) also noted that a firm’s dividend payout is important in the determination of its value, and dividend stability increases the value of the firm. hence, dividend is crucial to the shareholders as well as to the firm. the relationship between dividend payout and institutional investors has been empirically documented in the literature (for example, afza & mirza, 2011; huda & abdullah, 2014; manos, 2003; miko & kamardin, 2015; short, zhang, & keasey, 2002). however, evidence on the association between decision to pay dividend and institutional ownership is lacking in the academic literature particularly from the emerging market where the institutional setting is distinct from the developed markets. more so, as the decision to pay dividend is a prerequisite to the level of dividend and hence, the decision to pay or not to pay dividend should be determined in the light of institutional investors prior to considering the magnitude of the dividend payout (idris, ishak, & hassan, 2017). on the decision to pay dividend, companies are allowed by the regulators to pay dividends only out of its current profits or its revenue reserves (sec nigeria, 2013). likewise, firms are not allowed to borrow on the grounds for dividend payment. however, with regards to the institutional owners, the capital market regulators have mandated these investors (institutional investors and other large shareholders) to actively participate in the firms of which they invest (sec nigeria, 2011). the participation will allow them to influence positively in the financial policies of the firms. moreover, in terms of taxation, the institutional investors have no tax incentive better than other shareholders as dividends are paid net of withholding tax of 10%. therefore, it is unclear as to whether institutional investors may influence dividend decision or not. consequently, in nigeria, on average, the institutional investors have the highest percentage of shareholdings which accounts for about 48% of the total shares listed on the nigerian stock exchange market (abor & fiador, 2013). accordingly, these investors may have significant influence on the decision to pay dividend because of their primary objective which is return on equity in the form of dividend. furthermore, the institutional investors may pay much attention to the economic decision to pay dividens and ownership structures in nigeria : 27-44 29 benefits attached to their shareholdings in various firms (gedajlovic, yoshikawa, & hashimoto, 2005). this paper tends to empirically explore the potential relationship between decision to pay dividend and ownership structure using data set from nigeria. thus, the paper contributes to the dividend literature through examining the role played by institutional and managerial owners on determining whether to pay or not to pay a dividend. it also examines whether there is significant difference between dividend and non-dividend paying firms in relation to their institutional and managerial ownership. the paper is structured as follows. section 2 offers literature review and develops the research hypotheses. section 3 provides a detailed discussion on the data and methodology used in the study. the empirical findings of the paper are reported in section 4 while the conclusion of the paper is in section 5. 2. review of literature and research hypothesis several legal and corporate governance changes have taken place for the purpose of restructuring and empowering board of directors in handling corporate bodies or inspiring institutional shareholders activism (kapopoulos & lazaretou, 2007). institutional shareholders are believed to be more knowledgeable and capable of handling self-dealing issues in the firm (afza & mirza, 2011). they argued that the presence of the institutional investor may likely reduce the severity of agency conflicts among competing parties. invariably, institutional investors are found to be influential in various aspects of the corporate entity’s activities. these areas may include executive compensation (mancinelli & ozkan, 2006; victoravich, xu, & gan, 2013) enhancement of audit quality (han, kang, & rees, 2013) earnings management (hsu & koh, 2005), improving firm value (navissi & naiker, 2006) and share price volatility (dennis & strickland, 2002; rubin & smith, 2009). moreover, institutional investors do influence corporate payout policy (chang, kang, & li, 2016; miko & kamardin, 2015; short et al., 2002). 2.1 institutional ownership and decision to pay dividends the agency theorists have posited that firm managers are more interested in retaining substantial amount of corporate profits. the retained profits will allow them to have substantial amount of corporate resources within their reach or control and therefore, provide the managers an avenue for empire building (jensen & meckling, 1976). however, paying dividends may help to mitigate agency costs that may arise between shareholders and corporate managers. this is because payment of the dividend by the firm is likely to compel the firm to source for funding from the capital market and hence, expose the firm to monitoring by the market (easterbrook, 1984; rozeff, 1982). extant literature has suggested that institutional ownership is likely to reduce agency conflict by influencing managers to distribute free cash flow to the shareholders (chang 30 the international journal of banking and finance, vol. 14, 2018-2019: 27-44 et al., 2016; grinstein & michaely, 2005; jory, ngo, & sakaki, 2017; miko & kamardin, 2015; short et al., 2002; sindhu, hashmi, & haq, 2016). chang et al. (2016) examines how institutional owners influence dividend policy from the agency perspectives. the study establishes a significant positive association between institutional owners and dividend payout. they added that the positive association is more salient when there is weak external monitoring. further, the study argued that institutional owners may use dividend as a monitoring tool to address agency related problems subject to the financial performance of the firm. along this line, there is equally strong evidence supporting the argument that higher dividends increase the stake of institutional investors (jory et al., 2017). however, from the tax clientele view of dividends, allen, antonio, and welch (2000) posited that tax incentives attract more institutional shareholders in the u.s. market to invest in dividend paying firms than non-dividend paying firms. from the emerging markets, the evidence also supports the proposition that institutional investors and dividend payout are positively correlated. sindhu et al. (2016) observed that strong positive relationship prevails between institutional investors and dividend payouts from the pakistani market. the study showed that institutional investors used dividends as a mechanism for monitoring managers with a view to mitigate agency problem. the result is consistent with the previous evidence advanced by miko and kamardin (2015) from nigeria that also reveals statistically strong association between dividend payout and institutional ownership. the study suggested that institutional investors force managers to disgorge free cash flow from the firm such that managers may not have excess to embark on empire building or perquisite consumption. likewise, francis, hasan, john, and song (2011) found that decision to pay dividends by firms tends to increase when the shares held by institutional owners increases. recently, the result of jacob, jijo and lukose (2018) suggest that institutional ownership has significant positive effects on the propensity to pay dividends. overall, the demand for dividends by institutional investors is in line with the indirect monitoring of managers and free riding arguments (short et al., 2002). based on these theoretical explanations, the study hypothesized that; h1: positive relationship prevails between institutional ownership and decision to pay dividends. 2.2 managerial ownership and decision to pay dividends agency theory has suggested that dividend payouts could be used as a mechanism to reduce agency costs and to control particular set of investors from gaining far above the other set (rozeff, 1982). managerial ownership is used interchangeably with insider ownership, directors’ holdings among others (farinha, 2003; francis et al., 2011; sanda, mikailu, & garba, 2010; short et al., 2002). dividend policy is among the techniques that provide control in a firm against potential agency problems. the probability of managers owning a stake in a company may portray the likelihood of alleviating agency problems (jensen decision to pay dividens and ownership structures in nigeria : 27-44 31 & meckling, 1976). the argument is that, because the managers form part of the shareholders, they may not go against the interests of other shareholders by wasting the accumulated cash. this is because such an act may affect the entire shareholders of the firm. the arguments on the relationship between managerial ownership and dividend payout are based on a monitoring effect (florackis, kanas, & kostakis, 2015; francis et al., 2011; short et al., 2002). the monitoring hypothesis suggests that managerial ownership and dividend policy are inversely related, implying that managers are likely to pursue projects that maximize shareholder value. consistent with this argument, evidence from the nigerian market supports the agency theory prediction on the association between managerial ownership and dividend payout (dandago, farouk, & muhibudeen, 2015; miko & kamardin, 2015; ullah, fida, & khan, 2012). the studies found negative association between managerial ownership and dividend payout. the results are also in agreement with extant literature (al-amarneh & yaseen, 2014; rizqia & aisjah, 2013; sindhu et al., 2016). that inverse correlation exists between shares held by managers and dividend policy. hu and kumar (2004) also reported that firms with low insider ownership are more likely to pay dividends to the shareholders and francis et al. (2011) documented that the likelihood to pay dividends decreases with an increase in managerial shareholding. more recently, farooq and ahmed (2019) showed that insider ownership is negatively related to the likelihood of dividend payment; hence, consistent with the agency theory. h2: managerial ownership is negatively associated with decision to pay dividends. 3. methodology the sample for the data analysis of this paper consists of non-financial firms listed on the nigerian stock exchange for the period 2011 to 2015. the choice of period for instance, marked the implementation of the 2011 code of corporate governance and the end of the financial crisis; whereas, the choice of 2015 was due to availability of data used for the study. the study excludes financial firms for two reasons. firstly, financial firms are governed by financial regulatory authorities such as the central bank of nigeria (cbn), nigeria deposit insurance corporation (ndic) and nigerian insurance commission (naicom). these regulatory authorities for example cbn, requires banks to adequately provide for minimum capital adequacy ratio, cash reserve either low or moderate and a non-performing loan ratio not exceeding 5% prior to paying dividends to the shareholders. secondly, prior studies on decision to pay dividends excludes financial firms from the final sample of their investigations (see for example, baker, dutta, & saadi, 2008; chang et al., 2016; francis et al., 2011; idris et al., 2017; pucheta-martínez & lópez-zamora, 2017) and factors affecting dividend policy varies from financial to non-financial firms (baker et al., 2008). the data 32 the international journal of banking and finance, vol. 14, 2018-2019: 27-44 were collected from the annual report of individual firms. the annual reports were obtained simultaneously from individual websites of the firms and the nse website. the model of the study is as follows and is used in testing the hypotheses of the study. divdumi = β0 + β1instdumi + β2mangowsi + β3fsizei + β4retei + β5flevi + β6lnsalei + βjyeari + βkindustryi +εi besides the two variables of interest, institutional and managerial ownership, the study includes various control variables that were found to be important in the literature on the decision to pay dividends. firm size, retained earnings, leverage and sales and their definitions and sources are provided in table 3.1. the study also controls for year effect using five dummies. the use of the year dummy is important as 2011 was the year that marked the implementation of the new sec code of corporate governance and demanded that the institutional owners to actively participate in its enforcement of the industry effect. likewise, ten dummies have been used for the industry effect. this is because the decision to pay dividends may be affected by the industry as the number of firms within the industries is not evenly distributed. prior studies such as (francis et al., 2011; jiraporn, kim, & kim, 2011; liljeblom & maury, 2016; sharma, 2011) have considered year and industry effect while studying propensity to pay dividends. table 1. variable description variable description expected sign source divdum dummy variable that takes the value of 1 if the firm pays dividends and 0 if otherwise. (chang et al., 2016; pucheta-martínez & bel-oms, 2016; pucheta-martínez & lópez-zamora, 2017; sharma, 2011) divyld the dividend yield is estimated as dividend per share divided by the share price at the end of a year (chang et al., 2016; hu & kumar, 2004) instdum total institutional ownership ratio and takes the value of 1 if a firm has institutional investors and 0 if, otherwise. + (chang et al., 2016; francis et al., 2011; jory et al., 2017) mangows management ownership is the total percentage of equity owned by a firm’s directors. (francis et al., 2011; pucheta-martínez & lópez-zamora, 2017; sindhu et al., 2016) (continued) decision to pay dividens and ownership structures in nigeria : 27-44 33 variable description expected sign source fsize logarithm of total assets. + (pucheta-martínez & lópez-zamora, 2017; sharma, 2011) rete retained earnings to total capital. + (deangelo, deangelo, & stulz, 2006; francis et al., 2011; jiraporn et al., 2011) flev total debt divided by total assets. (chang et al., 2016; deangelo et al., 2006) lnsale logarithm of annual sales. -/+ (jeon, lee, & moffett, 2011; liljeblom & maury, 2016) 4. analysis of results table 2, panel a depicts the descriptive statistics of the variable used in the study and only the two variables of interest have been discussed in the summary statistics. also in table 2, panel b provides further summary statistics on the two independent variables. firms that paid dividends for the sample period on average accounted for 60.7% per. this indicated that more than half of the listed firms on the nse paid dividends. regarding the institutional investors, the average value was 48.6 % with a minimum and maximum of 0.00 and 98.2%, respectively. this statistic showed that there were a considerable number of institutional shareholders in some firms with a few firms having none. this was also the case for managerial ownership. among the sample firms in this study, managerial ownership was found to have an average value of 7% and a median of 1%. furthermore, the highest and lowest value for this variable was 0.00 to 70.8%, respectively. table 2. descriptive statistics panel a variable mean median min max std. dev. divdum 0.61 1.00 0.00 1.00 0.49 instow 0.49 0.54 0.00 0.98 0.27 instdum 0.90 1.00 0.00 1.00 0.29 mangows 0.07 0.00 0.00 0.71 0.15 fsize 7.04 6.97 5.78 8.36 0.71 rete 0.14 0.14 -0.67 0.57 0.31 flev 0.24 0.11 0.00 1.62 0.38 lnsales 6.87 7.00 4.00 9.00 0.89 (continued) 34 the international journal of banking and finance, vol. 14, 2018-2019: 27-44 panel a variable mean median min max std. dev. sales 39900000 7347873 26130 651000000 81600000 panel b (n=89) 2011 2012 2013 2014 2015 i n s t i t u t i o n a l own =1 number 78 80 81 81 82 percent 87.64 89.89 91.01 91.01 92.13 m a n a g e r i a l own =1 number 45 46 45 44 43 percent 50.56 51.69 50.56 49.44 48.31 note: divdum is decision to pay dividend and takes the value of 1 if the firm pays dividends and 0, if otherwise; instdum is total institutional ownership ratio and takes the value of 1 if a firm has institutional investors and 0, if otherwise; mangows is managerial ownership which is the total percentage of equity owned by a firm’s directors; fsize is logarithm of total assets ; rete is retained earnings to total capital; flev is total debt divided by total assets; lnsales logarithm of annual sales. the summary statistics reported in table 2, pane b, shows the data in relation to ownership variables. for the institutional investors, there has been a consistent increase. for instance, in 2015, institutional investors accounted for 92% as compared with 88% in 2011. this meant that there was a steady increase in the presence of institutional investors during the sample period. while institutional investors had increased, the managerial owners on the other hand, remain relatively fixed. evidently, the presence of managerial owners dropped during the study period from 51% in 2011 to 48% in 2015. this result could imply that agency problems may be addressed with the payment of dividends as suggested in prior studies. table 3. correlation 1 2 3 4 5 6 7 vif tol 1.divdum 1.00 2.instdum 0.16*** 1.00 1.21 0.83 3.mangows -0.27*** -0.36*** 1.00 1.39 0.72 4.fsize 0.35*** -0.05 -0.3*** 1.00 1.24 0.80 5.rete 0.49*** 0.15** -0.21*** 0.24*** 1.00 1.11 0.91 6.flev -0.10* 0.00 0.30*** -0.32*** -0.16*** 1.00 1.19 0.84 7.lnsales 0.09* 0.00 0.02 0.00 0.00 0.00 1.00 1.00 0.99 the variables have been defined in table 4.1 above. *, **and *** are significant at 10%, 5% and 1%, respectively. decision to pay dividens and ownership structures in nigeria : 27-44 35 the analysis of correlation is reported in table 3. the correlation analysis between the dependent and independent variables is to enable the study to ascertain the direction and strength of relationship. from the correlation matrix, the results showed a strong positive association between decision to pay dividends and the presence of institutional owners. it was found to be statistically significant at 1%. thus, this implied that firms with institutional owners may demand the payment of dividends. conversely, the association of decision to pay dividends and managerial ownership was found to be negative and was also statistically …. the result of the correction meant that decision to pay dividends and the management holding were inversely related. furthermore, the correlation also suggested that the presence of institutional owners could result in a low stake in managerial shares. the result pertaining to variance inflation factor (vif) test that is reported in table 3 demonstrates that none of the values for vif is equal or greater than 10 threshold (gujarati, 2004). similarly, the tolerance (tol) values in the study were also within the acceptable level that is below the value of 1. hence, the collinearity test provided additional evidence on the absence of multicollinearity that could endanger the findings of the study. table 4. univariate comparison dividend nonpayers (n=175) dividend payers (n=270) variable mean std. dev. mean std. dev. t-value instdum 0.85 0.36 0.94 0.24 -3.07*** mangows 0.12 0.19 0.04 0.10 5.18*** the variables have been defined in table 4.1 above. *, **and *** are significant at 10%, 5% and 1%, respectively a comparison between dividend non-payers and payers revealed further interesting results. both the mean and the standard deviation suggested that the dividend payers were distinct from non-payers. from the results in table 4, the nonpayers had a mean value of 85% which was lower than 94% for the dividend payers which was statistically significant. the summary statistics implied that the institutional owners were prevalent in the dividend paying firms. in addition to the result of the group mean, the group standard deviation of the non-dividend payers was also different from the dividend paying firms. the non-dividend paying firms had a higher standard deviation of 36.2% when compared with 23.7% for the dividend paying firms. pertinent to the managerial shareholding, the mean value for the non-payers was 11.9% which was much higher than 3.8% scored by the dividend payers. tests of mean differences in this case were statistically significant at 1%. the result could suggest that managers may be more inclined to non-dividend payers than paying firms. 36 the international journal of banking and finance, vol. 14, 2018-2019: 27-44 table 5. logit regression with robust standard errors panel a panel b expected sign coefficient z-statistics marginal effects z-statistics instdum + 0.71* 1.69* 0.10* 1.69 mangows -1.50* -1.70* -0.22* -1.71 fsize + 1.18 4.00*** 0.17*** 4.28 rete + 4.29 6.12*** 0.62*** 7.93 flev 0.29 0.88 0.04 0.89 lnsales -/+ 0.41 3.19*** 0.06*** 3.27 year yes yes yes yes industry yes yes yes yes constant -11.18 -4.99*** -0.04 0.39 observations 445 445 likelihood ratio 99.15*** pseudo-r2 33.5% correct prediction% 81.12% h-l test p-value 0.1589 the variables have been defined in table 4.1 above. *, **and *** are significant at 10%, 5% and 1%, respectively. the study performed regression analysis to get more insights on how ownership affected decision to pay dividends and control for factors that included firm size, retained earnings, leverage and sales in which prior studies had clearly highlighted their importance in dividend policy literature. table 5 shows the result of the robust logit regression estimation containing estimation based on coefficient and odd ratios along with their z-values. the model had a likelihood ratio chi-square value of 99.15 and was statistically significant at 1% with a nagelkerke pseudo r2 of 33.5% thus showing the goodness of fit of the model and confirmed the assertion that at least one of the coefficients in the model influenced the decision to pay dividends which was the dependent variable. the study also ran a classification test in order to determine the percentage of correctly predicted cases and how the model predicted group membership. hosmer, lemeshow, and sturdivant (2013) argued that the classification table was an intuitively appealing way that provided a summary of the results of a fitted logistic regression. in this regard, the percentage of group membership correctly predicted by the model was 81.12% which was far above 50% the bench mark for a failed model (pampel, 2000) therefore, suggesting the fitness of the model. lastly, the hosmer-lemeshow test (h-l test) also confirmed the decision to pay dividens and ownership structures in nigeria : 27-44 37 model fit of the logit regression as the h-l chi-square value of 11.83 with an insignificant p-value of 0.1589. table 5 provides the results of the individual variable of interest. the coefficient of the institutional investors was positive and statistically significant at 10%. furthermore, the results also showed that the presence of institutional investors (instdum) in a firm was more likely to make cash dividend payments which was in line with hypothesis one (h1) and the agency theory. this suggested that institutional investors helped address agency conflict by encouraging dividend payments as a monitoring device. thus, the result agreed with previous findings (chang et al., 2016; francis et al., 2011; grinstein & michaely, 2005; jacob et al., 2018; jory et al., 2017; miko & kamardin, 2015; short et al., 2002; sindhu et al., 2016) that firms with institutional owners were more likely to pay dividends. however, managerial ownership (mangows) had a negative and significant effect on the nse-firms’ decision to pay dividends. the logit regression revealed that the coefficient estimates on managerial ownership was negative and statistically significant at 10%. the result suggested that higher managerial shareholding was less likely to lead to more dividend payments. this result corroborated previous studies (farooq & ahmed, 2019; francis et al., 2011; hu & kumar, 2004; miko & kamardin, 2015; sindhu et al., 2016) which reported that the likelihood of paying dividends decreased with the increase in managerial shareholding, hence, this was consistent with hypothesis 2 (h2). the result could suggest that dividend was used in the nse to curtail the managers from wasting free cash since the managerial holding was on the decreasing path. it has been argued that dividend payments are important as it is part of the optimal monitoring or bonding tool that a firm may use to reduce agency costs (rozeff, 1982). alternatively, the result showed that managers in the nse preferred profit retention instead of disbursing it as dividends to the ordinary shareholders and this could likely lead to perquisite consumption the result of the control variables employed in the current study revealed firm size and retained earnings to be positive and statistically significant as expected and was consistent with prior evidence (for instance, jiraporn, kim, & kim, 2011; liljeblom & maury, 2016). thus, indicating that larger firms with higher retained earnings exhibited a higher likelihood of paying cash dividends. more so, the result of sales revenue was also found to be positive and statistically significant. hence, suggesting that firms with more sales revenue had a higher probability of paying dividends among the listed firms in nigeria. this result therefore, supported earlier findings on propensity to pay dividends. besides the reported results based on the coefficient of each variable, the study also estimated the marginal effect of the independent variables on outcome. this is to provide further support to the findings of logit regression results with marginal effects, which is depicted in table 4.3 panel b. according to williams (2012), the marginal effect, otherwise referred to as economic significance, was another way by which the effects of variables in nonlinear models such as logit regression analysis could be made more meaningful. in 38 the international journal of banking and finance, vol. 14, 2018-2019: 27-44 other words, the marginal effect provided a representative value for variables in nonlinear models. this paper used the average marginal effect because it was superior than other techniques (williams, 2012). the analysis of the marginal effect is reported in table 5. both the two variables of interest retained their signs and statistical significance indicating their impact on the decision to pay dividends among the nse firms. institutional ownership had a marginal effect of 10.3% which was positive and statistically significant at 10%. this meant that for every 1% change on the decision to pay dividends, it followed that, all things being equal, the institutional shareholding increased by 10.3%. the result implied that the presence of institutional investors exerted a greater influence on a firm to pay cash dividends and in turn forced firms to consider raising more funds for investment when necessary in the capital and hence, could result in more market monitoring which in turn agreed with the agency theory (rozeff, 1982). conversely, the marginal effect of the managerial shareholding although it was negative, was higher than the reported marginal effect for the institutional investors as depicted in table 5, panel b. the result showed that it was negative and statistically significant which meant that for 1% additional share of managerial ownership, the decision to pay dividends decreased by 21.8%. this result showed how important managerial shareholding was to firms which could consider paying cash dividends. the negative sign of the association between managerial shareholding and decision to pay dividends further, aligned with the argument advanced by rozeff (1982) that dividend payout was an optimal mechanism that could be used to reduce agency costs as the stake of the managers declined in the firm. 4.2 additional analysis the study also carried out two more additional analysis to ascertain the robustness of the findings (see for example, al-najjar & kilincarslan, 2016; chang et al., 2016; francis et al., 2011; hu & kumar, 2004). in the first specification, the study classified the firms into low and high based on their dividend yield and estimated the model using logit regression with robust standard errors specifications. the result was reported in table 4.3. interestingly, the model fitness was also found to be statistically significant at 1% with a chi-square value of 59.85 and pseudo r2 of 35.03%. moreover, the coefficient of the two key variables retained their signs with institutional shareholding becoming more significant than in the main model of table 3. the result showed that institutional shareholding was positive and statistically significant at 1%, indicating that institutional shareholders were concerned about firms that paid higher dividends. the results also confirmed the agency theory argument that institutional investors may use dividend as a form of control mechanism in the firm to avoid free riding problems. for the managerial ownership however, the coefficient was negative though statistically insignificant. but then the negative sign conformed to the agency theory as well which suggested that paying of dividend should be the alternative as the managers’ stake decreases in the firm. decision to pay dividens and ownership structures in nigeria : 27-44 39 table 6. additional analysis of dividend payers panel a panel b expected sign coefficient z-statistics coefficient t-statistics instdum + 2.04 2.93*** 0.018 1.98** mangows -0.94 -0.80 0.035 1.67* fsize + 1.06 2.66*** 0.000 0.02 rete + 4.62 4.89*** 0.012 1.45 flev 0.78 1.68* 0.003 0.60 lnsales -/+ 0.34 1.91* 0.000 0.12 year yes yes yes yes industry yes yes yes yes constant -11.85 -3.76*** -0.005 -0.16 observations 223 270 likelihood ratio 59.85*** pseudo-r2 35.03% correct prediction% 76.68% h-l test p-value 0.6259 r2 10.05 f-stat 2.12*** in the second stage of the additional test, the study used a continuous dependent variable known as dividend yield which was consistent with previous studies (for instance, chang et al., 2016; hu & kumar, 2004) with only dividend payers as the sample firms for the estimation. in this case the f-statistic was 2.12 which was significant at 1% with the r2 value of 10.05% thus, suggesting the fitness of the model. the result from this estimation showed that institutional shareholding was highly significant which supported the stated hypothesis of the study and prior evidence. while institutional shareholding remained positively significant, the coefficient of the managerial shares in the firm reversed from negative to positive and was statistically significant thus contradicting the earlier findings of the main results. the results may be interpreted as a form of managerial entrenchment. according to the entrenched view, a manager may use dividends to safeguard his position (farinha, 2003). therefore, he or she may be more likely to pay greater dividends so that he/she portrays the identity of a good manager who protects or aligns his/her interests with those of other shareholders. moreover, it is possible that the dividend payers had more committed and entrenched managers. jo and pan (2009) posited that managers may be committed to their responsibilities voluntarily to save their jobs which may in turn limit their opportunistic behaviour. 40 the international journal of banking and finance, vol. 14, 2018-2019: 27-44 consequently, the results of table 5 and 6 suggest that the presence of institutional investors in the firm may lead to the likelihood of making cash dividends in the firm; whereas, managerial shareholding may decrease the probability of paying dividends. 5. conclusion in this paper we analyse how institutional and managerial shareholdings affect the decision to pay cash dividends. a logit regression was used to estimate the effect these variables had on the decision to pay dividends. consequently, the results of this study suggested the importance of large scale ownership. the study provides that the presence of institutional investors in the firm may lead to a likelihood of making cash dividend payments. the likelihood of paying the dividends becomes more prevalent in firms with higher dividend yields. the results also showed that institutional investors may use dividends as a control tool to avoid free-riding problems in addressing agency conflicts. whereas, the relationship between managerial shareholdings and the probability of paying dividends were inversely related which also lend support to the agency theory. the policy implication of this finding is that regulatory bodies may consider all institutional and managerial owners irrespective of whether their holdings is 5% or below in the enforcement of good governance and policies such as taxation. moreover, further studies could consider combining these ownership variables with board independence and gender diversity. such an approach could provide more insights on how large owners may be influenced by board structures. this approach could also enhance the level of our understanding on ownership structures. in addition, it will be useful if future researchers consider various classes of institutional investors separately such as banks, discount houses, pension institutions and insurance firms in analysing their individual roles on firms’ decision to pay dividends. references abor, j., & fiador, v. (2013). does corporate governance explain dividend policy in sub-saharan africa? international journal of law and management, 55(3), 201–225. https://doi.org/10.1108/17542431311327637 afza, t., & mirza, h. h. (2011). institutional shareholdings and corporate dividend policy in pakistan. african journal of business management, 5(22), 8941–8951. https://doi.org/10.5897/ajbm11.564 al-amarneh, a., & yaseen, h. (2014). corporate governance and dividend policy in jordan. international journal of economics and finance, 6(4), 200–207. https://doi.org/10.5539/ijef.v6n4p210 al-malkawi, h. n., rafferty, m., & pillai, r. (2010). dividend policy: a review of theories and empirical evidence. international bulletin of business administration, 9(9), 171–200. decision to pay dividens and ownership structures in nigeria : 27-44 41 al-najjar, b., & kilincarslan, e. (2016). the effect of ownership structure on dividend policy: evidence from turkey. corporate governance: the international journal of business in society, in press. https://doi. org/10.1108/cg-09-2015-0129 allen, f., antonio, e. eb., & welch, i. (2000). a theory of dividends based on tax clienteles. the journal of finance, lv(6), 2499–2536. https://doi. org/10.1111/0022-1082.00298 baker, h. k., dutta, s., & saadi, s. (2008). impact of financial and multinational operations on manager perceptions of dividends. global finance journal, 19(2), 171–186. https://doi.org/10.1016/j.gfj.2007.11.002 baker, h. k., & weigand, r. (2015). corporate dividend policy revisited. managerial finance, 41(2), 126–144. https://doi.org/http://dx.doi. org/10.1108/mf-03-2014-0077 chang, k., kang, e., & li, y. (2016). effect of institutional ownership on dividends: an agency-theory-based analysis. journal of business research, 69(7), 2551–2559. dandago, k. i., farouk, m. a., & muhibudeen, l. (2015). corporate shareholding structure and dividend payout ratio of listed chemical and paint companies in nigeria. applied finance and accounting, 1(2), 47–54. https://doi. org/10.11114/afa.v1i2.789 deangelo, h., deangelo, l., & stulz, r. m. (2006). dividend policy and the earned/contributed capital mix: a test of the life-cycle theory. journal of financial economics, 81(2), 227–254. https://doi.org/10.1016/j. jfineco.2005.07.005 easterbrook, f. h. (1984). two agency-cost explanations of dividends. american economic review, 74(4), 650–659. https://doi.org/10.2307/1805130 farinha, j. (2003). dividend policy, corporate governance and the managerial entrenchment hypothesis: an empirical analysis. journal of business finance and accounting, 30(9–10), 1173–1209. https://doi.org/10.1111/ j.0306-686x.2003.05624.x farooq, o., & ahmed, n. (2019). dividend policy and political uncertainty: evidence from the us presidential elections. research in international business and finance, in press. https://doi.org/10.1016/j. ribaf.2019.01.003 florackis, c., kanas, a., & kostakis, a. (2015). dividend policy, managerial ownership and debt financing: a non-parametric perspective. european journal of operational research, 241(3), 783–795. https://doi. org/10.1016/j.ejor.2014.08.031 francis, b. b., hasan, i., john, k., & song, l. (2011). corporate governance and dividend payout policy: a test using antitakeover legislation. financial management, 40(1), 83–112. https://doi.org/10.1111/j.1755053x.2010.01135.x gedajlovic, e., yoshikawa, t., & hashimoto, m. (2005). ownership structure, investment behaviour and firm performance in japanese manufacturing industries. organization studies, 26(1), 7–35. https://doi. org/10.1177/0170840605046346 42 the international journal of banking and finance, vol. 14, 2018-2019: 27-44 grinstein, y., & michaely, r. (2005). institutional holdings and payout policy. journal of finance, 60(3), 1389–1426. https://doi.org/10.1111/j.15406261.2005.00765.x gujarati, d. n. (2004). basic econometrics. mcgraw-hill international (economic s). new york: mcgraw-hill international. https://doi. org/10.2307/2344828 hosmer, d. w., lemeshow, s., & sturdivant, r. x. (2013). applied logistic regression (3rd ed.). hoboken, new jersey: john wiley & sons, inc. hsu, g. c. m., & koh, p. s. (2005). does the presence of institutional investors influence accruals management? evidence from australia. corporate governance, 13(6), 809–823. https://doi.org/10.1111/j.14678683.2005.00472.x hu, a., & kumar, p. (2004). managerial entrenchment and payout policy. journal of financial and quantitative analysis, 39(04), 759. https://doi. org/10.1017/s0022109000003203 huda, n., & abdullah, m. n. (2014). relationship between ownership structure and dividend policy: empirical evidence from chittagong stock exchange. world review of business research, 4(3), 14–34. idris, a. a., ishak, r., & hassan, n. l. (2017). is there relationship between board structures and dividend policy: evidence from nigeria. journal of advanced research in business and management studies, 9(1), 10–20. jacob, c., jijo, j., & lukose, p. (2018). institutional ownership and dividend payout in emerging markets: evidence from india. journal of emerging market finance, (17), 1–29. https://doi.org/10.1177/0972652717751538 jensen, m. c., & meckling, w. h. (1976). theory of the firm : managerial behavior, agency costs and ownership structure. journal of financial economics, 3(4), 305–360. https://doi.org/http://dx.doi. org/10.1016/0304-405x(76)90026-x jeon, j. q., lee, c., & moffett, c. m. (2011). effects of foreign ownership on payout policy: evidence from the korean market. journal of financial markets, 14(2), 344–375. https://doi.org/10.1016/j.finmar.2010.08.001 jiraporn, p., kim, j. c., & kim, y. s. (2011). dividend payouts and corporate governance quality: an empirical investigation. financial review, 46(2), 251–279. https://doi.org/10.1111/j.1540-6288.2011.00299.x jo, h., & pan, c. (2009). why are firms with entrenched managers more likely to pay dividends? review of accounting and finance, 8(1), 87–116. jory, s. r., ngo, t., & sakaki, h. (2017). institutional ownership stability and dividend payout policy. managerial finance, 43(10), 1170–1188. kapopoulos, p., & lazaretou, s. (2007). corporate ownership structure and firm performance: evidence from greek firms. corporate governance: an international review, 15(2), 144–158. https://doi.org/10.1111/j.14678683.2007.00551.x karpavičius, s. (2014). dividends: relevance, rigidity, and signaling. journal of corporate finance, 25, 289–312. https://doi.org/10.1016/j. jcorpfin.2013.12.014 decision to pay dividens and ownership structures in nigeria : 27-44 43 liljeblom, e., & maury, b. (2016). shareholder protection, ownership, and dividends: russian evidence. emerging markets finance and trade, 52(10), 2414–2433. https://doi.org/10.1080/1540496x.2015.1073991 mancinelli, l., & ozkan, a. (2006). ownership structure and dividend policy: evidence from italian firms. the european journal of finance, 12(3), 265–282. https://doi.org/10.1080/13518470500249365 manos, r. (2003). dividend policy and agency theory: evidence on indian firms. south asia economic journal, 4(2), 275–300. miko, n. u., & kamardin, h. (2015). ownership structure and dividend policy of conglomerate firms in nigeria. academic journal of interdisciplinary studies, 4(2), 279–286. pampel, f. c. (2000). logistic regression: a primer. sage university series on quantitative applications in social sciences. thousand oaks, ca: sage publications. pucheta-martínez, m. c., & bel-oms, i. (2016). the board of directors and dividend policy: the effect of gender diversity. industrial and corporate change, 25(3), 1–25. https://doi.org/10.1093/icc/dtv040 pucheta-martínez, m. c., & lópez-zamora, b. (2017). how foreign and institutional directorship affects corporate dividend policy. investment analysts journal, 46(1), 44–60. https://doi.org/10.1080/10293523.2016 .1253136 rizqia, d. a., & aisjah, s. s. (2013). effect of managerial ownership, financial leverage, profitability, firm size, and investment opportunity on dividend policy and firm value. research journal of finance and accounting, 4(11), 120–130. rozeff, m. s. (1982). growth, beta and agency costs as determinants of dividend payout ratios. journal of financial research, 5(3), 249. https://doi. org/10.1111/j.1475-6803.1982.tb00299.x sanda, a. u., mikailu, a. s., & garba, t. (2010). corporate governance mechanisms and firms’ financial performance in nigeria. afro-asian j. of finance and accounting, 2(1), 22–39. https://doi.org/10.1504/ aajfa.2010.035193 sec nigeria. code of corporate governance (2011). sec nigeria. securities and exchange commission rules and regulation (2013). sharma, v. (2011). independent directors and the propensity to pay dividends. journal of corporate finance, 17(4), 1001–1015. https://doi. org/10.1016/j.jcorpfin.2011.05.003 short, h., zhang, h., & keasey, k. (2002). the link between dividend policy and institutional ownership. journal of corporate finance, 8, 105–122. sindhu, m. i., hashmi, s. h., & haq, e. u. (2016). impact of ownership structure on dividend payout in pakistani non-financial sector. cogent business & management, 15(1), 1–11. https://doi.org/10.1080/23311975.2016.1272 815 ullah, h., fida, a., & khan, s. (2012). the impact of ownership structure on dividend policy evidence from emerging markets kse-100 index 44 the international journal of banking and finance, vol. 14, 2018-2019: 27-44 pakistan. international journal of business and social science, 3(9), 298–307. victoravich, l. m., xu, p., & gan, h. (2013). institutional ownership and executive compensation: evidence from us banks during the financial crisis. managerial finance, 39(1), 28–46. https://doi. org/10.1108/03074351311283559 williams, r. (2012). using the margins command to estimate and interpret adjusted predictions and marginal effects. the stata journal, 12(2), 308– 331. . us monetary-fiscal policy mix evidence from a quartovariate vecm 40 international journal of banking and finance, volume 8 (number 2) 2011: pages 40-67 us monetary-fiscal policy mix evidence from a quatrovariate vecm george k. zestos, andrew n. geary, and kevin s. cooksey christopher newport university, indiana university and john hopkins university, united states ____________________________________________________________ abstract this study investigates the effectiveness of monetary and fiscal policies in the us by employing cointegration and a quatrovariate vector error correction model together with granger causality tests. two models are estimated: (i) nominal national income, the ten-year government bond yield, and two policy variables, the federal government deficit and the federal funds rate; (ii) real national income, and the other same three variables. monetary and fiscal policies are jointly ineffective in influencing nominal national income. however, monetary and fiscal policies are jointly effective in influencing real national income. in contrast to the first model, only monetary policy was found to be reactive to changes in real national income and the longterm interest rate. the asymmetric responses of the two policies to changes in real economic activity are attributed to the fact that monetary policy is much more efficient in promptly responding to changes in economic conditions than fiscal policy. keywords: fiscal policy mix, vecm, granger causality, real vs nominal income, interest rates, fed deficit, fed rate jel classification: e12, e51 _____________________________________________ 1. introduction economists are interested in knowing how monetary and fiscal policies influence economic activity. for this reason, they construct economic models analyzing such relations. most economic models, however, focus only on one of the two policies, assuming the other remains passive or neutral. this approach is inadequate because often both policies are assigned an zestos et al.: monetary-fiscal policy 41 active stance as countries pursue either expansionary (easy) or contractionary (tight) policies. the simultaneous employment of the two policies is prior research on the indonesian economy used stochastic frontier analysis (sfa) for evaluating a firm’s performance but on other than manufacturing sectors: agriculture (see daryanto, battese, and fleming (2002)) on technical efficiencies of rice farmers in west java; public and private referred to, in the economic literature, as monetary-fiscal policy mix or policy mix. the purpose of this paper is to investigate the joint effectiveness of us monetary and fiscal policies. the study also examines the response of the federal reserve system (fed) and the us fiscal authorities to changes in nominal and real national income. the project constitutes an empirical investigation based on cointegration and the estimation of a quatrovariate vecm. the estimated vecm is employed to test for granger econometric causality among four macroeconomic variables: a proxy for economic activity, a long-term interest rate and two policy variables, a short-term interest rate, and the us federal deficit. the results of the study are in agreement with studies which found fiscal policy to be less reactive in influencing output than monetary policy and can create contractionary ricardian effects. the paper presents evidence that the two policies jointly generate positive effects on real national income but do not affect nominal national income. the empirical results support the view that demand management policies influence the mean real gdp of the us economy. this is in contrast to the natural rate hypothesis model which supports that demand policies affect only the variance of real gdp. 1 in addition, the model examines the reaction of each policy to changes in economic activity, long-term interest rates, and shifts in the other policy. the empirical results are supported and reinforced by diagnostics such as cusum and cusum of squares tests. furthermore, impulse response functions and variance decomposition are utilized to project the separate long-run effects of monetary (fiscal) policy on nominal and real national income, twenty years into the future and beyond. the paper has shown that evidence exists to indicate the lack of fiscal policy efficacy. particularly, this study shows that fiscal policy in the us, as it was conducted in the 1955-2006 period, generated contractionary effects on real national income. such findings call for the revamping and re-launching of fiscal policy in the us in order to ensure that it consistently generates positive effects on the economy. 1 see de long , summers, mankiw and romer (1988). international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 42 this paper is organized as follows. in section 2, four episodes of policy mix from recent us history are discussed. section 3 discusses the relevant economic literature. section 4 describes the data, presents the stability properties of the variables and discusses the methodology of the paper. the empirical analysis and the results of the paper are presented in section 5. section 6 concludes the paper. 2. four episodes of us monetary fiscal policy mix president reagan’s administration adopted an expansionary fiscal policy during the first half of the 1980s; it pursued this policy by increasing spending and reducing taxes. during this period, the fed applied contractionary monetary policy by raising interest rates to curtail inflation. high interest rates attracted foreign capital inflows to the us, causing a dollar appreciation that consequently generated large current account deficits. this episode became known as the story of the us twin deficits, since many economists considered the two deficits highly interdependent. another important episode of fiscal-monetary policy mix in contemporary us history occurred during the clinton-greenspan period. in the 1990s, president clinton’s administration, with the cooperation of a republican dominated congress, was successful in reducing large and chronic government budget deficits. this was achieved with an accommodating expansionary monetary policy applied by the fed and its chairman, alan greenspan. such a policy-mix is associated with the longest business cycle expansion in the us history from 1991 to 2001, which was responsible for reducing the us unemployment rate down to four percent and the inflation rate close to two percent. as a result of this policy-mix, the us was able to generate federal budget surpluses for the years 1998-2001. another episode of policy-mix occurred when the fed, along with the us president and congress, responded to three negative shocks that struck the us at the turn of the 21 st century. economists identified three distinct events (shocks) responsible for the subsequent us zestos et al.: monetary-fiscal policy 43 recession. 2 the fed’s response to the three shocks was a drastic reduction in the target federal funds interest rate (rff), driving it down to one percent, a 42 year record low. such an aggressive expansionary monetary policy was accompanied by an expansionary fiscal policy. the us government reduced taxes and increased government spending to cope with the anticipated recession and finance the two wars that the us was simultaneously fighting in afghanistan and iraq. during the 2001-2003 period, expansionary monetary policy was very effective. it boosted consumption of automobiles, as automakers were able to promote sales by offering zero percent interest rates to finance purchases of new vehicles. expansionary monetary policy mainly contributed to the reduction of the long-term interest rates, inducing families to purchase homes, refinance existing homes, or borrow on existing mortgages and spend on a variety of other items. the expansionary monetary-fiscal policy mix during the 2002-2003 period was successful in reducing interest rates and simultaneously increasing stock and home prices beyond sustainable levels. a booming stock market and persistently increasing home prices generated a wealth effect, boosting consumption that led the us out of the recession. the consumer euphoria triggered by the monetary-fiscal policy mix, that led the us out of the 2002-2003 recession, is considered to be one of the main causes of the recent us mortgage financial crisis. the rationale of this explanation is based on the fact that a very expansionary us policy-mix during this period created a bubble in the housing and stock markets which encouraged and triggered drastic increases in household indebtedness. the most recent episode of us monetary-fiscal policy mix, the response to the us subprime mortgage crisis, began unfolding in 2006. this crisis has already caused millions of american families to lose their homes. furthermore, the crisis is far from over, as it still poses a serious threat to an unprecedented number of home-owners. the us subprime mortgage crisis has had detrimental effects on the entire us economy and has been transmitted to foreign 2 the abrupt decline in information technology investment and consumption, the corporate scandals of this period, and the september 11, 2001 terrorist attacks are considered the main causes of the recession. the three events created uncertainty and contributed to the formation of negative expectations regarding the future of the us economy. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 44 countries. during this period both monetary and fiscal authorities adopted an extraordinary expansionary stance. the us government, under the administration of george bush, adopted a massive fiscal plan to bail out several financial institutions at an initial estimated cost of 700 billion dollars. furthermore, on february 17, 2009, a new fiscal stimulus bill of 787 billion dollars was signed into law by president barack obama. the fed’s response to the crisis was swift, as it reduced the rff several times. on december 16, 2008, it reduced the rff to an unprecedented record low within the range of zero to 0.25 percent. it also reduced the discount rate to .50 percent. 3 3. literature review empirical studies investigating the relevance and importance of the rff as a monetary policy instrument in the us differ from study to study. for example, cook and hahn (1989) found the rff very influential in affecting other interest rates. similarly, bernanke and blinder (1992) have shown that the rff is a good indicator of monetary policy. a few other studies show that rff is not superior to other short-term interest rates. garfinkel and thornton (1995), for example, found the rff not to contain unique information regarding the conduct of monetary policy. kydland and prescott (1977) were the first to point out that the central banks that are not accountable for their conduct of monetary policy in their effort to increase output in the shortrun can fall victim to what is known as the time-consistency problem. this means that a highly expansionary monetary policy, aiming to boost economic activity in the shortrun, can trigger long-run inflationary pressure in the economy. consequently, a countercyclical monetary policy stance aimed at fighting a recession will inadvertently raise prices and nominal, long-term 3 european central banks followed the fed’s lead and they slashed their respective key interest rates several times. for example, the bank of england brought the bank rate to 0.5 percent, the lowest ever since its establishment in 1694. similarly, the ecb reduced its repo rate to 1 percent, the lowest since its establishment in 1998. the fed and the bank of england adopted unorthodox monetary policies (quantitative easing) to provide liquidity to their economies. similar policy was introduced by the ecb, in may 2010, which began buying national bonds of the weaker eurozone economies in order to protect these economies and save the euro. zestos et al.: monetary-fiscal policy 45 interest rates. as a result, an easy monetary policy can trigger inflation that may emerge during the expansionary phase of the business cycle. most of the early empirical term structure studies of interest rates have employed singleequation econometric estimation models to examine relations between short and long-term interest rates. 4 a few exceptions, however, exist in the literature, such as harvey (1997) and clarida and friedman (1983). the latter employed both short and long-term interest rates within the framework of an is-lm model. turnovsky (1989) criticized the term structure theory of interest rate studies because the vast majority of them employed partial equilibrium models, focusing only on the relation of shortand long-term interest rates in the absence of other macroeconomic variables. instead, turnovsky presented a stochastic macroeconomic model that included, in addition to the short and long-term interest rates, several other macroeconomic variables. brimmer and sinai (1986) used simulation analysis to study projections of the grammrudman-hollings act. this legislation aimed to promote fiscal discipline in the us. simulation analysis was also employed by ribe and beeman (1986) to investigate the long-run joint effects of a monetary and fiscal policy. some studies utilized the vector autoregression (var) model to study the effects of monetary and fiscal policies on the economy using time series variables. fackler and mcmillin (1989) employed a var model to investigate possible effects of federal debt on the us macroeconomy. using impulse response functions and variance decomposition, the two authors concluded that debt is not perceived as wealth. this finding is related to a topic known in the literature as ricardian equivalence. about two centuries ago, david ricardo observed that increases in government deficits financed with government bonds can trigger equivalent increases in private savings to enable taxpayers to pay the higher expected future taxes. under this scenario expansionary fiscal policy becomes neutral, implying that government bonds held by the public are not perceived as 4 these reduced form models, however, are valid only if the right-hand side variables are exogenous. it is possible for a right-hand side variable to also depend on the left-hand side variable or for a third variable which was omitted from the model to affect both variables in the equation. in these cases, the reduced form model is incorrectly specified, and thus cannot be employed to analyze the relationship of the term structure of interest rates. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 46 wealth. 5 this proposition has been extensively debated and investigated in empirical studies. most authors are convinced that ricardian equivalence effects exist, but no consensus is provided in the literature regarding the magnitude of these effects. recently, many macroeconomists hold the view that a contribution by john taylor, known as the taylor rule, accurately describes how monetary policy should be applied. the taylor rule suggests that monetary policy conducted by central banks is best described by a reaction function of the key interest rate to deviations of the actual from the potential output and the actual from the target inflation rates. after his (1993) multi-country study, taylor proposed that policy-makers following such a rule would achieve price and output stability. many studies, however, incorporate fiscal policy within one of the various versions of taylor’s rule. 6 recent advances in time series methodologies, such as cointegration, var and vecms, are currently very popular. kuttner (2002) employed a quatrovariate var model that included gdp, inflation, the rff, and a fiscal variable, which measured either with the actual government surplus or the structural primary balance. kuttner found that both fiscal variables are informative in predicting gdp 7 . kuttner (2002) also found that fiscal policy does not affect the rff; this finding allowed him to conclude that us monetary policy was conducted without constraint from fiscal policy. it is plausible that a country running significant chronic government deficits will accumulate such a large national debt that its central bank will find it necessary to monetize national debt. in this case, the central bank will lose control of the country’s inflation rate. a group of models, known as fiscal theory of price level (ftpl), deal with this issue. 8 granger causality tests performed within the framework of a var or a vecm reveal relations among variables in terms of strength of statistical significance and direction of causation. when a var or a vecm is employed, no explicit functional form relation of the 5 see barro (1974) 6 gonzalez and martinez (2003) for example, proposed two different rules, one for monetary policy and one for fiscal policy. 7 the results, however, were weak since the level of the statistical significance was only 10 percent. 8 see leeper (1991,2010), sims (1994), beetsma (2008). the main idea of these models was introduced by sargent and wallace (1983) zestos et al.: monetary-fiscal policy 47 variables is a priori required; therefore model specification error is avoided. 9 as the vecm employs only cointegrated variables, this suggests that standard asymptotic theory is applicable for statistical inference. a vecm permits a short-run dynamic adjustment following a shock as all variables return to their long-run values, according to the cointegration equation. 10 to study the effects of us and canadian monetary policies on nominal and real gdp and the reaction of the central banks to changes in the two variables, zestos et al. (2010), employed the methodology of cointegration and the estimation of a trivariate vecm. performance of granger causality tests, within the framework of estimated vecms, led the authors to conclude that both banks are effective in influencing proxies for nominal and real gdp. both central banks, however, were found to be either passive or procyclically reactive to real business cycle activity. claeys (2007) using the var approach and incorporating an explicit intertemporal budget constraint, analyzed the effect of us fiscal policy on employment and inflation. such a model forces the estimated parameters to take values that are consistent with a sustainable national debt. claeys found that us expansionary fiscal policy generated contractionary effects on us output and inflation. these results can be characterized as negative ricardian effects as they arise from the public’s expectation of higher future taxes. beetsma (2008), however, who surveyed relevant literature on the effects of fiscal policy, concluded that most studies provide evidence that expansionary fiscal policy has positive effects on output and consumption but negative on trade balance. 9 the method of cointegration produces an explicit estimated equation which depicts a long-run stable linear relation among these variables. according to engle and granger (1987) cointegrated variables always have an error correction representation. 10 futhermore, granger (1988) proved that if two variables are cointegrated, causality will always exist between them in at least one direction. starting with engle and granger (1987), the methodology of cointegration and the vecm were employed by several authors. johansen (1991, 1995) extended the cointegration technique to many variables. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 48 4. data description and methodology 4.1 data description a us time series data set was constructed for the period 1955-2006, consisting of 52 annual observations. annual data were chosen for this study because deficits (surpluses) are reported annually. in addition, annual data are not affected by short-run transitive and seasonal shocks, thus the analysis focuses exclusively on the fundamentals. 11 the data set includes four time series variables: a proxy for economic activity, a long-term interest rate, a monetary policy variable (a short-term interest rate), and a fiscal policy variable. the natural logarithms of nominal and real national incomes, lnni and lnrni, respectively, are employed as proxies for economic activity. 12 the ten-year us treasury government bond yield, rtb10, was utilized for the long-term interest rate. for the monetary policy variable, we employed the overnight, effective federal funds rate rff. the fed sets the target rff, which for many years has emerged as the most important monetary operating instrument employed by the fed for the conduct of monetary policy. 13 through open market operations the fed maintains the market or effective rff close to the target rate. a long-term interest rate is also included in the vecm because economists are convinced that monetary policy is mainly transmitted through its impact on long-term interest rates. fiscal 11 see nunes-correia and stemitsiotis (1993), and beetsma (2008). we purposely left out from the sample observations beyond 2006 since it is certain that such observations are outliers as both us monetary and fiscal policies were exceptionally expansionary causing a major structural break. 12 note that several proxies of nominal gdp and output were employed in the preliminary study all other proxies, however, either did not possess the required stability properties or were not cointegrated with the other variables of the vecm. 13 an exception to this policy occurred during the three year period of october 1979 to october 1982, a period during which the fed tried to control unborrowed reserves in its effort to reduce inflation. although rff was not always the main policy instrument during the entire sample period, rff was always considered as an informative indicator of the monetary policy stance. for practical purposes we will refer to both the target federal funds rate and the effective federal funds rate as rff. zestos et al.: monetary-fiscal policy 49 policy also affects interest rates because as the us treasury changes the supply of government bonds to finance deficits, or to reduce public debt, it affects interest rates. two different measures of the government deficit were utilized, the federal budget deficit (surplus) to gdp ratio bd, and the total deficit to gdp ratio, td. the latter includes the off budget items, thus providing a more accurate response of fiscal policy to business cycle conditions. the two fiscal measures, bd and td, consist of both discretionary revenues and spending along with government revenues and spending triggered because of changes in the business cycle. a few authors employ the structural balance as the fiscal policy variable. this is calculated by simply subtracting the cyclical component from the deficit (surplus) balance. the rationale for employing the structural fiscal balance is that the most relevant fiscal measure should include only the balance determined by the government as a result of discretionary policy. 14 data for the rff and the rtb10 were obtained from the federal reserve bank of st louis. the source of nominal national income and real national income is the bureau of economic analysis (bea). the two deficits to gdp ratios are from the economic report of the president. 4.2 stability properties of the time series variables prior to testing for evidence of cointegration among a set of time series variables, it is required that all variables are integrated of the same order. this requires that the stability properties of the variables must be investigated. the performance of the unit root test for a variable y is demonstrated below, in equation (1).  titi 1=i -t2t ++ y k y+t+=y  110 14 the remaining component of the fiscal balance is known as cyclical, and is captured by the automatic stabilizers. the congressional budget office, starting in 1962, produces the structural budget balance. the two fiscal measures employed in this study, bd, and td¸ exceed the span of the structural deficit measure by eight annual observations. (1) international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 50 where, βo, β1, β2 and  1 ,  2 … k are parameters to be estimated, and  t is the error of the regression assumed to be stationary with a zero mean and a constant variance. in table a (see appendix) the unit root results of all the time series variables are reported. according to the adf test, the vast majority of the results meet the stability requirements for the variables to be employed in the cointegration tests. the time series variables are non stationary in levels but stationary in first differences. all unit root results, however, according the phillips-perron (pp) test indicate that the hypothesis of nonstationarity cannot be rejected for the levels of all variables, but it can be rejected for the first differences 15 . since these variables are integrated of the same order we proceed to test for cointegration and the estimation of the quatrovariate vecm presented in section 4.3 below. 4.3 the vector error correction model (vecm) equations (2)-(5) below represent a quatrovariate vecm in four endogenous variables x, y, z and w. the variable x represents the two proxies for nominal or real gdp which is measured with the natural logarithms of nominal and real national incomes denoted by lnni and lnrni, respectively 16 . variable y represents the long-term interest rate, rtb10. variables z and w represent the two policy variables; z represents the fiscal policy variable measured either with the budget deficit or with the total deficit to gdp ratios, bd and td respectively. lastly, w represents rff, the monetary policy variable. a quatrovariate vecm  titi =1i i-t1i =1i i-t1i =1i i-t1i =1i 1-txt +++ w p z k y s +x r +t+x 1110 1111    (2) 15 the pp unit root test is considered more reliable than the adf test since it is robust in the presence of serial correlation and time dependent heteroskedasticity. 16 two vecms are estimated; the first employing the nominal national income, lnni, and the variables, rtb10, rff, and td, whereas the second vecm uses the real national income, lnrni, with the variables, rtb10, rff, and bd instead of td. zestos et al.: monetary-fiscal policy 51  2titi =1i i-ti =1i i-t2i =1i i-t2i =1i 1-tyt +++ w p z k y s +x r +t+=y   2 2 220 222 (3)  titi =1i i-ti =1i i-ti =1i i-ti =1i 1-tzt +++ w p z k y s +x r +t+=z 33 3 33330 333    (4)  titi =1i i-ti =1i i-ti =1i i-ti =1i 1-twt +++ w p z k y s +x r +t+=w 44 4 4 4 4 4 440 4    (5) the left-hand side variables of the vecm presented in equations (2) – (5) are expressed in terms of first differences. the right-hand side of each equation include an optimum number of lagged differences for each of the four variables. in addition to the lagged differences, each equation includes the one-period lagged error term of the cointegrating equation θt-1 as a right-hand side variable. furthermore, each equation includes a constant represented by α0, β0, γ0, and δ0 , in each of the four equations respectively and a time trend variable denoted by t, where, α1, β2, γ3, and δ4 are the constant coefficients of the time trend variable in each equation. a random error term  it , where i=1,2,3,4, is assumed to be a white noise, is also included in every equation. 4.4 granger causality tests three different types of granger causality tests are performed in every one of the four equations of the quatrovariate vecm. 17 a test for long-run non-causality is carried out by testing the null hypothesis that the coefficient of θt-1 is zero. this test examines whether the three right-hand side variables in each equation, which are cointegrated with the left-hand side variable, granger, cause this variable. in equation (2), this test is performed by testing the null hypothesis αx=0, versus the alternative αx ≠ 0. four short-run granger causality tests are also performed in each equation. three of these tests are carried out by setting the coefficients of all order-lagged differences of each of the right-hand side relevant variables equal to zero. for example, in equation (2) a test for short-run non-causality from y to x is carried out by testing whether the coefficients of the lagged differences of the y variable are all equal to zero, i.e., β11 = β12 = …..=β1s1 = 0. this is a wald f 17 based on granger (1988). international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 52 test for short-run non-causality from y to x since the lagged coefficients of y capture the shortrun dynamic adjustment of the system. a similar test for short-run-non-causality from z and w to x is also performed. lastly, a test for overall causality from each right-hand side variable to the left-hand side variable is performed by setting all the coefficients of each right-hand side variable and the coefficient of θt-1 jointly equal to zero. the null hypothesis for overall non-causality from y to x for example, is stated as αx = β11 = β12 = …….β1s1=0. similar tests are performed for the other two right-hand side variables, in each of the four equations. 5. empirical analysis 5.1 an estimated quatrovariate us vecm with nominal national income in this section, a quatrovariate vecm is estimated using natural logarithms of nominal national income, lnni, the ten year government bond yield, rtb10, the total deficit to gdp ratio, td, and the federal funds rate, rff. table 1: johansen cointegration test cointegration tests were performed according to johansen (1991, 1995) methodology. the number of cointegrating vectors was determined using the two maximum likelihood tests λmax and λtrace. according to table 1 above, there were three cointegrating vectors based on trace and two cointegrating vectors based on max. equation (6) below is the first cointegrating vector of the variables lnni, rtb10, td, and rff. lnni = 5.307445+1.271532 rtb10 + 1.419449 td – 0.185706rff +t t-stat (3.2) (1.85) (4.8) (-.30) (6) h0 eigenvalue max critical value 5% p-value trace critical value 5% p-value r = 0 0.57 38.02 28.58 0.00 97.42 54.07 0.00 r ≤ 1 0.54 35.83 22.29 0.00 59.40 35.19 0.00 r ≤ 2 0.24 12.63 15.89 0.15 23.57 20.26 0.02 zestos et al.: monetary-fiscal policy 53 the cointegration equation above indicates that lnni is positively related to rtb10 and td, but negatively to the rff. the negative relation between lnni and rff is plausible because the fed conducts monetary policy by reducing rff to induce increases in consumption and investment and to raise national income. on the contrary, the fed increases rff to discourage private spending and prevent the economy from overheating. cointegrating equation (6) also shows that td is positively related to lnni 18 . this relation indicates that an expansionary fiscal policy is associated with reduced national income. on the other hand a contractionary fiscal policy increases lnni. this finding is in line with the neoclassical view that expansionary fiscal policy leads to rising interest rates that crowds out private investment and generates contractionary fiscal ricardian effects. according to cointegrating equation (6), lnni and rtb10 are positively related. the comovement of lnni and rtb10 is empirically supported by time series data on business cycles activity and interest rates. during expansionary periods, economies experience both increases in prices and nominal interest rates. an explanation for this positive relationship is that during the expansionary phase of the business cycle, investors request a higher nominal interest rate to maintain a constant real rate of return. this is achieved by adding an inflation premium to rtb10. the positive relation between lnni and rtb10 is also explained by focusing on the bond market. during expansionary periods, the supply of bonds increases as governments and businesses choose to raise capital to finance various investment projects. an increase in the supply of bonds, ceteris paribus, leads to higher interest rates. 19 the estimated vecm is presented below in table 2 using the first cointegrating vector. 20 table 2 above presents the estimated vecm for lnni, rtb10, td and rff. according to equation (7), monetary and fiscal policies are jointly ineffective in influencing nominal 18 note that td is defined as government revenue minus expenditures to nominal gdp. this ratio is negative for the vast majority of the observations in the sample. in this model the total deficit to gdp ratio (td) was employed since td was cointegrated with the other three variables. 19 it is theoretically possible, however, for rtb10 not to increase if the demand for bonds increases proportionately with the supply of bonds, or if the demand for bonds is perfectly elastic with respect to the interest rate. 20 the vecm, however, was estimated with two and three cointegrating vectors, respectively. the results of these vecms were not economically plausible and could not be interpreted in any meaningful way; therefore, these results are not reported. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 54 national income. this result is supported because there is a lack of statistical evidence for long-run causality from the three variables rff, rtb10 and td to lnni. table 2: an estimated vecm with nominal national income (lnni) ***, **,* denote significance at 1%, 5%, and 10% levels, respectively. the figures in parentheses and brackets are t and f statistics, respectively. this is indicated by the statistically insignificant coefficient of θt-1 (0.005). similarly, a lack of statistical evidence exists for short-run and overall causality from each of the three right-hand side variables to lnni. this is shown by the low values of the f-statistic, below the summation terms of the three variables, and by the low f statistic values under the columns f2, f3, and f4. the latter three tests indicate a lack of overall granger causality, from any of the three variables to lnni. equation (8) indicates that no statistical evidence of any type of granger causality is statistically supported from lnni, rff , and td to rtb10. strong statistical evidence for long-run causality, however, is supported from lnni, rtb10, and rff to td in equation (9). this result indicates that fiscal authorities are reactive to changes in nominal national income and changes in the two interest rates. it also indicates that the fiscal authorities counter-cyclically (correctly) react to changes in national income. equation (9) further indicates that statistical evidence exists for short-run and overall causality from lnni to td at a one percent level of significance. such results indicate a very responsive fiscal policy in lnni. equation (10) supports statistical evidence only at a ten percent level of significance for long-run causality from lnni, rtb10 and td to rff. this finding suggests that the fed mildly reacts to changes in lnni, rtb10, and td. no statistical evidence, however, exists for short-run and overall dependent variable t-1    1 1 ln r i itni f1    1 1 10 s i tb itr f2    1 1 k i ittd f3     1 1 p i ff it r f4 (7) lnni .005 (1.14) 1.18 [10.6]*** 67.1*** .028 [.57] .56 .006 [.30] .46 -.01 [.87] .94 (8) rtb10 .25 (1.19) 8.90 [.63] .54 .12 [.59] .76 .006 [.25] .51 .78 [1.54] 1.7 (9) td .68 (4.59)*** 22.6 [6.6]*** 6.2*** 1.22 [1.87] 5.87*** 1.82 [4.72]*** 4.9*** .704 [1.83] 5.27*** (10) rff .75 (1.86)* 25.5 [1.11] .98 4.69 [.78] .84 2.14 [.69] .71 -1.73 [.67] 1.25 zestos et al.: monetary-fiscal policy 55 causality from lnni, rtb10 and td, to rff. these statistical tests suggest that the fed does not react to any short-run changes in any one of the three variables. the finding that rff is not granger caused by td is also supported by kuttner (2002) who found that fiscal policy does not affect rff. we conclude that us monetary policy during the sample period of 1955-2006 was exercised without constraint by fiscal policy. this is a plausible result because this period is characterized by fiscal stability as it is supported by historical us data of deficit and national debt to gdp ratios. by observing cointegrating equation (6) in conjunction with the vecm equation (7) it can be inferred that fiscal policy is very effective in influencing lnni, and it reacts countercyclically to changes in national income. similar claims, however, cannot be made for monetary policy by examining cointegration equation (6) in conjunction with the estimated equation (7) of the vecm. in this case, monetary policy is pro-cyclical 21 . to examine evidence of possible structural breaks in the model, the cusum and the cusum of squares tests were performed in each of the four equations. according to these two tests there is no evidence for structural break(s) in seven out of the eight tests performed in the four equations of the vecms 22 . despite the fact that fiscal policy is exercised counter-cyclically in regard to nominal national income (lnni), its long-run effects are contractionary. this finding was based on the results of the impulse response function (see figure 1 below). according to the generalized impulse response function, one standard deviation innovation on td reduces lnni in the first 20 years but afterwards lnni oscillates with cycles of increasing phase and amplitude. 23 the long-run effects of monetary policy, according to the impulse response function, on nominal income are positive, increasing lnni by approximately one percent. examining the variance decomposition of lnni, fiscal policy as measured with td accounts for a greater percentage contribution in affecting lnni than rff. fiscal policy is responsible for a greater contribution than monetary policy in affecting all other variables, other than rff, which are not reported in this study. 21 these results, however, are discounted because the coefficient of t-1 is statistically insignificant. 22 according to the cusum of squares test, a minor structural break was revealed in equation (10). 23 the graph of the impulse response function beyond 20 periods is not shown in figure 1. the cycles, however, appear when the impulse response function graph is increased to 100 periods (years). international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 56 figure 1: impulse response function and variance decomposition of lnni 5.2 an estimated quatrovariate vecm with real national income (lnrni) we now proceed with the estimation of the quatrovariate vecm, which employs the log of real national income (lnrni), rff, rtb10, and the budget deficit to gdp ratio, bd 24 . table 3 below reports the results of the johansen (1991, 1995) cointegration test. according to trace and max, there exists one cointegrating vector at the 1% level of significance. a second cointegrating vector is supported only by trace at the 10% (p=.09) level of significance. consequently, the vecm is estimated using only one cointegrating vector. 25 equation (11) below is the cointegrating equation of lnrni, the two interest rates rtb10, rff, and the budget deficit bd. lnrni = 7.706834+.033236*trend-.139390*rtb10-.035089*bd+.119452*rff + t t-stat (-7.46) (-4.34) (8.1) 24 replacing nominal national income with real national income in the vecm makes it possible to abstract from price effects (inflation) and focus on the real effects of monetary and fiscal policy. 25 the estimated vecm with the two cointegrated vectors did not produce meaningful results. response to generalized one s.d. innovations on lnni variance decomposition of lnni due to td and rff % (11) zestos et al.: monetary-fiscal policy 57 table 3: johansen cointegration test according to the cointegration equation (11), lnrni is inversely related to rtb10. this result is a theoretically correct behavioral relation because a decline in rtb10 induces increases in investment and consumption of durable goods and purchases of homes 26 . the negative relation of lnrni and rtb10, however, is exactly opposite the relation between nominal national income (lnni) and rtb10 obtained from the cointegrating equation (6) of the first estimated vecm. it is highly likely that the positive relation between lnni and rtb10 in cointegration (6) is driven by the presence of inflation, as both lnni and rtb10 are expressed in nominal terms. in cointegration equation (11), lnrni is inversely related to the government budget deficit, bd. an expansionary fiscal policy according to this relation increases lnrni, whereas a contractionary fiscal policy reduces lnrni. the relation between lnrni and bd is also opposite to the one between lnni and td as discussed in the previous vecm. as nominal and real national incomes are not perfectly correlated, it is possible that both relations are valid. cointegration equation (11) also indicates that rff is positively related to lnrni. 27 this relation, although counter-intuitive, is still plausible because the fed does not respond to changes in real national income, as the latter is not observed when the fed applies monetary policy through changes in rff. a countercyclical expansionary monetary policy implemented with a reduction in rff to reverse a decline in nominal national income is often followed by a rising rff 26 this claim can be made as rtb10 is closely related to corporate long-term bond yields, mortgage rates and all other long-term interest rates. 27 this is the most statistically significant relation in the cointegration equation as indicated by the t-statistic value of rff (8.1), the highest one in this equation. this implies that monetary policy is much more important than fiscal policy because the t-value of bd is only about half in magnitude (-4.54) of the t-value of rff. h0 eigenvalue max critical value 5% p-value trace critical value 5% p-value r = 0 0.64 51.69 28.58 0.00 84.38 54.07 0.00 r ≤ 1 0.32 19.79 22.29 0.16 32.69 35.19 0.09 r ≤ 2 0.15 8.70 15.89 0.46 12.89 20.26 0.37 international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 58 when real national income eventually increases. as a result, a positive long-run relation between lnrni and rff is possible. the relations of the variables in the cointegrating equation do not depict the final long-run relationship among these variables. the final long-run relation among the four variables can only be revealed if the cointegrating relation is analyzed in conjunction with the estimated vecm and particularly by focusing on the value of the one-period lagged error term t-1 and its estimated coefficient. it is interesting to note that a sign of the coefficients of the variables on the right-hand side of the cointegrating equation are identical with the respective coefficients shown under the summation terms in the estimated vecm. this indicates that longrun causality is related to short-run causality. table 4: an estimated vecm with real national income dependent variable trend c t-1    1 1 ln r i itrni f1    1 1 10 s i tb itr f2    1 1 k i itbd f3     1 1 p i ff it r f4 (12) lnrni -.0008 (-.28) .05 (2.5)** .31 (2.06)** -.74 (1.78) 3.12** -.11 1.91 1.59 -.03 (2.03) 2.06* .101 (2.6)** (2.55)** (13) rtb10 -.009 (-.65) 2.41 (1.98) * 16.3 (2.07)** -65 (1.84) 1.57 -6.06 (2.58)* 2.29* -1.7 (1.16) 1.12 5.9 (2.46)* 3.5** (14) bd -.003 (-.21) 1.65 (1.27) 12.02 (1.4) -49 (2.3)* 2.44* -4.9 (1.21) 1.51 -1.1 (1.45) 1.2 4.2 (2.2)* 2.04* (15) rff .01 (.71) 4.29 (1.99) 54.2 (3.89)*** -148 (1.79) 3.61** -16 (3.47)** 3.2** -5.4 (3.43)** 3.6*** 16.2 (4.91)*** 4.1*** ***, **,* denote significance at 1%, 5%, and 10% levels, respectively. the figures in parentheses and brackets are t and f statistics, respectively. equation (12) of the estimated vecm model indicates that monetary and fiscal policies are jointly effective in influencing real national income lnrni in the long-run. this is demonstrated by the positive coefficient, .31, of θt-1, the one-period lagged error term, which is significant at the 5% level. this result provides evidence for long-run causality from the three right-hand side variables rff, bd, and rtb10 to lnrni. the interpretation of this result is that monetary and fiscal policies are jointly effective in influencing real national income. this result is exceptionally important for policy purposes because it shows that an active policy mix has positive (negative) effects on real output. new keynesian models with “frictions,” such as a zestos et al.: monetary-fiscal policy 59 government budget constraint, support favorable effects of monetaryfiscal policy mix on output. 28 the effect on output, nonetheless, cannot be exclusively attributed to any one of the two policies by simply observing the estimated vecm equations (12)-(15). if, however, we examine the cointegrating equation together with equation (12), it is clear that only expansionary monetary policy affects positively real national income; in contrast, expansionary fiscal policy has a negative effect on output. 29 this implies that contractionary fiscal policy can increase lnrni. levy (2002) provides empirical evidence from the us that a sequence of government deficit reductions had positive effects on real national income during the clinton administration period. these results were robust to the type of deficit measure employed in levy’s model. furthermore, levy indicated that, contrary to keynesian theory, the omnibus reconciliation act of 1993 has resulted in surpluses which contributed to economic growth. similar results are supported by a group of models dealing with fiscal contractions adopted in response to severe economic or fiscal crises. 30 a possible alternative explanation of how contractionary fiscal policy can increase lnrni is also suggested here. since fiscal policy works with lags, it is possible for current deficits to be associated with increases in future lnrni but with decreases in current lnrni; i.e., government deficits can lead to real national income. such an explanation supports a modified version of the keynesian theory allowing for lagged effects between changes in fiscal balances and real national income. evidence is also supported for short-run and overall causality from rff to lnrni at the 5% significance level, as this can be seen by the two significant coefficients shown in the last two columns of equation (12). this result indicates that in the short-run an expansionary monetary 28 in this study an implicit government budget constraint is incorporated in the vecm as the government deficit must be in a permanent stable long–run linear relation with real national income and the long-term interest rate. 29 to see this, assume that you express equation (11) in terms of the lagged error term θt-1, you can do this by lagging all variables by one period, then a reduction of rff will make θt-1 positive. thus a positive value of θt-1 multiplied by the positive coefficient .31 of θt-1 in equation (12) of the vecm will increase lnrni. on the other hand, if bd increases (i.e. larger deficit) this will make θt-1 negative and when multiplied by the positive coefficient of θt-1 will result to a negative number implying that real national income declines. 30 these models show that large fiscal consolidations for high deficit countries can raise consumption and output, see perotti (1999) and giavazzi, and pagano (1990). international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 60 policy negatively affects lnrni. lastly, weaker evidence is supported at the 10% significance level for overall causality from bd to lnrni as shown by the reported statistic under f3. statistical evidence of long-run causality is supported in equation (13) from lnrni, bd and rff, to rtb10 at the 5% significance level, indicated by the coefficient of θt-1 in this equation. the meaning of this result is that joint fiscal and monetary policies affect long-term interest rates. equation (13) also provides evidence of short-run and overall causality from rff to rtb10 at 10% and 5% significant levels respectively, shown on the last two columns of this equation. no statistical evidence, however, is supported for short-run and overall causality from bd to rtb10. these results imply that only monetary policy in the short-run affects long-term interest rates. equation (14) indicates that no evidence for long-run causality from lnrni, rtb10, and rff to bd is supported. this is indicated by the coefficient of θt-1, which is statistically insignificant. this result suggests that fiscal policy is not reactive to changes in the real business cycle and the two interest rates. no short-run or overall causality is supported from rtb10 to bd, which indicates that fiscal policy is not reactive to changes in the long-term interest rates. short-run causality is only supported at the 10% significance level from lnrni to bd. this implies that fiscal policy mildly and counter-cyclically responds to changes in lnrni, in the short-run. this result is in line with the belief of the vast majority of economists who are convinced that fiscal policy is not very efficient in stabilizing output. fatas and mihov (2003), for example, demonstrated that fiscal policy is a source of instability that may even lead to negative economic growth. long-run causality is strongly supported in equation (15), indicating that lnrni, rtb10, and bd granger cause rff at a 1% significance level. this result indicates that the fed is reactive to changes in the three variables by changing rff. it can also be seen that strong statistical evidence exists for short-run and overall causality from lnrni, bd, 31 and rtb10 to rff. the results provide substantial statistical evidence that monetary authorities are very responsive to changes in any of the three variables lnrni, rtb10 and bd. the findings from the estimated vecm indicate that monetary and fiscal policies are effective in influencing real national income. monetary policy is much more reactive than fiscal 31 an expansionary fiscal policy triggers increases in rff to curtail raising prices. this is shown by the negative coefficient of the summation term bd. zestos et al.: monetary-fiscal policy 61 policy in responding to changes in the other three variables. these results are plausible because decisions regarding shifts and implementation of monetary policy are quick. figure 2: impulse response function and variance decomposition of lnrni in contrast to monetary policy, adoption and implementation of fiscal policy programs is slow because fiscal policy works with time lags. it is thus plausible for fiscal policy to be responding to changes in nominal national income but to consistently be missing its purported target of stabilizing real national income. to test for the possibility of structural breaks, model diagnostics were performed for this vecm. according to cusum and cusum of squares tests, seven out of eight tests provide no evidence of a structural break at the 95% confidence level 32 . thus, these test results provide evidence for parameter stability in the estimated vecm. to study the long-run effects of the two policies on national income we also employed the methodology of impulse response functions and variance decomposition. according to the generalized response of one standard deviation in bd, we found that it permanently reduced lnrni after small oscillation during the first ten years. this result supports the view that fiscal 32 in the third equation, which employs bd as a left-hand-side variable, there was evidence of a minor structural break. response to generalized one s.d. innovations on lnrni variance decomposition of lnni due to bd and rff % international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 62 policy generates recessionary effects. contrary to fiscal policy, monetary policy produces positive long-run effects on real national income, although it created more volatility in lnrni during the first 20 years. this finding is in agreement with the estimated vecm of table 4 in equation (15). looking at the variance decomposition, monetary policy is more influential in effecting real national income up to the first six years; thereafter, fiscal policy dominates having more lasting effects, as shown in figure 2 below. 6. conclusion the study found that fiscal and monetary policies are jointly ineffective in influencing nominal national income. a possible explanation of this unexpected result is that the two policies may counterbalance each other, as it is possible for fiscal and monetary policies to be noncoordinated and even pursue conflicting objectives for certain sub-periods. strong statistical evidence, nonetheless, exists supporting the hypothesis that monetary and fiscal policies are jointly effective in influencing real national income. similarly, statistical evidence is provided supporting a reactive monetary policy to changes in the real national income and to long-term interest rates. fiscal policy, on the contrary, is found to be passive. despite the fact that the fed cannot observe changes in real national income on a contemporaneous basis, it still reacts correctly (counter cyclically) to changes in it. this may be the case as the fed observes several other real variables such as the number of jobs lost, unemployment claims, and other leading economic indicators. the long-run effects of an expansionary fiscal policy on real national income are negative. this means that partial negative ricardian equivalence and some crowding out effects are present during the period of the study. contrary to the fed response, fiscal authorities were found to be passive to changes in real national income and to the long-term interest rate. the asymmetric responses of the two policies are attributed to the fiscal time lags which render fiscal policy inefficient in quickly responding to real business cycle fluctuations. employment of impulse response functions suggests that monetary policy generates mainly positive effects on both lnni and lnrni. fiscal policy, however, creates contractionary zestos et al.: monetary-fiscal policy 63 effects on lnni and lnrni. such partial results are in agreement with the empirical findings of the two estimated vecms. the technique of variance decomposition, in contrast to the impulse response function, attributes greater contribution to fiscal policy in affecting both nominal and real national income than monetary policy. this means fiscal policy effects cannot be underestimated as they explain approximately 15% of changes in both nominal and real national income. the policy lesson of this study is that economists and policy makers must seek new ways to improve fiscal policy to make it more efficient whenever this is possible (see leeper, 2010). since the study covers the period 1955-2006, it will be interesting to extend this research to cover the most recent years after 2006 in order to examine the effects of the monetary-fiscal policy mix on nominal and real national incomes and the reaction of the two policies to the subprime mortgage crisis. author information: george k zestos, the corresponding author, is a professor of economics at the christopher newport university while andrew geary and kevin cooksey are graduate students at indiana university and john hopkins university respectively. george wishes to express his gratitude to his university for a faculty grant; also to professor nicholas apergis of university of piraeus greece. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 64 table a1: appendix, unit root test results 33 adf phillips-perron level constant t p lags lags constant & trend t p lags lags constant t p lags llllllllllagslags constant & trend t p lags lagslalagslags rff -2.85 .06 1 -2.8 .2 1 -2.9 .21 6 -2.05 .56 7 -1.97 .3 2 -1.86 .66 2 rtb10 -1.62 .46 0 -1.28 .88 0 -1.72 .42 3 -1.31 .87 2 -1.93 .32 3 -1.08 .92 4 lnni -.97 .75 1 -.53 .97 1 -.84 .79 5 -.74 .96 5 -1.21 .66 0 .11 .99 0 bd -2.44 .14 0 -2.75 .22 0 -2.47 .13 2 -2.84 .19 2 -2.85 .06 1 -3.04 .13 1 td -2.51 .12 0 -2.51 .32 0 -2.54 .11 3 -2.6 .28 2 -2.97 .04 1 -2.88 .19 1 lnrni -.97 .75 2 -2.7 .21 1 -.65 .84 0 2.76 .21 1 -.711 .83 0 -2.38 .38 0 22 the augmented dickey-fuller (adf) tests and the phillips peron (pp) tests were performed on all the variables in their levels and first differences. the adf test was performed twice, first with only a constant and second including a constant and a linear trend. to assure that the error term in the unit root test equation is a white noise, a number of lagged differences of each tested variable were included in the equation. a unit root test is based on the null hypothesis: β2 = 0 versus the alternative β2 ≠ 0. the optimum number of lagged differences included in the adf unit root test equation was determined according to both the akaike information criterion (aic) and also according to the schwarz information criterion (sic). as a result the adf unit root test was performed twice, first based on the aic and second based on the sic. these results are presented in the first and second rows according to aic and sic criteria respectively for each variable in table 1, in the appendix. first difference constant t p lags lags constant & trend t p lags constant t p lags lagslags constant & trend t p lags lags rff -6.41 0 1 -6.42 0 1 -5.69 0 19 -6.01 0 22 -6.41 0 1 -3.46 .05 7 rtb10 -6.01 0 0 -6.16 0 0 -5.96 0 2 -6.17 0 1 -3.52 .01 3 -3.72 .03 3 lnni -2.42 .14 2 -4.7 .00 0 -4.82 .00 4 -4.86 .00 4 -4.65 .00 0 -4.7 .00 0 bd -6.9 0 0 -6.87 0 0 -7.24 0 6 -7.18 0 6 -6.12 0 1 -4.32 .00 5 td -6.75 0 0 -6.71 0 0 -7.15 0 7 -7.17 0 7 -6.75 0 0 -4.41 .00 5 lnrni -5.6 0 1 -5.6 0 1 -5.8 0 6 -5.8 0 7 -5.6 0 1 -5.6 0 1 zestos et al.: monetary-fiscal policy 65 references barro, r.j., (1974) are government bonds wealth? journal of political economy 82: 1095117 beetsma, roel, (2008). a survey of the effects of discretionary fiscal policy. rapport till finansplitisha radet. university of amsterdam, cepr and cesifo. january 29, 2008. bernanke, ben s., and alan s. blinder, (1992). the federal funds rate and the channels of monetary transmission. american economic review 82 (1992): 901-921 brimmer, andrew and sinai, allen; the monetary-fiscal policy mix: implications for the short-run. the american economic review, vol. 76, no. 2, papers and proceedings of the ninety-eighth annual meeting of the american economic association (may, 1986): 203-208 congressional budget office report, the cyclically adjusted and standardized budget measures, april 2008. claeys, peter, (2007). estimating the effects of fiscal policy under the budget constraint. research institute of applied economics. clarida, r. h., and b. m. friedman; why have short term interest rates been so high? brooking papers on economic activity, economic studies program, the brookings institution, 14 (1983): 553-586 cook, timothy, and thomas hahn, (1989). the effect of changes in the federal funds rate target on market interest rates in the 1970s. journal of monetary economics 24 (1989): 331351. de long, bradford., summers, lawrence h., mankiw, gregory n., and romer christina, d., (1998). how does macroeconomic policy affect output? brookings papers on economic activity, vol 1988, no. 2 (1988): 433-494. economic report of the president (2007) united states government printing office engle, robert f., and clive w. granger, (1987). cointegration and error correction: representation, estimation and testing. econometrica 55 (1987): 251-276 fackler, james s, and mcmillin,w. douglas, (2003). southern economic journal vol. no. 4. (apr., 1989): 994-1003. fatas, a., and mihov, i., (2003). the case for restricting fiscal policy discretion, quarterly journal of economics, 118 (2003): 1419-1447. garfinkel, michelle r., and daniel l. thornton, (1995). the information content of the federal funds rate: is it unique? journal of money, credit, and banking 27 (1995): 838-847 giavazzi, f., and pagano, m.; can severe fiscal adjustments be expansionary? in o. blanchard and s. fischer (eds.), nber macroeconomics annual, mit press, cambridge, ma. (1990). international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 http://www.jstor.org/stable/1818765 http://www.jstor.org/stable/1818765 http://www.jstor.org/action/showpublication?journalcode=amereconrevi 66 gonzalez, luis a., and martinez, pilar g., (2003). central bank independence: taylor rule and fiscal policy, working paper. granger, c.w.j., (1988). some recent developments in a concept of causality. journal of econometrics 39 (1988): 199-211 harvey, campbell, (1997). the relation between the term structure of interest rates and canadian economic growth. the canadian journal of economics, vol. 30, no. 1 (feb., 1997): 69-193 johansen, soren, (1988). estimation and hypothesis testing of cointegration vectors in gaussian vector autoregressive models. econometrica 24 (1988): 1551-1580. johansen, soren (199%). likelihood-based inference in cointegrated vector autoregressive models. oxford: oxford university press. kuttner, kenneth n; the monetary-fiscal policy mix: perspective from the u.s. october 1, 2002, working paper. kydland, finn, prescott, edward, (1977). rules rather than discretion: the inconsistency of optimal plans, journal of political economy. leeper, e., (1991). equilibria under “active” and “passive” monetary and fiscal policies. journal of monetary economics 27: 129-147. leeper, e.; monetary science, fiscal alchemy. kansas city fed (2010). retrieved from www.kansascityfed.org/publicat/sympos/2010/2010-08-16-leeper-paper.pdf levy d. mickey, (2001). don’t mix monetary policy and fiscal policy: why return to an old flawed framework. cato journal, vol. 21, fall: 277-283. mackinnon, james g., (1996). numerical distribution function for unit root and cointegration tests. journal of applied economics 11(6) (1996): 601-618. nunes-correia, jose; stemitsiotis, loukas, (1993). budget deficit and interest rates: is there a link? commission of the european communities directorate-general for economic and financial affairs, number 105, november 1993 perotti, r., (1999). fiscal policy in good times and bad. quarterly journal of economics 114: 1399-1436. ribe, frederick; beeman, william, (1986). the monetary-fiscal mix and long-run growth in an open economy. the american economic review, vol. 76, no. 2, papers and proceedings of the ninety-eighth annual meeting of the american economic association. (may, 1986): 209212. sargent, t., and wallace, n., (1981). some unpleasant monetarist arithmetic, federal reserve bank of minneapolis quarterly review fall, (1981): 1-17. taylor, john b., (1993). discretion versus policy rules in practice, carnegie-rochester conference series on public policy, 39: 195-214. zestos et al.: monetary-fiscal policy http://www.kansascityfed.org/publicat/sympos/2010/2010-08-16-leeper-paper.pdf 67 turnovsky, s.j., (1989). the term structure of interest rates and the effects of macroeconomic policy. journal of money, credit, and banking 21(3): 321-347. zestos, george, (2010). shaltayev, dmitriy; robert winder: us federal reserve and bank of canada causal effects on business cycles and central bank reactions. journal of business & economic studies, spring 2010. international journal of banking and finance, vol. 8, iss. 2 [2011], art. 3 the impact of underlying market closure on futures market: evidence from china the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 26 the impact of underlying market closure on futures market liquidity: evidence from china wang chun wei and alex frino university of sydney, australia ___________________________________________________________ abstract this study investigates the trading activity of chinese stock index futures, recently introduced at the open and close of the underlying trading. we document the impact of the underlying spot on the futures market liquidity as well as volatility as discussed in earlier works on market closure theory. our empirical results support previous literature on the impact of the underlying, particularly during the open session, as a contagion effect, which is clearly at play. we find significant u-shaped patterns in liquidity factors and intraday volatility during open and close trades in the morning. key words: market microstructure, market closure theory, liquidity, chinese futures market jel classification: g14, g15 ___________________________________________________________ 1. introduction chang et al. (1995) argue that the price linkage between derivatives and cash implies that the closure of the underlying cash market, the spot, is likely to have a profound impact of the trading of a derivative. this impact on derivative trading is explained using king and wadhwani’s (1990) contagion model, which suggests that price behaviour in one market affects the behaviour in related markets as traders can draw inferences from multiple markets. more precisely, we suggest information transfer from spot equities market to futures market. this information transfer would however be disrupted, when the underlying market closes and therefore trading behaviour in the derivative market would be clearly affected. early attempts to understand the effects of market closure effect focus primarily on intraday volatility measures. chang et al. (1995) examine intraday volatility and price changes of the s&p 500 futures traded on the cme at around nyse (i.e. underlying market) closure times. their research was possible due to the continuation of the futures contract for the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 27 15 minutes after nyse closure. they document two separate u-shaped patterns for futures intraday volatility, before and after underlying spot market closure. they note that before the closure of underlying spot, when the underlying asset is trading, the futures volatility generally tends to reflect the intraday patterns of stock volatility, i.e. u-shaped.1 in futures trading after the underlying market closure time, the volatility drops significantly before increasing at the close. this second smaller u-shaped occurrence is proof of the contagion model as described in king and wadhwani (1990). however, the u-shaped pattern itself is not explained via contagion. foster and viswanathan (1990) explain the u-shaped pattern as being attributable to strategic trading models. they state that informed traders prefer to transact during periods when liquidity traders are most likely to trade to reduce execution costs – this happens most at market open and close times. higher levels of liquidity trading at close may be due to several reasons, such as fund managers who might need to make trades at closing prices for fund purchases and redemption, and also investors wishing to hedge overnight positions. this explains the increase in volatility at futures closure time. daigler (1997) documents similar findings of a u-shaped volatility curve from futures contracts after underlying nyse closure. furthermore, in extension to chang et al. (1995), daigler (1997) finds that the intraday volatility of t-bond futures hold several u-shaped curves for respective day and night sessions, with the night sessions having a lower magnitude u-shaped curve than the day session. this is attributed to the lack of information flow when the underlying is closed. more recently, studies extending to asian markets have also been conducted. fong and frino (2001) document for hang seng index futures volatility changes around underlying market closure times. they find statistically significant declines in futures volatility following the close of the underlying market, followed by a rise at the futures close. this confirms results documented in the us. in this extension, another plausible explanation they raise for the pattern of intraday volatility is the decline of trading after underlying market closure, which would influence price volatility through bid-ask bounce. fong and frino (2001) argue that bid-ask bounce is less prevalent following underlying closure, therefore reducing any upward bias in the traditional volatility estimates of transaction prices. this was shown by 1 see mcinish and wood (1992) for intraday stock volatility patterns. international journal of banking and finance, vol. 9, iss. 2 [2012], art. 2 the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 28 taking the difference between quoted price volatility relative to transaction price based volatility measures as in chang et al. (1995). similarly, huang (2002) studies the intraday patterns of volume as well as volatility and examines the market closure theory on the taiwan stock index futures traded on the taiwan futures exchange taifex and singapore exchange derivatives trading sgx-dt. huang (2002) notes that the taifex opens and closes 15 minutes before and after underlying market trading, allowing the examination of the effect of an underlying cash market close on the futures market close. whilst the sgx-dt futures exhibits a slight increase in volatility at the end of trading, the volatility on the taifex futures, after a peak at the close of the underlying market, consistently drops towards the end of trading. it is reasoned that whilst sgx-dt futures have a large amount of market makers and scalpers closing their positions before futures market closure time, this is not the case for taifex. this is perhaps due to higher transaction costs and stringent restrictions on foreign investors in taifex trading. this paper contributes to the literature on market closure theory in two ways: (i) the impact of liquidity on underlying market closure; and (ii) the focus of china’s stock derivatives. we find scarce documentation of the impact of market closure on intraday liquidity, let alone with a focus on an emerging market such as china. research in this area is important for understanding of the intraday dynamics of futures markets and also for institutional and regulatory purposes such as knowing the underlying impact of extended after hours derivative trading and periods where trading halts exist in the underlying market. the remainder of this paper is structured as follows. first, we examine intraday price volatility and price changes, following traditional methods discussed in chang et al. (1995). this is necessary, as it allows us to compare chinese results with studies performed in the us. the second half of this paper drills deeper into liquidity factors such as trade frequency, size, spreads and depth around market closure for us to document effects of underlying market closure on liquidity. we then draw conclusions from our results and provide some explanation on the differences between our study on china and former studies on other markets. our results therefore are new for an important emerging market for futures and also for the reason we extend the tests by incorporating a number of innovations. the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 29 2. institutional detail and data this paper examines the trading behaviour of csi 300 futures contracts in china.2 these contracts are based upon a capitalisation weighted index of the 300 largest companies listed in both the shanghai and shenzhen stock exchanges. the csi 300 index linked futures contract was launched by the china financial futures exchange3 on april 16th, 2010. it is china's only listed financial derivative at the time of writing. the contract itself has been mock traded since october 30th, 2006. the trading platform is best described as an electronic open limit order book – that is orders that match prices are automatically traded while unexecuted orders are left in the order book. the underlying market trading hours for consecutive bidding in shanghai and shenzhen,4 are between 9:30am to 11:30am and from 1:00pm to 3:00pm. the pre-opening auction session runs from 9:15am to 9:25am, following which there is a 5 minute break prior to competitive bidding. trading hours for the index linked futures contracts are from 9:15am to 3:15pm with a lunch break from 11:30am to 1:00pm. similar to the futures contracts in hong kong, taiwan and singapore, the csi 300 contract possess the timing characteristic of opening 15 minutes prior to the cash market and closing 15 minutes after the cash market. this provides a perfect setting for examining the impact of underlying closure in both the morning and afternoon sessions of trading. the data used for this study consist both of intraday equal time spaced traded price series and tick data from thomson reuters tick history, with a sample period from april 16th, 2010 to november 30th, 2010. for the analysis completed on intraday price volatility, we work with a total of 15,192 observations which record traded open, close, high and low at 5 minute intervals. in order to analyse intraday liquidity (as discussed in a later section of this paper), we use tick data, recording each traded price, traded volume, best bid price and volume, and best ask price and volume with a time stamp to the nearest 1/100th of a second. there is a total of 3,952,959 observations for our analysis. 2 the underlying csi 300 index is a capitalisation weighted index determined by the daily turnover and market capitalisation of the 300 largest companies listed on both the shanghai and shenzhen stock exchange. the futures contract size is the index multiplied by a factor of 300 yuan. this factor is designed to limit participation by retail investors by making each individual contract reasonably expensive. they have a tick of 0.2 index points and a daily limit of a 10% deviation from previous settlement price. the last trading or settlement day is on the 3rd friday of the delivery month. 3 the china financial futures exchange was established with the country's two stock exhanges, shanghai and shenzhen stock exchange, in conjunction with three futures exchanges, shanghai futures exchange, dalian commodity exchange and zhengzhou commodity exchange on september 8th, 2006. 4 it is noted that shanghai and shenzhen are in the same time zone. the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 30 3. intraday volatility as discussed in chang et al. (1995), fong and frino (2001) and huang (2002) have focused on the study of intraday price volatility around underlying market closure times. since chinese stock index futures market is new, we begin by documenting their behaviour so as to provide information on what we know of previously studied markets. to understand the impact of market closure on intraday price volatility in china, we divide each trading day into 5 minute intervals. we will not go into specifics of choosing a 5 minute frequency as opposed to say a 15 minute frequency or 1 minute frequency, but as is suggested by chang et al. (1995), a 5 minute frequently adequately captures trading dynamics without being too data intensive. price volatility is measured using parkinson's (1980) extreme value variance which was employed by chang et al. (1995) due to its simplicity, (1) which is the difference between the high minus the low for the given time interval. wiggins (1991) argues parkinson's estimator is significantly more efficient than close-to-close estimators. we also apply the garman-klass estimate, which, as discussed in huang (2002), is considered to be a superior measure, incorporating the open and close price for the time interval as well. (2) the intraday price volatility measures are standardised per day for comparison i.e. where s is the sample standard deviation of the intraday volatility estimates. standardisation is required because different days are not directly comparable as the number of days to expiry has an effect on trading behaviour. with respect to the contagion model, there is an expectation that there will be a significant decline in the futures price volatility upon the closing of the underlying equity market. this can be tested using the null hypothesis, the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 31 furthermore, to examine the patterns exhibited in the 15 minute after closure, we test if there is any changes to intraday volatility immediately after the underlying closure, and whether there is a ramp up of volatility at futures closure. furthermore, let us compare at the closures, the difference in volatility at underlying closure to futures closure. this tests the significance of the impact of underlying closure in comparison to actual futures closure. unlike chang et al. (1995), we are interested in analysing the effects of the morning period of trading. this is because the futures contract in china trades 15 minutes prior to the underlying market opening which is not the case for cme traded s&p 500 futures. these hypotheses tests are performed, mirroring tests 1 to 4. h5 tests if the impact of the underlying opening on futures market volatility. h6 and h7 tests whether volatility remains constant in the first 15 minutes of trading, when the underlying is still closed, and h8 examines if there are any notable differences in volatility in the futures market between derivative opening and underlying opening. 4. results on intraday volatility international journal of banking and finance, vol. 9, iss. 2 [2012], art. 2 the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 32 in table 1, we present volatilities based on both parkinson and garman-klass measures. the t-statistics for each interval volatility estimate tests equality between the mean volatility for that interval on that weekday and the mean midday volatility . furthermore, table 2 presents twosample t-tests5 used to test the null hypothesis h1 to h8 for differences in intraday volatility due to market closure. table 1 documents standardised intraday volatilities at around underlying opening and closing. as expected, there is very little difference in the general trend of either garmanklass (gk) or parkinson (p) measures. we note that intraday volatility is significantly higher than the average midday volatility (i.e. a standardised volatility of 1.7460 and 1.5149 for gk and p respectively, compared to 0.0017 and 0.0196 for the midday average) and continues to fall lower in the next two 5 minute intervals. for example, gk records the lowest standardised volatility of -0.1154 at the 5 minute interval preceding the underlying market opening. we document that after the underlying market opens at 9:30am, the standardised intraday volatility shoots up to 0.2691 and 0.2586 for gk and p respectively. both measures show this to be statistically significant at the 0.05 level. this clearly illustrates a u-shaped pattern at the open, and confirms findings originally by chang et al. (1995) and later by fong and frino (2001) for hong kong and huang (2002) for taiwan are also applicable to china. information flow from the underlying market has clearly increased the volatility of china’s derivative market. table 2 further verifies the behaviour in futures trading in the morning session by performing our hypothesis tests for h5 to h8. we note in particular that tests h5, h7 and h8 are statistically significant at the 0.05 level, suggesting that the volatilities in these intervals tested are not equal. rejection of the null of h5 suggests that the volatility in the 5 minute period after underlying market opening is significantly higher than the volatility in the 5 minute period preceding underlying market opening. this is consistent with the view that information flow from the equity market should impact the futures market and generate higher intraday activity and subsequently volatility. results from h7 show a strong decline in 5 two sample t-tests with unequal population variance is defined as where and . the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 33 volatility in the 2nd 5 minute interval after futures market opening in comparison to the first 5 minute interval. significance of test 5 and 7 confirms with the u-shaped patterns in the volatility documented on hang seng index futures by fong and frino (2001) is also apparent with csi 300 equity linked futures. in the afternoon session around underlying market closure, we document in table 1 a steady decline in intraday volatility. it is important to note that the decline in volatility starts not at underlying market closure, but in the periods prior to market closure. this decline international journal of banking and finance, vol. 9, iss. 2 [2012], art. 2 the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 34 continues unabated at market closure and in periods after the market closure. however, in the last 5 minute interval at futures closure, we note an uptick in volatility (a standardised figure of -0.3510 under gk and -0.4700 under p in comparison to -0.4633 and -0.6679 in the previous 5 minute interval). however, this increase in volatility is still significantly lower than the average midday volatility in both garman klass and parkinson measures (0.0017 and 0.0196 for gk and p respectively). the lower volatility in the afternoon documented in table 1 in comparison to the midday average could be the direct result of the lack of new information in the underlying market at around closure. certainly contagion stops completely at 3:00pm and intraday volatility is unlikely to increase at all until futures closure, where liquidity traders may increase activity as explained by foster and viswanathan (1990). our hypothesis tests for the afternoon session documented in table 2 are less conclusive in comparison to the evidence from the morning session. we document a statistically significant increase in the last 5 minute interval, 3:10pm to 3:15pm, in comparison to the period 3:05pm to 3:10pm. this relationship is apparent in both gk and p measures, and is consistent with analysis conducted previously by chang et al. (1995), fong and frino (2001) and alsbed and al-khouri (2008). the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 35 whilst we cannot reject the null hypothesis for test 2 under gk, parkinson's measure for volatility shows a significant decline in volatility in period 3:05pm to 3:10pm compared to period 3:00pm to 3:05pm. this suggests that a significant u-shaped intraday volatility pattern is apparent at market closure. in conclusion, intraday volatility patterns in china are largely similar to findings by huang (2002) for taiwanese equity futures traded in taiwan and singapore and by fong and frino (2001) for the hang seng index futures. 5. intraday liquidity the focus to date has been on the impact of market closure on intraday volatility. here we extend these works with a particular focus on intraday liquidity. whilst various studies have tried to understand the aspects of intraday market liquidity, however to the best of our knowledge, very few have focused on the effects of the closure of the underlying asset on the liquidity of the respectively traded futures. academics have used various definitions for liquidity. schmukler et al. (2008) provide a general view on the definition of a liquid market as one in which ‘market participants can promptly execute large volume transactions without significant price impact’. liquidity is often associated with safety as it enables investors to sell in a timely manner with less slippage or price impact and therefore the fluctuations in liquidity in the intraday space is an important source of risk which needs to be studied. as argued by many academics, liquidity is multidimensional. kyle’s (1985) seminal work describes market liquidity in three distinct aspects – tightness, depth and resiliency. tightness is measured as the bid-ask spread, which is a proxy on the difference between the trade price and the actual price at execution. depth is the volume which can be traded at the current level, and is proxied as the sum of the volume of the best bid and ask. resiliency, discussed further in muranaga (1998) and muranaga and shimizu (1999), is essentially the elasticity of the bidask spread. early empirical works by admati and pfleiderer (1988) identified u-shaped patterns in the average volume of shares, suggesting heavy trading at the beginning and end of the day, and light trading during the lunch period. studies looking at intraday periodicities on liquidity in the form of a bid-ask spread was conducted by chan et al. (1995) on stock international journal of banking and finance, vol. 9, iss. 2 [2012], art. 2 the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 36 options traded on the cboe and their underlying assets on the nyse. like earlier works by mcinish and wood (1992), they document a u-shaped phenomenon for stock spreads. it can be argued that spreads are related to the perceived level of private information in the market. as large trades convey more information than small ones, we would expect bid-ask spreads to widen in times where levels of private information is high which arguably occurs at the opening and closing periods of the day. ranaldo (2001) concludes that large spread size is associated with intense trading and higher trade volume, showing that for the swiss exchange, intraday liquidity is characterised by a triple u-shaped curve due to the impact from market openings in the uk and us. in this section, we focus on the impact of intraday liquidity due to underlying market closure. therefore, instead of documenting variations across the day, we concentrate on changes in liquidity factors at the open and close of the underlying market. proxies for three factors on liquidity are measured to document the impact, which we discuss below. trading frequency: we document the number of trades in every 5-minute interval of trading on the futures contract is a proxy of how frequent it is traded during the day. trade frequency is expected is rise as the contract nears expiry, hence the observations are standardised each day to provide consistency for analysis. tightness: time-weighted bid-ask spreads as specified in mcinish and wood (1992) is used to give indication for market tightness. for a given time interval the time-weighted spread is defined as follows, (3) where basi is the quoted bid ask spread in the index as points, and t is the amount of time the spread had existed for, before a new bid or ask was quoted. in this paper, we aggregate tradeby-trade data into 5 minute interval time-weighted bid-ask spreads. we standardise the result, similar to chan et al. (1995), by subtracting the mean and dividing by the standard deviation for the day. depth: market depth is estimated via intraday volume. in particular, we are interested in aggregating the volume of the best bid and ask price as follows, (4) the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 37 similar to spreads, we time-weight depth into 5 minute intervals for analysis. (5) this measure provides a good proxy at the ability for an investor to transact at the best bidask quotes given his or her transaction size. likewise, for this paper, we calculate 5 minute time-weighted depth from using tick data. 5. results on intraday liquidity table 3 presents the standardised intraday spreads and depth of equity linked futures trading. focusing first on the standardised spreads, we document that bid-ask quotes start off extremely wide with a standardised spread of 1.0754 in the period 9:15am to 9:20am compared to the midday average of 0.0820. this is suggestive that at market open traders cannot agree on an equilibrium price. this spread decreases in a significant manner in the next 10 minutes. the next peak occurs after cash market opening, which arguably new information is feed into the futures market as by king and wadhwani’s (1990) contagion model. in table 4, we clearly have evidence to reject the null hypothesis for standardised spreads being equal in the morning session i.e. h5, h6, h7 and h8. a significant negative t statistic in h7 and a significant positive t statistic in h5 show a clear u-shaped pattern in the 15 minute period prior and post underlying trading. therefore, bid-ask spreads start decreasing as futures market opens and increases during the time around underlying opening. likewise for the afternoon session, the rejection of the null in h1, h2 and h3 in table 4 means that spreads decrease after underlying closure and starts increasing prior to futures closure. this matches results presented by fong and frino (2001) for hong kong and huang (2002) for taiwan. table 3 also presents results on standardised bid-ask depth. we document higher levels of bid-ask depth in the open and close of the trading day in comparison to the midday average. notably the standardised depth in the first 5 minutes and last 5 minutes of trading is 0.4539 and 0.6150 respectively, compared to the midday average of -0.0594. our tests h1 and h2 for depth in table 4 show a statistically significant increase in the depth after underlying market closure at 3:00pm. although we could not reject the null hypothesis for h3, the international journal of banking and finance, vol. 9, iss. 2 [2012], art. 2 the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 38 rejection of h4 at the 0.05 level clearly shows that the bid-ask depth in the last 5 minutes is much higher than in the period prior to underlying market closure. the increase in depth and the decrease in spreads towards the end of the day is suggestive of an increase in aggressive bidding by market participants. clearly, the risk of being picked-off towards the end of the trading day is lower due to the lack of information flow entering in the market. it is expected that any new information throughout the day would the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 39 have already been priced into the market. the reduced risk of being picked-off is further reflected in the declines in intraday volatility in around the same period (see table 1). in conclusion, our results on bid-ask spreads and depth illustrate both king and wadhwani (1990) and foster and viswanathan (1990) theories. thus, there is decline in information after underlying closure, reducing the spread, and an increase in spreads as the direct result of strategic trading at the end of the day. table 5 presents standardised trade frequency and trade size around underlying opening and closing. drawing attention to market open, trade frequency and size is notably higher than the midday average at the first 5 minutes of trading (9:15am to 9:20am) and similarly in the 5 minute period immediately after underlying opening. this observation along with our rejection of the null in tests h5 and h7 in table 6 suggests clearly a u-shaped pattern in the 15 minutes, where frequency and size start off high, decline and then rise again at the opening of the underlying. this behaviour is consistent with our findings in the bid-ask spread and intraday volatility. we also document a decrease in trade size and frequency is around the end of cash market close, as shown in table 5. furthermore, in the 15 minutes after cash market close, we observe that trade frequency and size increases as market participants hurry to trade prior to close. unfortunately, we find that we are unable to reject the null at the 0.05 level for any differences in standardised trade size at market closure. however, we draw attention to h3 in international journal of banking and finance, vol. 9, iss. 2 [2012], art. 2 the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 40 table 6 on standardised frequency, which documents a significant increase in trade frequency in the last 5 minutes of futures trading. this increase is explained well by foster and viswanathan (1990)'s strategic trading model as both informed and uninformed investors scramble to trade prior to closure. despite observing a u shaped pattern near market close (see table 5), what is different about china is that the decline in trades start before cash market closure. this would suggest market activity on the contracts falls towards the end of the day as there is no longer the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 41 any new information in the cash market. activity only rises in the last 5 minutes due to liquidity investors interested in hedging overnight exposures and informed traders trying to pick off liquidity trades at market close. table 6 clearly reflects that trading frequency and size is more effected by the opening of the underlying cash market in the morning session of trading than the underlying market closure that the end of the day. it is noted that both frequency and size declines after the initial futures opening and later increases after the start of underlying trading i.e. exhibiting the well documented u-shaped pattern. 6. conclusion in conclusion, this study documents several interesting characteristics on the impact of underlying market closure times on the intraday liquidity and volatility. we find evidence for china’s futures market that intraday volatility is highest at the open and exhibits a u-shaped curve with an increase at the opening of the underlying, which is consistent with the existing literature on market closure as in king and wadhwani's (1990) contagion model. similar to huang's (2002) assessment of taiwan, volatility falls dramatically towards the end of the day, especially after market closures. we also document a small uptick at the last 5 minute interval international journal of banking and finance, vol. 9, iss. 2 [2012], art. 2 the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 42 which can be explained using foster and viswanathan (1990)'s strategic trading model. in terms of liquidity, we find that bid-ask spreads decline prior to underlying market closure, and continue to decline after closure. bid-ask depth also increases in this period. this study seems to suggest that market participants are bidding increasingly aggressively towards the end of the trading day, comforted by the fact that the risk of being picked off with limit orders are low given the falls in intraday price volatility at the same time. we attribute the fall in volatility to be the direct result of the lack of new information in the underlying spot market around closure times. another finding is that morning effects on liquidity and volatility, when the underlying market open, were more statistically significant for the chinese market than the afternoon effects as documented by chang et al. (1995). author information: submitting authors: wang chun wei is staff member at the discipline of finance, business school, university of sydney. he is also working as a quantitative analyst at regal funds management as part of a capital markets crc grant. email: wang.wei@regalfm.com. alex frino is a professor at the discipline of finance, business school, university of sydney. he is also ceo of capital markets crc limited. mailto:wang.wei@regalfm.com the international journal of banking and finance, volume 9 (number 2), 2012: pages 26-43 43 references admanti, a. and p. pfleiderer (1988) a theory of intraday patterns: volume and price volatility, review of financial studies 1, 3-40 alabed, m.f. and r. al-khouri (2008) the pattern of intraday liquidity in emerging markets: the case of the amman stock exchange, journal of derivatives and hedge funds 14, ¾, 265-284 brock, w. and a. kleidon (1992) periodic market closure and trading volume: a model of intraday bids and asks, journal of economic dynamics and control 16, 451-489 chang, e.c., jain, p.c. and p.r. locke (1995) s&p 500 index futures volatility and price changes around the nyse close, journal of business 68, 61-84 daigler, r.t. (1997) intraday futures volatility and theories of market behaviour, the journal of futures markets 17, 45-74 foster, f.d. and s. viswanathan (1990) a theory of intraday variations in volume, variances, and trading costs in security markets, review of financial studies 3, 593-624 fong, k. and a. frino (2001) _stock market closure and intraday stock index futures market volatility: contagion, bid-ask bias or both?, pacific-basin finance journal 9, 219-232 garman, m. and m. klass (1980) on the estimation of security price volatilities from historical data, journal of business 53, 67-78 gourieroux, c., j. jasiak and g. le fol (1997) intraday market activity, working paper, crest, paris huang, y.c. (2002) trading activity in stock index futures markets: the evidence of emerging markets, the journal of futures markets 22, 983-1003 king, m.a. and s. wadhwani (1990) transaction of volatility between stock markets, review of financial studies 3, 5-33 kyle, a. (1985) continuous auctions and insider trading, econometrica 53, 1315-1335 mcinish, t. h. and r.a. wood (1990) an analysis of intraday patterns in bid/ask spreads for nyse stocks, journal of finance 47, 753-764 muranaga, j. (1998) dynamics of market liquidity of japanese stocks: an analysis of tick-by tick data of the tokyo stock exchange, working paper, institute for monetary and economics studies, bank of japan, tokyo muranaga, j. and t. shimizu (1999) market microstructure and market liquidity, working paper, institute for monetary and economics studies, bank of japan, tokyo ranaldo, a. (2001) intraday market liquidity on the swiss stock exchange, financial markets and portfolio management 15 issue 3, 309-327 schmukler, s., levy-yeyati, e. and n. van horen (2008) emerging market liquidity and crises, journal of the european economic association 6(2-3), 668-682 von wyss, r. (2004) measuing and predicting liquidity in the stock market, doctoral dissertation, st. gallen university wiggins, j.b. (1991) empirical tests of the bias and efficiency of the extreme-value variance estimator for common stocks, journal of business 64, 417-432 international journal of banking and finance, vol. 9, iss. 2 [2012], art. 2 mergers and acquisitions: the nigerian banking consolidation program the international journal of banking and finance, volume 8 (number 4) 2011: pages 19-46 mergers and acquisitions: the nigerian banking consolidation program chukwuma agu, *damilola olajide, *divine ikenwilo and anthony orji african institute for applied economics nigeria, *university of aberdeen scotland and university of nigeria nsukka ____________________________________________________________ abstract this paper examines the determinants of the exit behaviour of banks in the nigerian consolidation program during july 2004 and december 2005. we conceptualise the exit process in a flexible bivariate competing risks model to examine the importance of macroeconomic and industry-specific factors for both merged banks and failed banks jointly. the preliminary results suggest that bank-specific characteristics mattered more for preventing bank failure than they did for emergence of the m&a banks. second, the central bank of nigeria’s assistance was highly influential in preventing bank failure, and, for banks that benefited, the assistance increased their probability of being merged or acquired. also, we found no strong evidence suggesting that the prevailing macroeconomic conditions and industry-specific factors had influenced exit behaviour of banks during the consolidation exercise. we found evidence of structural dependence between failure and merger and acquisition hazards induced by cbn incentive. key words: bank merger and acquisition, nigeria, duration analysis jel classification: c41, g21, g34, n27 _____________________________________________ 1. introduction on july 6, 2004, the governor of the central bank of nigeria (cbn) announced a 13 point reform program expected to reposition the nigerian banking industry in its role of enhancing economic growth and development of the country. 1 an important component of the reform agenda was the recapitalization of banks with a minimum shareholders fund of n25 billion prescribed to be met by december 31, 2005. towards this end, the cbn encouraged banks to 1 for a review of the rationale for banking reform in nigeria, see imala (2005). agu et al.: mergers and acquisitions javascript:popup('contact.cgi?popup=yes&window=contact&context=ijbf&u=967156&article=1070&for=editor') 20 enter into mergers and acquisitions (m&a). 2 the cbn also provided a range of incentives to accompany the implementation of the program. for example, a number of committees were instituted, whose terms of reference were expected to assist weaker and problem banks in facilitating their m&as (cbn 2004, 2005). as at the end of the program on december 31, 2005, the number of banks operating in nigeria declined from 89 to 25, about 72 percent. nineteen banks emerged from m&a involving 70 banks and six banks remaining as stand-alone banks (i.e. banks that met the n25 billion without m&a). the remaining 13 banks did not scale through (or failed). accordingly, the emerging 25 banks jointly accounted for about 93.5 percent of the total market deposits, while the remaining 13 failed banks accounted for the rest (6.5%): see cbn (2006). the nigerian banking consolidation exercise represents an unprecedented change in institutional context in which bank m&a and failures occur. firstly, it represents a regulatory policy-induced bank consolidation accompanied by a framework of facilitating incentives that permit banks to exit the industry through either m&a or failure. this differs markedly from the conventional situation in which bank m&a and failures result from voluntary decisions of the affected banks. the exit process in the nigerian exercise was induced as part of an overall sectoral reform policy. also, the regulatory authority provided a range of incentives to assist banks in the exit process. this differs from the commonly observed situations where the regulatory authority merely approves the m&a decisions or liquidates failed banks. the different institutional context in which exit of banks had occurred raises certain questions of policy interest. these include: (i) how important were bank-specific characteristics in the emergence of m&a banks and those that failed? (ii) in what ways do these factors differ between m&a banks and failed banks? (iii) how important were the incentives provided by the cbn in facilitating exit among banks? (iv) how important were the prevailing macroeconomic and industry specific conditions in facilitating the outcome of the consolidation program? there had been no attempt to address these questions, despite their relevance to policy makers. this may be due largely to the generally held view that the exercise was intended to prevent mass failures (i.e. meeting the cbn capitalisation requirement). this is the reason for our research. meanwhile, the economics of bank consolidation treats m&a decisions as any other investment decisions of both acquiring and target banks driven by potential synergies that may 2 in this study, we use merger and acquisition interchangeably, hence we use the acronym m&a. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 21 arise from their mergers. empirical evidence distinguishes between bank-specific characteristics that increase the likelihood of a bank acquiring another bank, be a takeover target, or failed bank. a large, well-capitalised and efficient bank has greater likelihood of acquiring a smaller, less efficient, low capitalised bank. banks that are illiquid, have low quality assets or low capital, are more likely to fail. we do not know how important these and other bank characteristics were in the outcome of the nigerian consolidation exercise. the policy document indicates that the cbn incentive was specifically designed to assist weaker and problem banks through up to 80% debt write-off (cbn 2005). the idea behind the cbn incentive was based on the view that provision of this special forbearance to this group of banks by the cbn would improve their attractiveness to potential acquirers/investors, thereby preventing bank mass liquidation or failures and potential losses by the financial system. yet not all of the problem or weak banks actually received the cbn incentive. thus, understanding the role played by the cbn incentives in the consolidation exercise is of central importance for the outcomes as well as understanding the dimension of the difference between unassisted and assisted m&as. in the nigerian case, a counterfactual question of interest to policy makers is whether the merged banks would have failed had they not received cbn incentive. also, the fact that not all of the weaker/problem banks benefited from the cbn incentives raises a conceptual issue what underlies the selection process, which might not be publicly observable. 3 the goal of this paper therefore is to identify the determinants of exit behaviour of banks in the nigerian banking consolidation program. it hopes to identify bank-specific characteristics that may explain the emergence of the m&a banks side-by-side the failed banks; examine the cbn incentives that influence exit behaviour of banks; and determine whether the prevailing macroeconomic conditions and industry specific factors influenced the exit behaviour of banks during the consolidation program. understanding the interplay among these factors may help explain m&a decision of banks side-by-side failed banks in the nigerian banking consolidation program. we explore the view that m&a decisions are not necessarily made to prevent failure. thus, the emergence of m&a banks and failed banks in the nigerian banking consolidation should be systematically related to factors other than to meeting the capitalisation requirement. this paper is organised into six sections. the next section introduces the reader to the background; section 3 establishes a brief overview of the concepts and literature on the 3 we are especially grateful to the external review for pointing out this central issue. agu et al.: mergers and acquisitions 22 economics of the program. the methodology is explained in the next section and the results are presented in section 5 before ending this paper in section 6 with some conclusion. 2. banking in nigeria and the banking sector consolidation between 1892 when banking started in nigeria and 1952 when the legal framework for it was laid out, banking was largely an unregulated activity in nigeria. since 1952, there has been significant growth in size and structure of banks. financial liberalization led to a loosening of the conditions for granting banking license and consequently a sharp rise occurred in the number of banks between 1986 and 1993. by 1992, there were 120 banks with 3,300 branches up from 15 banks with 273 branches in 1970. a cbn stipulation that banks should have branches in major cities with cbn branches as a condition for direct cheque clearance led to a growth in branch expansion rate of 33.5 percent between 2001 and 2003. banking distress reduced the number to 89 by 2004 (with 26 banks collapsing in 1998 alone). the majority of banks were fragmented, small and marginal players with only about 10 of the banks controlling over 50 percent of total industry assets and deposits. at the initial periods of its growth, government ownership and participation in the industry was considered necessary for its survival. this was the case following the nationalization and indigenization programs of the mid-1970s. however, government presence in the industry fell drastically and, as can be seen from table 1 below, the number has remained low. with the consolidation, the little government presence got eliminated. prior to consolidation, the private banks mostly were tied to some families. about 11 banks had majority foreign interests; this number was reduced to 4 following the consolidation. post consolidation, the family strength has been largely watered down with 21 of the banks now quoted on the floor of the stock exchange and only four banks still retaining majority foreign holding. table 1: ownership structure of nigerian banks status 2000 2001 2002 2003 2004 2005 2006 privately 76 77 78 77 77 77 21 government 1 1 1 1 1 1 0 foreign 10 11 11 11 11 11 4 total 89 90 89 89 89 89 25 as at mid-2004 when the new cbn governor was appointed, the industry faced myriad challenges including operating within a slow and structurally impeded system, frequent changes international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 23 of policies about operations and government deposit management, periodic distress, weak credit regulation, poor management, macroeconomic and political instability, maturity mismatches, insider abuses, fraud and conflict of interest, general insecurity and corruption. prior to this time, cultural and business model rigidities meant voluntary m&as were uncommon with only 11 cases of partial consolidation between 1991 and 2001. the cbn consolidation changed this significantly. following the announcement for all banks to raise their capital base, the apex bank set up a monitoring committee to oversee the program. however, it did not seem to have a monopoly and/or competition monitoring process or commission. it did bring in and pay for technical assistance to the banks; with most of such assistance consisting of payments for merger and acquisition experts. it also worked with the securities and exchange commission to drastically reduce and in many cases, remove fees payable to the commission for such mergers and acquisitions. while banks handled software, operations and branch mergers, the apex bank allowed for a transition time for operations merger and regularization of employees for merged banks beyond the consolidation deadline. there was also special assistance that took the form of a special forbearance framework, which took effect on 6 april 2005. the special assistance had two components one of which is a write-off of 80 percent of debt owed cbn by the banks, subject to:  recovery of non-performing owner/insider related loans and advances within two months  injection of any shortfall in the banks’ capitalization to bring it up to a solvency status, also within two months; the cbn converted the balance of 20 percent of debt to a long term loan of a maximum of 7 years at 3 percent per annum with two years moratorium. the cbn also announced that further forbearance on the balance of 20 percent of the debt could be extended to the new owners after its acquisition and meeting the n25 billion capital base. the idea was to increase attractiveness of the banks concerned and accelerate their m&a through debt write-off. fortunately, the incentives were contingent upon recovery of nonperforming loans associated with owners and other insiders of the banks. this way, the cbn wanted to ensure that past mismanagement of banks were not rewarded. in its 2005 accounts, the cbn provided for these incentives by treating the loans to the affected banks as sunk and irrecoverable costs. the provision for the incentives to be contingent upon insider loan recovery led to only 11 banks actually benefitting from the agu et al.: mergers and acquisitions 24 incentive provision. besides, it started midway into the consolidation program and this may have impacted on the number of banks that benefited ultimately. the analysis in this paper is based on data from 89 banks over three years prior to the cbn’s consolidation program, 2001, 2002 and 2003. descriptive statistics (as in table 2) show that the average bank age was 16 years, with standard deviation of about 7 years difference between them. the youngest bank in the sample was 5 months old prior to announcement of consolidation while the oldest was 39 years old. the average bank had 33 branches, ranging from 3 branches to 351 branches across nigeria. the standard deviation of the number of branches was estimated to be 55. the average bank had a capital base of n4.22 billion naira with standard deviation of 7.46 billion naira, ranging from 0.1 billion naira to 38.6 billion naira. the average shareholders’ fund was estimated to be n1,350.77 million, with a standard deviation of 519.57 million naira. in terms of loans, loans to smes was 7.65 percent of total loans with a standard deviation of 2.55 while loans for agricultural purposes was estimated at an average of n223,553.20 million with a standard deviation of 17,147.86 million naira. finally, an average bank had 27.84 percent of their total loans as nonperforming prior to consolidation, with a standard deviation of 10.22 percent. the proportion of nonperforming loans ranged between 6.5 percent for one bank and almost half of total loans (46.55%) for another. table 2: descriptive statistics of banks prior to consolidation characteristic mean standard deviation range minimum maximum number of branches 33.48 54.58 3 351 bank age 16.14 6.72 0.48 39.49 percent of sme loans 7.65 2.55 0.89 13.30 agric loans (millions of naira) 223,553 17,147 200,856 242,185 percent of nonperforming loans 27.84 10.22 6.5 46.55 capital base (billions of naira) 4.22 7.46 0.10 38.60 shareholders’ fund (millions of naira) 1,350 519 430 3,500 there were three outcomes observed among the 89 banks post-consolidation. these banks were generally classified into three groups, reflecting the post-consolidation outcome. a majority of the banks either formed voluntary mergers or were forced into mergers to survive the consolidation rules. in all, 70 banks fall into this category, representing 78.65% of the total (89) banks. in addition, there were 6 banks which stood alone post consolidation, representing 6.74% of the total (89) banks. the remaining 13 banks (representing 14.61 percent of the 89 banks) international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 25 failed after the consolidation exercise. in all, we compare the age, number of branches, proportion of nonperforming loans, proportion of loans to small and medium enterprises (smes), shareholders’ fund as well as capital base (table 3). table 3: comparing banks by exit type characteristic standalone consolidated failed mean (standard deviation) number of branches 24.17 (21.77) 36.33 (60.50) 22.38 (17.19) bank age (years) 10.39 (7.32) 16.98 (6.86) 14.26 (3.12) percent of sme loans 5.84 (2.46) 7.49 (2.51) 9.35 (1.89) percent of nonperforming loans 9.84 (1.57) 27.16 (8.40) 39.96 (5.45) capital base (billions of naira) 16.38 (10.22) 3.59 (6.86) 2.00 (2.61) shareholders’ fund (millions of naira) 1,372.88 (217.13) 1,337.84 (481.75) 1,410.18 (766.38) 3. overview of relevant literature 3.1 the economics of bank consolidation the economics of bank consolidation is based on the theory of incentives for m&a. economic theory of incentives for m&a relies on two main points of view: the wealth maximisation of shareholders and managerial self-interest (o’keefe 1996). the wealth maximisation theory treats m&a decisions as any other investment decisions of both acquiring and target banks driven by potential synergies that may arise from their merger. for example, synergy is said to occur when the merged entity performs better than both of the banks individually or the long-term market value of the merged entity becomes greater than the simple sum of the individual banks. sources of such synergies include portfolio risk diversification, economies of scale and scope, expansion into new geographical markets, technology, market power, and so on. on the other hand, the notion of managerial self-interest is based on the view that m&a decisions can be driven by managers pursuing their self-interests rather than the interests of the owners or equity shareholders (hardlock, et al. 1999, ely and song 2000, kwan 2004), especially if managers are not well represented in the board of directors (o’keefe 1996). from this point of view, m&a decisions create agency conflicts that result from divergence between shareholders’ wealth maximization interests and managers’ interests. the managerial selfinterest theory predicts that where an m&a places managers at risk of job loss, those managers at risk have incentive to block m&a attempts or make m&a decisions that reduce their employment risk but which might not necessarily translate into increased wealth of agu et al.: mergers and acquisitions 26 shareholders. 4,5 ely and song (2000) provide a review of the us evidences that appear to support the managerial self-serving behaviour. 3.2 empirical evidence empirical studies analysing banking m&a and bank failures generally draw upon the above theories to identify bank financial and other specific characteristics to examine the likelihood that a bank will acquire another bank (acquirer), be a takeover target, be distressed, or fail. however, these studies vary widely in terms of their focus, institutional contexts as well as in analytical methods. hence, results are far from conclusive. we identify two strands of studies that allow us to highlight the major research questions we intend to address in this paper. studies analysing bank m&a in a developed countries the first strand of studies identified generally estimate logit models to examine bank characteristics that determine the likelihood of a bank being an acquirer or a takeover target using various us samples of banks. a feature of this strand of studies is that they do not analyse bank failures. 6 a major approach is to examine the relative importance of measures of profitability (e.g. earnings) and other bank characteristics in distinguishing between acquirers and target banks (hannan and rhoades 1987, amel and rhoades 1989, o’keefe 1996, hadlock et al. 1999, wheelock and wilson 2000). 7 results from this strand of studies are largely inconclusive. a group of the studies found negative relationship between a bank’s profitability and the likelihood of being acquired (amel and rhoades 1989; wheelock and wilson 2000). another group found no evidence supporting profitability measures and the likelihood of a bank acquisition (hannan and rhoades 1987; hadlock et al. 1999). in particular, this latter group found that acquirers are likely to be banks with large market shares, relatively larger (in asset size), adequate equity capital, sound financial position, high profit rates necessary to attract additional capital, well rated management quality and location. on the other hand, target banks are likely to be small in size, poorly managed, inefficient, low capitalised banks. rather than earnings, hadlock et al. (1999) found that the 4 this is necessarily so, to the extent that managerial compensation tends to increase with increased asset size and portfolio diversification arising from m&a. 5 o’keefer (1996) also considers the role of third party interests such as those of investment bankers and security dealers in facilitating mergers. these third parties can profit from a merger transaction even when the expected benefits to the acquiring bank’s shareholders are not obvious. 6 this may be due to the fact that large bank failures are relatively less prevalent in the usa. 7 hadlock et al. (1999) used matching technique. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 27 likelihood of being acquired is lower the larger the ownership stake of bank managers. the authors associated this finding with self-interested behaviour of bank managers. studies analysing bank failures, using samples from developing countries the second strand consists of a few recent studies analysing bank failures, using samples from developing countries. these studies generally do not analyse bank m&a. studies in this strand include dabos and escudero (2004) (argentina), sales and tannuri-pianto (2004) (brazil), and soyibo et al. (2004) (nigeria). dabos and escudero (2004) and sales and tannuripianto (2004) used duration models to examine the role played by bank-specific characteristics that increase the probability of bank failures. they found that banks that are illiquid, low quality assets or low capital are more likely to fail. in addition to bank-specific characteristics, they found also that sectoral shocks (eg. decline in agricultural and energy prices), macroeconomic factors, and industry-specific factors were also important. soyibo et al. (2004) used logit models for both cross-sectional and panel data of nigerian banks to examine specific conditions that accounted for bank distress during 1990s. in their cross-sectional logit estimation, they found that the likelihood of bank distress/failure decreases with bank’s earnings/profitability and less risky loan portfolio (measured as higher rates of proportion of agricultural loans in their loan portfolio). the likelihood of distress increased in government owned banks and in older generation banks. the authors found similar results in their panel data logit estimation, particularly for operating efficiency (i.e. using of costly funds), ownership, and older generation. in addition, they found that poorly capitalised banks and banks not quoted on the stock exchange market tend to be in distress. in the next section, we show some limitations of the theoretical and associated empirical evidence. specifically, we explore the concepts of unassisted and assisted bank m&a in order to understand the influence of the regulatory environment on exit of banks. the aim is to show how regulatory authorities such as the cbn can influence m&a decisions. this is necessary for a clear understanding of the major research issues arising from the nigerian banking consolidation program. 3.3 the concepts of unassisted and assisted bank consolidation the concepts of unassisted and assisted bank m&a are central to understanding the influence of the regulatory environment on m&a decisions of banks. banking m&a are potentially agu et al.: mergers and acquisitions 28 beneficial to the extent that they yield synergies, reduce the cost of doing business, and the sector’s competitiveness are enhanced locally and globally (soludo 2004, kwan 2004). however, the synergies potentially arising from m&a can also raise issues of public policy interest. banking consolidation through m&a will alter the banking market structure, which may raise antitrust concerns. antitrust concerns arise if consolidation results in concentration of the local market in the hands of a few large banks and increases their market power. increase in market power in turn may increase prices of banking services (olajide 2005). also, emergence of a few megabanks increases potentials for systemic risks arising from exposure to default risk (kwan 2004). unassisted m&a take place under a high degree of regulatory oversights in which an m&a decision is jointly undertaken by the controlling directors and shareholders of acquiring and target banks. in this case, the banking regulator merely approves the m&a transaction, having satisfied that the m&a is unlikely to raise public concerns. on the other hand, m&a decisions are assisted when the regulatory authority assists in facilitating m&a between acquiring and target banks. such assistance has the benefit of reducing uncertainties and costs associated with information asymmetries (adverse selection) during merger negotiations. assistance can take several forms including facilitating liquidation procedures, writing-off of debts, and compensating the creditors/depositors of the failing bank. 8 ely and song (2000) compared motivations for mergers in unassisted and assisted m&a in us banks during 1990s using different measures of m&a decision. they found that merger motive differ significantly between unassisted and assisted mergers, and how m&a decision was measured. when m&a decision was measured by the value of asset acquired, they found evidence supporting the wealth maximisation hypothesis for unassisted mergers relative to when measured by the number of completed m&a transactions. however, the authors found no evidence supporting either the wealth maximisation or the managerial self-interest hypothesis in assisted mergers, irrespective of how m&a decision was measured. the above paragraphs show potential influence of banking regulatory authorities on m&a decisions; by facilitating it, by preventing bank failures, as well as ensuring that public interests are protected. this raises important questions that are central to understanding the determinants 8 in addition to these options, the federal deposit insurance corporation (fdic) in the us also uses the purchase and assumption transaction option, whereby it sells some or all of the failing bank’s operations intact to a financially sound bank. for details of fdic’s assistance options, see o’keefe (1996). international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 29 of exit behaviour of banks, particularly in a developing country context. the theoretical literature identifies motivations for bank m&as. the empirical evidences, though largely inconclusive, are generally consistent with the wealth maximisation motives of m&a decisions. this is whereby a large, well-capitalised and efficient bank acquires a smaller, less efficient, low capitalised bank. in this case, the acquiring bank is able to correct any deficiencies of the target bank, thereby improving performance in the emerging bank. in addition to macroeconomic and industry specific factors, evidence from developing countries shows that characteristics that make a bank an acquisition target can also increase the likelihood of bank failure. however, certain conceptual and related empirical questions limit the extent to which results from these studies can be generalised to explain banking exit behaviour process in recent consolidation programs across a number of developing countries, particularly the nigerian consolidation program. we highlight some of these related questions below. 3.4 different institutional contexts in which bank consolidation occur in contrast to bank consolidation through m&a occurring under high degree of regulatory oversights (unassisted m&as), recent experiences mainly from developing countries (eg; turkey, malaysia, argentina, mexico, brazil, and nigeria) demonstrate that banking consolidation can be effectively induced by regulatory or reform policy. 9 in a developing country context, banking consolidation through m&a form an important component of broader banking sector reforms, which result from deliberate regulatory policy designed to reform the banking sector and prevent banking crisis that may result in mass failures. in this context, bank consolidation is part of a wider national objective of ensuring an efficient, developmental and growth oriented banking with greater incentive to improving overall economic performance and society welfare (imala 2005). the context of regulatory policyinduced bank consolidation raises important empirical questions that leave gaps in existing understanding of determinants of exit behaviour of banks. there is need to understand whether the resulting m&a are intended to prevent bank failures or the merging banks pursuing their wealth maximisation interests. comparing evidences from the identified strands of studies, it would appear that characteristics that make a bank an acquisition target also tend to increase the likelihood of bank failure. whether this is indeed the 9 see uchendu (2005) and ahmad et al. (2007) for malaysian and ogubunka 2005 for turkish experiences. the united states (us) federal legislation permitting interstate branching from mid-1997 also provide an example of a regulatory policyinduced banking consolidation (kwan 2004). but this does not provide incentive for bank failure. agu et al.: mergers and acquisitions 30 case is purely an empirical question that we seek to examine in this paper. as an implication of the wealth maximisation theory, that a bank becomes a target for m&a does not necessarily imply that such a bank will fail. however, a bank can also be acquired just before it fails. therefore, we seek to understand those bank characteristics that may explain the emergence of the m&a banks side-by-side failed banks. if the consolidation exercise were intended to prevent bank failures (i.e. meeting the minimum capitalisation requirement) then bank specific characteristics might not matter. this leads to our first hypothesis: hypothesis 1: bank specific characteristics did not matter for the emergence of the m&a banks. the role of incentives provided by the cbn the ely and song (2000) study (cited earlier) showed the importance of the role of the regulatory authority in facilitating m&a decisions relative to sheer wealth maximisation motives of acquiring banks. the result suggests the need to understand the role played by the cbn incentives in the observed exit outcomes, particularly given the different context in which nigerian banking consolidation took place. understanding the role played by the cbn incentives is of interest to policy makers, particularly in future design of implementation strategies that may accompany similar reform policies that affect other industries in the financial sector (eg. the insurance industry). a reform policy accompanied by a framework of incentives to facilitate m&a has the benefit of reducing uncertainties and potential (search) costs associated with information asymmetries (adverse selection) during merger negotiations. by eliminating uncertainties and reducing search costs, m&a negotiations are facilitated and should reduce the likelihood of failures (see borek et al. 2002). for example, improving the financially condition of banks in distress through debt writeoff can eliminate concerns for current liabilities of the target bank, thereby facilitating its merger with other banks. in the present paper therefore, we seek to address the counterfactual question of whether the merged banks would have failed had they not benefited from the cbn incentives. this leads to our second hypothesis: hypothesis 2: the banks that benefited from the cbn incentives increased their probability of being merged or acquired. in order to evaluate this hypothesis, we shall treat the exit outcomes as potentially dependent. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 31 macroeconomic conditions and industry factors with contagion effects on the banking industry. unlike the us studies that focused mainly on m&a and bank specific characteristics, evidence from developing countries show the importance of macroeconomic conditions and industry specific factors having contagion effects on the banking industry in determining bank failures. largely because these studies focused on only on bank failures however, the direction in which these additional factors tend to also influence m&a decisions is not obvious. in their study of swiss firms, bueher et al. (2005) found the impact of macroeconomic conditions on mergers and voluntary liquidations. the nigerian banking consolidation program provides an opportunity to also test this hypothesis in the banking industry. understanding the role played by macroeconomic conditions and contagion effects factors relative to other factors (cbn incentives and bank characteristics) are of interest to policy makers, particularly in timing of banking restructuring policy. for example, a period of business cycle in the economy may delay m&a decision among banks, thereby increasing the probability of failures, especially when there is time limit in which such m&a decisions have to be made. our third hypothesis is: hypothesis 3: the prevailing macroeconomic conditions and industry specific factors had no effect on the exit behaviour of banks during the consolidation exercise. 4. methodology in this section, we show how we hope to test the stated hypotheses and to a larger extent address the research questions. our main interest is not only to identify bank characteristics and other factors that determine bank m&a side-by-side bank failures, but also to examine whether those explaining m&a are different from those explaining bank failures in some systematic way. more importantly, we recognise the cbn incentive as a central variable that is potentially driving the observed exit outcomes. thus, we consider the cbn ‘judgement’ underlying its incentive provision as a latent or unobserved random variable which induces dependence between the m&a and the failed banks. if this unobserved variable is not taken into account, the error terms in the model are potentially correlated. therefore, we specify a semi-parametric bivariate competing risks model with a flexible specification of the outcome-specific baseline hazards. this specification allows for unrestricted correlation across the stochastic disturbances in the agu et al.: mergers and acquisitions 32 competing risks model. in the remaining part of this section, we present the model and associated data requirement. 4.1 the analytical model we model exit behaviour of banks using a flexible bivariate competing risks model originally proposed by han and hausman (1990). several extensions of the model had been applied in different settings (for example, rosholm and svarer (2001), sueyoshi 1992, fallick and ryu 2007, and others). the presentation of the model follows the standard competing risks framework as in kalbleisch and prentice (1978). a model of competing risks requires definition and a specification of the distribution for a set of observable quantities. the followings are defined: i) there are n number of banks at risk of exit, i = 1, 2,…………, n ii) j is the distinct exit mode, j (1, 2,…….,j). iii) t  0 is the time until event j occurs, given that the event has not occurred before. j is unobserved if t is censored. iv) x is a vector of bank characteristics and other determinants. both continuous and discrete are included, x = x. assuming discrete time periods t = 1,2,….t, the outcome-specific hazard function for the i th bank with x vector of covariates specified for j th outcome is stated to include the corresponding proportional hazard specification.     );||,( );( lim 0 xttjjtttp xth jj i )exp()( 0 jji j i xth  j=1,2,…j (1) where j i h (.) gives the hazard rate for the i th bank from exit mode j at time t, given the vector of covariates x. the lower part of eq. (1) shows the proportional hazard specification corresponding to the outcome-specific hazard function. )( 0 th j i  0 is the outcome-specific baseline hazard, through which the form of dependence between m&a and failed banks can be examined. for the purpose of incorporating unobserved heterogeneity underlying the cbn incentive provision, the j th outcome baseline hazard is specified in the log form of the integrated hazard: international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 33 jijji j oi xdtth jt   )(log 0 (2a) which can be re-written as jijji j i xt  )( 0 (2b) where )()(log 0 0 0 tdtth jj i jt  denotes the log form of the integrated baseline hazard for the i th bank;  jt j iji dtxth 0 );(log is nonnegative outcome-specific error term, and its distribution is assumed to take an extreme value form. following han and hausman (1990), the log of the integrated outcome-specific baseline hazards is assumed to be constant in each time period, so that j t j t  )( 0 , (t=1, 2, …t). therefore, the hazard functions corresponding to the two exit modes (j=1, 2) by the i th bank can be specified as a bivariate competing risks model of the form: ii t it xdtth 111 0 1 0 1 )( 1    ; (3) and ii t it xdtth 222 0 2 0 2 )( 2    (4) where 1 t  and 2 t  are the hazards of exit by m&a and bank failure, respectively; (j=1, 2) where 1= m&a and 2= bank failure; t is the discrete time period within which the baseline hazard is assumed to be constant. x1 and x2 are vectors of covariates explaining exit by m&a and bank failure, respectively; 1 and 2 are vectors of parameters to be estimated for m&a and bank failure, respectively. 1221 )1( )( )1( ),( 22 2 0 11 1 0 11 1 0     ddf i i i xt xt xt      (5) where  ))1(( )1()( )1()( 11 1 011 0 1 0 2 0 2 0     i xt tt tt          .  ensures that the required relationship between the observed and latent failure time obtains for each discrete time interval. the term ))1(( 22 2 0   i xt is based on the assumption that for a given distribution of the error term for m&a, 1, the implied hazard rate for bank failure is greater than the implied hazard rate of m&a. in deriving this term, the evaluation points of support are assumed to agu et al.: mergers and acquisitions 34 change linearly over the discrete interval (han and hausman 1990). the linearity assumption ensures computational simplicity as it serves as one of the identifying assumptions (sueyoshi 1992). the linearity assumption is reasonable in our application given the potential presence of tied observations in our data. tied observations are present as exit of banks and censoring coincide with the end of the period stipulated by the cbn (december 31 2005). the term ),( 21 f denotes the density function, which allows for possible correlation, thus permitting dependence between the error terms of the model. since j=1, 2; the specification in eqs. (3) and (4) provide estimates of two sets of hazard functions j t ( ), which gives a flexible parametric counterparts of both bank failure and m&a hazards. 4.2 inferences in application, inference is based on the specification of a log likelihood function corresponding to the bivariate specification of the competing risks model in eqs (3) and (4). for the data setup, we assume that the n banks give rise to data which are observed at k discrete periods ti1 < ti2 < -------ti, k-1, and are identical for all n banks. we also observed either m&a, failure, or the observation on the i th bank is censored (i.e. stand alone banks) at time ti,ki. we assumed that data on x1 and x2 corresponding to time ti,ki , k = 1, 2,…., k-1 reflect measurable bank characteristics and other variables that may explain exit behaviour of banks. these variables may vary across intervals of time, but they are assumed to be constant within discrete time intervals. the censoring indicator di takes the value of 1 if exit mode is observed for the i th bank and the value 0 otherwise; dji = 1 if the i th bank exits by exit outcome j at time ti,ki and 0 otherwise. we threat the stand-alone banks as censored (i.e. banks that meet the capitalisation requirement on their own), as exit mode j is not observed for this group of banks. the log likelihood function can be stated as:           122 2 0 11 1 0 11 1 0 ))1(( 2121 )( )1( 1 1 ),(log)1()(     i i i xt xt xt i n i t t jij ddfddlogl +      211 1 0 22 2 0 22 2 0 )1( 2121 )( )1( ),(log     i i i xt xt xt i ddfd (6) where international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 35   ti xt  )))((( 11 1 011  ;   ti xt  /)))((( 22 2 022  ; and          )()( )()( 1 0 1 0 2 0 2 0 tt tt t    for t = 2, ….,k-1 with 1 = k = 1 the log likelihood function in eq. (6) is a function of j  , where ),,( j j ojj   . inference follows from the inverse of the hessian matrix of the log likelihood function where a finite vector parameterises the stochastic terms in the specification (han and hausman 1990). the bivariate competing risks model proposed here allows for considerable flexibility in the specification of the baseline hazard, while maintaining a parametric form for the function of explanatory variables. also, the model can easily accommodate features of discrete data such as tied observations and unobserved heterogeneity (cameron and trivedi 2005). ‘flexible’ of the model lies in the fact that the parametric assumption of the density function f does not impose any specific parametric forms on the outcome-specific baseline hazards. thus, distribution assumptions that have been used can be also be employed, including a discrete bivariate distribution (fallick and ryu 2007, rosholm and svarer 2001) and bivariate log-normal distribution (hans and hausman 1990). also, the flexible specification approach allows a straightforward introduction of unobserved heterogeneity underlying with cbn incentives, without the necessity of multiple integrations that would have arisen from the cox proportional hazard model proposed previously. hans and hausman (1990) show that the non-parametric specification of the unobserved heterogeneity is convenient, easier to estimate, and yields an asymptotic (normal) estimator that is consistent with standard large sample. 4.3 identification a key issue in dependent competing risks models is identification of the key parameters. lee (2005) shows how a transformed model can be used to identify important features of a dependent competing risks model, such as j, j 0  , and the joint distribution of ),( 21  . the key identification condition for the bivariate competing risks model is that the number of continuous variables is at least as great as the number of competing risks in the model, even if the xj covariates are identical (hans and hausman 1986, 1988). this identification result and others have been applied in several studies, especially in empirical studies of unemployment durations (for example; hans and hausman 1990; fallick and ryu 2007, sueyoshi 1992). in the present agu et al.: mergers and acquisitions 36 application, this identification implies that at least two covariates of xj in our dataset are continuous. in estimation of eq. (3) and (4), we have identical explanatory variables; x1 = x2, consisting of both continuous and discrete variables. other less stringent identification conditions include sufficient variations in the covariates; covariates are not perfectly collinear, and the j th baseline hazards, j 0  , are not perfectly related (cameron and trivedi 2005). 4.4 data requirement this paper seeks to use the competing risks duration models to investigate the determinants of exit behaviour of banks in the nigerian consolidation program. there is no consensus as to the type of data that provide the best estimates or the type of variables that should be considered in an analysis of bank m&a and failures. studies cited earlier typically use both cross-sectional and panel data of banks. in their application of logit models, soyibo et al. (2004) found similar results using both cross-sectional and panel data. however, because duration analysis is based on specific time covered in the study, panel data have been found to provide better estimates in duration models generally relative to crosssectional data. also, using both time-varying covariates rather than only non-time varying covariates had been found to increase the predictive power of models using panel data relative to similar cross-section models (sales and tannuri-pianto 2005). the only study that has compared banking failure and acquisition (wheelock and wilson 2000) used explanatory variables representing only time-varying bank-specific characteristics. in the present paper therefore, we propose to use a panel data of all the 89 banks operating in nigeria during the period january 2001 – december 31 2005. data for the period 2001-2004 will be used to estimate the competing risks model. using the data from three years before the program will allow us to account for performance history of the individual banks. 5. estimation methods and findings estimations were carried out at two stages. the first stage involved estimating two models using the standard competing risks models, consisting of bank level model (‘model 1’), and controlling industry specific and economy wide effects in ‘model 2’. in both models, bank failure and m&a were treated as independent risks; and the key dependence variable (cbn incentive) was international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 37 included as an explanatory variable in the failure hazard equation. table 4 presents estimation results for the independence case. at the second stage of the analysis, we focused on the structural dependence among the determinants. the cbn assistance is the key variable inducing the structural dependence between failure and m&a. thus, we modelled the cbn assistance explicitly as a function of the political/ethnic considerations underlying the consolidation programme and other relevant variables. the cbn assistance and failure probabilities were jointly estimated, allowing the failure probability to affect cbn assistance. explicit specification of a cbn assistance equation has an added advantage of attempting to examine potential heterogeneity underlying cbn incentives directly. tables 4 and 5 present estimation results. 5.1 independent competing risks case bank characteristics: age of bank (bankage) has a quadratic effect on failure hazard; with failure hazard increasing in relatively younger banks and decreases in older banks. the critical age (inflection point) is at approximately 18.5 years in ‘model 1’, suggesting that banks below 18.5 years were more likely than older banks to fail. the average age of failed banks is approximately 14 years, which is below the sample average of approximately 16 years. thus, the increasing part of the quadratic function is relevant for failure hazard. a quadratic effect was also observed on the m&a hazard, but the effect is statistically not different zero. the number of branches is a measure of size of banks. however, this has no effect on either failure or m&a probabilities. as expected, the share of small-medium enterprises loans in a bank’s total loans (smeloans) increased the risk of failure and decreased the m&a probability. comparing the magnitudes of the coefficients, the extent to which increasing smeloans increases the likelihood of failure appears stronger than the extent to which it decreases the likelihood of m&a. the risk of bank failure decreases with shareholders fund, but has no effect on m&a probability. however, the capital base of the bank significantly decreases the probability of m&a, but has no effect on the risk of bank failure. the share of private ownership provides an indication of efficiency of operations (ref.). share of private ownership strongly reduced the risk of failure but did not increase the probability of agu et al.: mergers and acquisitions 38 m&a either. however, the decreasing effect of share of private ownership on risk of failure was much stronger than the decreasing effect on probability of m&a. cbn assistance (cbnassist) is key variable of interest in the analysis, as it provides an indication of the cbn attempt to assist weaker banks towards m&a. as expected, cbn assistance strongly reduced the risk of bank failure and increased the probability of m&a. comparing the magnitudes of the effects, the extent to which cbn assistance reduced bank failure was significantly larger than the extent to which it increased m&a probability. the share of non-performing loans in total loans provides an indication of the bank’s credit risk. as expected, higher non-performing loans significantly increased the risk of bank failure, but no statistically significant effect on the probability of m&a. however, the coefficient of the interaction of non-performing loans with cbn assistance showed that the risk of failure declined significantly and probability of m&a increased significantly if the bank were to be cbn assisted, compared to when the bank was not assisted. the state of origin of bank ceos was used to proxy the geo-political/ethnic dimensions of the consolidation programme and to some extent capture the heterogeneity between banks. compared to the baseline category of bank ceos (south-west), the risk of failure increased significantly if the bank ceo hailed from the south-south or south-east geo-political zone, while the risk of bank failure decreased significantly if the bank ceo hailed from the northern geopolitical/ethnic zone or foreigner. however, the state of origin of bank ceo has no effect on the probability of m&a. the above results changed dramatically when the state of origin of bank ceos was interacted with non-performing loans. compared to the baseline category, the risk of failure reduced significantly if the ceo hailed from the south-south or south-east geo-political zone, while the risk of bank failure increased significantly if the bank ceo hailed from the northern geo-political/ethnic zone. the state of origin of banks generally has no effects on the probability of m&a. industry specific and economy wide characteristics: following the literature on determinants of bank failure and m&a, we included index of manufacturing production and agricultural loans for the period 2001-2003 to control for industry and economic wide covariates, respectively (model 2). however, the inclusion of these factors did not change the results. there was no statistical evidence that these factors were influential in determining the exit behaviour of international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 39 table 4: determinants of exit of banks: independence case model 1 model 2 variables failure m&a failure bank failure bank characteristics: bankage 2.446*** 0.0330 2.446*** 0.0330 (0.595) (0.0267) (0.595) (0.0267) bankage sq. -0.0659*** -0.000249 -0.0659*** -0.000249 (0.0155) (0.000584) (0.0155) (0.000584) critical age a 18.544*** (0.385) 66.208 (104.79) 18.544*** (0.385) 66.196 (104.72) branches 0.0125 -0.000741 0.0125 -0.000741 (0.00852) (0.000736) (0.00852) (0.000736) smeloans 0.615*** -0.0444*** 0.615*** -0.0444*** (0.124) (0.0163) (0.124) (0.0163) shareholders fund -0.940*** -0.0525 -0.940*** -0.0523 (0.352) (0.123) (0.352) (0.123) capital base -0.107 -0.0728* -0.107 -0.0729* (0.108) (0.0380) (0.108) (0.0380) share of private ownership -2.874*** -0.255*** -2.874*** -0.255*** (0.676) (0.0881) (0.676) (0.0882) cbn assist -20.90*** 0.408*** -20.89*** 0.408*** (1.510) (0.149) (1.514) (0.149) non-performing loans 0.279*** -0.00442 0.279*** -0.00442 (0.0732) (0.00858) (0.0732) (0.00858) non-performingloans*cnb assist -0.1903** 0.0159** -0.1903** -0.1903** (0. 059) (0.0060) (0.0060) (0. 059) bank ceo (ss/se) 5.592** 0.0587 5.592** 0.0587 (2.480) (0.374) (2.480) (0.374) bank ceo (north) -31.18*** 0.271 -31.18*** 0.271 (7.760) (0.287) (7.760) (0.287) bank ceo (foreign) -4.390* -0.871** -4.390* -0.871** (2.346) (0.437) (2.346) (0.437) non-performing loans*ceo(ss/se) -0.178*** 0.000206 -0.178*** 0.000201 (0.0624) (0.0129) (0.0624) (0.0129) non-performing loans*ceo(north) 0.683*** -0.0164 0.683*** -0.0164 (0.173) (0.0123) (0.173) (0.0123) non-performing loans*ceo (for.) 0.0737 0.0159 0.0737 0.0159 (0.0638) (0.0157) (0.0638) (0.0157) industry and economy wide variables: index of manufacturing production -0.00107 -0.00498 (0.746) (0.343) agric loan 0.0293 0.136 (23.33) (10.72) n 267 267 267 267 model fit statistics: deviance 321.41 2296.88 321.41 2296.88 aic 351.41 2326.88 355.41 2330.88 aib 405.22 2380.69 416.39 2391.87 notes: robust standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1. a critical age is obtained by using the stata v.10 command nlcom on estimated (bankage) / (-2*bankage sq.). aic: akaike’s information criterion; bic: bayesian information criterion; deviance: -2*log likelihood. agu et al.: mergers and acquisitions 40 banks in the consolidation programme. 10 this result may be reflecting the fact that there were little variations in these variables during the period considered in this analysis. 5.2 structural dependence between failure and m&a hazards induced by cbn incentive in the joint estimation of the failure and cbn assistance probabilities as described earlier, we also compared the hans and hausman (1990) and extended by rosholm and svarer (2001) approach by incorporating the failure hazard into the cbn assistance (‘model 3’) with fukumoto (2005) by explicitly estimate the dependence parameter (‘model 4’). table 5 presents estimation results for the structural dependence case. the quadratic effect of bank age remained consistent. however, the critical age at which a bank is more likely to fail was reduced slightly from the independent competing risks case. interestingly, bank age became influential in determining whether a bank received cbn assistance. compared with the mean age of 16 years in the data, the critical age at which a bank was more likely to receive cbn assistance is approximately below 26 years. this is consistent with the mean age of approximately 18 years for banks that received the cbn incentive (comparable to the critical age below which banks were more likely to fail). in terms of structural dependence between failure and m&a therefore, this result seems to suggest that more banks that would have failed by virtue of their age above 18 years were less likely to do so if they receive cbn assistance. other strong predictors of bank failure in the structural dependence case include vulnerability of banks to external shocks (as measured by sme loans), non-performing loans (much stronger than in the independent competing risks case), especially if the bank ceo hailed from the northern geo-political/ethnic zone of the country. the remaining results remained generally consistent with the independent competing risks case, though the magnitudes of the coefficients appear to differ. on the other hand, strong predictors of cbn assistance include age of bank and vulnerability to external shocks. there was also a geo-political/ethnic dimension to the probability of cbn assistance. compared to the baseline category, banks whose ceo hailed from the south-south or south-east geo-political/ethnic zone were strongly less likely to receive 10 we also experimented with other variables including such as forbearance and public confidence did not change the result. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 41 cbn assistance. if non-performing loans were considered however, banks whose ceo hailed from the northern geo-political/ethnic zone were strongly more likely to receive cbn assistance. table 5: determinants of bank failure with structural dependence model 3 model 4 variables failure cbn assist failure cbn assist bankage 1.117* 0.228*** 1.053* 0.360*** (0.594) (0.0817) (0.548) (0.139) bankage sq. -0.0315* -0.00438** -0.0293** -0.00781** (0.0164) (0.00186) (0.0146) (0.00329) critical age a 17.736*** (0.872) 26.021*** (2.679) 17.987*** (1.021) 23.025*** (1.445) smeloans 0.487*** 0.0869* 0.500*** 0.0883* (0.134) (0.0483) (0.132) (0.0537) shareholders fund -1.482*** -1.550*** (0.505) (0.497) capital base -0.142 -0.136 (0.139) (0.137) share of private ownership -1.031* -1.034* (0.606) (0.627) non-performing loans 8.175*** 0.469 8.491*** 0.313 (2.015) (0.537) (1.907) (0.528) bank ceo (ss/se) -41.82*** -1.307*** -63.24*** -31.03** (12.58) (0.310) (10.47) (12.65) bank ceo (north) -17.25** -0.263 -17.27*** -0.626 (6.768) (0.510) (6.559) (0.482) bank ceo (foreign) 0.274*** 0.282*** (0.0691) (0.0663) non-performing loans*ceo(ss/se) -0.266*** -0.0141 -0.276*** -0.00966 (0.0629) (0.0158) (0.0589) (0.0158) non-performing loans*ceo(north) 0.958*** 0.0369*** 1.620*** 0.918** (0.281) (0.0110) (0.284) (0.369) non-performing loans*ceo(for.) 0.416** 0.414** 0.0241* (0.185) (0.179) (0.0145) dependence: dependence parameter -12.21*** -1.516 (1.906) (1.614) failed (exittype=2) -15.35*** (4.042) constant -12.63* -4.435*** -12.05* -5.589*** (6.602) (1.033) (6.429) (1.639) n 267 267 267 267 deviance 211.76 200.42 aic 255.76 252.42 aib 334.68 345.69 robust standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1. a critical age is obtained by using the stata v.10 command nlcom on estimated (bankage) / (-2*bankage sq.). aic: akaike’s information criterion; bic: bayesian information criterion; deviance: -2*log likelihood. agu et al.: mergers and acquisitions 42 from the south-south or south-east geo-political/ethnic zone were strongly less likely to receive cbn assistance. if non-performing loans were considered however, banks whose ceo hailed from the northern geo-political/ethnic zone were strongly more likely to receive cbn assistance. in the raw data, the mean non-performing loans was approximately 26.3% in the baseline category, where the bank ceo hailed from the south-west compared with approximately 30.4% and 27.8% where the bank ceo hailed from the south-south/east and the north, respectively. therefore, it can be concluded that the cbn assistance on the basis of the geo-political/ethnic appeal of the bank ceos was sensitive to the magnitude of non-performing loans they have approved. the coefficient of the dependence parameter showed a strong negative correlation between cbn assistance and bank failure (‘model 3’). similar results were obtained when the failure hazard was allowed to affect cbn assistance, with strong negative correlation (‘model 4’). also, the fact that the significance of the coefficient of dependence parameter disappeared in ‘model 4’, suggesting that allowing the failure hazard to affect the cbn assistance adequately captured the structural dependence. finally, comparing the deviance, aic, and aib fit statistics across all the estimated models, it appears that taking into account the structural dependence induced by the cbn assistance provide better fit for the data. on the basis of deviance and aic criteria, allowing the failure hazard to affect the cbn assistance appears to provide a better fit for the structural dependence models. if the aib criterion is also considered, it is not immediately clear which of the structural dependence models provides the best fit. 6. conclusion this result conforms to expectation that cbn incentive provision prevented banks that would otherwise have failed. the influences on cbn assistance also became clearer, with older banks more likely to receive cbn assistance (more than the critical age below which the risk of failure increased). six out of the nine banks that the cbn assisted were aged below 18 years. also, it appeared that the cbn was sympathetic to those banks that were more vulnerable to external shocks by the proportion of sme loans in total loans. in the past, the cbn had encouraged banks to increase their sme loans to support the development and growth of smes in nigeria (sanusi international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 43 2003). the average sme loans by cbn-assisted bank was approximately 12% points higher than the overall industry average. this may explain the positive association. also, there was a clear geo-political/ethnicity gradient in the exit behaviour of banks. looking at the state of origin of bank ceos alone, it would appear that the banks from the northern geo-political/ethnic zone were favoured or prevented from failure. but this was true only if the non-performing loans were not considered, because the risk of failure increased significantly higher than if the ceo hailed from the south-west once the non-performing loans were considered. the reverse was the case for the banks from the south-south or south-east geopolitical/ethnic zone, in which the risk of failure reduced significantly. thus, it appears that the non-performing loans factor (the credit risk of banks) affected the northern banks relative to their southern counterparts. finally we found no evidence supporting the argument by ezeoha (2007) that the economy wide situation in the country would influence the outcomes of the consolidation programme. author information: chukwuma agu, the submitting author, is a staff member of the african institute for applied economics, park avenue, gra, enugu p.o. box 2147, enugu, nigeria. he may be contacted at e-mail: shookslife@yahoo.com or at phones (234)-4225-6644; (234)-803598-0798 (cell). the co-authors of this article are: damilola olajide and divine ikenwilo at the university of aberdeen, scotland; and anthony orji at the university of nigeria, nsukka, nigeria. agu et al.: mergers and acquisitions mailto:shookslife@yahoo.com 44 references ahamd, r., ariff, m., and skully, m., (2007). factors determining mergers of banks in malaysia’s banking sector multinational finance journal vol. 11 (nos 3-4): 177-194. amel, d. f., and rhoades, s. a., (1989). empirical evidence on the motives for bank mergers, eastern economic journal 15: 95-139. andrade, g., m. mitchell; stafford, e., (2001). new evidence and perspectives on mergers, the journal of economic perspectives vol. 15 (2): 103-120. berg, van der (2005). competing risks models, working paper 25, department of economics, free university, amsterdam. borek, t., buehler s., and schumtzler, a., (2002). weddings with uncertainty prospects mergers under asymmetric information? working paper 0213, socioeconomic institute, university of zürich buehler, s., c. kaiser, jaeger, f., (2005). merge or fail? the determinants of mergers and bankruptcies in switzerland, 1995-2000, working paper no. 0506, university of zurich, switzerland. cameron, c. a. and trivedi, p.k., (2005). microeconometrics: methods and applications, cambridge university press: ch.19.2.3: 646. central bank of nigeria (2004). guidelines and incentives on consolidation in the nigerian banking industry, press release, governor’s office, august 25. central bank of nigeria (2005). banking sector consolidation: special incentives to encourage weaker banks, press release, governor’s office, april 11. central bank of nigeria (2005). banking sector consolidation: special incentives to encourage weaker banks, press release: april 11. central bank of nigeria (2006). bank consolidation: 25 banks meet december 31 st deadline, press release, governor’s office, jan. 3. cox, d.r., (1972). regression models and life tables (with discussion), journal of the royal statistical society 34b: 187-220. cox, d.r., (1975). partial likelihood, biometrika 62:269-276. dabos, m., and escudero, w.s., (2004). explaining and predicting bank failure using duration models: the case of argentina after the mexican crisis, revista de analisis economico vol. 19 (1): 31-49. diamond, p.a., and hausman, j.a., (1984). the retirement and unemployment behaviour of older men, in h. aaron and g. burtless (eds.) retirement and economic behaviour, brookings institute, washington, dc. ely, d.p., and song, m.h., (2000). acquisition of large depository institutions in the 1990s: an empirical analysis of motives, the quarterly review of economics and finance, vol. 40 (4): 467-484. ezeoha, abel, (2007). structural effects if banking industry consolidation in nigeria: a review, journal of banking regulation 8: 159-176. fallick, b.c., and ryu, k., (2003). the recall and new job search of laid-off workers: a bivariate proportional hazard model with unobserved heterogeneity, the review of economics and statistics 89(2): 313–323. glick, r., and hutchison, m., (1999). banking and currency crises: how common are twins’, pacific basin working paper 99-07, federal reserve bank of san francisco. han, a., and hausman, j.a., (1986) semiparametric estimator of duration and competing risk models, working paper, mit. han, a., and hausman, j.a., (1990). flexible parametric estimation of duration and competing risk models, journal of applied econometrics vol. 5, no. 1: 1-28. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 http://www.soi.unizh.ch/research/wp/wp0506.pdf http://www.soi.unizh.ch/research/wp/wp0506.pdf 45 han, a. and hausman, j.a., (1988). identification of continuous and discrete competing risks models, mimeo, mit. hannah, t.h., and rhoades, s.a., (1978). acquisition targets and motives: the case of the banking industry, the review of economics and statistics vol. 69 (1): 67-74. hardlock, c., houston, j., and ryngaert, m., (1999). the role of managerial incentives in bank acquisitions, journal of banking and finance 23: 221-249. harhoff, d., stahl k. and woywode, m., (1998). legal form, growth and exit of west german firms--empirical results for manufacturing, construction, trade and service industries, the journal of industrial economics vol. 46, (4): 453-488. imala, i. o., (2005). challenges of banking sector reforms and bank consolidation in nigeria, bullion vol. 29 (2) central bank of nigeria: 26-37 kwan, s., (2004). banking consolidation’, frbsf economic letter, 15, 18 june; http://www.frbsf.org/publications/economics/letter/2004/el2004-15.pdf. lee, s., (2005). identification of a competing risks model with unknown transformation of latent failure times, working paper cwp17/05, the institute of fiscal studies, ucla. lillard, l. a., (1993). simultaneous equations for hazards: marriage duration and fertility timing, journal of econometrics vol. 56 (1-2): 189-217. o’keefe, j.p., (1996). banking industry consolidation: financial attributes of merging banks, fdic banking review vol. 9 (1): fall. ogubunka, u. m., (2005). banking sector reforms and bank consolidation: the experience of turkey, bullion, vol. 29 (2) central bank of nigeria: 17-25. olajide, d., (2005). the changing banking environment in nigeria and emerging public policy issues, discussion paper, initiative of public policy analysis, no. 10-2005, central bank of nigeria. peel, m. j., and wilson, n., (1989). the liquidation/merger alternative. some results for the uk corporate sector, managerial and decision economics vol. 10 (3): 209-220. prentice, r. l., and kalbfleisch, j.d., (1980). the analysis of failure times in the presence of competing risks, biometrics vol. 34 (4): 541-554. rosholm, m., and svarer, m., (2001). structurally dependent competing risks, economics letters vol. 73 (2): 169-173. sales, a. s., and tannuri-pianto, m.e., (2005). the use of duration models to explain bank failures in brazil (1994-1998), research paper, the central bank of brazil. sanusi, j.o., (2003). overview of government’s efforts in the development if smes and the emergence of small and medium industries equity investment scheme(smieis), presented at the national summit on smieis organised by the bankers’ committee and lagos chamber of commerce and industry(lcci), lagos, 10 th june. soludo, c.c., (2004). consolidating the nigerian banking industry to meet the development challenges of the 21st century, being text of address delivered at the special meeting of the bankers’ committee, july 6, cbn headquarters, abuja. soyibo, a., alashi, s. o., and ahmad, m.k., (2004). a positive and normative analysis of bank supervision in nigeria, african economic research consortium research paper no. 145, nairobi. thomson, j.b., (1991). predicting bank failures in the 1980s, economic review 27, federal reserve bank of cleveland: 9-20. uchendu, a.o., (2005). banking sector reforms and bank consolidation: the malaysian experience, bullion, vol. 29 (2) central bank of nigeria: 12-16. whalen, g., (1991). a proportional hazards model of bank failure: an examination of its usefulness as an early warning tool, economic review q 1, federal reserve bank of cleveland: 21-31. agu et al.: mergers and acquisitions http://www.frbsf.org/publications/economics/letter/2004/el2004-15.pdf http://econpapers.repec.org/article/eeeeconom/ 46 wheelock, d.c., and wilson p.w., (2000). why do banks disappear? the determinants of us bank failures and acquisitions, the review of economics and statistics vol. 82 (1): 127-138. international journal of banking and finance, vol. 8, iss. 4 [2011], art. 2 rational expectations, irrational exuberance: linkage between u.s. investors and pacific-basin stock returns the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 67 rational expectations, irrational exuberance: linkage between u.s. investors and pacific-basin stock returns rahul verma the university of houston, united states of america abstract we shed new light on the relevance of rational expectations and irrational exuberance of u.s. individual and institutional investors on pacific-basin stock returns. we find insignificant effects of irrational exuberance and significant effect of rational expectations on asian markets with varying degrees of intensity. there are greater responses of hong kong, malaysia, philippines, and singapore while weaker linkages with taiwan, thailand, and korea. overall evidence suggests that rational expectations of institutional investors are transmitted to a greater extent than those of individual investors. these results are consistent with the view that international effects of the u.s. market can be attributed to rational investor sentiments. key words: stock returns, investor sentiment, var model, asia pacific markets jel classification: g12, g14, c22 1. introduction the central task in financial economics is to identify the systematic risks that drive asset prices and expected returns (campbell, 2000; cochrane, 2000). however, in recent years there has been a growing debate on the possible linkages between the behavioral aspects of investors and stock prices. financial economics has become more receptive to imperfect rational explanations, and investor psychology has emerged as a major determinant of asset prices. after decades of study, the sources of risk premium in purely rational models are well understood; while, dynamic psychology based asset pricing theories are still in ijbf 68 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 the infancy stage. this debate surrounding asset pricing has identified two prime suspects in setting stock prices: fundamentals and investor sentiments. the theoretical framework describing the role of investor sentiments in determining stock prices is provided by researchers such as black (1986), trueman (1988), delong, shleifer, summers and waldman [dssw henceforth] (1991, 1990), shleifer and summers (1990), lakonishok, shleifer and vishny (1991), campbell and kyle (1993), shefrin and statman (1994), palomino (1996), barberis, shleifer and vishny (1998), daniel, hirshleifer and subramanyam (1998) and hong and stein (1999). a direct implication of these studies is certain groups of investors (noise traders), who often do not make investment decisions based on a company’s fundamentals, are capable of affecting stock prices by way of unpredictable changes in their sentiments. following the ‘noise trader model’ of dssw (1990), several empirical studies examine the influence of investor sentiments on stock prices (brown and cliff, 2004a, 2004b; lee et al. 2002; fisher and statman, 2000; clarke and statman, 1998; solt and statman 1988; de bondt, 1993). overall, these studies provide evidence in favor of strong co-movements between investor sentiment and the stock market returns recognizing the existence of individual investor sentiments, as well as institutional investor sentiments. the previous research mainly focuses on the effect of investor sentiments on the u.s. market while less attention has been given to its relevance in the international context. for example, little has been done to understand the degree of the relationship between the u.s. individual and institutional investor sentiments and pacific-basin stock returns. given strong empirical evidence on the strengthening response of pacific-basin stock markets to the u.s. market over time (soydemir, 2005; kim, 2003; ratanapakorn and sharma, 2002; janakiramanan and lamba, 1998; park and fatemi, 1993) it is important to analyze whether the expectations of the u.s. investors is an important player in propagating u.s. stock market movements abroad. we shed new light on the relevance of the rational expectations and irrational exuberance of the u.s. investors in determining stock returns of hong kong, malaysia, philippines, singapore, taiwan, thailand, and korea. using the investor sentiments data at the individual and institutional level, provided by american association of individual investors and investors intelligence and the vector auto regression (var) models we find the following results: first, we do not find any significant effect of irrational exuberance of the u.s. investors on pacific-basin stock market returns. second, we find significant relationship of varying degrees of strength between the rational expectations of the u.s. individual and institutional investors and asian stock returns except in the cases of taiwan and thailand. third, there are greater effects of rational expectations of the institutional investors than individual investors on theses stock returns. these findings suggest that the international effects of the u.s. stock market can be attributed to fundamental trading and not to noise trading in the u.s. market. these results are consistent with the view that investor sentiment in the u.s. is an important player in propagating u.s. stock market the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 69 movements abroad. therefore it is important for policymakers to consider such spillover effects in their international policy making decisions and for investors in their portfolio allocation decisions involving stock markets movements. this remainder of this paper is organized as follows: section two reviews the existing literature on investor sentiments and stock prices while section three presents the model. section four presents the data. section five reviews the econometric methodology. section six presents empirical findings. section seven concludes. 2. previous work on investor sentiments and stock prices the concept of investor sentiments, noise trading and its role in the financial markets is first given by black (1986). black (1986) labels non rational investors as “noise traders”, who have no access to inside information, and act irrationally in response to news that conveys little information about fundamentals. however, there are two opposing views in the literature on the relevance of noise traders in determining stock prices. based on friedman (1953) and fama (1965), it is argued that noise traders are irrelevant and cannot survive since they are driven out of the market by rational arbitrageurs. for example, west (1988) states “there is little direct evidence that trading by naïve investors plays a substantial role in stock price determination”. on the contrary, black (1986) and trueman (1988) argue that noise traders induce necessary liquidity in the market, and therefore provide incentives for informed investors to trade. the notable work of dssw (1990, and 1991) models the influence of noise trading on equilibrium prices. they argue that noise traders falsely believe that they have special information about the future prices and the unpredictability of their sentiments brings an additional risk in the market. they may get pseudo signals from analysts, brokers, consultants and irrationally believe that these signals carry information. their ‘noise trader’ model shows that a nonfundamental factor exists in the form of investor sentiments that is priced in equilibrium. furthermore, noise traders as a group can earn expected returns higher than rational investors and can also survive in terms of wealth gain in the long run, due to unpredictability in their sentiments. campbell and kyle (1993) model the competitive interaction between noise and informed traders and its consequent effect on stock prices. shleifer and summers (1990) present an alternative to the efficient markets paradigm that stresses the role of investor sentiments and limited arbitrage in determining stock prices. they show that the assumption of limited arbitrage is more plausible as a description of risky asset markets than the assumption of complete arbitrage on which market efficiency hypothesis is based. this implies that changes in investor sentiments are not fully countered by arbitrageurs therefore may affect stock returns. similarly, shefrin and statman (1994) show the interaction between noise and informed traders and present the behavioral capital asset pricing theory. they argue that in contrast to information, sentiments of noise traders’ act as a second driver which takes the market away from efficiency. 70 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 lakonishok, shleifer and vishny (1991) find that in small markets institutional investors influence prices. along the same lines, palomino (1996) extends the dssw (1990) model for an imperfectly competitive market and shows that noise traders may earn higher return and obtain higher expected utility than rational investors. wang (2001) examines the dynamics of non-rational investors find that bullish sentiments can survive while bearish sentiment cannot survive in the long run. overall, these models suggest that the unpredictability in investor sentiments of noise traders acting as a group can introduce a systematic risk that is priced in markets. following these predictions, several empirical studies have examined the role of investor sentiments on stock pricing. these studies have either used indirect measures or direct measures of investor sentiments. studies using indirect measures include the following proxies: close-ended fund’s discount (gemmill and thomas, 2002; baker & wurgler, 2005; sias, starks and tinic, 2001; neal and whitney, 1998; swaminathan, 1996; elton, gruber and busse, 1998; chan, kan and miller, 1993; lee, shleifer and thaler, 1991); market performance based measures (brown and cliff, 2004a); trading activity based measures (brown and cliff, 2004a; neal and whitney, 1998); derivative variables (brown and cliff, 2004a); dividend premium (baker and wurgler, 2005); and ipos related measures (baker & wurgler, 2005; brown and cliff, 2004a). overall these studies do not provide a consensus on whether the proxies chosen are appropriate measures of investor sentiment and also show mixed results in their debate on the linkages between sentiments and stock returns. studies using direct measures employ sentiment surveys data that indicate the expectations of market participants. research related to individual investors sentiments find strong co-movements with stock market returns (brown and cliff, 2004a; de bondt, 1993) and mixed results regarding its role in short term predictability of stock prices (brown and cliff, 2004a; fisher and statman, 2000). similarly, studies examining institutional sentiments find strong co-movements with stock market returns (brown and cliff, 2004a) and mixed results regarding its short run implications on stock prices (brown and cliff, 2004a; lee, jiang and indro, 2002; clarke and statman, 1998; solt and statman, 1988). also, brown and cliff (2004b) examine the long run implications of institutional investor sentiments and find strong relationships with long horizon stock returns. overall, these studies provide powerful and consistent empirical support for the hypothesis that stock prices are affected by individual and institutional investor sentiments. 3. model previous studies suggest that some shifts in investor sentiments are fully rational i.e., expectations based on the risk factors, while some are irrational exuberance, induced by the noise (baker and wurgler, 2005; brown and cliff, the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 71 2004b, shleifer and summers, 1990). hirshleifer (2001) also relates expected returns to both risks and investor mis-valuation. when an investor is bullish or bearish, then this could be a rational reflection of future period’s expectation or irrational enthusiasm or a combination of both. therefore it is quite possible that international stock returns are affected by both rational (risk based) and irrational (noise) components of the u.s. investor sentiments. we follow the approach of baker and wurgler (2005) to capture the irrational component of investor sentiments by regressing sentiment indicators to a set of risk factors and computing the residuals. accordingly, we formulate equations (1) and (2): (1) (2) where γ 0 and θ 0 are constants, γ j and θ j are the parameters to be estimated; ζ t and t are the random error terms. sentt 1t and sentt 2t represent the shifts in sentiments of individual and institutional investors respectively at time t. fund jt is the set of fundamentals representing rational expectations based on risk factors that have been shown to carry non-redundant information in conditional asset pricing literature. the fitted values of equations (1) and (2) capture the rational component of sentiments (i.e. and ). on the other hand the residual of equations (1) and (2) capture the irrational component of sentiments (i.e. ζ t and t ). next, we analyze the extent to which pacific-basin stock returns are affected by rational expectations and irrational exuberance of the u.s. individual and institutional investors. accordingly, the sentiment variables are decomposed into the rational and irrational components based on equations (1) and (2) and included in the return generating process as: (3) where α 0 is a constant while α 1 , α 2 , α 3 and α 4 are the parameters to be estimated; ρ t is the random error term. r it is the returns for the ith pacificbasin stock market at time t. specifically, the parameters α 1 and α 2 capture the effects of rational expectations on part of individual and institutional investors respectively; while α 3 and α 4 capture the effects of irrational exuberance of individual and institutional investors respectively. we also place importance on jointly modeling the sentiments of individual and institutional investors to avoid misspecification. specifically, shocks originating from sentiments of one class of investors not considered might mistakenly be seen as a disturbance originating from sentiments of another class of investors in the analysis. ∂ 1̂tsentt 2̂tsentt ∂ 0 1 1 2 2 3 4 ˆ ˆ it t t t t tr sentt senttα α α α ξ α � ρ= + + + + + 1 0 1 j t j jt t j sentt fundγ γ ξ = = + +∑ 2 0 1 j t j jt t j sentt fundθ θ ϑ = = + +∑ 72 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 4. data we obtain all data in monthly intervals from october 1988 to april 2004. to measure sentiments of market participants, we employ survey data similar to the ones used in the literature. the institutional investors participate in the market for living while the individual investors’ primary line of business is outside the stock market (brown and cliff, 2004a). our choice of individual investor sentiment index is based on brown and cliff (2004a), fisher and statman (2000) and debondt (1993) which use the survey data of american association of individual investor (aaii). beginning july 1987, aaii conducts a weekly survey asking for the likely direction of the stock market during the next six months (up, down or the same). the participants are randomly chosen from approximately 100,000 aaii members. each week, aaii compiles the results based on survey answers and labels them as bullish, bearish or neutral. these results are published as ‘investor sentiment’ in monthly editions of aaii journal. the sentiment index for individual investors is computed as the spread between the percentage of bullish investors and percentage of bearish investors (bullbear). since this survey is targeted towards individual investors, it is primarily a measure of individual investor sentiments. our choice of institutional investor sentiment index is based on brown and cliff (2004a, 2004b), lee et al. (2002), clarke and statman (1998) and solt and statman (1988) which use the survey data of investors intelligence (ii), an investment service based in larchmont, new york. ii compiles and publishes data based on a survey of investment advisory newsletters. to overcome the potential bias problem towards buy recommendation, letters from brokerage houses are excluded. based on the future market movements the letters are labeled as bullish, bearish or correction (hold). the sentiment index for the institutional investor is found by calculating the spread between the percentage of bullish investors and percentage of bearish investors. because authors of these newsletters are market professionals, the ii series is interpreted as a proxy for institutional investor sentiments. we analyze the response of seven pacific-basin stock markets, korea, malaysia, philippines, singapore, taiwan, thailand and korea. the market variable identified for these countries are the major indexes in their respective stock markets. the continuously compounded returns are computed from the local currency denominated stock market indexes obtained from the datastream. we include the following variables as risk factors that have been shown to carry non-redundant information in the asset pricing literature: (i) economic growth (fama, 1970; schwert, 1990) measured as the monthly changes in the industrial production index (ii) short term interest rates (campbell, 1991) measured as the yield on one month u.s. treasury bill (iii)economic risk premia (ferson and harvey, 1991; campbell, 1987) measured as the term structure of interest rates (difference in monthly yields on three month and one month treasury bills (iv) future economic expectations variables (fama, 1990) measured as the term spread (yields spread on the 10 year u.s. treasury bond the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 73 and three month treasury bill) (v) business conditions (fama and french, 1989; keim and stambaugh, 1986) measured as the default spread (difference in yields on baa and aaa corporate bonds)(vi) dividend yield (hodrick, 1992; fama and french, 1988; campbell and shiller, 1988a, 1988b) measured as the dividend yield for the value weighted center for research in security prices (crsp) index over the past 12 months (vii) inflation (sharpe, 2002; fama and schwert, 1977) measured as the monthly changes in the consumer price index (viii) excess returns on market portfolio(lintner, 1965; sharpe, 1964) measured as the value-weighted returns on all nyse, amex, and nasdaq stocks minus the one-month treasury bill rate (ix) premium on portfolio of small stocks relative to large stocks (smb) (fama and french, 1993). smb (small minus big) is the average return on three small portfolios minus the average return on three big portfolios (x) premium on portfolio of high book/market stocks relative to low book/market stocks (hml) (fama and french, 1993). this fama/french benchmark factor is constructed from six size/book-to-market benchmark portfolios that do not include hold ranges and do not incur transaction costs. hml (high minus low) is the average return on two value portfolios minus the average return on two growth portfolios. (xi) momentum factor (umd) (jegadeesh and titman, 1993). umd (up minus down) is the average return on the two high prior return portfolios minus the average return on the two low prior return portfolios (xii) currency fluctuation (elton and gruber, 1991) measured as the changes in 15-country trade weighted basket of currencies. the data on economic growth, business conditions and inflation are obtained from datastream; short term interest rates, economic risk premium, future economic variables and currency fluctuations are obtained from federal reserve bank of st. louis; dividend yield and excess return on market portfolio from crsp; and smb, hml and umd from kenneth french data library at tuck school of business, dartmouth college. table 1 reports the descriptive statistics of the above mentioned variables. the mean of sntt 1 and sntt 2 are approximately 11% and 9% respectively. this suggests both individual and institutional investors have been bullish during most of the sample period. interestingly, individual investors have been more bullish than institutional investors. the standard deviations of pacific-basin stock markets are very high indicating their extremely volatile nature during the sample period. among these markets hong kong seems to have provided the highest return to investors, while the mean returns in case of thailand and korea are negative. most of the variables relating to the risk factors have shown less variability as compared to the investor sentiments and asian stock returns. table 1: descriptive statistics the variables are individual investor sentiments (sentt 1 ), institutional investor sentiments (sentt 2 ), economic growth(iip), short term interest rates (t30), economic risk premiums (t90-t30), future economic variables (b10-t30), business conditions (baa-aaa), dividend yield (div.), inflation (inf), excess returns on market portfolio (r m ), premium 74 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 on portfolio of small stocks relative to large stocks (smb), premium on portfolio of high book/market stocks relative to low book/market stocks (hml), momentum factors (umd), currency fluctuations (usd), and stock market returns on hong kong (hong kong), malaysia (malaysia), philippines (philippines), singapore (singapore), taiwan (taiwan), thailand (thailand), and korea (korea). mean median maximum minimum std. dev. skewness kurtosis sentt 1 0.1143 0.1200 0.5100 -0.3500 0.1760 -0.0863 2.6626 sentt 2 0.0896 0.1100 0.3640 -0.3420 0.1413 -0.5373 2.9513 iip 0.0096 0.0147 0.1011 -0.1094 0.0389 -0.5279 3.5607 t30 0.0026 0.0032 0.0199 -0.0121 0.0052 -0.1152 3.2783 t90-t30 0.0043 0.0041 0.0080 0.0021 0.0013 0.4793 2.9139 b10_t30 0.0004 0.0004 0.0017 -0.0003 0.0004 0.8185 3.7719 baa-aaa 0.0071 0.0078 0.0549 -0.0440 0.0181 -0.0562 3.1558 div 0.0078 0.0073 0.0144 0.0053 0.0018 1.1020 4.3580 inf 0.0127 0.0153 0.1141 -0.1437 0.0408 -0.4639 3.9027 r m 0.0026 0.0023 0.0103 -0.0012 0.0021 0.9335 4.3616 smb 0.0031 0.0077 0.0994 -0.1655 0.0414 -0.7543 4.3240 hml -0.0012 -0.0028 0.2138 -0.1626 0.0382 1.0244 11.0803 umd 0.0024 0.0009 0.1367 -0.1205 0.0363 0.4417 5.3273 usd 1.1658 1.3200 18.2100 -25.1300 4.5224 -0.7366 11.7315 hong kong 0.0120 0.0139 0.2645 -0.3482 0.0872 -0.2028 5.0505 malaysia 0.0005 0.0006 0.2895 -0.2784 0.0991 0.0815 4.2293 philippines 0.0060 -0.0016 0.3317 -0.2989 0.0974 0.2858 4.6905 singapore 0.0048 0.0057 0.2484 -0.2107 0.0757 0.0797 4.3745 taiwan 0.0014 0.0007 0.3324 -0.1746 0.0910 0.7308 4.0350 thailand -0.0057 -0.0092 0.2843 -0.2817 0.1087 0.2910 3.5018 korea -0.0015 -0.0097 0.3945 -0.3181 0.0998 0.5246 4.7757 5. econometric methodology we choose the var modeling technique (sims, 1980) as an appropriate econometric methodology to investigate the postulated relationships. the rationale for doing so lies in the arguments of brown and cliff (2004a & 2004b) and lee et al. (2002) which suggest that stock market returns and investor sentiments may act as a system. our approach is also consistent with studies such as soydemir (2005), kim (2003), ratanapakorn and sharma (2002), janakiramanan and lamba (1998), and park and fatemi (1993) which have employed the var models to analyze the linkages between the u.s. and pacificbasin stock markets. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 75 the var specification allows the researchers to do policy simulations and integrate monte carlo methods to obtain confidence bands around the point estimates (doan, 1988; genberg et al. 1987; hamilton, 1994). the likely response of one variable at time t, t+1, t+2 etc. to a one time unitary shock in another variable at time t can be captured by impulse response functions. as such they represent the behavior of the series in response to pure shocks while keeping the effect of other variables constant. since, impulse responses are highly non-linear functions of the estimated parameters, confidence bands are constructed around the mean response. responses are considered statistically significant at the 95% confidence level when the upper and lower bands carry the same sign. thus var model captures the dynamic feedback effects in a relatively unconstrained fashion and is therefore a good approximation to the true data generating process. we express the var model as: (4) where, z(t) is a column vector of variables under consideration, c is the deterministic component comprised of a constant, a(s) is a matrix of coefficients, m is the lag length and ε(t) is a vector of random error terms. table 2: cross-correlations of variables relating to fundamentals the variables are economic growth (iip), short term interest rates (t30), economic risk premiums (t90), future economic variables (b10), business conditions (baa), dividend yield (div), inflation (inf), excess returns on market portfolio (r m ), premium on portfolio of small stocks relative to large stocks (smb), premium on portfolio of high book/market stocks relative to low book/market stocks (hml), momentum factors (umd), and currency fluctuations (usd). b10 baa iip hml inf r m div smb t30 t90 umd usd b10 1.00 baa 0.00 1.00 iip -0.15 -0.39 1.00 hml 0.06 -0.06 0.05 1.00 inf -0.02 0.16 -0.15 0.00 1.00 r m 0.26 -0.01 -0.10 -0.56 -0.17 1.00 div 0.32 0.02 -0.11 -0.47 -0.16 0.97 1.00 smb -0.20 -0.04 -0.05 -0.50 0.00 0.17 -0.04 1.00 t30 0.14 0.40 -0.28 -0.06 0.26 -0.07 0.02 -0.13 1.00 t90 0.32 0.25 -0.19 -0.15 0.09 0.14 0.15 -0.01 0.19 1.00 umd 0.20 -0.06 0.00 -0.20 -0.10 0.03 -0.03 0.20 0.01 -0.14 1.00 usd -0.03 -0.09 0.17 0.19 -0.14 -0.16 -0.15 0.00 0.00 -0.04 0.00 1.00 1 m s z(t) c a(s)z(t m) (t)ε = = + − +• 76 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 table 3: effects of fundamentals on individual and institutional investor sentiments the variables are individual investor sentiments (sentt 1 ), institutional investor sentiments (sentt 2 ), economic growth (iip), short term interest rates (t30), economic risk premiums (t90), future economic variables (b10), business conditions (baa), dividend yield (div), inflation (inf), excess returns on market portfolio (r m ), premium on portfolio of small stocks relative to large stocks (smb), premium on portfolio of high book/market stocks relative to low book/market stocks (hml), momentum factors (umd), and currency fluctuations (usd). variables sentt 1 sentt 2 b10 -0.96 (0.88) 0.49 (0.71) baa -29.93*** (8.35) -4.82 (8.27) iip 1.30 (2.77) -2.22 (2.20) hml 1.44*** (0.53) 1.14*** (0.46) inf -18.28*** (6.60) -8.35 (6.06) r m -6.75** (3.29) -3.60 (2.66) div 8.32*** (3.31) 4.50* (2.71) smb 2.78*** (0.80) 1.99*** (0.66) t30 7.47 (13.57) -6.83 (11.51) t90 -13.11 36.71 -31.70 (31.64) umd 0.00 (0.00) 0.00 (0.00) usd 0.00 (0.01) 0.00 (0.01) c 0.29*** (0.07) 0.15** (0.07) r-squared 0.304 0.161 ssr 3.190 2.48 akaike info criterion -0.839 -1.090 schwarz criterion -0.578 -0.829 f-statistic 4.989 2.186 prob(f-statistic) 0.000 0.015 0 1 j it i ij jit it j sentt fundλ λ ξ = = + +• the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 77 we first decompose the sentiments variables into rational and irrational components. in particular, we estimate two separate ordinary least square (ols) regressions based on equations (1) and (2). to check the presence of multicollinearity, we estimate the cross-correlations between the variables related to fundamentals. the results of the cross correlations are reported in table 2. the low correlations among most of the variables suggest that multicollinearity is not a major issue. table 3 reports the regression results based on equations (2) and (3). individual investor sentiments are significantly related to business conditions, inflation, dividend yield, excess returns on market, smb, and hml. similarly, the sentiments of institutional investor sentiments are significantly related to dividend yield, smb and hml. these results are consistent with the arguments of baker and wurgler (2005), brown and cliff (2004b) and shleifer and summers (1990) that investor sentiments may contain a combination of both rational and irrational components and not necessarily only noise. 6. estimation results before proceeding with the main results, we first check the time series properties of each variable by performing unit root tests. table 4 reports the results of unit root tests using augmented dickey fuller (adf) test (dickey and fuller, 1979, 1981) and kwiatkowski, phillips, schmidt, and shin (1992) (kpss test). based table 4: unit root test results the variables are rational sentiments of individual investors (rational 1 ), rational sentiments of institutional investors (rational 2 ), irrational sentiments of individual investors (irrational 1 ), irrational sentiments of institutional investors (irrational 2 ), and stock market returns on hong kong (hong kong), malaysia (malaysia), philippines (philippines), singapore (singapore), taiwan (taiwan), thailand (thailand), and korea (korea). adf test kpss test rational 1 -4.019 0.112 rational 2h -5.714 0.107 irrational 2 -6.337 0.098 institutional_ir -3.989 0.153 hong kong -6.994 0.188 malaysia -6.129 0.195 philippines -6.764 0.092 singapore -6.349 0.126 taiwan -7.081 0.096 thailand -6.537 0.114 korea -6.421 0.134 test critical values: 1% level -3.469 0.739 5% level -2.878 0.463 10% level -2.575 0.347 78 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 on the consistent and asymptotically efficient aic and sic criteria (diebold, 2003) and considering the loss in degrees of freedom, the appropriate number of lags is determined to be two. in the case of the adf test, the null hypothesis of nonstationarity is rejected. in the kpss test, the null hypothesis is that each series in stationary. we fail to reject the null hypothesis in the case of kpss test. the inclusion of drift/trend terms in the adf and kpss test equations does not change these results (dolado, jenkinson, and sosvilla-rivero, 1990). given that the series are stationary in nature, we estimate a set of five variable var model for each of the seven pacific-basin stock countries. each var model is composed of rational and irrational components of the u.s. individual and institutional investor sentiments and the stock returns of pacificbasin country being studied1. 6.1 variance decomposition the decomposition of variance gives a quantitative measure to the causal relationship indicating how much the movement in one variable can be explained by other variables in terms of the percentage of the forecast error variance. table 5 (panels a through g) shows the results of the innovation accounting procedure and reports the 1 through 10 day ahead forecast error variance of pacific-basin stock market returns accounted for by innovations in rational and irrational sentiments of the u.s. individual and institutional investors. in the case of hong kong (panel a), total sentiments accounts for approximately 14% of the total forecast error variance. the major portion of such influences is due to rational sentiments of institutional investors which accounts for approximately 9%. however, in the case of malaysia (panel b), total sentiments seem to have lesser influence on stock returns as compared to honk kong (approximately 11%). also, the rational sentiments of individual and institutional investors account for approximately 5% and 4% respectively, which are much greater than forecast error variances accounted by irrational sentiments. panel c reports similar decomposition for philippines stock market returns. the variance due to the total sentiments averages approximately between 12-13%, of which rational sentiments of institutional investor has the highest contribution (approximately 8%). similar to the case of korea, there is less influence of rational sentiments of individual investors. there is relatively very strong effect of rational sentiments of institutional investors in the case of singapore (panel d), as it accounts for 18-19% of the total forecast error variance. the effect of other components of sentiments is much less. panel e reports the forecast error variance in the case of taiwan. the total sentiments accounts for approximately 8-9%, which is the least among all the pacific-basin stock markets. similar to the earlier findings, the rational sentiments 1 our approach is similar to the one employed by park and fatemi (1993) which estimate a four variable var model for each of the seven pacific-basin countries instead of one model including all the variables to avoid irrelevant feedback relationships among stock markets. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 79 of institutional investors account for the maximum variance (approximately 4%). similar to malaysia, rational individual investor sentiments accounts for greater proportion of the total variance (approximately 3.5%). the results are very similar for thailand (panel f), where the rational institutional and individual investor sentiments account for approximately 4% and 3% respectively. likewise, in the case of korea (panel g), rational sentiments of two classes of investors have approximately similar contribution. overall, the decomposition of forecast error variances of pacific-basin stock returns consistently suggests much higher effect of rational sentiments than the irrational sentiments. specifically, in all the cases, rational sentiments of institutional investors is the most endogenous variable. these findings imply that the rational expectations of institutional investors are an important player in propagating the u.s. stock market movements in the pacific-basin region. table 5: decomposition of forecast error variances of asian stock returns the variables are rational sentiments of individual investors (rational 1 ), rational sentiments of institutional investors (rational 2 ), irrational sentiments of individual investors (irrational 1 ), irrational sentiments of institutional investors (irrational 2 ), and stock market returns on hong kong (hong kong), malaysia (malaysia), philippines (philippines), singapore (singapore), taiwan (taiwan), thailand (thailand), and korea (korea). panel a: decomposition of hong kong by innovations in period rational 1 rational 2 irrational 1 irrational 2 total sentiments hong kong 1 1.9598 0.2250 1.0710 0.3259 3.2558 96.4183 2 1.8287 5.9193 1.1767 1.8388 8.9248 89.2364 3 1.9357 8.1152 1.3355 1.7897 11.3863 86.8240 4 2.1002 8.8842 1.3669 1.7740 12.3513 85.8746 5 2.0978 9.0597 1.3839 1.7761 12.5415 85.6824 6 2.0981 9.0652 1.3898 1.8371 12.5530 85.6099 7 2.0977 9.0693 1.3973 1.8539 12.5643 85.5817 8 2.0982 9.0712 1.4070 1.8601 12.5765 85.5634 9 2.0988 9.0735 1.4151 1.8619 12.5874 85.5507 10 2.0995 9.0747 1.4212 1.8631 12.5954 85.5415 table continues on the next page 80 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 panel b: decomposition of malaysia by innovations in period rational 1 rational 2 irrational 1 irrational 2 total sentiments malaysia 1 0.1231 0.0159 0.0000 0.7613 0.9003 99.0997 2 4.4359 3.4134 0.0581 1.3181 9.2256 90.7744 3 4.1158 4.1651 0.2746 1.5794 10.1349 89.8651 4 4.2303 4.5481 0.5493 1.5791 10.9067 89.0933 5 4.1876 5.0159 0.5658 1.5929 11.3622 88.6378 6 4.2917 5.0893 0.5663 1.5905 11.5377 88.4623 7 4.3044 5.1322 0.5808 1.5904 11.6078 88.3922 8 4.3318 5.1457 0.5806 1.5897 11.6477 88.3523 9 4.3355 5.1576 0.5810 1.5893 11.6633 88.3367 10 4.3402 5.1625 0.5811 1.5893 11.6730 88.3270 panel c: decomposition of philippines by innovations in period rational 1 rational 2 irrational 1 irrational 2 total sentiments philippines 1 0.5410 1.1955 0.0040 1.0546 2.2541 97.2048 2 1.0577 7.2477 0.6001 1.0808 8.9286 90.0136 3 2.1692 7.5731 0.8720 1.3869 9.8320 87.9988 4 2.2722 8.0619 0.9584 1.4514 10.4717 87.2561 5 2.2930 8.4228 0.9587 1.4567 10.8382 86.8688 6 2.3287 8.5021 0.9574 1.4576 10.9171 86.7542 7 2.3483 8.5228 0.9583 1.4571 10.9382 86.7135 8 2.3573 8.5282 0.9582 1.4568 10.9432 86.6994 9 2.3608 8.5316 0.9581 1.4569 10.9467 86.6925 10 2.3623 8.5335 0.9583 1.4571 10.9489 86.6888 panel d: decomposition of singapore by innovations in period rational 1 rational 2 irrational 1 irrational 2 total sentiments singapore 1 0.1703 0.6844 0.1512 0.0686 1.0745 98.9255 2 1.5614 14.6185 0.1918 1.3938 17.7656 82.2345 3 2.5504 17.6776 0.1820 1.3821 21.7922 78.2078 4 2.6344 18.6832 0.1793 1.3699 22.8668 77.1332 5 2.6584 19.3584 0.2282 1.3560 23.6010 76.3990 table continues on the next page the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 81 6 2.6557 19.3540 0.2468 1.4530 23.7095 76.2905 7 2.6573 19.3594 0.2711 1.4755 23.7633 76.2367 8 2.6605 19.3557 0.2876 1.4903 23.7941 76.2059 9 2.6625 19.3613 0.3033 1.4928 23.8199 76.1801 10 2.6669 19.3615 0.3142 1.4957 23.8383 76.1617 panel e: decomposition of taiwan by innovations in period rational 1 rational 2 irrational 1 irrational 2 total sentiments taiwan 1 1.7413 0.2104 0.1842 0.5202 2.6561 97.3439 2 3.1608 4.1845 0.6226 0.6517 8.6196 91.3804 3 3.5231 4.1651 0.6250 1.0546 9.3678 90.6322 4 3.5875 4.1551 0.6269 1.0869 9.4564 90.5436 5 3.5893 4.1535 0.6308 1.1201 9.4938 90.5062 6 3.5877 4.1522 0.6413 1.1501 9.5313 90.4687 7 3.5878 4.1512 0.6498 1.1654 9.5541 90.4459 8 3.5876 4.1505 0.6560 1.1757 9.5699 90.4301 9 3.5878 4.1502 0.6602 1.1821 9.5803 90.4197 10 3.5881 4.1500 0.6631 1.1861 9.5873 90.4127 panel f: decomposition of thailand by innovations in period rational 1 rational 2 irrational 1 irrational 2 total sentiments thailand 1 0.8681 0.3833 1.5703 0.0913 0.8681 0.3833 2 1.2603 2.3268 1.5154 1.2825 1.2603 2.3268 3 3.0258 3.8247 1.6956 3.0017 3.0258 3.8247 4 3.3520 4.3644 1.8083 2.9728 3.3520 4.3644 5 3.4016 4.6906 1.8057 2.9724 3.4016 4.6906 6 3.4531 4.7648 1.8053 2.9696 3.4531 4.7648 7 3.4806 4.7984 1.8037 2.9707 3.4806 4.7984 8 3.4954 4.8113 1.8032 2.9698 3.4954 4.8113 9 3.5022 4.8190 1.8037 2.9692 3.5022 4.8190 10 3.5058 4.8228 1.8044 2.9690 3.5058 4.8228 panel g: decomposition of korea by innovations in period rational 1 rational 2 irrational 1 irrational 2 total sentiments korea 1 4.0825 0.9269 0.0120 1.0277 5.0214 93.9509 table continues on the next page 82 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 2 4.0046 0.8984 0.7324 1.0486 5.6354 93.3160 3 3.8057 3.9798 0.9001 2.6300 8.6855 88.6845 4 3.7764 4.3255 0.9843 3.3019 9.0861 87.6119 5 3.7690 4.4325 0.9838 3.3463 9.1853 87.4684 6 3.7698 4.4903 0.9834 3.3466 9.2435 87.4099 7 3.7687 4.5185 0.9835 3.3475 9.2707 87.3818 8 3.7683 4.5267 0.9837 3.3483 9.2787 87.3731 9 3.7683 4.5292 0.9837 3.3484 9.2812 87.3704 10 3.7683 4.5302 0.9837 3.3483 9.2822 87.3694 6.2 impulse response function next, we analyze the impulse response functions to shed light on the significance and duration of the effect of shock in rational and irrational sentiments of individual investors to pacific-basin stock returns. it is well known theoretically that traditional orthogonalized forecast error variance decomposition results based on the widely used choleski factorization of var innovations may be sensitive to variable ordering (pesaran and shin, 1996; koop, pesaran and potter, 1996; pesaran and shin, 1998). to mitigate such potential problems of misspecifications, we employ the recently developed generalized impulses technique as described by pesaran and shin (1998) in which an orthogonal set of innovations which does not depend on the var ordering. the generalized impulse responses from an innovation to the jth variable are derived by applying a variable specific cholesky factor computed with the jth variable at the top of the cholesky ordering. these generalized impulses can capture the effect of unanticipated components and therefore can be regarded as an appropriate choice for this study. figures 1 a through 1d plot the impulse responses of hong kong to rational and irrational sentiments of the u.s. individual and institutional investors. the response to rational sentiments for both individual and institutional are significant in the second month and becomes insignificant thereafter (figures 1 and 1c). however, the effect of the irrational component of sentiments is not significant in both the cases (figures 1 b and 1d). figures 2a through 2d plot the impulse responses of malaysia to rational and irrational sentiments of the u.s. investors. similar to the findings of hong kong, the responses to rational sentiments are significant in the second month and become insignificant thereafter (figures 2a and 2c). on the other hand the effect of irrational component of sentiments remains insignificant throughout (figures 2b and 2d). similarly in the case of philippines, the responses of stock returns to the rational sentiments are significant the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 83 (figures 3a and 3c) while insignificant to the irrational components (figures 3b and 3c). the results of impulse responses of singapore are somewhat similar in that rational (irrational) sentiments have significant (insignificant) influences on stock market returns (figures 4a through 4d). positive significant effect of the shocks of rational sentiments in the case of hong kong, malaysia, philippines, and singapore are consistent with earlier studies which find moderate linkages between these markets and the u.s. stock market movements. figure 1: response of hong kong to the u.s. individual and institutional investor sentiments the dashed lines on each graph represent the upper and lower 95% confidence bands. when the upper and lower bounds carry the same sign the response becomes statistically significant. * on each graph, “percentage returns” are on the vertical and “horizon” is on the horizontal axis. figures 5a through 5d plot the responses of taiwan to the one unit shock in the rational and irrational sentiments of the u.s. individual and institutional investors. unlike the results of hong kong, malaysia, philippines, and singapore . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 . 1 0 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 1 a r e s p o n s e o f h o n g k o n g t o r a t i o n a l s e n t i m e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 . 1 0 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 1 b r e s p o n s e o f h o n g k o n g t o i r r a t i o n a l s e n t i m e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 . 1 0 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 1 c r e s p o n s e o f h o n g k o n g t o r a t i o n a l s e n t i m e n t s o f i n s t i t u t i o n a l i n v e s t o r s . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 . 1 0 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 1 d r e s p o n s e o f h o n g k o n g t o i r r a t i o n a l s e n t i m e n t s o f i n s t i t u t i o n a l i n v e s t o r s 84 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 there are insignificant effects of rational components of both class of investors. similarly, the irrational investor sentiments have insignificant effects on stock returns. we find similar results in the case of thailand, where both the rational and irrational components of sentiments have insignificant effects (figure 6a through 6d). however, in the case of korea there is significant effect of only rational sentiments of institutional investors (figure 7a). we find insignificant effect of rational sentiments of individual investors (figure 7c). consistent with our earlier results there are insignificant effects of the irrational sentiments for both class of investors (figures 7b and 7d). these insignificant results in the case of taiwan, thailand, and korea are consistent with earlier studies such as park and fatemi (1993) which find little linkages of these markets with the u.s. figure 2: response of malaysia to the u.s. individual and institutional investor sentiments the dashed lines on each graph represent the upper and lower 95% confidence bands. when the upper and lower bounds carry the same sign the response becomes statistically significant. * on each graph, “percentage returns” are on the vertical and “horizon” is on the horizontal axis. . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 2 a r e s p o n s e o f m a la y s i a t o r a t i o n a l s e n t i m e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 2 b r e s p o n s e o f m a l a y s i a t o i r r a t io n a l s e n t i m e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 2 c r e s p o n s e o f m a l a y s i a t o r a t i o n a l s e n t i m e n t s o f i n s t i t u t i o n a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 2 d r e s p o n s e o f m a l a y s i a t o r a t i o n a l s e n t i m e n t s o f i n s t i t u t i o n a l i n v e s t o r s the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 85 figure 3: response of philippines to the u.s. individual and institutional investor sentiments the dashed lines on each graph represent the upper and lower 95% confidence bands. when the upper and lower bounds carry the same sign the response becomes statistically significant. * on each graph, “percentage returns” are on the vertical and “horizon” is on the horizontal axis. in all, the results of the variance decomposition and impulse responses strongly suggest that irrational exuberance of the u.s. investors, whether individual and institutional are not transmitted to the pacific-basin region. however, the rational both classes of investors are transmitted to asian markets with varying degrees of intensity. also, there are somewhat greater effects of the rational expectations of the u.s. institutional investors than those of the individual investors. a significant development in emerging markets is that individual investors have increasingly delegated the management of their assets to professional fund managers (griffith-jones and cailloux, 1998). such institutionalization has increased the sensitivities of emerging markets to the behavior of international . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 3 a r e s p o n s e o f p h i l i p p in e s t o r a t i o n a l s e n t i m e n t s o f i n d iv i d u a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 3 b r e s p o n s e o f p h i l i p p i n e s t o i r r a t i o n a l s e n t i m e n t s o f i n d i v id u a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 3 c r e s p o n s e o f p h i li p p i n e s t o r a t i o n a l s e n t im e n t s o f i n s t it u t i o n a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 3 d r e s p o n s e o f p h i l ip p i n e s t o i r r a t i o n a l s e n t i m e n t s o f i n s t i t u t i o n a l i n v e s t o r s 86 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 institutional investors. moreover, it is much easier for domestic institutional investors engage in herding behavior compared to individual investors since similar information circulates among funds allowing them to follow more easily other institutions’ decisions (nofsinger and sias, 1999). also, due to the high transaction costs of investing in emerging markets, closed-end country funds have emerged as one of the most popular means of international investments by the u.s. individual investors. these factors may explain the greater responses of pacific-basin stock markets to the u.s. institutional investor sentiments as compared to the u.s. individual investor sentiments. figure 4: response of singapore to the u.s. individual and institutional investor sentiments the dashed lines on each graph represent the upper and lower 95% confidence bands. when the upper and lower bounds carry the same sign the response becomes statistically significant. * on each graph, “percentage returns” are on the vertical and “horizon” is on the horizontal axis. . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 4 a r e s p o n s e o f s in g a p o r e t o r a t i o n a l s e n t i m e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 4 b r e s p o n s e o f s i n g a p o r e t o i r r a t i o n a l s e n t i m e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 4 c r e s p o n s e o f s i n g a p o r e t o r a t i o n a l s e n t im e n t s o f i n s t i t u t i o n a l i n v e s t o r s . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 4 d r e s p o n s e o f s i n g a p o r e t o i r r a t i o n a l s e n t i m e n t s o f i n s t it u t i o n a l i n v e s t o r s the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 87 figure 5: response of taiwan to the u.s. individual and institutional investor sentiments the dashed lines on each graph represent the upper and lower 95% confidence bands. when the upper and lower bounds carry the same sign the response becomes statistically significant. * on each graph, “percentage returns” are on the vertical and “horizon” is on the horizontal axis. . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 . 1 0 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 5 a r e s p o n s e o f t a i w a n t o r a t i o n a l s e n t i m e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 . 1 0 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 5 b r e s p o n s e o f t a i w a n t o i r r a t i o n a l s e n t i m e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 . 1 0 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 5 c r e s p o n s e o f t a i w a n t o r a t i o n a l s e n t i m e n t s o f i n s t i t u t i o n a l i n v e s t o r s . 0 4 . 0 2 . 0 0 . 0 2 . 0 4 . 0 6 . 0 8 . 1 0 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 5 d r e s p o n s e o f t a i w a n t o i r r a t i o n a l s e n t i m e n t s o f i n s t i t u t i o n a l i n v e s t o r s 88 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 figure 6: response of thailand to the u.s. individual and institutional investor sentiments the dashed lines on each graph represent the upper and lower 95% confidence bands. when the upper and lower bounds carry the same sign the response becomes statistically significant. * on each graph, “percentage returns” are on the vertical and “horizon” is on the horizontal axis. . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 . 1 6 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 6 a r e s p o n s e o f t h a il a n d t o r a t i o n a l s e n t im e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 . 1 6 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 6 b r e s p o n s e o f t h a i l a n d t o i r r a t i o n a l s e n t im e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 . 1 6 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 6 c r e s p o n s e o f t h a i l a n d t o r a t i o n a l s e n t i m e n t s o f i n s t i t u t i o n a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 . 1 6 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 6 d r e s p o n s e o f t h a i la n d t o i r r a t io n a l s e n t i m e n t s o f i n s t i t u t io n a l i n v e s t o r s the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 89 figure 7: response of korea to the u.s. individual and institutional investor sentiments the dashed lines on each graph represent the upper and lower 95% confidence bands. when the upper and lower bounds carry the same sign the response becomes statistically significant. * on each graph, “percentage returns” are on the vertical and “horizon” is on the horizontal axis. expectations of 7. conclusion in this study, we investigate whether the rational expectations and irrational exuberance of the u.s. individual and institutional investors are propagated to pacific-basin stock markets, hong kong, malaysia, philippines, singapore, taiwan, thailand, and korea. we employ the investor sentiments data at the individual and institutional level, provided by american association of individual investors and investors intelligence and the vector auto regression (var) models to investigate the postulated relationships. the results of variance decomposition and impulse response functions strongly suggest that there are distinct effects of the rational and irrational investor sentiments on these international stock market returns. . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 7 a r e s p o n s e o f k o r e a t o r a t i o n a l s e n t i m e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 7 b r e s p o n s e o f k o r e a t o i r r a t i o n a l s e n t i m e n t s o f i n d i v i d u a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 7 c r e s p o n s e o f k o r e a t o r a t i o n a l s e n t i m e n t s o f i n s t i t u t i o n a l i n v e s t o r s . 0 4 . 0 0 . 0 4 . 0 8 . 1 2 1 2 3 4 5 6 7 8 9 1 0 f i g u r e 7 d r e s p o n s e o f k o r e a t o i r r a t i o n a l s e n t i m e n t s o f i n s t i t u t i o n a l i n v e s t o r s 90 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 we find strong evidence of insignificant effect of irrational exuberance of the u.s. investors on these set of pacific-basin stock market returns. however, the rational expectations of both the u.s. individual and institutional investors have significant effect on these asian markets with varying degrees of intensity. we find greater effects in the case of hong kong, malaysia, philippines, and singapore while a weak linkage with the stock returns of taiwan, thailand, and korea. the overall evidence suggests that institutional investor sentiments are transmitted internationally from the u.s. stock market to a greater extent than the individual investor sentiments; and that the international effects of the u.s. stock market can be attributed to the sentiments induced by fundamental trading and not to noise trading. these results are consistent with the view that besides economic fundamentals, the international effects of the u.s. stock market can be attributed to rational investor sentiments. therefore it is important for policymakers to consider such spillover effects in their international policy corporation and for investors in their portfolio allocation decisions involving stock markets movements. author statement: rahul verma, college of business, university of houstondowntown. e-mail: vermar@uhd.edu. i like to record my thanks to the editors for receiving comments on the paper and making changes to the paper. the mistakes are my own responsibility. references baker, m., and wurgler, j. (2005). investor sentiment and cross-section of stock returns. journal of finance, forthcoming. barberis, n., shleifer, a., and vishny, r. w., (1998). a model of investor sentiment. journal of financial economics, 49: 307-343. black, f., (1986). noise. the journal of finance, 41(3): 529-543. brown, g. w., and cliff, m. t. (2004a). investor sentiment and the near-term stock market. journal of empirical finance 11(no. 1): 1-27. brown, g. w., & cliff, m. t., (2004b). investor sentiment and asset valuation. journal of business, (forthcoming). campbell, j. y., (1987). stock returns and the term structure. journal of financial economics, 18: 373-399. campbell, j. y. (1991). a variance decomposition for stock returns. economic journal, 101, 157-179. campbell, j. y., (2000). asset pricing at the millennium. journal of finance, 55: 1515-1567. campbell, j. y., and kyle, a. s., (1993). smart money, noise trading, and stock price behavior. review of economic studies, 60: 1-34. campbell, j. y., and shiller, r. j., (1988a). the dividend-price ratio and expectations of future dividends and discount factors. review of financial studies, 1, 195-228. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 91 campbell, j. y., and shiller, r. j., (1988b). stock prices, earnings and expected dividends. journal of finance, 43: 661-676. chen, n., kan, r., and miller, m., (1993). are the discounts on closed-end funds a sentiment index? journal of finance 48: 795-800. clarke, r. g., and statman, m., (1998): bullish or bearish? financial analysts journal may/june, 63-72. cochrane, j. h. (2000). asset pricing. (princeton university press, princeton, nj). daniel, k., hirshleifer, d., and subrahmanyam, a., (1998). investor psychology and security market underand overreactions. journal of finance 53: 1839-1886. debondt, w., (1993). betting on trends: intuitive forecasts of financial risk and return. international journal of forecasting 9: 355-371. delong, shleifer, j., summers, a., and waldmann, r., (1990). noise trader risk in financial markets. journal of political economy, 98: 703-738. delong, shleifer, j., summers, a., and waldmann, r., (1991). the survival of noise traders in financial markets. journal of business, 64 (1): 1-19. dickey, d. a., and fuller, w. a., (1979). distribution of the estimators for autoregressive time series with a unit root. journal of the american statistical association 74: 427-431. dickey, d. a., and fuller, w. a., (1981). likelihood ratio statistics for autoregressive time series with a unit root. econometrica, 49: 10571072. diebold, f. x., (2003). elements of forecasting. south western college publishing. doan, t., (1988). rats user’s manual. evanston, illinois: var econometrics. dolado, j. j., jenkinson, t., and sosvilla-rivero, s., (1990). cointegration and unit roots. journal of economic surveys 4: 249-273. elton, e., j, gruber, m., j, and busse, j., a., (1998). do investors care about sentiments? journal of business, 71: 477-500. elton, e. j., and gruber, m. j. (1991). modern portfolio theory and investment analysis. john wiley and sons, inc., fourth edition. fama, e. f., (1965). the behavior of stock market prices. journal of business, 38 (no. 1): 34-105. fama, e. f., (1970). efficient capital markets: a review of theory and empirical work.???????????? fama, e. f. (1990). term structure forecasts of interest rates, inflation, and real returns. journal of monetary economics 25: 59-76. fama, e. f., and french, k. r. (1988). dividend yields and expected stock returns. journal of financial economics 22: 3-25. fama, e. f., and french, k. r. (1989). business conditions and expected returns on stocks and bonds. journal of financial economics, 25: 23-49. fama, e. f., and french, k. r., (1993). common risk factors in the returns on stocks and bonds. journal of financial economics 33: 3-56. 92 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 fama, e. f., and schwert, g. w., (1977). asset returns and inflation. journal of financial economics, 5: 115-146. ferson, w. e., and campbell, r. h., (1991). the variation in economic risk premiums. journal of political economy 99: 385-415. fisher, k. l., and statman, m., (2000). investor sentiments and stock returns. financial analysts journal, march/april (16-23). friedman, m., (1953). essays in positive economics. university of chicago press, chicago. gemmill, g., and thomas, c. d., (2002). noise trading, costly arbitrage, and asset prices: evidence from closed-end funds. the journal of finance, 6: (2571-2594). genberg, h., salemi, m. k., and swoboda, a. (1987). the relative importance of foreign and domestic disturbances for aggregate fluctuations in open economy: switzerland. journal of monetary economics 19: 45-67. gervais, s., and odean, t. (2001). learning to be overconfident. review of financial studies 14: 1-28. hamilton, j. d., (1994). time series analysis. princeton university press(princeton, nj). hirshleifer, d. (2001). investor psychology and asset pricing. journal of finance, lvi(4): 1533-1597. hodrick, r., (1992). dividend yields and expected stock returns: alternative procedures for inference and measurement. review of financial studies 5, 357-386. hong, h., lim, t., and stein, j. c., (2000). bad news travels slowly: size, analysts coverage and the profitability of momentum strategies. journal of finance, 55, 265-292. janakiramanan, s., and lamba, a. s., (1998). an empirical examination of linkages between pacific-basin stock markets. journal of international financial markets, institutions and money, 8, 155-173. jegadeesh, n., and titman, s. (1993). returns to buying winners and selling losers: implications for stock market efficiency. journal of finance, 48, 65-91. kim, s.-j. (2003). the spillover effects of us and japanese public information news in advanced asia-pacific stock markets. pacific-basin finance journal, 11, 611-630. koop, g., pesaran, m. h., and potter, s. m. (1996). impulse response analysis in non linear multivariate models. journal of econometrics, 74, 119-147. kwiatkowski, d., phillips, p. c. b., schmidt, p., and shin, y. (1992). testing the null hypothesis of stationarity against the alternative of a unit root: how sure are we that economic time series have unit root. journal of econometrics, 54, 169-178. keim, d. b., and stambaugh, r. f. (1986). predicting returns in the bond and stock markets. journal of financial economics, 17, 357-390. lakonishok, j., shleifer, a., and vishny, r. w. (1991). do institutional investors destabilize stock prices? evidence on herding and feedback trading. working paper, nber, 3846. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 93 lee, c., shleifer, a., and thaler, r. (1991). investor sentiment and the closedend fund puzzle. journal of finance, 46, 75-109. lee, w. y., jiang, c. x., and indro, d. c., (2002). stock market volatility, excess returns, and the role of investor sentiments. journal of banking & finance, 26, 2277-2299. lintner, j., (1965). security prices, risk, and maximal gains from diversification. journal of finance. neal, r., and wheatley, s., (1998). do measures of investor sentiments predict stock returns? journal of financial and quantitative analysis, 34(523547). palomino, f. (1996). noise trading in small markets. journal of finance, 51(4), 1537-1550. park, j., and fatemi, a. m., (1993). the linkages between the equity markets of pacific-basin countries and those of the u.s., u.k., and japan: a vectorautoregression analysis. global finance journal, 4(1): 49-64. pesaran, m. h., and shin, y., (1996). cointegration and speed of convergence to equilibrium. journal of econometrics 71: 117-143. pesaran, m. h., and shin, y. (1998). generalized impulse response analysis in linear multivariate models. economics letters 58: 17-29. ratanapakorn, o., and sharma, s. c., (2002). interrelationships among regional stock indices. review of financial economics, 11(91-108). schwert, g. w., (1990). stock returns and real activity: a century of evidence. journal of finance, 45: 1237-1257. sharpe, s. a., (2002). reexamining stock valuation and inflation: the implications of analysts’ earnings forecasts. the review of economics and statistics (forthcoming). ??? sharpe, w. f. (1964)., capital asset prices: a theory of market equilibrium under conditions of risk. journal of finance 425-442. shefrin, h., and statman, m. (1994). behavioral capital asset pricing theory. the journal of financial and quantitative analysis 29(3): 323-349. shleifer, a., and summers, l. (1990). the noise trader approach to finance. journal of economic perspectives, 4(2): 19-33. sias, r., w, starks, l., t, & tinic, s., m., (2001). is noise trader risk priced? the journal of financial research, 24(3): 311-329. sims, c. (1980). macroeconomic and reality. econometrica, 48, 1-49. solt, m. e., and statman, m., (1988). how useful is the sentiment index? financial analysts journal, september/october (45-55). soydemir, g. (2005). differences in the price of risk and the resulting response to shocks: an analysis of asian markets. journal of international financial markets, institutions and money, 15: 285-313. swaminathan, b. (1996). time-varying expected small firm returns and closedend fund discounts. review of financial studies 9: 845-887. trueman, b. (1988). a theory of noise trading in securities markets. the journal of finance, 43(1): 83-95. 94 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 67-94 wang, f. a.,(2001). overconfidence, investor sentiment, and evolution. journal of financial intermediation 10: 138-170. west, k. (1988). bubbles, fads and stock price volatility tests: a partial evaluation. the the journal of finance, 43(3): 639-565. international journal of banking and finance 9-1-2008 rational expectations, irrational exuberance: linkage between u.s. investors and pacific-basin stock returns rahul verma recommended citation cetak a4 ijbf6(2)09.indd the international journal of banking and finance, 2008/09 vol. 6. number 2: 2009: 95-109 95 exchange rate behaviour of east european transitional economies catherine s. f. ho university technology mara, malaysia abstract eastern european countries, which are candidates for accord into the exchange rate mechanism (erm) and the eventual move towards euro, fi nd exchange rate management a tedious challenge. this paper examines the underlying factors that move exchange rates and helps us to contribute towards streamlining policies and strategies in moving these countries forward. the new fi ndings on exchange rate determinants for this group of transitional economies are based on parity factors as well as non-parity factor effects. the evidence that emerges from this paper is that non-parity factors including economic growth rate, current account and capital fl ows are signifi cantly correlated with exchange rates. the results are robust whichever data set is used, high-frequency and low-frequency data sets. keywords: exchange rates, prices, trade and capital fl ows, foreign debt, reserves, growth jel classifi cation: f31, f32, c33, e31, f43 1. introduction eastern european countries have plans to join and some are already very close to being in the european union (eu). some of them are considering membership in exchange rate mechanism (erm) and eventually are likely to adopt the euro as their currency. it is essential that monetary and infl ationary policies are in line with the other more developed eu countries prior to the likely merger with the euro zone. in this context of former soviet satellites seeking to fully incorporate into a different exchange rate mechanism, it is vital that a study is done to understand the factors that affect their exchange rate movements based on their current separate currencies. insuffi cient attention to the role of exchange rate movements during the pre-erm period will impede the effectiveness of their macroeconomic policies. czech republic and poland fi xed their exchange rates at the beginning of the transition from a centrally planned to a market economy, but later gradually ijbf ht tp :// ijb f.u um .e du .m y 96 exchange rate behaviour of east european transitional economics: 95-109 moved from pegs to the current more fl exible currency regime. these transition countries have experienced relatively high infl ation and marked changes in their currencies. it is part of the consequences of opening their economies to external trade and investments, and the strong productivity growth in the traded sector. romania has had a fl oating regime since 1991 but they started to increase management of their exchange rates since then and are currently under a managed fl oat with euro dollar as a reference currency. slovenia, hungary and slovakia have managed fl oats with target monetary growth and interest rates in line with the other eu countries.1 empirical evidence on the behaviour of exchange rates across the world has been mixed. this paper tests the two parity theorems jointly using panel estimation methods both time series and cross-sectional tests at the same time that maximizes the use of information in a data set. the tests are done within a cluster of countries that are heavily trade-related as eastern european economies. many researchers have pointed out that exchange rate is determined by a number of fundamentals and omission of some of these variables is perhaps the main reason why prior results yielded mixed fi ndings (meese and rogoff: 1983, 1988, frankel and rose: 1996a, and muscatelli, spinelli and trecroci: 2007) and these studies were not able to support parity theorems. much effort at studying exchange rates between pairs of countries has yet provided little consistent fi ndings and so newer approaches using multi-country framework as in this research paper using pooled time series panel data methodology for a group of countries is needed as are already evident in recent literature. it is the aim of this study to extend the literature by looking systematically at the contributions of parity and non-parity variables. the resulting fi ndings can be expected to lead to improved understanding of the dynamics of how exchange rates are determined by factors beyond traditional parity conditions. the remainder of this paper is divided into fi ve sections. the next section contains a brief overview of the current literature relevant to this study. section three describes the methodology, followed by the presentation of the fi ndings in section four. the paper ends in section fi ve with a conclusion. 2. literature on exchange rates more central banks across eastern europe are adopting monetary adjustment towards the european monetary system in order to be included as a party to the union. there exist many issues that need to be assessed in each respective economy prior to the merging of the currency region. during this transition stage, it is vital that major hurdles other than monetary and infl ationary expectations controls be in place but exchange rate stability and ability of each economy to join at an appropriate exchange rate is of utmost importance. moreover, eastern european countries have experienced changes in political and economic transformations in the last decade or two. their exchange rate credibility is vital 1 imf: annual report, 2006 and imf: “exchange arrangements and exchange restrict”, october 2007. ht tp :// ijb f.u um .e du .m y exchange rate behaviour of east european transitional economics: 95-109 97 for attracting international trade and investment to speed up the overall growth and development of these nations. understanding exchange rate movements and the effects of fundamentals will alleviate mismanagement of exchange rates as these countries prepare for a common currency. the concept of purchasing power parity (ppp) is nothing new and it is a very long term expectation of how currency achieves equilibrium exchange rate in the long run. it has been sometimes rated as a poor tool even in the long run (murray and papell; 2005, kuo and mikkola; 2001, lothian and taylor; 2000, macdonald; 1993 and rogoff; 1996). with years of high exchange rate volatility in the world following the breakdown in 1973 of the bretton woods system, studies to verify the ppp have failed to show that the ppp holds in the short term.2 the apparent lack of evidence on ppp under the fl oating regimes that followed the breakdown in most cases motivated the development of the sticky price hypothesis, which is an over-shooting exchange rate model of dornbusch (1976). however with longer time series and more sophisticated methodology, studies have shown some signifi cant results consistent with the sticky price hypothesis.3 according to ppp, the exchange rate will adjust relative rapidly to the ratio of domestic and foreign price levels according to equation (1) with exchange rate defi ned as domestic currency units in terms of one unit of foreign currency. where the change in exchange rate e is a function of price differentials, where j represents country, t represents time period, p represents prices, d domestic and f foreign. (1) another theory that comes to affect exchange rate of a currency is the interest rate parity (irp) theory which suggests that the interest rate differential between any two currency areas adjust the exchange rates of the two areas. this may be more relevant to emerging nations when current world fi nancial market is increasingly liberalized and therefore subjected to global interest rate infl uence.4 in theory, the foreign exchange market should be in equilibrium when deposits of all currencies offer the same rate of return provided that open economic and 2 adler and lehman (1983) found that the deviations from ppp follow a random walk without reverting back to ppp for 43 countries. studies including juselius and macdonald (2004), and sertelis and gogas (2004) found contradictory or at best weak evidence on ppp. 3 parsley (2007), mark and sul (2001), schnabl and baur (2002), and smallwood (2007) found considerable evidence for long run relation and concluded that fundamentals play a signifi cant role in determining exchange rates. 4 studies that provided evidence include mark (1995), ehrmann and fratzscher (2005) and chaboud and wright (2005) found measures of long run expected changes in exchange rates highly correlated with interest rate differentials. however recently, bekaert, wei and xing (2007) found mixed results for interest parity. ln ln d t jt j j jtf t j p e a b p ht tp :// ijb f.u um .e du .m y 98 exchange rate behaviour of east european transitional economics: 95-109 capital movements are permitted. according to the uncovered interest rate parity, the ratio of changes in exchange rate e, within a time period t, is a function of domestic interest rate id, and foreign interest rate if. (2) there are other variables which are correlated with exchange rate movements as predicted by mainstream economic theories. inclusion of these variables should assist researchers in their attempt to explain exchange rate behaviour.5 in recent years there has been increased interest in these factors, given the confl icting empirical evidence on price parity theories. exchange rate determination has not only been linked to parity conditions as in cassel (1918), keynes (1923) and fisher (1930), or trends in productivity as in balassa (1964) and samuelson (1964) but to other international trade, capital and investment issues. studies of fi nancial crises in latin america and east asia have been motivated by an interest in the roles of balance of payments, trade and capital fl ows. trade liberalization has introduced volatility in the balance of payments, and the increase in current account fl ow directly affects currencies. trade in goods accounts for a large proportion of gdp in asia and the european developing countries. sudden outfl ow of capital including portfolio and foreign direct investment is another major concern when it drastically affects exchange rates as were seen during the fi nancial crises of brazil, east asia, and mexico. it is well documented that the recent currency crises were due to vast changes in these variables, including kim (2000) and kaminsky and reinhart (1999). recent studies by lee and chinn (2006), baharumshah and masih (2005), and bergin (2006) found relation between current account, output growth, infl ation and exchange rate. in relation to trade and capital fl ows, international reserves act as a mean to defend a country’s currency and provides credibility to currency value and therefore is another factor affecting exchange rate determination.6 on the part of the government management, calvo, leiderman and reinhart (1994) on latin american countries, and marini and piersanti’s (2003) study covering asian countries found that a rise in current and expected future budget defi cits generated appreciation in exchange rates. hsiao and hsiao (2001) found a unidirectional causality from short-term external debt/international reserves ratio to exchange rates in korea. similar to martinez (1999) on mexico, frankel 1 1 1 d t t f t t e i e i 5 frankel and rose (1996b) on current account and government budget defi cits; calvo, leiderman and reinhart (1994) on capital fl ows, infl ation and current account defi cits; and aizenman and marion (2002) on reserve and credibility; hagen and zhou (2007); and others. 6 korea’s usable reserve fell from us$28 billion to a mere us$6 billion when their currency went on a free fall in december 1997: aizerman and marion (2002). brazil’s reserves fell from us$75 billion to less than half of that before the currency collapsed in 1998: dornbusch and fisher (2003). ht tp :// ijb f.u um .e du .m y exchange rate behaviour of east european transitional economics: 95-109 99 and rose (1996b) studied a large group of developing countries and found that the level of debt, foreign direct investment, foreign interest rates, foreign reserves and growth rates affect exchange rates signifi cantly. papell and theodoridis’s (1998) study on openness to trade, exchange rates and prices found stronger evidence of ppp for countries with less exchange rate volatility and shorter distance from other countries but not for countries with greater openness to trade. using a panel of oecd countries, canzoneri, cumby and diba (1999) found that when relative productivity of traded goods grew more rapidly in italy and japan than in germany, both lira and yen appreciated in real terms against deutschemark. macdonald and wojcik’s (2003) study on eu accession countries found that productivity, as well as private and government consumption signifi cantly affect exchange rate behaviour. frommel, macdonald and menkhoff (2005), vlaar (2007) and bailey, millard and wells (2001) found evidence of a relationship between productivity, growth and exchange rates. chinn (2000) provided evidence on productivity explanation for long-run real exchange rate movements in east asia: japan, indonesia, korea, malaysia and philippines but not for singapore, china, thailand and taiwan. 3. data and methodology the data relating to exchange rates between individual countries, and the united states (u.s.) dollar as the foreign unit are observed at end of observation periods. the international financial statistics (ifs) is the major source for these data for both quarterly and yearly time series. price variable of consumer price index (cpi) of individual countries; treasury-bill and money market rates are used to arrive at the interest differentials between countries. u.s. short term treasury-bill rate is the foreign interest rate for measuring interest differentials between countries. changes in exchange rates, prices and interest differentials are calculated using natural logarithm. the non-parity current and capital fl ow variables include: trade balance (trade) from imports and exports of goods, current account balance (cur), balance of payments (bop) from overall balance, capital fl ows include both infl ows and outfl ows of foreign direct investment (fdi) and portfolio investments (pt), total reserves (tres) as well as foreign debt (fd). monetary expansion data is broader money7 (m2) which includes both money and quasimoney. growth rate (prodty) is measured by change in gross domestic product (gdp) per capita. the set of dummy variables includes exchange regimes which are grouped into three categories: free-fl oat, exchange band/managed, and fi xed regime. trade openness is measured by total trade (ttrade), that is, the sum of total imports and exports, as a proportion of gdp. incomplete data are sourced from datastream, world bank as well as individual country’s central banks and statistical departments. the countries included in this study consist of eight in the eastern european region: czech republic, hungary, poland, romania, russia, slovak, slovenia and turkey and data relate to the period 1994-2004 with quarterly and yearly data. the reasons behind the choice of countries are the high level of inter-trade between countries in the similar geographical region and the availability of ht tp :// ijb f.u um .e du .m y 100 exchange rate behaviour of east european transitional economics: 95-109 information on these nations. this study investigates exchange rates until 2004 because it recognizes that some of these eastern european countries have moved into the eu with other exchange rate arrangement since then. 4. pooled series panel model as the sample data are limited to just over a decade of transition since the most useful time series are available only on an annual basis: similar studies typically use panel estimation for a larger number of these economies. both seemingly unrelated regression (sur) and fi xed effect (fe) pooled data model are employed to investigate exchange rate behaviour. sur allows cross-sectional variations in the data set, and thus yields robust estimates of the test statistics according to zellner, 1962. as a system of equations, this method can be applied here rather than estimating the equation in one cross section thus losing the time series variations in each country, which would be wasteful as it would leave out information in the data set. sur is estimated using generalized least squares algorithm. since sur technique utilizes information on the correlation between the error terms, the resulting estimates are more precise than estimates from least squares: it also yields lower standard errors and higher r². more recent studies have also concentrated on longitudinal data set. these panel data sets are more oriented toward cross-sectional analyses. panel data provides a richer environment for the development of estimation techniques with robust test results. it allows the use of time-series cross-sectional data to examine issues that could not be studied in either cross-section or time-series settings alone. by allowing cross-sectional variation or heterogeneity to affect estimations, the resulting estimates are robust. we use the fi xed effect approach here because it permits the constant term to be the country-specifi c variations in the regression as stated in greene, 2003. this is referred to as the least squares dummy variable (lsdv) model. the random effect model is not appropriate for our tests. we also assume that the issue of ambiguous relationship may be minimized through the use of instrumental-variables (iv) regression. the hausman (1978) test statistics proposed by davidson and mackinnon (1993) for endogeneity is applied. in summary, this econometric model recognizes the relationships between fundamentals and exchange rate movements. the analysis of the determinants of exchange rates is carried out by estimating the pooled and panel data regressions as follows, where the subscripts i and t index country and time, respectively. (3) the fi xed effect approach allows the constant term to vary from one crosssection unit to another (the lsdv model). this helps to control for unobserved ( ) ( ) * *it i it it it it it p i er ad b c dtrade ecur ebop p i reit it it it it itfinfdi gotfdi hinpt jotpt kfdebt mt s pr reit it it it it itn od pbdgt qttrade rtmy s gime ht tp :// ijb f.u um .e du .m y exchange rate behaviour of east european transitional economics: 95-109 101 components of country heterogeneity (through having country-specifi c constant terms) that may in fact drive both exchange rates and other country characteristics included in the regressions. diagnostic tests including unit root tests and variance infl ation factor (vif) that shows how the variance of an estimator is infl ated by the presence of multicollinearity are shown in tables 1 and 2. some data tests were conducted to ensure that the results are robust. this included the stationarity tests using the latest kpss procedure. as is well known, this test is a better test than the earlier tests such as the adf tests such that the resulting interpretation is robust in a small as well as large sample. the test results as is evident from the statistics in the table, the variables in our tests are stationary, and that the results from both tests suggest that the variables are stationary with the exception of two variables entering the regression. table 1: non-parity variables vif and tolerance measure eastern europe variables vif tolerance lnp 1.679 0.596 lni 1.807 0.553 trade/gdp 3.873 0.258 cur/gdp 3.863 0.259 bop/gdp 2.796 0.358 infdi/gdp 1.184 0.845 outfdi/gdp 1.096 0.912 inpt/gdp 1.992 0.502 otpt/gdp 1.234 0.811 tres/im 2.061 0.485 bgt/gdp 1.331 0.751 tmy/gdp 1.961 0.510 prod 2.052 0.487 fd/gdp 1.382 0.724 ttrade/gdp 1.731 0.578 regime 1.587 0.630 * vif values of more than 10 shows signifi cant multicollinearity. ht tp :// ijb f.u um .e du .m y 102 exchange rate behaviour of east european transitional economics: 95-109 table 2: unit root tests for parity and non-parity variables eastern europe variables adf test kpss test t-stats model (lag) kpss statistic lner -11.76*** c(0) 0.431* lnp -1.51 c(4) 0.158 lni -6.93*** c(2) 0.415* trade/gdp -6.35*** c(7) 0.131 cur/gdp -8.58*** c(6) 0.131 bop/gdp -15.24*** c(2) 0.151 infdi/gdp -26.67*** c(0) 0.112 outfdi/gdp -8.86*** c(7) 0.098 inpt/gdp -15.96*** c(2) 0.100 outpt/gdp -7.52*** c(7) 0.381* tres/im -14.30*** c(0) 0.034 bdgt/gdp -16.99*** c(2) 0.083 tmy/gdp -11.48*** c(6) 0.134 prodty -11.13*** c(6) 0.167 fd/gdp -14.70*** c(2) 0.062 ttrade/gdp -8.35*** c(6) 0.248 note 1: critical values for adf tests at 10,5 and 1% levels of signifi cance are respectively, -2.59, -2.90 and –3.53 with a constant and –3.17, -3.48 and –4.09 with a constant and a deterministic trend. critical values for kpss tests at 10, 5 and 1% levels of signifi cance are respectively, 0.35, 0.46 and 0.74 with a constant and 0.12, 0.15 and 0.22 with a constant and a linear trend. note 2: for the adf tests, the unit root null is rejected if the value of the adf t-statistics is less than the critical value. for the kpss tests, the null of stationarity is rejected if the value of the kpss statistic is greater than the critical value. *, ** and *** denote statistical signifi cance at 10, 5 and 1% level. the critical values for the adf tests are from mackinnon (1991). 5. findings and interpretation from the results using quarterly data, and the data for one to three-year intervals are summarized in table 3. the coeffi cient for interest rate is statistically signifi cant for all time period intervals from the short to long run. the price coeffi cient does not have the expected sign as predicted by the ppp and it is insignifi cant for all time periods. thus, for the region of ee transitional/ developing countries, the interest parity is holding but the price parity is not found to be statistically signifi cant, probably due to the slow adjustment of prices ht tp :// ijb f.u um .e du .m y exchange rate behaviour of east european transitional economics: 95-109 103 t ab le 3 : s u r a nd f ix ed e ff ec ts r es ul ts f or e as te rn e ur op e (c o n ti n u ed ) e as te rn e ur op e q ua rt er ly y ea rl y 2 y ea rl y 3 y ea rl y s u r ♣ ♦ f ix ed e ff ec ts ♠ # s u r ♣ ♦ f ix ed e ffe ct s♠ # s u r ♣ ♦ f ix ed e ffe ct s♠ # s u r ♣ ♦ f ix ed e ffe ct s♠ # o bs er va ti on 14 7 14 7 42 42 27 27 18 18 in te rc ep t .0 29 (2 .3 4) ** .0 50 (2 .5 6) * -. 01 1 (0. 26 ) .0 09 (0 .1 1) .2 48 (3 .8 6) * .4 27 (3 .5 9) ** .3 99 (3 .7 9) * -3 0. 17 9 (5. 72 )* * p ar it y p ri ce -. 00 2 (0. 58 ) -. 01 0 (2. 43 )* * .0 23 (1 .1 7) -. 02 4 (0. 73 ) -. 02 2 (1. 11 ) -. 13 5 (4. 67 )* -. 09 9 (1. 11 ) -. 01 7 (0. 10 ) in te re st .1 77 (7 .9 8) * .1 65 (8 .1 0) * .8 57 (9 .6 5) * .5 03 (2 .8 3) * 2. 01 0 (1 2. 09 )* 1. 57 5 (3 .3 9) ** 3. 02 9 (8 .2 9) * 2. 91 3 (2 .9 1) ** * n on p ar it y t ra de / g d p -. 45 0 (2. 31 )* * -. 30 9 (1. 80 )* ** -. 23 6 (0. 23 ) .2 52 (0 .2 6) .2 27 (0 .1 8) 2. 47 9 (2 .0 5) ** * b o p / g d p .0 18 (0 .2 3) .0 02 (0 .0 2) .5 31 (1 .2 1) -. 20 4 (0. 47 ) 1. 58 7 (1 .8 9) ** * -1 .5 12 (2. 40 )* ** 2. 00 1 (3 .1 5) * 3. 61 8 (4 .0 3) ** * c ur / g d p .2 74 (1 .4 9) .2 29 (1 .5 2) -1 .0 32 (0. 91 ) -. 10 7 (0. 16 ) -1 .1 71 (0. 78 ) -1 .3 59 (1. 04 ) -. 00 1 (0. 70 ) -2 .1 47 (5. 72 )* * in f d i/ g d p .2 47 (1 .4 7) .3 12 (1 .9 7) ** -. 90 0 (0. 93 ) .0 44 (0 .0 5) -. 12 8 (0. 15 ) -2 .4 85 (1. 74 ) -. 00 1 (3. 09 )* 4. 62 1 (5 .7 2) ** o u tf d i/ g d p .0 06 (0 .0 2) .0 28 (0 .1 6) -2 .6 40 (0. 84 ) 10 .7 78 (1 .4 1) -1 .2 69 (0. 26 ) -2 5. 62 4 (3. 50 )* * in p t/ g d p -. 14 5 (1. 25 ) -. 09 3 (0. 92 ) -. 03 4 (0. 05 ) -. 59 4 (0. 69 ) -2 .0 85 (1. 64 ) -1 .6 11 (3. 66 )* * .0 01 (0 .9 6) -4 .4 13 (5. 72 )* * t a b le c o n ti n u es o n t h e n ex t p a g e ht tp :// ijb f.u um .e du .m y 104 exchange rate behaviour of east european transitional economics: 95-109 o ut p t/ g d p -. 10 9 (0. 31 ) -. 03 1 (0. 10 ) .6 48 (0 .5 0) 1. 93 2 (2 .4 0) ** -1 .2 49 (0. 55 ) -3 .1 22 (6. 07 )* t r es / im .0 03 (0 .0 9) -. 01 1 (0. 25 ) .2 57 (0 .7 9) .2 89 (1 .0 9) -. 07 6 (0. 31 ) .7 64 (3 .4 9) ** -. 36 4 (1. 12 ) -1 .6 18 (2. 25 ) f or d t/ g d p -. 01 4 (0. 10 ) -. 06 9 (0. 94 ) -2 .7 22 (2. 13 )* * -. 11 0 (0. 07 ) p ro dt y 19 .7 15 (1 .0 0) -1 01 .7 19 (2. 64 )* -8 3. 97 8 (2. 66 )* -1 41 .4 19 (5. 09 )* -1 04 .2 65 (2. 70 )* * 45 .5 32 (0 .4 5) -1 34 .2 42 (4. 96 )* -2 12 .9 36 (8. 60 )* b dg t/ g d p -. 14 7 (2. 14 )* * -. 06 9 (0. 94 ) 2. 70 3 (5 .2 4) * 1. 78 1 (1 .8 9) ** * t m y/ g d p -. 66 0 (15 .2 0) * -. 53 8 (4. 75 )* -. 67 0 (1. 87 )* ** -. 22 9 (0. 56 ) -. 23 7 (0. 53 ) -2 .1 91 (1. 90 ) r eg im e .0 04 (0 .3 8) .0 02 (0 .2 1) .0 66 (2 .2 5) ** .0 95 (2 .0 4) ** * -. 06 4 (1. 43 ) -. 12 6 (2. 72 )* * t t ra d e / g d p .0 64 (1 .1 5) .1 14 (1 .8 4) ** * -. 19 9 (1. 27 ) .0 06 (0 .0 3) -. 31 8 (3. 14 )* -. 35 7 (1. 00 ) -. 35 4 (1. 03 ) 1. 12 7 (1 .2 9) a dj r ² 0. 80 8 0. 82 3 0. 93 7 0. 95 2 0. 96 2 0. 96 9 .9 99 0. 79 6 f -p ro b 0. 00 0 0. 00 0 0. 00 0 0. 00 0 0. 00 0 0. 00 0 0. 00 0 0. 16 7 t he n um be r of o bs er va ti on s fo r s u r a nd f ix ed e ff ec ts i s lo w er d ue t o h un ga ry w it ho ut f ul l da ta u nt il 1 99 5, p ol an d w it ho ut p or tf ol io o ut fl ow s da ta un ti l an d in co m pl et e da ta f or s lo va ki a fr om 2 00 1. ♣ p oo le d g en er al l ea st s qu ar es w it h c ro ss -s ec ti on s u r t ha t es ti m at es a f ea si bl e g l s s pe ci fi ca ti on co rr ec ti ng f or b ot h cr os sse ct io n he te ro sc ed as ti ci ty a nd c on te m po ra ne ou s co rr el at io n. ♠ f ix ed e ff ec ts p oo le d g l s w it h cr os s se ct io n w ei gh ts w he re e vi ew s es ti m at es a f ea si bl e g l s s pe ci fi ca ti on a ss um in g th e pr es en ce o f cr os sse ct io n he te ro sc ed as ti ci ty . # w it h w hi te ’s c ro ss -s ec ti on s ta nd ar d er ro rs & co va ri an ce c or re ct io n by t re at in g po ol ed r eg re ss io n as a m ul ti va ri at e re gr es si on w it h an e qu at io n fo r ea ch c ro ss s ec ti on a nd c om pu ti ng w hi te -t yp e ro bu st s ta nd ar d er ro rs f or t he s ys te m o f eq ua ti on s. ♦ w it h cr os sse ct io n s u r ( p c s e ) us in g p an el c or re la te d s ta nd ar d e rr or m et ho do lo gy s ta nd ar d er ro rs & c ov ar ia nc e co rr ec ti on . n um be rs i n pa re nt he se s ar e tst at is ti cs . *, * *, * ** r ep re se nt 1 % , 5% , 10 % s ig ni fi ca nc e le ve l re sp ec ti ve ly . f -p ro b re pr es en ts f -p ro ba bi li ty v al ue s an d a dj r ² re pr es en ts a dj us te d r -s qu ar ed v al ue s. t ab le 3 : s u r a nd f ix ed e ff ec ts r es ul ts f or e as te rn e ur op e ht tp :// ijb f.u um .e du .m y exchange rate behaviour of east european transitional economics: 95-109 105 to exchange rates (and the short data period available from 1994). since most of these countries are newly formed, when a lengthier time period is available in the future, later studies might be able to furnish us with more theoretically consistent results. the role of non-parity fundamentals cannot be ignored both in the short as well as the longer term. quarterly trade balance is statistically signifi cant in affecting exchange rates in the shorter period where improvement in trade improves the value of the domestic currency. government’s monetary expansion also plays an important part in infl uencing exchange rates in the shorter term when monetary expansion is needed to sustain rapid economic growth resulting in a positive effect on domestic exchange rates. trade openness has a negative effect with exchange rates in the shorter period quarterly intervals but it has a positive relationship with exchange rates in the longer period of two-year intervals. growth rate is also a statistically signifi cant variable in the determination of exchange rates where faster growth rates strengthen the domestic exchange rates especially after taking into consideration individual country effects in the fi xed effect model. fiscal budget balance is marginally signifi cant in infl uencing exchange rates at oneyear intervals however budget surplus corresponds to a fall in currency value which is inconsistent with theoretical understanding. the other fundamentals including balance of payments, capital fl ows, and accumulation of reserves are insignifi cant in affecting exchange rates in the shorter term. in the longer term for this region of developing countries in eastern europe, interest parity continues to hold and higher interest rates deteriorate the value of the domestic currency in accordance with the prediction of irp. the coeffi cients for current account and portfolio infl ows are positively related to exchange rates and are statistically signifi cant in three-year intervals. conversely, foreign debt and accumulation of reserves have incorrect signs but are also insignifi cant in determining exchange rates in this region. growth and interest rates continue to be statistically signifi cant in the longer period in affecting the currency values of these countries. on the other hand, monetary expansion and trade openness do not continue to be infl uential in the longer run. in summary, (1) interest rate, (2) current account balance, (3) capital fl ows and (4) growth rate are major driving forces of exchange rates in the longer term. however in the interim periods of one and two-year intervals, other nonparity variables budget, monetary expansion and trade openness are marginally signifi cant in determining exchange rates. 6. conclusion the results presented in this paper constitute a modest contribution to extend the literature on exchange rate behaviour for the region of trade-related transition economies in eastern europe by considering the extent to which parity and nonparity variables infl uence the movements of exchange rates systematically. these ht tp :// ijb f.u um .e du .m y 106 exchange rate behaviour of east european transitional economics: 95-109 results are robust as these are obtained from pooled panel data to test the theories for a group of trade-related countries. we fi nd that non-parity fundamentals (1) growth rate, (2) current account balance and (3) capital fl ows are signifi cant drivers of exchange rates. that these factors are so important to this region is highlighted by these tests in contrast with results for other regions such as asean countries. we believe the tests developed in this study led to improved results, helped to identify new fundamentals that are related to exchange rates while the puzzle of the short term versus long term behaviour is made obvious by applying different data frequencies from quarterly to several years. these transition economies would be well equipped to fi nd an appropriate exchange rate stability to be admitted into the eu and also eventually to be able to manage their currency fl uctuations within the regional currency area. last but not least, this study may be constrained by the lack of data in terms of both length of time series and quality and reliability: our most reliable source is the ifs of the international monetary fund. we also assume that exchange rates have been free to adjust according to macroeconomic fundamentals and central bank intervention is relatively minor. even though some of these countries are having managed or pegged exchange rates, the results are still useful as a starting point for an investigation of the behaviour of their exchange rates. a permanent peg system is never a solution to infl ationary problem indefi nitely in the long run for small open economies. in the case of transition economies, there should be a more fl exible exchange rate arrangement once infl ation has been eliminated with strong fi nancial and banking system in place. author statement: the author is an associate professor in the faculty of business management, university technology mara, 40450 shah alam, malaysia. telephone: (603) 5544-4843 email: catherine@salam.uitm.edu.my. references adler, m., and lehman, b. (1983). deviations from purchasing power parity in the long run. journal of finance, 38(5), 1471-1487. aizenman, j., and marion, n. (2002). reserve uncertainty and the supply of international credit. journal of money, credit, and banking, 34(3), 631649. baharumshah, a. z., and masih, a. m. m. (2005). current account, exchange rate dynamics and predictability: the experience of malaysia and singapore. journal of international financial markets, institutions and money, 15, 255-270. bailey, a., millard, s., and wells, s. (2001). capital fl ow and exchange rates. bank of england quarterly bulletin, london, 41(3), 310-318. balassa, b. (1964). the purchasing-power parity doctrine: a reappraisal. the journal of political economy, 72(6), 584-596. bank for international settlements, annual reports and press release: 2006 and 2007. ht tp :// ijb f.u um .e du .m y exchange rate behaviour of east european transitional economics: 95-109 107 bekaert, g., wei, m., and xing, y. (2007). uncovered interest rate parity and the term structure. journal of international money and finance, 26, 1038-1069. bergin, p. r. (2006). how well can the new open economy macroeconomics explain the exchange rate and current account. journal of international money and finance, 25, 675-701. calvo, g. a., leiderman, l., and reinhart, c. m. (1994). the capital infl ow problem: concepts and issues. contemporary economic policy, 12(3), 54-60. canzoneri, m. b., cumby, r. e., and diba, b. (1999). relative labour productivity and the real exchange rate in the long run: evidence for a panel of oecd countries. journal of international economics, 47(2), 245-266. cassel, g. (1918). abnormal deviations in international exchanges. the economic journal, 28(112), 413-415. chaboud, a. p., and wright, j. h. (2005). uncovered interest parity: it works, but not for long. journal of international economics, 66(2), 349-362. chinn, m. d. (2000). the usual suspects? productivity and demand shocks and asia-pacifi c real exchange rates. review of international economics, 8(1), 20-43. davidson, r., and mackinnon, j. g. (1993). estimation and inference in econometrics. new york: oxford university press. dornbusch, r. (1976). expectations and exchange rate dynamics. the journal of political economy, 84(6), 1161-1176. dornbusch, r., and fisher, s. (2003). international fi nancial crisis. cesifo working paper, 926. ehrmann, m., and fratzscher, m. (2005). exchange rates and fundamentals: new evidence from real-time data. journal of international money and finance, 24(2), 317-341. fisher, i., (1930). the theory of interest. new york: macmillan. frankel, j. a., and rose, a. k. (1996a). a panel project on purchasing power parity: mean reversion within and between countries. journal of international economics, 40, 209-224. frankel, j. a., and rose, a. k. (1996b). currency crashes in emerging markets: an empirical treatment. journal of international economics, 41, 351366. frommel, m., macdonald, r., and menkhoff, l. (2005). do fundamentals matter for the d-mark/euro-dollar? a regime switching approach. global finance journal, 15, 321-335. greene, w. h. (2003). econometric analysis (5th ed.). new jersey: prentice hall. hagen, j. v., and zhou, j. (2007). the choice of exchange rate regime in developing countries: a multinomial panel analysis. journal of international money and finance, 26, 1071-1094. hausman, j. a. (1978). specifi cation tests in econometrics. econometrica, 46(6), 1251-1271. ht tp :// ijb f.u um .e du .m y 108 exchange rate behaviour of east european transitional economics: 95-109 hsiao, f. s. t., and hsiao, m.-c. w. (2001). capital fl ows and exchange rates: recent korean and taiwanese experience and challenges. journal of asian economics, 12, 353-381. international monetary fund, annual report and research papers: 2006, 2007. juselius, k., and macdonald, r. (2004). international parity relationships between the usa and japan. japan and the world economy, 16, 17-34. kaminsky, g. l., and reinhart, c. m. (1999). the twin crises: the causes of banking and balance-of-payments problems. the american economic review, 89(3), 473-500. keynes, j. m. (1923). a tract on monetary reforms. london: macmillan. kim, y. (2000). causes of capital fl ows in developing countries. journal of international money and finance, 19, 235-253. kuo, b.-s., and mikkola, a. (2001). how sure are we about purchasing power parity? panel evidence with the null of stationary real exchange rates. journal of money, credit, and banking, 33(3), 767-789. lee, j., and chinn, m. d. (2006). current account and real exchange rate dynamics in the g7 countries. journal of international money and finance, 25, 257-274. lothian, j. r., and taylor, m. p. (2000). purchasing power parity over two centuries: strengthening the case for real exchange rate stability. journal of international money and finance, 19, 759-764. macdonald, r. (1993). long-run purchasing power parity: is it for real? the review of economics and statistics, 75(4), 690-695. macdonald, r., and wojcik, c. (2003). catching up: the role of demand, supply and regulated price effects on the real exchange rates of four accession countries. cesifo working paper, 899. marini, g., and piersanti, g. (2003). fiscal defi cits and currency crises. centre for international studies, research paper series, 15. mark, n. c. (1995). exchange rates and fundamentals: evidence on long-horizon predictability. the american economic review, 85(1), 201-218. mark, n. c., and sul, d. (2001). nominal exchange rates and monetary fundamentals: evidence from a small post-bretton woods panel. journal of international economics, 53, 29-52. martinez, j. d. l. c. (1999). mexico’s balance of payments and exchange rates: a cointegration analysis. the north american journal of economics and finance, 10, 401-421. meese, r. a., and rogoff, k. (1983). empirical exchange rate models of the seventies: do they fi t out of sample? journal of international economics, 14, 3-24. meese, r. a., and rogoff, k. (1988). was it real? the exchange rate-interest differential relation over the modern fl oating-rate period. the journal of finance, 43(4), 933-948. murray, c., and papell, d. (2005). the purchasing power parity puzzle is worse than you think. empirical economics, 30, 783-790. ht tp :// ijb f.u um .e du .m y exchange rate behaviour of east european transitional economics: 95-109 109 muscatelli, v. a., spinelli, f., and trecroci, c. (2007). macroeconomic shocks, structural change and real exchange rates: evidence from historical data. journal of international money and finance, 26, 1403-1423. papell, d. h., and theodoridis, h. (1998). increasing evidence of purchasing power parity over the current fl oat. journal of international money and finance, 17(1), 41-50. parsley, d. c. (2007). accounting for real exchange rate changes in east asia. journal of international money and finance, 26, 1355-1377. rogoff, k. (1996). the purchasing power parity puzzle. journal of economic literature, 34(2), 647-668. samuelson, p. (1964). theoretical notes on trade problems. the review of economics and statistics, 46(2), 145-154. schnabl, g., and baur, d. (2002). purchasing power parity: granger causality tests for the yen-dollar exchange rate. japan and the world economy, 14, 425-444. serletis, a., and gogas, p. (2004). long-horizon regression tests of the theory of purchasing power parity. journal of banking and finance, 28, 19611985. smallwood, a. d. (2007). measuring the persistence of deviations from purchasing power parity with fractionally integrated star model. journal of international money and finance, 26, 1-16. toma, r. (2007). exchange rate arrangement in central and eastern european countries evolutions and characteristics. munich personal repec archive, paper no. 1967. vlaar, p. j. g. (2007). gdp growth and currency valuation: the case of the dollar. journal of international money and finance, 26, 1424-1449. world bank. (2006). world development indicators, database 2006. zellner, a. (1962). an effi cient model of estimating seemingly unrelated regressions and tests for aggregation bias. journal of american statistical association, 348-368. ht tp :// ijb f.u um .e du .m y ht tp :// ijb f.u um .e du .m y impact of macroeconomic and endogenous factors on non performing bank assets 27 the international journal of banking and finance, volume 9 (number 1) 2012: pages 27-47 impact of macroeconomic and endogenous factors on non performing bank assets vighneswara swamy indian business school, hyderabad, india ____________________________________________________________ abstract determinants of default risk of banks in emerging economies have so far received inadequate attention in the literature. using panel data techniques, this paper seeks to examine the impact of macroeconomic and endogenous factors on non-performing assets during 1997-2009. the findings reveal some interesting inferences contrary to the perception of few opinion makers. lending rates have been found to be not so significant in affecting the non-performing loans, which is contrary to the general perception. such assets have a negative and significant influence, indicating that large banks may have better risk management procedures and technology which definitely allows them to finish up with lower levels of non-performing assets compared to smaller banks. further, this study suggests that private banks and foreign banks have advantages in terms of their efficiencies in better credit management that contains non-performing assets. that indicates that bank privatization can lead to better management of default risk. key words: banks, risk management, ownership structure, financial markets, non-performing assets, lending policy, macro-economy, central banks jel classification: g21, g32, e44, e51 _____________________________________________ 1. introduction global financial crisis, which has developed into the most severe crisis of the post-world war ii crisis, has hit the real economy on a devastating scale resulting in the collapse of financial markets and institutions. moreover, output per capita is projected to slide down in countries representing three-quarters of the global economy. the consequent deterioration in the economic environment has led to a rise in the overall level of stress in the banking sectors. commercial bank loan charge-offs in the us and europe may exceed the levels reached during the 1991–1992 recession, even though they should remain below the levels experienced in the us during the great depression. 28 on a thorough analysis of the crisis, financial stability has once again emerged as an important area of concern in the financial systems across the globe. financial stability is widely accepted as a situation in which financial system is capable of satisfactorily performing its three key functions simultaneously: (1) efficient and smooth facilitation of the inter-temporal allocation of resources from the surplus economic units to the deficit economic units; (2) managing the forward looking financial risks with appropriate pricing; and (3) to be prepared all the time to absorb the financial and real economic surprises and shocks. the core function of the payment system should therefore not fail as well. counterparty risk, being an important risk in the financial system more particularly of the banking system, poses a bigger challenge in order to achieve financial stability. counter-party risk is an outcome directly related to the non-performing assets1 [npas] of a financial institution. even though npas are permanent phenomenon in the balance sheets of the financial institutions, if not contained properly, they eventually lead to crisis that can pose big threats of contagion that can engulf the financial health of the system. the problem of npas in the financial sector has been an matter of concern for all economies and resolution of npas has become synonymous to functional efficiency of financial intermediaries. as per the global financial stability report of international monetary fund, (imf, 2009), identifying and dealing with distressed assets, and recapitalizing weak but viable institutions and resolving failed institutions are stated as the two of the three important priorities which directly relate to npas. in view of this, it is essential to identify and understand the determinants of npas, which is the motive for this paper. this paper is organised into four sections. the next section is a very brief overview of the theoretical considerations along with a short review of latest literature. in section 3 discussion on factors relevant for npas is presented. in section 4, the reader will find an attempt to model this research on npas in an economy. the results are presented in section 5 showing possible impact of macroeconomic and endogenous determinants on npas. policy issues and conclusions follow in the following sections. 1 loans the bank anticipates will have difficulty in recovering. npas mainly arise due to the default of the borrower which involves his inability or unwillingness in meeting the commitments to the loan. non-performing assets (npas) or bad loans, as they are commonly called, have been a menace for the banking sector across the world. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 2 29 2. theoretical considerations the critical role of financial and banking development in economic growth in any economy has been established by many researchers (levine, 2004 and singh, 2005). goldsmith (1969 ) is adamant in pointing to excessive credit growth as the harbinger of all crises. the role of how financial crisis could destabilise that role is described in ariff (2012). these pointers should be judged against the state of affairs that there is as yet a credible theory of financial market equilibrium (allan and gale, 2007). in the process of providing credit assistance to the investment activities and projects in the economy, financial institutions face inherent risks in the form of default risk which results in build-up of non-performing assets (npas) that have a negative effect on the profitability of the financial institutions. typically a credit transaction involves a contract between two parties: the borrower and the creditor (bank) are subject to a mutual agreement on the ‘terms of credit’.2 optimising decision pertaining to the terms of credit could differ from the borrower to that of the creditor. as such, the mutual agreement between the borrower and the creditor may not necessarily imply an optimal configuration for both. the most important reason for default3 could be mismatch between borrower’s terms of credit and creditor’s terms of credit. however, a common perspective is that both the cases of defaulter and nonperformer imply similar financial implications, that is, financial loss to banks. moreover, regulatory and supervisory process does not focus on such a distinction between defaulter and non-performer as far as prudential norms are concerned. a synoptic look at the available literature conveys some useful insights into the determining factors of npas across countries. quite a lot of economies have experienced such distressed debt cycles. npa levels and capital to risk (weighted) assets ratio (crar) of developing and advanced countries presented in table 1 and the provisions to npas and return on assets (roa) of developing and advanced countries captured in table 2 explain the differences in the levels as well as the approaches towards npa management in different countries. 2 the ‘terms of credit’ are defined over five critical parameters, viz, amount of credit, interest rate, maturity of loans, frequency of loan servicing and collateral. 3 a ‘default’ entails violation of the loan contract or the agreed terms of the contract, while a non-performing loan entails that the borrower does not renege from the loan contract but fails to comply the repayment schedule due to evolving unfavourable conditions. 30 table 1: npa levels and crar of developing & advanced countries country crar npa/tl 2002 2006 2007 2008 2002 2006 2007 2008 developing economies china 8.4 8.2 26 7.5 6.7 2.5 india 12 12.4 12.3 13 10.4 3.5 2.5 2.3 indonesia 20.1 21.3 19.3 16.8 24 13.1 4.1 3.5 korea 11.2 12.8 12.3 10.9 2.4 0.8 0.7 1.1 south africa 12.6 12.3 12.8 12.5 2.8 1.2 1.4 2.6 advanced economies australia 9.6 10.4 10.2 10.9 0.4 0.2 0.2 0.5 canada 12.4 12.5 12.1 12.7 1.6 0.4 0.7 1.1 france 11.5 10.1 4.2 3.2 2.7 germany 12.7 12.9 5 4 2.7 italy 11.2 10.7 10.4 6.5 5.3 4.6 japan 9.4 13.1 12.9 12.3 7.4 2.5 1.5 1.5 united kingdom 13.1 12.9 12.6 2.6 0.9 0.9 united states 13 13 12.8 12.5 1.4 0.8 1.4 2.3 source: global financial stability report, april 2009, imf table 2: provisions to npas and roa of developing & advanced countries country provisions to npas roa 2002 2006 2007 2008 2002 2006 2007 2008 developing economies china 39.2 115.3 0.9 1 india 58.9 56.1 52.6 0.8 0.9 0.9 1 indonesia 130 99.7 87.7 98.5 1.4 2.6 2.8 2.6 korea 89.6 175.2 199.1 155.4 0.6 1.1 1.1 south africa 46 0.4 1.4 1.4 1.8 advanced economies australia 106.2 204.5 183.7 87.2 1.4 1 0.9 canada 41.1 55.3 42.1 34.7 0.4 1 0.9 1.3 france 58.4 58.7 61.4 0.5 0.4 germany 77.3 0.1 0.5 0.2 italy 46 49.5 0.5 0.8 0.8 japan 30.3 26.4 24.9 -0.7 0.4 0.2 0.3 united kingdom 75 0.4 0.5 0.4 united states 123.7 137.2 93.1 84.7 1.3 1.3 0.8 0.3 source: global financial stability report, april 2009, imf. bank regulation based crar of select countries is captured in table 3 and figure 1. in the usa, the nonand sub-performing loans resolution was embedded into the savings and loans crisis from 1989 to 1994. in japan, the npa cycle began in 1997 and in china and the international journal of banking and finance, vol. 9, iss. 1 [2012], art. 2 31 figure 3: comparison of gross domestic product to bank credit in india source: compiled by author based on data from reserve bank of india publications rest of asia deal with npas sub-prime loans (spls) since 1999. the origin of the chinese npas crisis can be traced to political issues. during the centrally planned economy from 1949 onwards, loans were granted by state owned banks to state-owned companies without proper credit due diligence at predetermined standardised conditions by the government. especially, in the overheated economy of the 1990s domestic credits extended enormously and grew by 30 percent year on year between 1992 and 1995 (chen, 2004; sprayregen et al., 2004). table -3: bank regulatory capital to risk-weighted assets [in percent] no. country 2002 2003 2004 2005 2006 2007 1 china -5.9 -4.7 2.5 4.9 8.4 8.2 2 hong kong sar 15.3 15.4 14.9 15.2 13.4 14.3 3 india 12.7 12.9 12.8 12.3 12.3 13 4 indonesia 22.3 19.4 19.3 21.3 19.3 16.8 5 korea 11.1 12.1 13 12.8 12.3 10.9 6 brazil 16.6 18.8 18.6 17.9 18.9 18.4 7 russia 19.1 19.1 17 16 14.9 16.8 8 france 11.5 11.9 11.5 11.4 10.9 . . . 9 germany 12.7 13.4 13.2 12.2 12.5 . . . 10 south africa 12.6 12.4 14 12.7 12.3 12.2 11 australia 9.6 10 10.4 10.4 10.4 10.3 12 canada 12.4 13.4 13.3 12.9 12.5 12.1 13 japan 9.4 11.1 11.6 12.2 13.1 12.9 14 united states 13 13 13.2 12.9 13 12.8 15 united kingdom 13.1 13 12.7 12.8 12.9 . . . source: global financial stability report, april 2009, imf. 32 figure 1: bank regulatory capital to risk-weighted assets source: global financial stability report, april 2009, imf. a typically high leverage in the country shown by a debt/gross domestic product (gdp) ratio of 146 percent may be an indicator for the problem of npas (see ernst & young, 2001). high leverage was experienced in the real estate sector, particularly during the rise of the japanese real estate bubble in the 1980s. with the burst of the bubble in the 1991 and the dramatic economic slowdown, real estate values waned tremendously in the case of japan. as an aftermath, borrowers defaulted on the debt service and lenders had to sign big losses. lacking regulations and tax incentives by the japanese government as well as insufficient equity reserves of the banks to compensate write-offs of distressed debt, banks tried to deal with the problem by a wait-and-see approach. barseghyan (2004) identifies a link between the japanese government’s reluctance to solve the bad loan problem and the economic slowdown. he opines that the governments behaviour deteriorated the economic situation of japan and affirms this hypothesis by a normative study. in the case of thailand, the causes for npas include factors like liberalized capital and current account, a legal system that made credit recovery time consuming and difficult, real estate speculations, steep rise in interest rates and inability of the banks to assess the credit risk. to overcome the problem the financial sector restructuring plan (1998) focused on capital support facilities for bank recapitalization and setting up of asset management corporations (devakula, pridiyathorn, 2001). korean causes for distressed loans were like directed credit (kang, moon-soo, 2001), the “compressed growth policy” which backfired when slowing demand and rising input costs placed severe stress on their profitability, their international journal of banking and finance, vol. 9, iss. 1 [2012], art. 2 33 lack of monitoring and contagion effects. these issues were attempted to be countered with measures like; creation of the korea asset management corporation (kamco) and a npa fund to support the acquisition of npas (bama, 2002) and corporate restructuring vehicles (crvs) and debt/equity swaps were made use of to ease the resolution of bad loans. indian banking system has endured the stress of global financial crisis as revealed in the advancement in the capital to risk-weighted assets ratio (crar). on the whole crar of all (81) scheduled commercial banks (scbs) improved to 13.2 per cent at end-march 2009, continuing to be considerably well above the required minimum of 9.0 per cent. the gross npas to gross advances ratio (gnpa) continued to be at 2.3 per cent as at end-march 2009 as it was at end-march 2008. the roa also remained unchanged at 1.0 per cent at endmarch 2009 over its level at end-march 2008 indicating no deterioration in efficiency with which banks deployed their assets. the return on equity (roe) improved from 12.5 per cent at end-march 2008 to 13.3 per cent as at end-march 2009, signifying increased efficiency in the capital usage by banks. in india, as in most other countries, npas are only an indicator of loan performance. figure 2: gnpa levels of bank groups in india source: compiled by author based on data from reserve bank of india publications. the degree to which it measures actual performance of banks depends on the quality of accounting, auditing, regulation and supervision and the amount of ‘ever greening’ of weak loans, through restructuring, which is an incessant problem in india to judge from the numerous circulars against the practice, which the reserve bank of india (rbi) has issued against it over the last decade. although npas have been substantially reduced since regulation was tightened in 1993, especially in the public sector banks (psbs), the 34 momentum has recently slowed down and the levels of npas remain high compared to international standards (refer o figure-2). he further argues that the problems of npas have a sizeable overhang component, resulting from infirmities in the prevailing practices of debt recovery, inadequate legal provisions for foreclosure and bankruptcy and hitches in the execution of court decrees. the problem is exacerbated by the regulatory provisions for loan classification vis-à-vis international best practices. even though public sector banks have recorded surges in profitability, efficiency in asset quality and intermediation costs in the 1990s, they continue to linger with high interest rate spreads astonishingly with lower rates of return, and higher operating costs. in an advancing rate regime, prime borrowers would switch over to other possibilities such as capital markets, internal accruals, etc., for their fund requirements. under such conditions, banks would have no alternative but to mitigate the quality of borrowers which increase the probability of generation of npas (bhattacharya, 2001). 3. factors relevant for npas 3.1 macroeconomic factors most of the empirical evidence suggests that banks’ npas as closely linked to the economic activity. in other words, macroeconomic factors matter: downturns/slowdowns in the economy, recessions, low rate of savings, weak markets, depressions in industrial production, reduction in per capita income levels and most importantly the inflation levels in the economy. a fair amount of the academic literature has dealt with determinants of banking crisis, which is the most severe of the consequences of bad loans in a banking system which is of valuable understanding as a backdrop for the study of npas.4 dermiguc-kunt and detragiache (2000) employed a multivariate logit framework to develop an early warning system for banking crisis and a ratings system for bank fragility. beck, demirguc-kunt, and levine (2005) examined the inter-linkage between bank concentration and banking system fragility where they have established that higher bank concentration is associated with lower profitability. lis, et.al.,(2000) have found that gross 4 non-performing assets is used interchangeably with non-performing loans in this comment. npas are measured on either gross basis or on net basis (net of provisions). while the gross npas reflects the quality of loans made by the banks, net npas shows the actual burden of the banks. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 2 35 domestic product growth, bank size and capital had negative effect on npas while loan growth, collateral, net interest margin, debt-equity, market power and regulation regime had a positive impact on npas. resti and sironi (2001) examined corporate bond recovery rate referring to bond default rate, macroeconomic variables such as gdp and growth rate, amount of bonds outstanding, amount of default, return on default bonds, and stock return wherein it was established that default rate, amount of bonds, default bonds, and economic recession had negative effect, while the gdp growth rate, and stock return had positive effect on corporate recovery rate. lis et.al.,(2000) used a simultaneous equation model in which they explained bank loan losses in spain using a host of indicators, which included gdp growth rate, debt-equity ratios of firms, regulation regime, loan growth, bank branch growth rates, bank size (assets over total size), collateral loans, net interest margin, capital-asset ratio (car) and market power of default companies. they found that gdp growth (contemporaneous, as well as one period lag term), bank size, and car, had negative effect while loan growth, collateral, net-interest margin, debt-equity, market power, regulation regime and lagged dependent variable had positive effect on problem loans. sergio (1996) in a study of non-performing loans in italy found evidence that, an increase in the riskiness of loan assets is rooted in a bank’s lending policy adducing to relatively unselective and inadequate assessment of sectoral prospects. interestingly, this study refuted that business cycle could be a primary reason for banks’ npas. the study emphasised that increase in bad debts as a consequence of recession alone is not empirically demonstrated. however, according to bloem and gorter (2001) npas may be caused by wrong economic decision or by plain bad luck. there are many internal and external factors affecting npas in india. while the internal factors might be taking up new projects, promoting associate concerns, business failure, cost overruns during the project implementation stage, strained labour relations, inefficient management, technology related problems, product obsolescence etc., the external factors include gdp growth, default in other countries, high inflation, accidents and natural calamities. further, it is observed that while there is a positive correlation between the factors such as gdp growth induce the bank credit. procyclicality is observed in the case of comparison of gdp growth to npa levels (figure-3 and figure-4). das and ghosh (2003) established relationship between nonperforming loans of india’s public sector banks in terms of various indicators such as; asset size, credit growth and macroeconomic condition and operating efficiency indicators. 36 bercoff, giovanniz and grimardx (2002) in their study of argentinean banks tried to measure npas by using the various bank related parameters as well as macroeconomic parameters. bank specific parameters in their study were ratio of networth to net assets, bank’s exposure to peso loans, and type of banks such as foreign, private or public. macroeconomic factors in this study were credit growth, reserves adequacy, foreign interest rate and monetary expansion. they have established that variables such as operating cost, exposure to peso loans, credit growth, and foreign interest rate had a negative effect on npas. the macroeconomic variables such as money multiplier and reserve adequacy had a positive impact on npas. figure 4: pro-cyclicality of npas, comparison of gross domestic product to gross npa level – indian banking source: compiled by author based on data from reserve bank of india publications. chen et al. (1998) study the relationship between the risks and the ownership structure, and it appears that a negative correlation exists between the managers’ shareholdings and the risks faced by the financial institution. that means that if the managers’ shareholding percentage increases, the financial institution will reduce its own risk behaviour. while berger and de young (1995) mention that a management team with poor operating capability is unable to correctly appraise the value of collateral, which means that it is difficult for it to follow up on its supervision of the borrower, its poor credit-rating technology will result in management being unable to control and supervise the operating expenses efficiently, thus leading to a significant increase in npls. wahlen (1994) also points out that unexpected changes in the npl ratio may indicate that expected future loan losses are relatively nondiscretionary and negatively related to bank stock returns. therefore, we have considered the various bank groups in indian banking based on their ownership structures for the analysis. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 2 37 ownership pattern can also affect the bad loan levels significantly. in times of downturn the government would often turn to banks for financial resources through policy loans for the state owned enterprises (soes). projects financed by these policy loans gave rise to growing default rates (huang, 1999). the biased lending behaviour of the banks to soes is supported by other research findings as well (lu et al., 2001). in case of taiwanese banks the rate of non-performing loans decreases as the government shareholding in bank goes higher up to 63.51 per cent, while thereafter it increases (hu et al., 2002). few studies have also indicated a relationship between the size of the bank and the level of bad loans. bank’s sizes are often found negatively related to the rate of non-performing loan (hu et al., 2002). bodla and verma (2006) have emphasised that financial sector reforms have brought in greater competition among the banks and have brought their profitability under pressure. accordingly, banks face a number of challenges such as; worrying level of npas, stringent prudential norms, increasing competition, growing customer expectations, increasing pressure on profitability, liquidity and credit risk management, assets-liability management, shrinking size of spread, rising operating expenditure and so on. however, singh (2005) argues that globalization of operations and new technologies development are taking place swiftly and these has given rise to the increase in resource productivity, increasing level of profitability, deposits and credits and thereby decrease in npas. 3.2 endogenous factors the literature on these issues identifies determinants of banks risk taking that can be translated into a tractable empirical specification by measuring the effect of observable variables like; capital adequacy, credit growth, operational efficiency, branch spread and others. rajaraman, bhaumik and bhatia (1999) have explained the variations in npas across the indian banks through differences in operating efficiency, solvency and regional concentration. again rajaraman and vasishstha (2002) in their empirical study have proved that significant bivariate relationship exists between npas of the public sector banks and the inefficiency problems. das (1999) has contrasted the different efficiency measures of public sector banks by applying data envelopment analysis model and concluded that the level of npas has significant negative relationship with efficiency parameters. kwan and eisenbis (1997) have examined the relationship between problem loans and bank efficiency by employing granger-causality technique and found that high level of problem loans cause banks to increase spending on monitoring working out and / or selling 38 off these loans and possibly become more diligent in administering the portion of their existing loan portfolio that is currently performing. ranjan and dhal (2003) attempted an empirical analysis of the npas of public sector banks in india and probed the response of npas to terms of credit, bank size, and macroeconomic condition and found that terms of credit have significant effect on the banks’ non performing assets in the presence of bank size and macroeconomic shocks. they also found that alternative measures of bank size could give rise to differential impact on npas. 4. empirical specification and estimation in this section, we introduce the methodology adopted for the empirical analysis. the objective here is to identify the determinants of npas in indian banking. accordingly, there is a need to estimate a relationship of the following form using the panel data consisting of different groups of banks in indian banking industry (such as state bank group, nationalised banks, old private banks, new private banks and foreign banks) data across a period from 1997 to 2009. the equation can be; it + it ------------------> (1) where, α represents the intercept, β1,………βn represent the respective regression coefficients for explanatory variables x1 …… xn. for estimating the determinants of npa, the equation can be; npait = ƒ [met eit] -------------------- (2) where, met represent the macro-economic variables and eit represent the endogenous (industry specific) variables that are believed to determine the level of npas. then, the equation would be; gnpait = α + β1 gdpgrt + β2 pcnnpt + β3 mcapt + β4 lrt + + β5 iipgrt + + β6 inflat + β7 svgrt + β8 assetit + β9 carit + β10 cdrit + β11 cofit + + β12 roait + β13rusubrait + β14 credgrit + β15 pslit + β16 oerit + + it ----------------------------------------- (3) international journal of banking and finance, vol. 9, iss. 1 [2012], art. 2 39 here, the explanatory variables x1 …… xn are represented by the macro-economic variables such as gdp growth rate (gdpgr), inflation rate (infla), index of industrial production (iipgr), savings growth rate (svgr), growth rate in per capital income in nnp (pcnnp) and market capitalisation growth rate (mcap) and endogenous variables such as bank assets (natural log) (asset), capital adequacy ratio (car), credit to deposit ratio, bank lending rates (lr), operating expenses to total assets (oer), ratio of priority sector loans to total loans (psl), ratio of rural and semi urban branches to total bank branches and return on assets (roa) that are supposed to determine the npas in the indian context. while ‘i’ represents the category of bank group, ‘t’ represents the year and it represents the unexplained residual. this equation is estimated using the regression analysis considering gross npa (gnpa) which is calculated as the ratio of gross non-performing assets to total advances and as regressand. gdpgr is involved as a determinant in view of its all-pervading effect in the economy that may have a say in causing the npas. it controls for the macroeconomic conditions that owing to the business cycles in the economy have a significant role to play in causing defaults in loan repayments. it is also because of the reasoning that as the gdp increases the amount of npas decrease. infla is considered as a macroeconomic determinant as it is one of the aspects related to indian economy which affects the banks overall performance especially the level of npas in the banking system. this is because when reserve bank of india (rbi) takes some steps related to interest rates to control inflation, the defaulters list also grows for the banks with the rising interest rates. further, savings levels in an economy explain the savings propensity as well as the economic surpluses available which has a relation to the repayment capacity of the borrowers of the banking sector in general. in view of this, sgr is considered as a determinant. in addition, in a growing economy like india, capital markets attract a whole lot of investors as well stimulate the capital formation in the country which has a bearing on the performance of the organised industrial sector. in view of this logic, bombay stock exchange market capitalisation growth rate (mcap) is considered as a determinant. it is argued that non priority sector is the prime contributor to the npas. to include this viewpoint in the study, bank assets (asset) is taken as control for whether the bigger banks are more vulnerable to the problem of npas than their smaller counterparts are. car was also considered as a determinant in view of the logic that the higher the capital of the banks 40 the lower is the level of npas. it was also due to the fact that as capital base of the banks increases confidence of the bank also increases and gets reflected in their performance thus leading to effective recovery of bank loans and bringing down the level of npas. roa is considered as an endogenous determinant because of the fact that profitability of banks would have a close relation with its npas. it is obvious in general that the more profitable banks would have less npas. in order to capture the aggressiveness in lending activity of the banks which can lead to npas, credit to deposit ratio (cdr) is considered as an endogenous variable. cost of funds for the banks cause significant strategic decisions in the area of bank lending. in order to account this argument, cost of funds (cof) is also considered as a determinant. growth in bank credit is also one of the factors that can determine the emergence of npas. table 4: summary statistics of the variables variable empirical definition mean std. dev. gnpa gross npa to total advances 6.797 4.710 asset bank assets (natural log) 12.432 1.091 car capital adequacy ratio (capital to risk weighted assets) 12.487 1.275 cdr credit deposit ratio 65.284 12.998 cof cost of funds in percentage 6.224 1.760 credgr bank credit growth rate (growth in real advances) 20.780 10.962 gdpgr growth rate of real gross domestic product 6.977 1.902 iipgr index of industrial production 6.808 2.155 infla inflation levels 5.762 1.499 lr bank lending rates 14.225 1.813 oer operating expenses to total assets 0.024 0.008 pcnnp per capita income at nnp 5.323 2.303 psl priority sector loans to total loans 28.803 8.519 roa return on assets of banks 0.941 1.604 rusubra ratio of number of rural and semi-urban branches to total bank branches 0.395 0.267 svgr savings growth rate 16.094 7.317 source: compiled from the results of the analysis by author in view of this, credit growth rate (credgr) is considered as one of the determinants. in the area of bank lending the lending rates play a significant part. the cheaper the rates the much of the operating expenses in the bank are believed to be towards employing the work force and related resources for credit deployment and recovery. accordingly, operating expenses ratio is also considered as a variable. proportion of rural and semi-urban international journal of banking and finance, vol. 9, iss. 1 [2012], art. 2 41 table 5: determinants of npas variables model-1 model-2 model-3 constant 0.0000 (39.899)** 0.0005 (41.648)** 0.0000 (42.482)** asset 0.0001 (-2.214)** 0.0475 (-1.975)* 0.0004 (-2.127)** car 0.6811 (0.069) 0.7787 (-0.052) 0.5983 (0.074) cdr 0.0000 (-0.195)** 0.0000 (-0.154)** 0.0000 (-0.179)** cof 0.0022 (-0.833)** 0.0000 (-1.286)** 0.0004 (-1.279)** credgr 0.7566 (-0.007) 0.2547 (-0.018) 0.1481 (-0.036) gdpgr 0.4074 (0.550) 0.2547 (-0.018) 0.2936 (0.602) iipgr -na0.0486 (-0.721)* -na mcap -na-na0.0219 (3.710)* infla 0.8574 (-0.057) 0.1425 (0.870) 0.2620 (0.541) lr 0.1397 (0.654) 0.1368 (0.389) 0.0950 (0.621) oer 0.0000 (151.898)** 0.0000 (161.721)** 0.0001 (129.149)** pcnnp 0.1586 (-0.775) 0.0993 (-0.312) 0.0206 (-1.271)* psl 0.0264 (-0.094)* 0.0958 (-0.083) 0.0212 (-0.104)* roa 0.0000 (0.449)** 0.0033 (0.326)** 0.0013 (0.385)** rusubra 0.0914 (6.220) 0.1666 (4.353) 0.1039 (5.096) svgr 0.1389 (-0.053) 0.2248 (0.078) 0.0166 (-0.165)* sbdummy 0.7744 (0.263) nbdummy 0.3470 (1.103) olddummy 0.0341 (-2.248)* -na-na newdummy 0.0001 (-1.999)** -na-na fdummy 0.0000 (-3.374)** -na-na r square 0.869 0.886 0.879 adjusted r square 0.818 0.842 0.828 note: ** at 1% level of significance, * at less than 0.05 level of significance. source: compiled from the results of the analysis by author. 42 bank branches (rusubra) has been considered as a determinant to examine whether the location of banks i.e, rural and semi-urban areas matter in causing npas in banking. the more is the recovery rate, the higher the rates the higher the defaults. in order to account this argument, lending rates are considered. the more aggressive are the banks in their lending they may end up in pushing riskier loans and thereby end up in higher level of npas. however, there is a contention that as banks concentrates on credit management they may have developed expertise in managing the credit risk and hence may sometimes exhibit lower level of npas. therefore the role of lending aggressiveness in causing increase in npas is still hazy. ratio of priority sector lending to total bank lending (psl) was included as a determinant in order to account for the argument that the priority sector loans are responsible for the most number of defaults (refer figure-5). figure 5: priority sector loans to total bank credit in india source: compiled by author based on data from reserve bank of india publications. summary statistics of the explanatory variables is furnished in table 4 in the annexures. balanced panel data is used for estimation by employing the eviews tools for detailed analysis. annual data on state bank group, nationalised banks group, old private banks group, new private banks group and foreign banks group for the period from 1996-97 to 2008-09 is obtained from the robust data base of reserve bank of india (various issues of statistical tables relating to banks in india, the report of trend and progress of banking in india and the published annual audited accounts of individual banks). the choice of the period is dictated by several considerations. the first is the availability of published data on the variables considered in the study. secondly, the year 1996-97 marks the rigorous regime of the prudential norms as a result of the ‘first generation’ reforms programme initiated in 1991, so that it would be useful to examine the impact of various international journal of banking and finance, vol. 9, iss. 1 [2012], art. 2 43 determinants and the behaviour of different banking sector in terms of npas through the initiation of the reform process. further, the said period covers the significant period of postliberalisation in indian banking. 5. results and discussion table 6: nature and strength of the impact of macroeconomic and endogenous determinants on npas explanatory variable model-1 model-2 model-3 asset negative and significant negative and significant negative and significant car insignificant insignificant insignificant cdr negative and significant negative and significant negative and significant cof negative and significant negative and significant negative and significant credgr negative and significant negative and significant negative and significant gdpgr not significant insignificant insignificant iipgr -----------negative and significant ------------ mcap ----------------------positive and significant infla insignificant insignificant insignificant lr insignificant insignificant insignificant oer positive and significant positive and significant positive and significant pcnnp insignificant negative and significant negative and significant psl negative and significant negative and significant negative and significant roa positive and significant positive and significant positive and significant rusubra insignificant insignificant insignificant svgr insignificant insignificant negative and significant olddummy negative and significant -na-na newdummy negative and significant -na-na fdummy negative and significant -na-na source: compiled from the results of the analysis by author. the result of the analysis is presented in table 6 and the nature and strength of the impact of macroeconomic and endogenous determinants on npas are furnished in table 6. gdp growth rate has no significance on the npas whereas as expected per capita income has 44 negative significance on npas. the coefficient of asset has turned out to be negative and very significant indicating that large banks may have better risk management procedures and technology which definitely allows them to finish up with lower levels of npas. it was also observed that cdr is negatively associated with bad loans signifying that higher the cdr the lower tends to be the level of npas. as an alternative macro-economic variable, we employed the index of industrial production (iipgr) instead of gdpgr in model-2. the results indicate that the coefficient on this variable is negatively significant as conjectured. further, as another variant of the aforesaid specification, we introduce the market capitalisation ratio (mcap) in model-3 with a view to capture the transition from a bank based to market based financial system. the result shows that the coefficient is positively significant at 2 per cent level implying that transition to market orientation has impinged on the problem loans as the surpluses tend to move into the booming markets as investments and thereby affecting the repayments of bank loans. as is theoretically well established when the roa has increased, it is resulting in lesser amount of problem loans. accordingly, the analysis has found that roa is strongly associated with the npas negatively. cost of funds (cof) was found to be significantly associated with the npas negatively to evidence our viewpoint that as the cost of funds increase the banks tend to be very cautious and choosy in their lending thus leading to decrease in npas. lending rates have been found to be not so significant in affecting the npas contrary to the general perception. the rest of the explanatory variables exhibit theoretically expected relationships with npas and are self-explanatory as detailed in the columns 1, 2 and 3 of table-6 which explains the nature and strength of the impact of endogenous determinants on npas. the final subject of interest in this study was whether the npas are in any way affected by the ownership styles of the banks. this issue was investigated by introducing the ownership dummies (sbdummy for state bank group of banks, nbdummy for nationalised banks, olddummy for old private banks, newdummy for new private banks and fdummy for foreign banks. the results summarized in table-5 indicate that private banks (both old and new) and foreign banks appear to manage their npas efficiently. state bank group and nationalised banks appear to lag behind their private counter parts in npa management. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 2 45 6. policy implications and conclusion several policy implications can be gleaned from this analysis. favourable macroeconomic conditions facilitate in npa management. first, as the banks grow in size, they tend to control the npa owing to efficiency in their management. in this background there is a case for consolidation of banks in the public sector to reap this potential of efficiency in scale of operations. larger banks have exhibited better credit risk management with lower npa levels. secondly, priority sector lending by banks is found to be not much significant in contributing for npas in contrast to the perception of some urban bankers that psl cause npas. this supports the contention that branch expansion in rural and semi urban areas for extending priority sector credit is a viable proposition and there need not be aversion on this by the policy makers as well as the industry heads. thirdly, ownership of banks is an interesting issue that has been quite often debated. this study has established that private banks and foreign banks have advantages in terms of their efficiencies in better credit management in containing the npas which indicates that bank privatization can lead to better management of default risk. the above findings infer that better credit risk management practices need to be taken up for bank lending. adequate attention should be paid to those banks with low operating efficiency and low capitalisation as also to macroeconomic cycles that appear to be playing some role in npa management. the state owned banks need to be toned up with adequate measures to sharpen their npa management practices. these findings assume crucial importance in view of the significance. author information: vighneswara swamy is a professor at the ibs-hyderabad, india. he may be contacted via e-mail: vighneswar@ibsindia.org mailto:vighneswar@ibsindia.org 46 references allan, f., and gale, r., (2007). understanding financial crisis, oxford university press, oxford, united kingdom. altman, e.l., andrea resti and andrea sironi, (2001). analysing and explaining default recovery rates, a report submitted to the international swaps & derivatives association. ariff, m., (2012). the origin of the global financial crisis: an alternative view, in m. ariff, j. farrar and a. khalid (eds) regulatory failure and the global financial crisis, edward elgar publishing uk-usa. bama b., (2002). npl securitisation in china – anatomy of the deal, structured product group, icici research bangalore, 2002. barseghyan, l., (2004). non-performing loans, prospective bailouts, and japan’s slowdown, cornell university, new york, ny, usa. beck t.a, demirguc-kunt and r. levine (2005). bank concentration and fragility: impact and mechanics, nber working papers 11500, national bureau of economic research inc. berger, a.n., and r de young (1995). problem loans and cost efficiency in commercial banks, journal of banking & finance 21: 849–870. bhattacharya, h., (2001). banking strategy, credit appraisal & lending decisions, oxford university press, new delhi. bodla b.s., and richa verma (2006). determinants of profitability of banks in india: a multivariate analysis, journal of services research, vol. 6 (2: (october 2006 – march 2007). chen, j., (2004). non-performing loan securitization in the people’s republic of china, stanford university, palo alto, ca, usa. chen, c.r., t.l. steiner and a.m. whyte (1998). risk-taking behavior and management ownership in depository institutions, the journal of financial research 21: 1–16. das, abhiman, (1999). efficiency of public sector banks: an application of data envelopment model, prajnan vol. 28 (2, september 1999). das a., and ghosh s., (2003). determinants of credit risk, paper presented at the conference on money, role and investment held at nottingham treat university in nov 2003. dermiguc-kunt, a., and e. detragiache (2000). monitoring banking sector fragility: a multivariate logit approach, world bank economic review, vol. 14 (2): 287 307. devakula, pridiyathorn (2001). monetary policy in thailand: current challenges and prospects, apfa 2001 conference, bangkok, 24 july 2001. ernst & young (2001). non-performing loan report: asia 2002, ernst & young, new york. fernandez de lis s, j martinez-pages and surina, (2000). credit growth, problem loans, and credit risk provisioning in spain, working paper no. 0018, banco de espana goldsmith, r. w (1969), financial structure and development, new haven, ct: yale university press. hu, jin-li, yang li and chiu, yung-ho (2002). ownership and non-performing loans: evidence from taiwanese banks, proceedings of international conference, national taiwan university. huang, yiping, (1999). dealing with bad loans of the chinese banks, china update 1999 conference papers, november, national centre for development studies. imf (2009). global financial stability reportresponding to the financial crisis and international journal of banking and finance, vol. 9, iss. 1 [2012], art. 2 47 measuring systemic risks’, world economic and financial surveys, international monetary fund, washington dc, april 2009. kang, moon-soo (2001). financial restructuring in korea, korea development institute october 2001. kwan, s., and r. eisenbis (1997). bank risk, capitalization and operating efficiency, journal of financial services research 12: 117-131. levine, r., (2004). finance and growth: theory, evidence & mechanism’, edited by aghion, p and durlauf, s. (eds), handbook of economic growth, amsterdam: northholland, pp.81, in reforming corporate governance in southeast asia, by khai leong ho (2005), published by institute of southeast asian studies. lis, s.f. de, j.m. pages, and j. saurina (2000). credit growth, problem loans and credit risk provisioning in spain banco de españa — servicio de estudios, documento de trabajo no. 0018. lu, ding, thangavelu, shandre, m., and qing, hu, (2001). the link between bank behavior and non-performing loans in china, working paper no. 0108, national university of singapore. rajaraman, bhaumik and bhatia (1999). npa variations across indian commercial banks, some findings, economic and political weekly vol. 37 (3-4): 16-23. rajaraman indira and vasishstha (2002). non performing loans of public sector banks – some panel results, economic and political weekly (february 2002). ranjan, rajiv, and sarat chandra dhal (2003). non-performing loans and terms of credit of public sector banks in india: an empirical assessment, reserve bank of india occasional papers, vol. 24 (3): winter 2003. sergio, m., (1996). non-performing bank loans: cyclical patterns and sectoral risk, review of economic conditions in italy, rome: jan-jun 1996, issue 1. singh, c., (2005). financial sector reforms and state of indian economy, indian journal of economics & business vol. 4 (1): 88-133. sprayregen, j., friedland, j., miller, n., and li, c., (2004). non-performing loans in china: a potential win-win opportunity for foreign investors and china’s economy, global restructuring & insolvency review, pp. 38-9. wahlen, j., (1994). the nature of information in commercial bank loan loss disclosures, the accounting review 69: 455–478. interest rate convergence in bangladesh 81 the international journal of banking and finance, volume 9 (number 1) 2012: pages 81-97 interest rate convergence in bangladesh dewan m. rahman and kohinur akter university of dhaka and brac university, bangladesh ____________________________________________________________ abstract financial liberalization, a widely-accepted policy paradigm since the 1980s, aims to remove financial repression and thus establish an efficient financial sector as a prerequisite for financial development. interest rate convergence is one of the obvious outcomes of the interest rate liberalization, a crucial part of financial liberalization. uniformity of interest rate should be there in a free and competitive financial market. this paper examines the state of interest rate convergence by measuring the degree of convergence in the financial market of bangladesh, a success case of financial liberalization initiated in the 1990s. key words: financial sector reform programs, private commercial banks, nationalized commercial banks, foreign commercial banks, specialized banks, interest rate convergence jel classification: o16, p22 _____________________________________________ 1. introduction the financial liberalization programs appeared in the developing countries in the 1980s and 1990s as part of the worldwide approach toward giving markets a greater role in development through strengthening the financial sector: see ariff and khalid (2005). governments used financial system as an instrument of allocating financial resources through directed credit at below market interest rate. interest rate on deposits was kept very low to maintain costs of loans low. these together limited the mobilization and efficient allocation of financial resources, which in turn slowed economic growth according to the fathers of financial liberalization (mckinnon 1973; shaw 1973). low interest rates discouraged the mobilization of finance, and so bank deposit growth 82 slowed in in the major countries. financial intermediation was discouraged while the inappropriate legal frameworks and information structures supported the existence of inefficient private bonds and equity markets: see gupta and karapatakis (2006). the earliest policy changes as financial liberalization focused on interest rates. by freeing both deposit and loan interest rate ceilings, the advocates of liberalization suggested that credit market can operate to maximize both the quality and efficiency of investment (mckinnon 1973; shaw 1973). hypothetically, if market is perfectly competitive, the market based interest rate structure should allow the interest rate in the financial sector to converge. therefore, convergence of interest rate serves as a criterion for perfectly competitive market where the financial intermediaries would maintain the same level of interest rates, which would be market determined under competition. convergence refers to the idea where financial intermediaries move in the same direction together to a certain level of interest rate which is determined by the market. as a part of the financial liberalization, bangladesh initiated the financial sector reform programs (fsrps) at the beginning of the 1990s under guidance from the world bank. one of the objectives of this was to provide better return on deposits and thereby allocate credits efficiently in the financial market by moving towards a market based interest rate regime from the then administered interest rate regime, hoping thereby to promote economic growth through increased investment spending. therefore, this implementation of fsrp should finally lead to a more competitive credit market and thus, convergence of interest rate will occur with the passage of time. from a general point of view, lack of an efficient intermediation and interest rate setting was believed to be a major hurdle for development. in this connection, this study investigates whether interest rate liberalization has been able to create a competitive environment in the financial market through the convergence of interest rate, which can be used as an indicator of a more efficient financial market leading to more efficient allocation of financial resources. we examine convergence in different ways on the formal credit market comprised with four players: the nationalized commercial banks (ncbs), the private commercial banks (pcbs), the specialized banks (sbs) and the foreign commercial banks (fcbs). international journal of banking and finance, vol. 9, iss. 1 [2012], art. 4 83 this study presents the pre and post financial liberalization interest rate structure and also its impact on creating an efficient financial market. our study also reveals the degree of competition and the structure of the credit market currently available in bangladesh. the policy implications of interest rate convergence as an instrument of financial liberalization are discussed from its positive and negative contributions. the rest of this paper is divided as follows. the next section is a brief statement of the literature, although it is not exhaustive. section 3 explains the methodology employed to conduct statistical significance of the parameters estimated. the results and interpretations are to be found in sections 4 and 5 while the paper ends with concluding remarks in section 6. 2. literature review since the mid-1980s several developing countries liberalized their financial systems with the common aim as in this case. this liberalization has been to provide greater scope to market forces in the determination of interest rates and in the allocation of credits. one crucial question that needs to be addressed is whether the financial reforms that have been implemented have led to an improvement in the allocation of resources through the set up of convergence in the interest rate. it is curious that financial liberalization in general involves replacing one deeply flawed system, characterized by heavy government intervention, with another system with different flaws. financial system in the developing countries consists primarily of commercial banks and specialized banks, and in some cases, cooperative societies, savings banks and loan associations. mckinnon and shaw (1973) argue that, the system of financial repression allows the financial markets to work under the administration of the government. that leads to the existence of interest rate ceilings, directed credits and subsidized credits to priority sectors, thus the overall result is not to encourage competition in the setting of interest rates in the process of providing credits to the most productive entrepreneurs. under this situation, they explain that financial repression in the form of low or negative real interest rates has severe consequences on the quantity and quality of investment. as in the financial sector, especially for the banks, deposits are the main 84 source of funds, at low interest rate the supply of fund is reduced for the financial institutions. gupta and karapatakis (2006) argue that this reduces the quantity of credits in the market. they state that the low real lending interest rate under financial repression also creates excess demand for credits because virtually all investment opportunities seem profitable at this low interest rates in an administered regime. regulations of interest rates leads to financial repression, which leads to overall market inefficiency, severe imperfections, information asymmetry, undesired government interventions and distortion of the financial market equilibrium. eicher and hull (2004) conclude through a theoretical model that for countries which had severe market imperfections, liberalization should lead to higher growth. akpan (2004) in his study of financial liberalization and endogenous growth in the case of nigeria has shown that liberalization results in a positive impact between broad money (m2), investment and the real deposit rate on economic growth. this is because more mobilization of funds, innovation of new instruments and development of legal framework for the financial sector occurred with the liberalization policy. alike, in many countries liberalization policies proved to be successful. but some problems also have been identified in the previous studies on the impact of financial liberalization. ghosh (2005) argues that in many cases, the social and economic effects of financial liberalization have been especially adverse for the poor and for farmers and workers, who have not only suffered more precarious conditions even during a so-called “financial boom”, but have typically also been the worst affected during a financial crisis or the subsequent adjustments. she also concludes that, the extreme forms of liberalization are neither effective nor necessary, and that a large variety of alternative measures, as well as varying degrees of liberalization, is not only possible but can also be observed in several more successful developing countries. galindo, schiantarelli and weiss (2002) through the informal ocular econometric exercise and the comparison of mean values of efficiency index in the preand post liberalization regimes, suggest that financial liberalization has led to an improvement in the efficiency with which investment funds have been allocated. both the indices have improved for many (although not for all) countries, following the introduction of financial reform. but as a whole, financial liberalization in the 1990s in developing international journal of banking and finance, vol. 9, iss. 1 [2012], art. 4 85 countries has contributed positively to the growth of those countries. sarr (2000), luc (2003), ghosh (2005), akpan (2004) and karapakatis (2006) separately in their empirical studies have concluded with the same results. as part of the financial liberalization, interest rate liberalization has been the basic toolto promote financial development. mckinnon and shaw (1973) argue that output and economic growth can be best facilitated by freeing both deposit and loan rate ceilings so that credit markets can operate to maximize both quantity and efficiency of investment. thus, if this takes place, the banking sector would face a free and competitive environment where the interest would be charged by the market mechanism. they showed that liberalization would remove repression with low interest rate on loans and high deposit interest rates. sarr (2000) also indicated that this would encourage competition among the financial institutions with benefits to consumers in terms of reduced net margin. in this view, suzuki (2003) argues, in his study on competition in banking sector of japan, that many industrialized countries have now reduced or even removed governmental controls over interest rates, fees, commissions, ownership, portfolios and the boundaries between different financial businesses, on the ground that they are seen as less important from a prudential perspective but rather harmful from a competition perspective. bikker (2003) also suggest market imperfections by different ways would cause allocational inefficiency, and so detract from the prosperity which society derives from banking services. as interest rate liberalization along with financial liberalization takes place, free competition results where consumers and the financial institutions would match their needs for finance. mehran and laurens (1995) explain that in recent years, many developing and transition countries have allowed market forces to play a greater role in their economies and in the financial sector. this means liberalizing interest rates so that they are allowed to be set by the market, and developing financial markets so that credits can be allocated more efficiently. satterthwaite and shneyerov (2005) show in their analysis on convergence to perfect competition in an bargaining market that the presence of many players removes imperfections and results in convergence to perfect competition with small frictions that appear to be robustly almost efficient. interest rate liberalization first increases the interest rates in the short term but in the long term it moves toward convergence. 86 therefore when the market is free to act, from the inception of financial linearization, interest rates should converge. this has been seen from different analysis on the convergence of interest rate between central and eastern european countries (cee): holz (2004) concludes that, since the end of the 1990s, in some of the cee countries, namely the czech republic, hungary, slovenia and the baltic states, there has been a remarkable interest rate (long-term bond yield) convergence from high doubledigit levels now close to those of emu member countries. though this was a crosscountry investigation in the european union, the conclusion is that financial market free from controls should promote convergence of interest rates. the speed of convergence may vary depending on the availability of an environment of competition in the financial market. honohan (2003) in his study on interest rate changes under liberalization in developing countries also showed evidence in support that the speed of convergence in interest rate is positively related with liberalization. he also shows that, liberalization means an increased short-term volatility in both real and nominal money market interest rates. treasury bill rates and bank spreads were evidently the most repressed, and they showed the greatest increase as liberalization progressed. but this convergence of interest rate would take place only if market is efficient. mehran and laurens (2005) stated that uncompetitive banking systems, inadequate regulatory frameworks, and borrowers that are insensitive to interest rates undermine the efficiency of market-based credit allocation and disrupt the transmission of monetary policy signals, with adverse consequences for macroeconomic policy. when these conditions prevail, interest rates are not likely to move to their market-clearing levels. the financial liberalization leads also to another outcome by squeezing the interest spread of financial intermediaries. ghosh (2005) explains price competition squeezes spreads and forces financial firms (including banks) to depend on volumes to ensure returns. but honohan showed that, while quoted bank spreads in industrial countries contracted again somewhat during the late 1990s, spreads in developing countries remained much higher, presumably reflecting both market power and the higher risk of lending in the developing world with less growth opportunities. ahmed and islam (2006) state that high spread to some degree leads to institutional inefficiencies. they also argue that this is the result of the government’s interventionist policies of the past. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 4 87 therefore, interest rate liberalization should bring low spread for the banks, which will make the liberalization policy a success. 3. methodology 3.1 data selection our analysis is based on monthly interest rates over the reform period, up to 31 june 2006 from january 1990. we collect data from 4 ncbs, 4 sbs dedicated to agricultural and industrial lending, 26 domestic private commercial banks (pcbs) and 16 foreign commercial banks (fcbs). the data source is economic trends, a monthly publication of the bangladesh bank (bb). we focus on interest rate data for all scheduled banks during 1990 to 2006 and also those which withdrew from bangladesh during the course of this period. the records interest rate on deposits relate to seven categories: urban saving deposits (savings-ur); rural savings deposits (savings-ru); and fixed deposits with 3 to 6 months term (fd3-6m); 6 to 12 months term (fd6-2m); 12 to 24 months term (fd12-24m) and 24 to 36 months term (fd24-36m). interest rates on lending are recorded in fourteen categories – agricultural loan (agri), term loan, working capital loan for jute sector (wcap_jute), working capital for other sector (wcap-oth), loan for jute trade (jute-trade), loan for jute export (jute-export), loan for other export (other-exp), other commercial loans (other-comm), housing loan (housing), loan for residential housing in urban area (ur-house-res), loan for commercial housing in urban area (ur-com-res), special term loan (spec-term), other special loan (spec-other) and loan for other activities (other). data set from january 1990 to june 2006 is set up because the interest band was withdrawn by the central bank from 1990 to initiate the liberalization of the financial sector. 3.2 analysis approach to test the convergence of interest rate we follow the following approach – mean difference analysis: we utilize the mean difference analysis technique for every analysis to test the convergence. mean difference analysis of interest rate shows 88 whether any significant difference exists between mean interest rates. the analysis is completed in three stages: the first horizon (january 1990 – march 1992): in january 1990, efforts were taken to liberalize the financial market. the market oriented interested policy was taken but still up to march 1992, government continued to determine the interest rates for 11 sectors. the second horizon (april 1992 – july 1999): in april 1992, partial liberalization was initiated by withdrawing interest rate ceilings from all categories except 3 namely agriculture, export and small and medium term loan. this partial liberalization continued till july 1999. the third horizon (august 1999 – june 2006): in june 1999, interest rates in the 3 categories also were free and the ceiling was removed ushering a full interest rate liberalization thereby. from that moment to until june 2006 and forward, the financial market is enjoying full liberalization. to test the convergence, we test the difference between two means with respect to ncbs vs pcbs, ncbs vs sbs, ncbs vs fcbs, sbs vs pcbs, sbs vs fcbs, fcbs vs pcbs by testing the direction towards a certain level of interest rate. as our sample size is large, we incorporated a two-tailed hypothesis test about the difference between two means with the assumption of normal distribution. we assume in our null hypothesis (h0) that there exists no significant difference between two means. to test this hypothesis we calculate the standard error of the difference between two means, standard error = 21 ˆ xx −σ = 2 2 2 1 2 1 nn σσ + (1) we use the standard error to find out the normalized z value 21 ˆ 0 )11()21( xx hxx z − −−− = σ µµ (2) where, international journal of banking and finance, vol. 9, iss. 1 [2012], art. 4 89 x1 = mean interest rate of range 1, x2 = mean interest rate of range 2, n1 = no. of observations of range 1, and n1 = no. of observations of range 1. the confidence level is set at 95% for which z value calculated is compared with the observed values to accept or reject the null hypothesis. while we accept the hypothesis, it means that there is no significant difference and is we reject, it indicates a significant difference between means, suggesting movement away from convergence. 4. result and interpretation 4.1 convergence between ncb and scb the test results are summarized for different institutions in table 1. table 1: convergence in rates across different institutions convergence in savings savings rate ncb vs scb periods 1990-1992 1990-1992 1992-1999 1992-1999 1999-2006 1999-2006 institutions ncb sb ncb sb ncb sb mean 12.031 12.291 7.362 7.873 7.019 7.220 variance 2.600 1.732 2.822 2.988 3.101 2.643 z -3.228 -9.755 -3.898 convergence in lending rate: ncb vs scb 1990-1992 1990-1992 1992-1999 19921999 19992006 1999-2006 ncb sb ncb sb ncb sb mean 12.541 13.433 12.712 12.820 11.867 12.218 variance 9.287 8.846 5.390 4.637 5.385 5.828 z -7.845 -0.179 -5.930 90 ii) convergence between ncb and pcb: convergence in savings rate: ncb vs pcb time period 1990-1992 1990-1992 1992-1999 1992-1999 1999-2006 1999-2006 ncb pcb ncb pcb ncb pcb mean 12.031 12.227 7.361 8.351 7.019 8.456 variance 2.600 1.965 2.822 2.446 3.101 2.470 z -2.875 -25.466 -36.222 convergence in lending rate: ncb vs pcb 1990-1992 1990-1992 1992-1999 1992-1999 1999-2006 1999-2006 ncb pcb ncb pcb ncb pcb mean 12.541 13.839 12.712 13.558 11.867 13.116 variance 9.287 7.324 5.390 4.394 5.385 6.137 z -17.1577892 -17.481 -25.433 iii) convergence between ncb and fcb: convergence in savings rate: 1990-1992 1990-1992 1992-1999 1992-1999 19992006 1999-2006 ncb fcb ncb fcb ncb fcb mean 12.03 10.68 7.36 7.46 7.02 7.36 variance 2.60 3.45 2.82 2.94 3.10 2.68 z 16.59 -2.31 -7.62 convergence in lending rate: 1990-1992 1990-1992 1992-1999 1992-1999 1999-2006 1999-2006 ncb fcb ncb fcb ncb fcb mean 12.541 13.733 12.712 12.676 11.867 11.899 variance 9.287 7.506 5.390 4.754 5.385 4.881 z -13.686 0.693 -0.582 international journal of banking and finance, vol. 9, iss. 1 [2012], art. 4 91 iv) convergence between scb and pcb: convergence in savings rate: 1990-1992 1990-1992 1992-1999 1992-1999 1999-2006 1999-2006 sb pcb sb sb sb pcb mean 12.291 12.227 7.873 8.351 7.220 8.456 variance 1.732 1.965 2.988 2.446 2.643 2.470 z 0.979 -10.721 -33.212 convergence in lending rate: 1990-1992 1990-1992 1992-1999 1992-1999 1999-2006 1999-2006 scb pcb scb pcb scb pcb mean 13.433 13.839 12.820 13.558 12.218 13.116 variance 8.846 7.324 4.637 4.394 5.828 6.137 z -3.537 -1.225 -22.013 v) convergence between scb and fcb: convergence in savings rate: 1990-1992 1990-1992 1992-1999 1992-1999 1999-2006 1999-2006 sb fcb sb fcb sb fcb mean 12.291 10.681 7.873 7.459 7.220 7.362 variance 1.732 3.453 2.988 2.940 2.643 2.681 z 20.451 8.669 -3.303 convergence in lending rate: scb vs fcb 1990-1992 1990-1992 1992-1999 1992-1999 1999-2006 1999-2006 sb fcb sb fcb sb fcb mean 13.433 13.733 12.820 12.676 12.218 11.899 variance 8.846 7.506 4.637 4.754 5.828 4.881 z -2.443 0.239 6.765 vi) convergence between pcb and fcb: convergence in savings rate: 1990-1992 1990-1992 1992-1999 1992-1999 1999-2006 1999-2006 pcb fcb pcb fcb pcb fcb mean 12.227 10.681 8.351 7.459 8.456 7.362 variance 1.965 3.453 2.447 2.940 2.470 2.681 z 23.419 27.720 40.333 92 convergence in lending rate: 1990-1992 1990-1992 1992-1999 1992-1999 1999-2006 1999-2006 pcb fcb pcb fcb pcb fcb mean 13.839 13.733 13.558 12.676 13.116 11.899 variance 7.324 7.506 4.394 4.754 6.137 4.881 z-value 1.204 23.994 36.145 the tables provide a summary – a lengthy one – of the convergence in the market parameters as a result of the reforms undertaken. there are six panels of results with test statistics comparing the results before and after the reforms undertaken. for example, the first panel shows the convergence in savings and in lending between the ncb and scb. as is evident from the two parts of the panel (i) in table 1, there is significant convergence in the comparisons between the national and state commercial banks. similarly, as is evident from panel (ii), convergence is illustrated for ncb and pcb (private banks). the results in the rest of the panels are about convergence in the rates in different parts of the sector. for example, in panel (vi) the reader finds the results on lending rates and savings rates. in the next section, the reader will note a discussion of the significance of these results. 4.2 summary of the results table 2 is a summary of the on liberalization effects in the country. these results show across the board significant changes from the liberalization measures put in place for the economy. except in four cases, there is overall significance statistics across the financial sector. the four are shown in italics for lending, savings rates in the earlier years when the reforms had not taken roots. during the latest period after a period over which the financial institutions learned to cope with the reforms, there is strong effect showing significant impact of the reforms in the latest period over 199-2006. 5. policy implications having analyzed all the results above it can be concluded that interest rate convergence does not yet take place in the financial market of bangladesh. though one of the very usual outcomes of financial liberalization is interest rate convergence, we see on bangladesh, international journal of banking and finance, vol. 9, iss. 1 [2012], art. 4 93 financial liberalization has failed to produce this outcome during the last fifteen years. this is because what we find could not take place due to asymmetric information, several market imperfections, existence of govt. control on the ncbs, pcbs, fcbs and sbs. as a whole, interest rate convergence, proceeding to a uniform table 2: summary of the results for different degrees of liberalization, bangladesh 1990-1992 1992-1999 1999-2006 ncb and sb : saving significant difference between means significant difference between means significant difference between means ncb and sb : lending significant difference between means no significant difference between means no significant difference between means ncb and pcb: saving significant difference between means significant difference between means significant difference between means ncb and pcb: lending significant difference between means significant difference between means significant difference between means ncb and fcb: saving significant difference between means significant difference between means significant difference between means ncb and fcb: lending significant difference between means no significant difference between means significant difference between means sb and pcb: savings no significant difference between means significant difference between means significant difference between means sb and pcb: lending significant difference between means no significant difference between means significant difference between means sb and fcb: saving no significant difference between means significant difference between means significant difference between means pcb and fcb: saving significant difference between means significant difference between means significant difference between means pcb and fcb: lending no significant difference between means significant difference between means significant difference between means 94 interest rate, which is the result of free and competitive financial market still not in bangladesh. to have the full benefit of financial liberalization in terms of efficient financial market the much known problems should be solved. this study reveals that the bangladesh financial market is still has high degree of market imperfection by the way of information asymmetry. the banking sector yet does not communicate all information that should be made available and government control also discourages much information not to be published in the market. this is because government is having good degree of control over the banking sector and using the banks for its many politically biased and unproductive purpose in bangladesh. information asymmetry also is available regarding the performances of the banks in the financial market which leaves the market out of equilibrium. therefore the banking sector and the government both should allow more information to be in the market so that market becomes more competitive and consumers are benefited. this social desirable behavior would finally contribute to further development. government directed credits are common phenomena in bangladesh which have been reduced after liberalization in 1990 but in many cases government directly and indirectly facilitating the white elephants and thus distorting the efficiency of the market. subsidies are available granted to different sectors (e.g., agriculture, education) which also contributing in credit allocation inefficiencies. therefore directed credits should be removed to maximum level and subsidy should be withdrawn and this policy would bring more competition which is desired by the liberalization procedure. in bangladesh financial market, liberalization has been failed to produce desired result due to the very poor legal framework. of course the bangladesh bank has formulated new policies and laws regarding the banking sector, money laundering etc. which have positive impact in the market. though initiatives are being taken but not enough, we see the rate of default and non-performing advances are alarmingly high. legal framework and the administration over the banking sector should be more prudent to have an efficient and competitive market. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 4 95 6. conclusion the study of interest rate convergence is a desired outcome of interest rate liberalization as a part of financial liberalization. interest rate convergence is a process where market tends to be more competitive and a market set interest rate prevails at least in the long run. this paper finds that in bangladesh interest convergence is not taking place after even though financial reforms measures have been in 1990s. during last 15 years after initiating the financial liberalization measures have not produced the desired result of a competitive financial market, especially in the banking sector, due to existence of severe market imperfections. therefore, convergence of interest rate along with the competitive market can result successfully beneficial for the financial development if these market imperfections are wiped out. after 1990, during these 15 years, how much financial liberalization have contributed to the economic development of bangladesh, might become the another important focus of study. author information: dewan mostafizur rahman is staff member in the department of finance, dhaka university, bangladesh. he may be contacted via e-mail: sajibdewan@gmail.com while kohinur akter is a staff member at the brac university, dhaka, bangladesh. references abdourahmane sarr, march 2000, financial liberalization, bank market structure and financial deepening: an interest margin analysis, imf working paper, monetary and exchange affairs department, wp/00/38. ahmed, shamim and islam, ezazul, july 2006, interest rate spread in bangladesh: an analytical review, policy note series pn 0701, policy analysis unit (pau), research department, bangladesh bank. akpan david b., 2004, financial liberalization and endogenous growth: the case of nigeria, african institute for economic development and planning (idep), 94 pgs. ariff, m., and khalid, a., (2005). liberalization and economic growth in asia: 21st century challenges. edward elgar publishing, united kingdom and usa, may, 2005; hard & soft cover issues. bhattacharya, debapriya and chowdhury, toufic a, april, 2003, financial sector reforms in bangladesh: the next round, occasional paper series, paper 22, center for policy dialog publication. bikker, j.a., february 2003, testing for imperfect competition on eu deposit and loan markets with bresnahan’s market power model, research series supervision no. 52, section banking and supervisory strategies, directorate supervision, de nederlandsche bank, amsterdam, the netherlands. mailto:sajibdewan@gmail.com 96 black, b., r. kraakmen and a. tarassova, 2000, russian privatization and corporate governance: what went wrong? stanford law review 52: 1781. bokros, lajos; fleming, alexander; votava, cari , spetember 2001, financial transition in europe and central asia: challenges of the new decade, world bank publications, 296 pgs. caprio, g. and l. summers, 1996, finance and its reform: beyond laissez-faire in stability of the financial system, eds. d. papadimitriou, macmillan press, new york. caprio, gerard, 1997, financial reform in developing countries: a case of the missing model? in bruce smith and alison harwood, eds. sequencing financial reform, the brookings institution and policy research discussion paper 1574, january 1996. den valle, clemente, and piero ugolini, 2003, the development of domestic markets for government bonds, in litan et al., the future of capital markets in developing countries. dooley m. 1986, an analysis of external debt positions of eight developing countries through 1990, journal of development economics, vol. 21, no. 2. dooley, m., 1996, a survey of literature on controls over international capital transaction, international monetary fund staff reports, 43(4). dornbusch, rudiger, and sebastian edwards, 1991, the macroeconomics pf populism in latin america, chicago, university of chicago press. edward j. kane & asli demirguc-kunt, 2001. deposit insurance around the globe: where does it work?, nber working papers 8493, national bureau of economic research, inc. eicher, theo and hull, leslie, 2004, financial liberalization, openness and convergence, j. int. trade & economic development 13:4 443–459, issn 09638199 print/issn 1469-9559 online, taylor & francis ltd. financial sector overview, may 1 2006, volume 1, publication of bangladesh bank. financial sector reforms in bangladesh: the next round, august, 2002, report no. 49, centre for policy dialogue. ghosh, jayati, october 2005, the economic and social effects of financial liberalization: a primer for developing countries, department of economic and social affairs, desa working paper no. 4, st/esa/2005/dwp/4. gupta, rangan and karapatalis, andreas g, 2006, financial liberalization: a myth or a miracle cure? dissertation at the university of connecticut, by karapatakis (1992) extended by rangan gupta with independent study at university of pretoria, department of economics. holz, michael, january 2006, interest rate convergence in cee countries towards emu levels, bond market performance and the perspective of emu membership, universität trier, fachbereich iv – volkswirtschaftslehre, insbesondere geld, kredit, währung. honohan, p., and a. shi, 2003, deposit dollarization and the financial sector, in j hanson, p. hansonb, and g. majnoni, eds. (2003), globalization and national financial systems, new york and oxford: oxford university press for the world bank. honohan, patrick, how interest rates changed under financial liberalization (a cross-country review), development research group, the world bank. islam, e. md. and n. begum, 2004, high lending rates in bangladesh: an analytical review, bank parikrama, bibm, vol. 29-29, 100-119. international journal of banking and finance, vol. 9, iss. 1 [2012], art. 4 http://ideas.repec.org/s/nbr/nberwo.html 97 jahan, s., 2005, interest rate policy in bangladesh, policy note: world bank institute (wbi), washington, d.c. la porta, rafael/ lopez de silames, florencio/ schleifer, andrei, 2002 government ownership of banks, journal of finance, 57 (1), pp. 265-301. laeven luc, 22-mar-03, does financial liberalization reduce financing constraints? financial management association, article – spring 2003. mahmud, w., 2004, macroeconomic management from stabilization to growth?, economic and political weekly, vol. 35, no. 36, pp. 4023-32. mehran, hassanali and laurens, bernard, 1995, title interest rate liberalization and money market development: selected country experiences, article presented in an international seminar sponsored by the imf's monetary and exchange affairs department and the people's bank of china and held in beijing in august 1995 and published in 1996 edited by hassanali mehran, bernard laurens, and marc quintyn (washington: imf). satterthwaite, mark and shneyerov, artyom january 11, 2005, convergence to perfect competition of a dynamic matching and bargaining market with two-sided incomplete information and exogenous exit rate, kellogg school of management, northwestern university and department of economics, university of british columbia respectively. savastano, m., 1992, the pattern of currency substitution in latin america: an overview, revista de análisis económico 7(1) . schiantarelli, fabio; galindo, arturo; and weiss, andrew; april 2002, does financial liberalization improve the allocation of investment?: micro evidence from developing countries, inter-american development bank, research department, working paper # 467. sherif, k., m. borish, and a. gross, state owned banks in the transition: origins, evolution, and policy response, world bank. suzuki, kenji, february 2001, competition policy in the japanese banking sector: support big bang? european institute of japanese studies, stockholm school of economics, sweden, working paper no. 113. the bank of latvia, http://www.bank.lv/eng/main/statistics/converg european central bank, http://www.ecb.int financial system of bangladesh, http://www.bangladesh-bank.org. http://www.bank.lv/eng/main/statistics/converg http://www.ecb.int/ http://www.bangladesh-bank.org/siteupdate1.php international asset pricing models: the case of asean stock markets the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 117 international asset pricing models: the case of asean stock markets chee-wooi hooy and kim-leng goh university science malaysia and the university of malaya abstract this paper is about the role of economic grouping as it affects international capital asset pricing models, icapm. the conventional icapm is extended to include the economic grouping, regional and world factors. inclusion of the economic grouping factor increases the explanatory power of the asset pricing models. data on asean (indonesia, malaysia, philippines, singapore and thailand) stock markets are used in tests of the proposed models. the economic grouping factor turned out to be most important while the regional factor is least important for asset pricing in these stock markets. while four of the markets have higher systematic risk exposure to the economic group, the singapore market, the largest market, exhibits higher exposure to world risk. the segmentation of emerging markets offers a possible explanation for these results. keywords: capm, garch, integration, market risk, trading bloc jel classification: f360, g120 1. introduction the decade of the 1990s witnessed a surge in regional trade agreements.1 some proponents of trade regionalism argue that, since trading blocs involve lesser number of participants in the process of liberalization, blocs offer a more efficient way of moving towards globalization. frankel et al. (1995) and frankel and wei (1998) suggest that the recent trade regionalisms are likely to be welfareimproving. the formation of regional trading blocs not only induces intra-trade among the member countries leading to higher economic integration, but also stimulates active capital mobilization through fdi and portfolio investment flows across capital markets. this is usually supported by the regulation and policy cooperation in the financial system and the increased likelihood of coordination in monetary and exchange rate policies. the experience of european countries 1 http://www.cid.harvard.edu/cidtrade/issues/regionalism.html. ijbf 118 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 in the formation of the european union (eu) provides a supporting case (fratzscher, 2002). when the formation of a trading bloc opens up trade opportunities, newly created economic activities boost prosperity of the trading region while also attracting capital inflows from member countries. what started as a trading bloc often leads to wider cooperation among eonomic grouping. improved liquidity and market depth encourage monetary integration and a closer economic linkage among the member countries, in particular in the financial sector. many studies in the literature focused on the issues of portfolio diversification, market liberalization and contagion effect of an economic or financial crisis in their analysis of stock market linkages among member countries of groups. one of the early studies is by lessard (1973) who investigated the diversification opportunity in an investment union formed by four latino countries. that study documented evidence that the stock markets of the union members have closer relation compared to that of non-member countries. the formation of the eu has generated research interests on its capital markets some two decades later. akdogan (1992), for example, found evidence of integration among the stock markets of eu members, suggesitng increasing relaxation of capital controls among the union members. that result is further supported by studies of johnson and soenen (1993) and johnson et al. (1994) using covariance and correlation analysis. a number of studies investigated linkages of markets in other regional blocs. soydemir (2000) documented that the mexican stock market is weakly linked to the markets of argentina and brazil, but has strong ties with the us market. the result reveals the simple fact of different trading bloc memberships – mexico is a member of nafta (north american free trade agreement) that includes the us, while the other two latino markets are members of mercosur (mercado comun del cono sur). the paper concludes that stock market interdependence is a result of economic convergence. similarly, chen et al. (2002) established evidence of a long-run equilibrium relationship among six latino markets, attributing their findings to the formation of mercosur. although mexico is included in their analysis, the degree of exogeneity of its market is higher, and has a weaker linkage with the other markets. more recently, johnson and soenen (2003) reported significant simultaneous responses that exist between canada and mexico (both nafta members) and among argentina, brazil, chile (which are members of mercosur) and peru. the literature provides clear indication that stock markets are systematically linked to their counterparts belonging to the same economic grouping. such linkages would in turn affect the way assets are priced in these markets. on the latter aspect, however, there is little evidence in the literature documenting how economic grouping affects the asset pricing mechanism. this study investigated the economic grouping effects on asset pricing in the case of asean (association of southeast asian nations). the asean free trade area, or afta, was established with the aim of promoting intra-regional trade. subsequent developments led to wider economic cooperation among the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 119 asean members. the concept of an asean investment area was endorsed in order to lower barriers to intra-regional investment, to liberalize regulations, and to streamline incentives to be offered to boost regional investments. efforts are also underway for moving towards a higher degree of monetary and financial integration by fostering closer linkages through cross-border movements within the asean securities markets.2 in the capital market, for example, the ftse/ asean index series was recently launched as an indicator to the performance of the five oldest stock markets in asean, i.e., the stock markets of indonesia, malaysia, philippines, singapore and thailand. this study includes these five countries, which are also the founding members of asean. the different size and the degree of openness of the five markets offer an interesting case study. the markets of malaysia and singapore are relatively larger in size, matching the ratio of market capitalization to gdp of the markets in some industrialized countries, and their degree of financial openness is also much higher (rillo, 2004) compared to the other three markets. by these measurements, singapore has the most developed market. this study also proposes a direct modeling of the economic grouping impact using the international asset pricing model (icapm), while accounting for the time-varying nature of asset prices. the contribution of this paper is threefold. first, we show that the international asset pricing model, to which an economic grouping factor is added, has a higher explanatory power than the conventional icapm models. second, a dynamic analysis adopted in this study indicates that the pricing mechanism in a more developed market within the economic group is far more stable. lastly, this study offers new evidence that the stock markets of asean have a higher tendency to converge within the economic group than to the world market. the rest of this paper is organized as follows. section 2 introduces the modified icapm model to study economic grouping framework and the methodology for analysis. section 3 reports the results and section 4 concludes the paper. 2. models and methodology the icapm offers a theoretical framework for the pricing of risky assets in a fully integrated world of financial markets. the conditional expected return of a national equity market is exposed to the movement in returns of the world portfolio given by the following process: (1) where r it and r wt represent the returns of market-i and global portfolio, respectively, is the world risk-free rate, and all expectations are taken with respect to the information set available at time t-1, ; ω t-1 . this version of one2 http://www.asean.org. i );|r,rcov(r)|r(e twtitwtfttit ∀�=−� −− 11 δ 120 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 factor icapm assumes that the expected excess return of an individual market above an international risk-free rate is proportional to the country specific and non-diversifiable risk in the world market. the global portfolio return is the only source of systematic risk that affects the return of each individual market in model (1). some researchers including errunza and losq (1985), campbell and hamao (1992), davidson et al. (2003) and bekaert et al. (2005) have added the returns of dominant markets in the region into the icapm to reflect the regional effects. in this paper, the region refers to asia. we consider a two-factor model given as: (2) where r rt represents the return to a regional portfolio. apart from exposure to the world market risk, model (2) includes movements in the markets within the region as an additional source of systematic risk that affects the return of each individual market. although the regional influence is included, the effect due to formation of economic groupings is not considered in the two-factor model. we propose a three-factor model specified as follows: (3) where r gt , the returns to the portfolio of the economic group, provides a new factor to capture the risk exposure to the grouping where country-i is a member. following the above discussion, a number of different versions of icapm are obtained. for ease of exposition, the conditional expectation operation is dropped. three versions of one-factor icapm model are considered. first is the version that relates to the world market systematic risk that is given by: (4) where er i denotes the excess return of market-i and er w is the global portfolio excess return. the coefficient β i w captures the systematic risk of marketi due to exposure to the world market, as suggested in model (1). the other two single-factor models account for the exposure to market risk of the region and market risk of the economic grouping. these models are stated as (5) and (6) where er r and er g represent the excess returns of the regional portfolio and economic grouping portfolio, respectively. henceforth, we refer to models (4), (5) and (6) as the w-capm, r-capm and g-capm, respectively. i );|r,r(cov)|r,r(cov r)|r(e twtitwttrtitrtfttit ∀�+�=−� −−− 111 δδ )|r,r(cov)|r,r(cov r)|r(e trtitrttgtitgtfttit 111 −−− �+�=−� δδ i );|r,r(cov twtitwt ∀�+ −1δ itwt w iiit erer εβα ++= itrt r iiit erer εβα ++= itgt g iiit erer εβα ++= the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 121 three possible specifications of two-factor icapm can be formed by including the exposure to market risks of the region and world, economic grouping and world, and, economic grouping and region. the specifications of the two-factor models are as follows: (7) (8) (9) the models are referred to as rw-capm, gw-capm and gr-capm, respectively in this paper. the three-factor icapm model incorporates the economic grouping, regional and global effects. the model, denoted as grw-capm, is as follows: (10) we conduct a preliminary analysis of models (4) to (10) using ols estimation in a static framework. the best model is selected based on the akaike information criterion (aic) and schwarz criterion (sc). the parameter stability of the selected model is examined using the cusum of squares test (brown et al., 1975) and the n-step forecast test. evidence of recursive residuals outside the error bounds determined at the 5 per cent level is taken to indicate parameter or variance instability. the selected model is then re-estimated using the generalized autoregressive conditional heteroscedasticity (garch) model of bollerslev (1986) to account for temporal dependence in unconditional residuals which can be induced by time-varying volatility. the conditional expectation underlying models (1) to (3) is stated as follows: where, the conditional variance σ2 it follows the specification which can be written as: (11) this simple specification of garch(1,1) is found to be generally sufficient for empirical modeling (engle and ng, 1993). to account for non-normal conditional residual distribution, we apply the robust consistent variance-covariance estimator suggested by bollerslev and wooldridge (1992). a dynamic framework is used whereby the model is estimated using rolling regression method. the window is set at 3 years. this study is based on monthly data collected for the period january 1988 and november 2004. the morgan stanley capital international (msci) country itwt w irt r iiit ererer εββα +++= itwt w igt g iiit ererer εββα +++= itrt r igt g iiit ererer εββα +++= itwt w irt r igt g iiit erererer εβββα ++++= ),(n~| it1-tit 20 σε � 2 1 2 1 2 −− ++= t,it,iit βσαεωσ 122 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 index is used in the computation of returns of the individual country. the world portfolio is proxied by the msci all country world index, a free float-adjusted market capitalization index computed from the stock exchanges of 49 leading markets. the msci all country asia index based on indices of 12 leading asian markets is used to compute the returns on the regional portfolio. the return on the portfolio of the asean group for a member country is computed from an equal weighted market index of the remaining four members. the three-month treasury bill rates of us, downloaded from the website of the federal reserve, represent the risk-free rate used in this study. 3. results table 1: presents some summary statistics and the correlation matrix of the market returns. table 1: summary statistics and correlation matrix of returns summary statistics indonesia malaysia philippines singapore thailand region world mean 0.004 0.004 0.002 0.005 0.002 -0.001 0.005 std. dev. 0.149 0.093 0.098 0.073 0.121 0.063 0.042 skewness 0.422 -0.207 -0.009 -0.478 -0.381 0.002 -0.566 kurtosis 7.042 6.415 4.609 5.152 4.628 3.468 3.785 jarque-bera 144.206 100.091 21.910 46.912 27.319 1.855 16.045 (p-value) (0.000) (0.000) (0.000) (0.000) (0.000) (0.396) (0.000) no. of observations 203 203 203 203 203 203 203 correlation indonesia malaysia philippines singapore thailand region world indonesia 1 malaysia 0.479 1 philippines 0.500 0.549 1 singapore 0.505 0.658 0.617 1 thailand 0.466 0.566 0.634 0.655 1 region 0.231 0.391 0.348 0.528 0.438 1 world 0.261 0.424 0.409 0.622 0.467 0.785 1 note: std. dev. standard deviation. with the exception of the regional returns, the jarque-bera test rejects the normality assumption of the unconditional distribution of all the series. the returns of the indonesian market have the highest standard deviation while the return variability of the singapore market is the lowest. the asean markets the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 123 experienced more volatile movements than the global market. all correlation coefficients are positive, indicating co-movements of the same direction. malaysia and singapore are the most closely related markets compared to other markets. the indonesian market has the lowest correlation not only with its asean counterparts, but also with the regional and world markets. being the most developed and liberalized stock market, the singapore market has the highest correlation with the regional and world markets. 3.1. static estimation the results of the ols estimation of the w-, rg-, rw-, gw-, grand grwcapm models stated in equations (4) to (10) are reported in table 2. note that the economic grouping factor is significant at 1% in all the models across all five countries. the inclusion of the economic grouping factor also increases the adjusted-r2, i.e., models with this factor has better explanatory power than models without. the world and regional returns are not always significant in explaining returns of the individual markets. while regional returns are significant in the one-factor r-capm model, they are not significant with the incorporation of the global returns in the rw-capm model. similarly, the regional factor is also not significant in the grw-capm model. this suggests that the world market movements have higher influence on the asean stock markets compared to regional dynamics. the economic grouping dynamics have higher impact than the world market movements. of the grw-capm models, the economic grouping is always significant, but the world factor is only significant in two out of the five models. the magnitude of the betas suggests that the economic grouping factor has a dominant role in equity pricing of the asean markets. the presence of economic grouping factor reduces the magnitude of the world and regional betas when the one-factor model is extended to the twoand three-factor models. the exposure to systematic risks due to economic grouping is also consistently higher than those due to the regional and world factors. the only exception is singapore, where exposure to the world factor is higher than the economic grouping factor. the above findings are further supported by the model selection result. the gw-capm model outperforms the other models in four of the markets, while the gr-capm setting is the preferred model for thailand. the results are indicative that inclusion of the economic grouping returns has enhanced the goodness-of-fit and explanatory power of the conventional icapm model. the diagnostic results, however, suggest that the ols estimation suffers from arch effects. this problem is taken into account in the next section. two tests are conducted to show that the selected models experienced parameter instability. the results of the cusum of squares and n-step forecast tests are illustrated in figure 1. both the tests consistently show that the parameters in the selected models are not stable over time in all but the philippine market. some instability occurred in the early 1990s, but instability is most predominantly around the period of the 1997 financial crisis. to deal with this problem, the rolling window estimation is conducted. 12 4 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 table 2: icapm models-static estimations model constant group region world q(12) q2(12) normality arch lm f adj r2 aic sc indonesia w-capm 0.001 0.778 24.627 32.905 180.521 4.026 31.378 0.131 -1.040 -1.007 (0.889) (0.000)*** (0.017)** (0.001)*** (0.000)*** (0.000)*** (0.000)*** r-capm 0.006 0.621 20.557 28.920 156.448 3.627 27.277 0.115 -1.022 -0.989 (0.531) (0.000)*** (0.057)* (0.004)*** (0.000)*** (0.000)*** (0.000)*** g-capm 0.001 0.937 15.509 40.940 418.920 7.695 119.114 0.369 -1.360 -1.327 (0.867) (0.000)*** (0.215) (0.000)*** (0.000)*** (0.000)*** (0.000)*** rw-capm 0.002 0.170 0.601 23.293 31.813 181.006 3.950 15.871 0.128 -1.032 -0.983 (0.807) (0.557) (0.043)** (0.025)** (0.001)*** (0.000)*** (0.000)*** (0.000)*** gw-capm 0.004 1.147 -0.391 13.693 41.485 374.164 8.270 63.281 0.381 -1.375 -1.326 (0.679) (0.000)*** (0.027)** (0.321) (0.000)*** (0.000)*** (0.000)*** (0.000)*** gr-capm 0.001 1.095 -0.266 13.072 41.079 375.583 8.198 62.218 0.377 -1.369 -1.320 (0.881) (0.000)*** (0.050)** (0.364) (0.000)*** (0.000)*** (0.000)*** (0.000)*** grw-capm 0.003 1.152 -0.070 -0.323 13.359 41.442 370.343 8.314 42.034 0.379 -1.366 -1.300 (0.708) (0.000)*** (0.749) (0.248) (0.343) (0.000)*** (0.000)*** (0.000)*** (0.000)*** t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 1 25 malaysia w-capm 0.000 0.876 31.640 167.200 59.203 7.642 115.337 0.361 -2.105 -2.072 (0.030)** (11.795) (0.002)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** r-capm 0.006 0.705 21.594 113.650 32.372 6.126 98.538 0.326 -2.051 -2.018 (0.344) (0.000)*** (0.042)** (0.000)*** (0.000)*** (0.000)*** (0.000)*** g-capm 0.003 0.747 12.129 82.995 16.221 4.714 263.557 0.565 -2.489 -2.457 (0.596) (0.000)*** (0.435) (0.000)*** (0.000)*** (0.000)*** (0.000)*** rw-capm 0.002 0.222 0.644 26.060 153.060 53.899 7.384 59.162 0.365 -2.106 -2.058 (0.798) (0.120) (0.000)*** (0.011)** (0.000)*** (0.000)*** (0.000)*** (0.000)*** gw-capm 0.001 0.632 0.247 14.478 123.060 20.584 6.333 139.751 0.579 -2.516 -2.467 (0.813) (0.000)*** (0.017)** (0.271) (0.000)*** (0.000)*** (0.000)*** (0.000)*** gr-capm 0.003 0.646 0.195 16.734 115.290 21.925 5.977 139.419 0.578 -2.515 -2.466 (0.590) (0.000)*** (0.008)*** (0.160) (0.000)*** (0.000)*** (0.000)*** (0.000)*** grw-capm 0.002 0.627 0.100 0.148 15.779 123.570 21.892 6.318 93.247 0.578 -2.510 -2.444 (0.722) (0.000)*** (0.402) (0.382) (0.202) (0.000)*** (0.000)*** (0.000)*** (0.000)*** table 2: icapm models-static estimations (continued) 12 6 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 table 2: icapm models-static estimations (continued) model constant group region world q(12) q2(12) normality arch lm f adj r2 aic sc philippines w-capm -0.004 1.040 17.229 13.430 2.440 1.062 144.366 0.415 -1.986 -1.953 (0.550) (0.000)*** (0.141) (0.339) (0.295) (0.395) (0.000)*** r-capm 0.003 0.799 13.036 6.769 0.256 0.493 105.099 0.340 -1.865 -1.832 (0.634) (0.000)*** (0.366) (0.873) (0.880) (0.917) (0.000)*** g-capm -0.001 0.894 10.113 25.129 0.121 2.270 329.290 0.619 -2.415 -2.382 (0.791) (0.000)*** (0.606) (0.014)** (0.941) (0.011)** (0.000)*** rw-capm -0.003 0.087 0.949 16.619 12.645 1.704 0.986 72.086 0.413 -1.977 -1.928 (0.611) (0.600) (0.000)*** (0.164) (0.395) (0.427) (0.464) (0.000)*** gw-capm -0.003 0.732 0.342 10.408 27.965 1.653 2.378 182.228 0.642 -2.472 -2.423 (0.497) (0.000)*** (0.000)*** (0.580) (0.006)*** (0.437) (0.007)*** (0.000)*** gr-capm -0.001 0.792 0.193 9.841 22.971 0.210 1.910 172.397 0.629 -2.437 -2.388 (0.803) (0.000)*** (0.006)*** (0.630) (0.028)** (0.900) (0.036)** (0.000)*** grw-capm -0.004 0.737 -0.081 0.422 10.547 28.879 2.156 2.497 121.282 0.641 -2.464 -2.399 (0.432) (0.000)*** (0.531) (0.018)** (0.568) (0.004)*** (0.340) (0.005)*** (0.000)*** t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 1 27 singapore w-capm 0.001 0.948 12.226 101.370 36.305 6.307 293.565 0.592 -2.882 -2.849 (0.849) (0.000)*** (0.428) (0.000)*** (0.000)*** (0.000)*** (0.000)*** r-capm 0.007 0.741 12.061 59.878 6.159 5.096 204.807 0.502 -2.684 -2.652 (0.119) (0.000)*** (0.441) (0.000)*** (0.046)** (0.000)*** (0.000)*** g-capm 0.005 0.675 15.281 23.029 41.912 1.881 411.017 0.670 -3.095 -3.063 (0.205) (0.000)*** (0.226) (0.027)** (0.000)*** (0.039)** (0.000)*** rw-capm 0.002 0.138 0.804 10.676 93.288 32.790 6.304 148.372 0.593 -2.882 -2.833 (0.688) (0.164) (0.000)*** (0.557) (0.000)*** (0.000)*** (0.000)*** (0.000)*** gw-capm 0.002 0.454 0.503 16.363 47.638 45.667 2.775 329.416 0.765 -3.429 -3.380 (0.613) (0.000)*** (0.000)*** (0.175) (0.000)*** (0.000)*** (0.002)*** (0.000)*** gr-capm 0.005 0.507 0.345 25.006 40.655 76.921 2.978 283.661 0.737 -3.316 -3.268 (0.167) (0.000)*** (0.000)*** (0.015)** (0.000)*** (0.000)*** (0.001)*** (0.000)*** grw-capm 0.002 0.452 0.040 0.464 17.547 47.577 49.807 2.828 218.899 0.764 -3.421 -3.355 (0.557) (0.000)*** (0.615) (0.000)*** (0.130) (0.000)*** (0.000)*** (0.001)*** (0.000)*** table 2: icapm models-static estimations (continued) 12 8 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 table 2: icapm models-static estimations (continued) model constant group region world q(12) q2(12) normality arch lm f adj r2 aic sc thailand w-capm -0.005 1.216 30.032 103.230 10.534 4.376 135.878 0.400 -1.612 -1.580 (0.538) (0.000)*** (0.003)*** (0.000)*** (0.005)*** (0.000)*** (0.000)*** r-capm 0.003 0.996 18.874 61.432 1.676 3.271 121.346 0.373 -1.568 -1.536 (0.694) (0.000)*** (0.092)* (0.000)*** (0.432) (0.000)*** (0.000)*** g-capm -0.002 1.101 16.263 36.122 297.860 3.102 309.758 0.605 -2.029 -1.996 (0.721) (0.000)*** (0.180) (0.000)*** (0.000)*** (0.001)*** (0.000)*** rw-capm -0.002 0.386 0.814 24.646 86.510 5.621 4.064 71.119 0.410 -1.623 -1.574 (0.751) (0.063)* (0.000)*** (0.017)** (0.000)*** (0.060)** (0.000)*** (0.000)*** gw-capm -0.004 0.908 0.390 15.154 41.754 148.087 3.378 169.663 0.625 -2.078 -2.029 (0.450) (0.000)*** (0.009)*** (0.233) (0.000)*** (0.000)*** (0.000)*** (0.000)*** gr-capm -0.002 0.917 0.339 14.570 44.631 144.553 3.879 172.567 0.629 -2.089 -2.040 (0.734) (0.000)*** (0.005)*** (0.266) (0.000)*** (0.000)*** (0.000)*** (0.000)*** grw-capm -0.003 0.896 0.248 0.142 14.816 44.698 134.620 3.799 114.994 0.629 -2.082 -2.017 (0.617) (0.000)*** (0.116) (0.471) (0.252) (0.000)*** (0.000)*** (0.000)*** (0.000)*** note: figures in the parentheses are the p-values. * denotes significance at the 0.10 level; ** denotes significance at the 0.05 level; and *** denotes significance at the 0.01 level. the coefficient significance test is based on the white (1980) heteroskedasticity consistent covariance estimates. aic and sc denotes the akaike and schwarz information criteria. the underlined values show the model chosen using adjusted (adj) r2, aic and sc. q(12) and q2(12) are the lm tests for serial dependence in the residuals and squared residuals at 12 lags, respectively. normality is the jacque-bera test for normality. arch lm is the test for presence of arch effects at 12 lags. f indicates the overall test for model significance. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 129 figure 1: tests for parameter instability cusum of squares test n-step forecast test indonesia malaysia philippines singapore thailand 13 0 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 table 3: rolling regression estimates of the gw-capm model for indonesia constant mean group world constant variance arch garch q(12) q2(12) normality arch lm adj r2 88/1-90/12 0.012 0.422 -0.806 0.001 -0.112 1.183 11.397 12.177 8.458 0.335 0.156 (0.420) (0.100)*** (0.011)** (0.701) (0.572) (0.000)*** (0.495) (0.432) (0.015)** (0.963) 89/1-91/12 -0.013 1.416 -1.351 0.007 0.910 -0.047 10.782 3.082 6.805 0.479 0.128 (0.271) (0.000)*** (0.033)** (0.018)** (0.381) (0.007)*** (0.548) (0.995) (0.033)** (0.889) 90/1-92/12 -0.023 1.273 -0.724 0.001 -0.277 1.134 14.560 20.293 2.015 0.699 0.406 (0.007)*** (0.000)*** (0.000)*** (0.001)*** (0.208) (0.000)*** (0.266) (0.062)* (0.365) (0.727) 91/1-93/12 -0.019 0.666 0.203 0.000 -0.201 1.167 5.712 13.836 1.345 0.926 0.358 (0.066)* (0.000)*** (0.155) (0.000)*** (0.348) (0.000)*** (0.930) (0.311) (0.510) (0.554) 92/1-94/12 -0.005 0.535 0.567 0.002 -0.027 0.531 13.250 13.024 1.520 0.901 0.567 (0.637) (0.000)*** (0.005)*** (0.699) (0.800) (0.671) (0.351) (0.367) (0.468) (0.572) 93/1-95/12 0.003 0.907 0.169 0.000 0.333 0.595 14.430 17.321 0.303 0.912 0.516 (0.770) (0.000)*** (0.402) (0.639) (0.038)** (0.087)* (0.274) (0.138) (0.859) (0.564) 94/1-96/12 0.008 1.100 0.134 0.000 -0.166 1.070 6.654 7.642 1.987 0.883 0.657 (0.142) (0.000)*** (0.345) (0.445) (0.481) (0.006)*** (0.880) (0.812) (0.370) (0.585) 95/1-97/12 0.021 1.262 -0.147 0.000 1.286 0.056 11.733 7.999 1.965 0.965 0.626 (0.000)*** (0.000)*** (0.390) (0.120) (0.002)*** (0.558) (0.467) (0.785) (0.374) (0.527) 96/1-98/12 0.014 1.052 0.673 0.000 0.514 0.752 12.593 7.764 1.195 0.367 0.244 (0.070)* (0.000)*** (0.001)*** (0.835) (0.112) (0.001)*** (0.399) (0.803) (0.550) (0.951) 97/1-99/12 0.019 1.358 -0.137 0.002 0.473 0.594 18.482 17.371 1.015 0.772 0.350 (0.442) (0.000)*** (0.756) (0.408) (0.032)** (0.000)*** (0.102) (0.136) (0.602) (0.669) 98/1-00/12 -0.039 1.255 -0.633 -0.001 -0.082 1.048 13.353 14.038 1.795 1.084 0.318 (0.001)*** (0.000)*** (0.008)*** (0.000)*** (0.634) (0.000)*** (0.344) (0.298) (0.408) (0.450) 99/1-01/12 -0.007 1.459 -0.553 0.002 -0.083 0.853 6.882 6.210 1.929 0.532 0.532 (0.655) (0.000)*** (0.006)*** (0.416) (0.392) (0.005)*** (0.865) (0.905) (0.381) (0.853) 00/1-02/12 0.009 1.077 -0.318 0.002 -0.239 1.057 12.274 5.939 2.129 0.395 0.395 (0.512) (0.000)*** (0.012)** (0.318) (0.052)* (0.000)*** (0.424) (0.919) (0.345) (0.937) 01/1-03/12 0.020 0.802 -0.177 0.001 -0.259 1.131 4.433 7.066 0.890 0.411 0.154 (0.042)** (0.000)*** (0.033)** (0.000)*** (0.027)** (0.000)*** (0.974) (0.853) (0.641) (0.929) 02/1-04/11 0.022 0.982 -0.008 0.002 0.036 0.702 7.179 22.320 1.574 2.747 0.394 (0.156) (0.011)** (0.987) (0.801) (0.744) (0.500) (0.846) (0.034)** (0.455) (0.060)* full sample 0.004 1.232 -0.258 0.000 0.306 0.763 8.348 18.773 356.015 7.839 0.359 (0.524) (0.000)*** (0.094)* (0.500) (0.027)** (0.000)*** (0.757) (0.094)* (0.000)*** (0.000)*** t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 1 31 table 4: rolling regression estimates of the gw-capm model for malaysia constant mean group world constant variance arch garch q(12) q2(12) normality arch lm adj r2 88/1-90/12 0.001 0.364 0.827 0.000 -0.219 1.124 11.312 6.797 0.603 0.631 0.619 (0.894) (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.502) (0.871) (0.740) (0.780) 89/1-91/12 0.002 0.440 0.573 0.000 0.216 0.544 13.432 7.770 1.067 0.210 0.709 (0.706) (0.000)*** (0.001)*** (0.528) (0.292) (0.295) (0.338) (0.803) (0.586) (0.994) 90/1-92/12 0.001 0.595 0.361 0.000 -0.223 1.149 6.748 10.178 1.335 0.981 0.771 (0.765) (0.000)*** (0.000)*** (0.000)*** (0.001)*** (0.000)*** (0.874) (0.600) (0.513) (0.516) 91/1-93/12 0.008 0.646 0.242 0.001 -0.141 0.592 13.635 5.892 0.468 0.371 0.708 (0.000)*** (0.000)*** (0.036)** (0.000)*** (0.336) (0.010)** (0.325) (0.921) (0.792) (0.949) 92/1-94/12 0.000 0.809 0.106 0.001 0.071 0.485 19.398 10.390 0.512 0.463 0.640 (0.971) (0.000)*** (0.682) (0.634) (0.605) (0.614)*** (0.079)* (0.582) (0.774) (0.899) 93/1-95/12 -0.004 1.037 0.174 0.000 0.931 0.216 26.146 14.322 0.951 0.651 0.606 (0.259) (0.000)*** (0.136) (0.128) (0.002)*** (0.049)** (0.010)** (0.281) (0.622) (0.764) 94/1-96/12 0.005 0.615 0.353 0.000 -0.175 1.008 5.369 10.832 1.454 0.570 0.530 (0.312) (0.000)*** (0.000)*** (0.000)*** (0.039)** (0.000)*** (0.945) (0.543) (0.483) (0.826) 95/1-97/12 0.010 0.888 0.056 0.000 0.050 1.187 10.460 9.527 1.040 0.564 0.643 (0.225) (0.000)*** (0.808) (0.000)*** (0.779) (0.000)*** (0.576) (0.657) (0.594) (0.831) 96/1-98/12 0.012 0.842 -0.185 0.000 0.253 0.854 8.034 7.032 1.006 0.325 0.535 (0.257) (0.000)*** (0.475) (0.946) (0.258) (0.000)*** (0.782) (0.856) (0.605) (0.967) 97/1-99/12 0.000 1.041 -0.713 0.001 -0.205 1.106 8.320 13.270 1.413 0.717 0.563 (0.969) (0.000)*** (0.009)*** (0.000)*** (0.089)* (0.000)*** (0.760) (0.350) (0.493) (0.712) 98/1-00/12 0.003 0.794 -0.637 0.007 -0.272 0.684 8.041 12.549 0.002 0.659 0.297 (0.768) (0.000)*** (0.011)** (0.000)*** (0.015)** (0.000)*** (0.782) (0.403) (0.999) (0.758) 99/1-01/12 0.033 0.400 0.380 0.000 -0.116 1.143 10.719 15.445 1.147 1.465 0.221 (0.001)*** (0.000)*** (0.001)*** (0.000)*** (0.642) (0.000)*** (0.553) (0.218) (0.564) (0.267) 00/1-02/12 0.023 0.325 0.411 0.000 -0.143 1.071 7.696 14.116 1.846 1.583 0.317 (0.001)*** (0.000)*** (0.000)*** (0.166) (0.642) (0.006)*** (0.808) (0.293) (0.397) (0.227) 01/1-03/12 0.019 0.418 0.317 0.002 -0.156 0.354 3.663 6.761 0.865 0.377 0.552 (0.013)** (0.000)*** (0.013)** (0.309) (0.059)* (0.624) (0.989) (0.873) (0.649) (0.946) 02/1-04/11 -0.002 0.310 0.551 0.001 0.232 0.166 8.658 7.758 1.412 0.492 0.683 (0.747) (0.008)*** (0.000)*** (0.368) (0.181) (0.813) (0.732) (0.804) (0.494) (0.878) full sample 0.002 0.522 0.398 0.000 0.159 0.797 7.872 9.846 0.700 0.722 0.565 (0.654) (0.000)*** (0.000)*** (0.270) (0.042)** (0.000)*** (0.795) (0.629) (0.705) (0.729) 13 2 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 table 5: rolling regression estimates of the gw-capm model for philippines constant mean group world constant variance arch garch q(12) q2(12) normality arch lm adj r2 88/1-90/12 -0.010 0.541 0.274 0.003 -0.141 0.567 12.357 19.316 2.794 1.548 0.229 (0.198) (0.000)*** (0.010)** (0.177) (0.018)** (0.116) (0.417) (0.081)* (0.247) (0.239) 89/1-91/12 0.008 0.616 0.385 0.003 -0.137 0.618 15.643 15.265 6.775 0.882 0.399 (0.363) (0.000)*** (0.004)*** (0.015)** (0.056)* (0.002)*** (0.208) (0.227) (0.034)** (0.585) 90/1-92/12 0.009 0.738 0.424 0.003 -0.096 0.543 7.235 3.996 5.634 1.360 0.570 (0.320) (0.000)*** (0.030)** (0.327) (0.085)** (0.438) (0.842) (0.984) (0.060)* (0.309) 91/1-93/12 0.014 1.091 0.122 0.001 -0.397 1.077 9.193 13.672 2.164 0.935 0.626 (0.011)** (0.000)*** (0.303) (0.147) (0.014)** (0.000)*** (0.686) (0.322) (0.339) (0.548) 92/1-94/12 0.012 1.078 0.001 0.001 -0.054 0.766 15.482 16.378 1.347 1.533 0.698 (0.222) (0.000)*** (0.997) (0.797) (0.738) (0.460) (0.216) (0.175) (0.510) (0.244) 93/1-95/12 -0.004 1.358 -0.490 0.001 -0.310 1.004 14.182 17.805 0.450 1.361 0.750 (0.367) (0.000)*** (0.000)*** (0.000)*** (0.001)*** (0.000)*** (0.289) (0.122) (0.799) (0.308) 94/1-96/12 0.007 1.115 -0.292 0.000 -0.189 1.117 18.227 12.471 0.101 1.342 0.677 (0.213) (0.000)*** (0.015)** (0.000)*** (0.352) (0.000)*** (0.109) (0.409) (0.951) (0.317) 95/1-97/12 -0.012 0.916 0.302 0.000 -0.032 1.139 10.994 5.671 0.066 0.239 0.542 (0.227) (0.000)*** (0.271) (0.320) (0.931) (0.000)*** (0.529) (0.932) (0.967) (0.990) 96/1-98/12 0.009 0.796 0.210 0.000 -0.119 1.142 11.632 6.463 0.296 0.283 0.640 (0.332) (0.000)*** (0.372) (0.551) (0.563) (0.000)*** (0.476) (0.891) (0.862) (0.980) 97/1-99/12 -0.023 0.610 0.671 0.006 -0.104 0.044 16.154 7.253 0.151 0.501 0.667 (0.055)* (0.000)*** (0.007)*** (0.334) (0.319) (0.969) (0.184) (0.840) (0.927) (0.874) 98/1-00/12 -0.028 0.588 0.908 0.000 -0.194 1.166 8.545 9.897 1.139 2.177 0.696 (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.090)** (0.000)*** (0.741) (0.625) (0.566) (0.104) 99/1-01/12 -0.034 0.663 0.521 0.002 0.662 -0.300 6.669 14.126 2.244 0.541 0.713 (0.000)*** (0.000)*** (0.000)*** (0.017)** (0.002)*** (0.125) (0.879) (0.293) (0.326) (0.847) 00/1-02/12 -0.016 0.790 0.317 0.005 0.659 -0.659 10.109 10.568 0.951 0.638 0.595 (0.000)*** (0.000)*** (0.000)*** (0.001)*** (0.000)*** (0.000)*** (0.606) (0.566) (0.622) (0.774) 01/1-03/12 -0.026 0.934 0.342 0.000 -0.203 1.177 12.926 5.538 1.830 0.348 0.610 (0.003)*** (0.000)*** (0.009)*** (0.000)*** (0.189) (0.000)*** (0.374) (0.938) (0.401) (0.959) 02/1-04/11 -0.023 0.912 -0.034 0.001 -0.305 1.190 12.214 10.476 1.857 0.826 0.551 (0.001)*** (0.000)*** (0.851) (0.000)*** (0.209) (0.000)*** (0.429) (0.574) (0.395) (0.628) full sample -0.003 0.732 0.341 0.003 -0.004 0.334 10.383 27.781 1.656 2.380 0.637 (0.493) (0.000)*** (0.000)*** (0.938) (0.934) (0.969) (0.582) (0.006)*** (0.437) (0.007)*** t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 1 33 table 6: rolling regression estimates of the gw-capm model for singapore constant mean group world constant variance arch garch q(12) q2(12) normality arch lm adj r2 88/1-90/12 0.002 0.378 0.615 0.001 -0.207 0.770 16.504 11.278 1.042 0.594 0.720 (0.691) (0.000)*** (0.000)*** (0.004)*** (0.284) (0.002)*** (0.169) (0.505) (0.594) (0.808) 89/1-91/12 0.006 0.400 0.631 0.000 0.337 0.507 12.093 8.062 3.799 0.501 0.816 (0.286) (0.000)*** (0.000)*** (0.484) (0.467) (0.300) (0.438) (0.780) (0.150) (0.874) 90/1-92/12 0.003 0.427 0.567 0.000 -0.167 1.070 9.140 5.987 0.636 0.585 0.911 (0.330) (0.000)*** (0.000)*** (0.000)*** (0.003)*** (0.000)*** (0.691) (0.917) (0.727) (0.814) 91/1-93/12 0.004 0.427 0.570 0.000 -0.114 1.207 9.054 5.107 1.691 0.730 0.877 (0.245) (0.000)*** (0.000)*** (0.000)*** (0.696) (0.000)*** (0.698) (0.954) (0.429) (0.702) 92/1-94/12 0.002 0.399 0.569 0.000 -0.164 1.160 4.780 13.290 2.028 1.308 0.839 (0.571) (0.000)*** (0.000)*** (0.000)*** (0.399) (0.000)*** (0.965) (0.348) (0.363) (0.332) 93/1-95/12 0.004 0.512 0.470 0.000 -0.171 1.138 4.627 18.196 1.273 0.844 0.824 (0.262) (0.000)*** (0.000)*** (0.000)*** (0.233) (0.000)*** (0.969) (0.110) (0.529) (0.614) 94/1-96/12 -0.001 0.501 0.402 0.000 -0.185 1.144 14.895 14.963 0.421 1.363 0.778 (0.576) (0.000)*** (0.000)*** (0.000)*** (0.471) (0.000)*** (0.247) (0.243) (0.810) (0.308) 95/1-97/12 -0.012 0.261 0.825 0.000 0.034 1.169 12.082 15.700 1.042 1.129 0.696 (0.049)** (0.000)*** (0.000)*** (0.000)*** (0.888) (0.000)*** (0.439) (0.205) (0.594) (0.424) 96/1-98/12 -0.011 0.309 0.771 0.000 0.295 0.749 13.477 14.261 2.244 0.430 0.657 (0.103) (0.038)** (0.000)*** (0.293) (0.253) (0.000)*** (0.335) (0.284) (0.326) (0.919) 97/1-99/12 0.012 0.426 0.729 0.000 0.280 0.630 9.280 12.295 0.786 0.522 0.684 (0.143) (0.000)*** (0.000)*** (0.473) (0.286) (0.015)** (0.679) (0.422) (0.675) (0.860) 98/1-00/12 0.018 0.504 0.808 0.001 0.561 0.192 6.761 13.868 1.424 0.938 0.637 (0.002)*** (0.000)*** (0.000)*** (0.063)* (0.010)** (0.188) (0.873) (0.309) (0.491) (0.545) 99/1-01/12 0.006 0.372 0.409 0.004 0.280 -0.452 19.134 9.508 1.998 0.338 0.578 (0.467) (0.000)*** (0.004)*** (0.078)* (0.261) (0.254) (0.085)* (0.659) (0.368) (0.962) 00/1-02/12 0.002 0.239 0.432 0.001 1.144 -0.139 17.495 6.482 0.627 0.603 0.537 (0.693) (0.000)*** (0.000)*** (0.001)*** (0.001)*** (0.005)*** (0.132) (0.890) (0.731) (0.801) 01/1-03/12 -0.003 0.309 0.446 0.002 0.547 -0.384 9.632 12.749 0.610 0.520 0.644 (0.459) (0.000)*** (0.000)*** (0.004)*** (0.024)** (0.021)** (0.648) (0.388) (0.737) (0.862) 02/1-04/11 -0.001 0.402 0.567 0.000 -0.145 1.089 9.320 10.980 0.439 0.777 0.811 (0.709) (0.000)*** (0.000)*** (0.000)*** (0.493) (0.000)*** (0.675) (0.531) (0.803) (0.665) full sample 0.002 0.401 0.572 0.000 0.174 0.799 11.058 10.061 3.139 0.864 0.759 (0.520) (0.000)*** (0.000)*** (0.217) (0.034)** (0.000)*** (0.524) (0.611) (0.208) (0.585) 13 4 t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 table 7: rolling regression estimates of the gw-capm model for thailand constant mean group world constant variance arch garch q(12) q2(12) normality arch lm adj r2 88/1-90/12 0.005 0.676 0.844 0.002 -0.262 0.914 9.417 8.857 1.549 0.484 0.498 (0.504) (0.000)*** (0.000)*** (0.000)*** (0.054)* (0.000)*** (0.667) (0.715) (0.461) (0.886) 89/1-91/12 0.008 0.800 0.544 0.002 -0.379 0.829 14.638 17.344 2.027 0.564 0.620 (0.450) (0.000)*** (0.000)*** (0.000)*** (0.090)* (0.000)*** (0.262) (0.137) (0.363) (0.831) 90/1-92/12 0.003 0.884 0.316 0.001 -0.298 1.079 16.278 10.486 1.761 0.992 0.617 (0.668) (0.000)*** (0.094)* (0.000)*** (0.001)*** (0.000)*** (0.179) (0.573) (0.414) (0.509) 91/1-93/12 0.014 1.068 -0.118 0.002 -0.348 0.991 14.697 6.019 1.151 0.579 0.504 (0.061)* (0.000)*** (0.300) (0.000)*** (0.013)** (0.000)*** (0.258) (0.915) (0.562) (0.819) 92/1-94/12 0.000 1.075 -0.215 0.003 -0.265 0.564 22.854 17.516 0.550 0.805 0.638 (0.949) (0.000)*** (0.023)** (0.001)*** (0.001)*** (0.049)** (0.029)** (0.131) (0.759) (0.643) 93/1-95/12 -0.004 1.198 -0.161 0.000 -0.162 1.096 13.749 19.978 1.111 0.637 0.755 (0.291) (0.000)*** (0.057)* (0.000)*** (0.409) (0.000)*** (0.317) (0.068)* (0.574) (0.775) 94/1-96/12 -0.007 1.206 -0.116 0.000 -0.036 1.144 10.173 5.117 2.798 0.443 0.676 (0.226) (0.000)*** (0.386) (0.670) (0.763) (0.000)*** (0.601) (0.954) (0.247) (0.911) 95/1-97/12 -0.010 1.191 -0.182 0.000 -0.140 1.307 9.568 6.138 1.007 0.245 0.554 (0.017)** (0.000)*** (0.320) (0.000)*** (0.489) (0.000)*** (0.654) (0.909) (0.604) (0.989) 96/1-98/12 -0.049 0.641 1.532 0.009 -0.107 0.498 10.731 15.678 6.493 0.665 0.563 (0.000)*** (0.000)*** (0.005)*** (0.317) (0.138) (0.474) (0.552) (0.206) (0.039)** (0.753) 97/1-99/12 -0.042 0.680 1.348 0.008 -0.100 0.506 10.550 14.459 2.705 0.381 0.583 (0.003)*** (0.000)*** (0.009)*** (0.475) (0.157) (0.582) (0.568) (0.272) (0.259) (0.944) 98/1-00/12 -0.017 1.009 0.671 0.001 -0.174 0.945 20.034 8.504 1.490 0.674 0.539 (0.208) (0.000)*** (0.005)*** (0.000)*** (0.000)*** (0.000)*** (0.066)* (0.745) (0.475) (0.746) 99/1-01/12 -0.020 1.231 0.266 0.006 0.188 -0.185 60.674 6.622 0.938 0.760 0.704 (0.163) (0.000)*** (0.173) (0.238) (0.405) (0.829) (0.000)*** (0.882) (0.626) (0.679) 00/1-02/12 -0.002 1.162 0.172 0.000 -0.175 1.096 15.731 12.514 0.919 0.665 0.630 (0.832) (0.000)*** (0.208) (0.107) (0.278) (0.000)*** (0.204) (0.405) (0.632) (0.753) 01/1-03/12 0.022 0.677 0.407 0.000 -0.152 1.086 7.768 6.339 1.771 0.418 0.511 (0.016)** (0.000)*** (0.008)*** (0.305) (0.476) (0.000)*** (0.803) (0.898) (0.413) (0.925) 02/1-04/11 0.020 0.519 0.621 0.002 0.095 0.348 8.272 5.803 10.098 0.282 0.653 (0.044)** (0.004)*** (0.001)*** (0.632) (0.731) (0.763) (0.764) (0.926) (0.006)*** (0.979) full sample 0.000 0.880 0.335 0.000 0.103 0.848 10.265 12.251 29.167 1.001 0.617 (0.972) (0.000)*** (0.007)*** (0.355) (0.115) (0.000)*** (0.593) (0.426) (0.000)*** (0.450) t h e in te rn a ti o n a l jo u rn a l o f b a n ki n g a n d f in a n ce , 2 00 8/ 09 v ol . 6 . n um be r 1: 2 00 8: 1 17 -1 40 1 35 table 8: rolling regression estimates of the gr-capm model for thailand constant mean group region constant variance arch garch q(12) q2(12) normality arch lm adj r2 88/1-90/12 0.000 0.866 0.353 0.005 0.291 -0.458 20.305 9.956 1.463 1.637 0.430 (0.933) (0.000)*** (0.003)*** (0.005)*** (0.056)* (0.042)** (0.062)* (0.620) (0.481) (0.212) 89/1-91/12 0.018 0.858 0.359 0.001 -0.341 1.054 10.837 11.527 2.675 0.572 0.601 (0.023)** (0.000)*** (0.000)*** (0.000)*** (0.054)* (0.000)*** (0.543) (0.484) (0.262) (0.824) 90/1-92/12 0.011 0.943 0.159 0.001 -0.292 1.127 21.705 11.456 2.349 0.804 0.619 (0.145) (0.000)*** (0.090)* (0.000)*** (0.052)* (0.000)*** (0.041)** (0.490) (0.309) (0.644) 91/1-93/12 0.010 0.999 -0.074 0.002 -0.320 0.808 18.156 7.207 1.436 0.371 0.513 (0.144) (0.000)*** (0.204) (0.000)*** (0.000)*** (0.000)*** (0.111) (0.844) (0.488) (0.949) 92/1-94/12 0.008 1.234 -0.133 0.000 -0.020 1.050 12.102 13.001 0.494 0.673 0.642 (0.304) (0.000)*** (0.214) (0.000)*** (0.916) (0.000)*** (0.438) (0.369) (0.781) (0.747) 93/1-95/12 -0.006 1.224 -0.182 0.000 -0.162 1.099 17.121 26.717 1.592 0.840 0.769 (0.141) (0.000)*** (0.002)*** (0.000)*** (0.448) (0.000)*** (0.145) (0.008)*** (0.451) (0.617) 94/1-96/12 -0.011 1.148 -0.143 0.000 -0.081 1.178 7.821 6.170 2.210 0.529 0.671 (0.022)** (0.000)*** (0.108) (0.002)*** (0.316) (0.000)*** (0.799) (0.907) (0.331) (0.855) 95/1-97/12 -0.013 1.154 -0.194 0.000 -0.088 1.228 8.336 5.517 0.403 0.272 0.553 (0.001)*** (0.000)*** (0.045)** (0.714) (0.525) (0.000)*** (0.758) (0.938) (0.818) (0.983) 96/1-98/12 -0.032 0.725 0.948 0.001 -0.145 1.051 14.262 16.407 7.825 3.039 0.570 (0.005)*** (0.000)*** (0.000)*** (0.000)*** (0.270) (0.000)*** (0.284) (0.173) (0.020)** (0.038)** 97/1-99/12 -0.014 0.675 1.011 0.008 -0.129 0.552 12.756 17.292 3.743 0.879 0.590 (0.416) (0.000)*** (0.002)*** (0.031)** (0.134) (0.129) (0.387) (0.139) (0.154) (0.588) 98/1-00/12 -0.022 0.717 0.920 0.005 -0.145 0.607 11.455 4.333 0.949 0.766 0.602 (0.089)* (0.000)*** (0.000)*** (0.000)*** (0.001)*** (0.000)*** (0.490) (0.977) (0.622) (0.674) 99/1-01/12 -0.021 1.169 0.405 0.004 0.206 0.054 50.657 8.980 1.140 0.791 0.724 (0.134) (0.000)*** (0.039)** (0.231) (0.343) (0.939) (0.000)*** (0.705) (0.565) (0.654) 00/1-02/12 -0.015 1.274 0.273 0.010 0.419 -0.823 14.436 10.586 0.896 0.483 0.640 (0.082)* (0.000)*** (0.033)** (0.000)*** (0.000)*** (0.000)*** (0.274) (0.565) (0.639) (0.887) 01/1-03/12 0.018 0.575 0.524 0.000 -0.197 1.125 8.849 2.688 3.089 0.521 0.525 (0.049)** (0.000)*** (0.000)*** (0.416) (0.559) (0.003)*** (0.716) (0.997) (0.213) (0.861) 02/1-04/11 0.018 0.701 0.439 0.002 0.230 0.251 8.806 4.628 10.061 0.258 0.622 (0.054)* (0.000)*** (0.028)** (0.529) (0.285) (0.741) (0.719) (0.969) (0.007)*** (0.985) total 0.001 0.933 0.247 0.000 0.100 0.852 9.383 14.302 32.996 1.226 0.621 (0.857) (0.000)*** (0.010)** (0.357) (0.142) (0.000)*** (0.670) (0.282) (0.000)*** (0.268) 136 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 -2 -1 . 5 -1 -0 . 5 0 0 . 5 1 1 . 5 2 8 8 / 1 9 0 /1 2 8 9 / 1 9 1 / 1 2 9 0 /1 9 2 / 1 2 9 1 /1 9 3 /1 2 9 2 / 1 9 4 /1 2 9 3 / 1 9 5 / 1 2 9 4 /1 9 6 / 1 2 9 5 /1 9 7 /1 2 9 6 / 1 9 8 /1 2 9 7 / 1 9 9 / 1 2 9 8 /1 0 0 / 1 2 9 9 /1 0 1 /1 2 0 0 / 1 0 2 / 1 2 0 1 / 1 0 3 / 1 2 0 2 /1 0 4 / 1 1 eco n o m ic g ro up w orld -2 -1 .5 -1 -0 .5 0 0 . 5 1 1 . 5 2 8 8 /1 9 0 /1 2 8 9 / 1 9 1 / 1 2 9 0 / 1 9 2 / 1 2 9 1 /1 9 3 / 1 2 9 2 /1 9 4 /1 2 9 3 / 1 9 5 / 1 2 9 4 / 1 9 6 / 1 2 9 5 /1 9 7 / 1 2 9 6 /1 9 8 /1 2 9 7 / 1 9 9 / 1 2 9 8 / 1 0 0 / 1 2 9 9 /1 0 1 / 1 2 0 0 /1 0 2 /1 2 0 1 / 1 0 3 /1 2 0 2 / 1 0 4 / 1 1 econ o m ic g ro u p w o rld 3.2. rolling regression estimation the gw-capm model is reestimated using the garch(1,1) specification and a rolling window of three years. a total of 15 windows are considered. the grcapm model is also estimated for thailand following the model selection results from the last section. the estimates are given in tables 3 to 8. the diagnostic results suggest that the models are generally well specified. the arch-lm test and q-statistic for the squared residuals indicate that the problem of arch effects has been adequately dealt with. the jarque-bera normality test shows no evidence against the normality assumption in most of the cases. the economic grouping variable is significant in all 15 windows as well as the full sample period for all the five markets. in contrary, the world factor, (and regional factor for thailand) is not always significant. for singapore, however, the exposure to world risk is significant for all the windows. for the other four markets, it is interesting to note that the world market returns are not significant in many instances prior to and during the 1997 crisis. the betas obtained from the rolling estimates of the gw-capm model are plotted in figure 2. these estimates are generally not stable, thus supporting the use of rolling windows. singapore is the only market with stable betas, and instability is only observed around the 1997-8 financial crisis years. bigger fluctuations are seen in the other markets, in particular, indonesia. it is also clear that the financial crisis caused higher instability in beta estimates of four markets. figure 2: world and economic group betas of the gw-capm model indonesia malaysia the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 137 -2 -1 .5 -1 -0 .5 0 0 . 5 1 1 . 5 2 8 8 /1 9 0 /1 2 8 9 / 1 9 1 / 1 2 9 0 / 1 9 2 / 1 2 9 1 /1 9 3 / 1 2 9 2 /1 9 4 /1 2 9 3 / 1 9 5 / 1 2 9 4 / 1 9 6 / 1 2 9 5 /1 9 7 / 1 2 9 6 /1 9 8 /1 2 9 7 / 1 9 9 / 1 2 9 8 / 1 0 0 / 1 2 9 9 /1 0 1 / 1 2 0 0 /1 0 2 /1 2 0 1 / 1 0 3 /1 2 0 2 / 1 0 4 / 1 1 econ o m ic g ro u p w o rld figure 2: world and economic group betas of the gw-capm model (continued) philippines singapore thailand risk exposures to the economic grouping and world generally move in opposite direction. markets with low world betas are likely to have a relatively high exposure to economic grouping risks, and vice versa. specifically, the exposure to the economic grouping factor is relatively higher in indonesia, malaysia, philippines and thailand except for a small number of windows. the singapore market has a different behaviour, where the exposure to the world factor is more prominent. nevertheless, the economic grouping beta for singapore is only -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 88/190/12 89/191/12 90/192/12 91/193/12 92/194/12 93/195/12 94/196/12 95/197/12 96/198/12 97/199/12 98/100/12 99/101/12 00/102/12 01/103/12 02/104/11 eco no m ic g ro u p w o rld -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 88/190/12 89/191/12 90/192/12 91/193/12 92/194/12 93/195/12 94/196/12 95/197/12 96/198/12 97/199/12 98/100/12 99/101/12 00/102/12 01/103/12 02/104/11 eco n om ic g rou p w o rld 138 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 slightly lower than the world beta, suggesting that the influence of the former cannot be overlooked. the five markets have different reactions to the financial crisis. while malaysia and indonesia markets saw a drop in exposure to the world risk, the magnitude of the world beta for the other three markets has increased during around the financial crisis period. these three markets re-anchored themselves to the global market but returned to the pre-crisis position after 1999. the case of malaysia, however, is different. the exposure to the economic grouping risk has dropped since the crisis. this supports the evidence provided by goh et al. (2005) that degree of exogeneity of the malaysia market has increased within the asean group since the implementation of the capital control policy by the malaysian government in the last quarter of 1998. they show a reduction in the contemporaneous movements between the malaysia and the other four asean stock markets since the crisis. 4. conclusion this article tested different specifications of the icapm model and proposed the economic grouping factor as an additional variable in the asset pricing. using data on five stock markets of asean, the importance of exposure to systematic risks in this economic group is shown. we found that the economic grouping factor, if included, increases the explanatory power of the icapm model. some evidence is found that the exposure to world and regional risks has reduced in importance when the economic grouping risks are taken into account. this article also shows that the pricing mechanism is not stable over time. the singapore market, which is perhaps most developed market, exhibited relatively stable pricing behaviour compared to any of the other four other markets (indonesia, malaysia, philippines and thailand). the exposure to world risk is generally higher in the singapore market, but the other four markets have higher exposure to the economic grouping risk. however, the effect of the economic grouping factor on the singapore market is rather sizeable and not negligible. given that exposure to economic grouping behaviour is not be neglected in international asset pricing model, the findings offer some explanation for segmentation in emerging markets. the higher exposure to movement in returns of the economic group comes together with the reduction in the impact of global market movements on the individual markets, and hence there is lower degree of integration into the world market. author statement: the submitting author is kim-leng goh: e-mail: klgoh@ um.edu.my. author statement: the submitting author is kim-leng goh: email: klgoh@um.edu.my at the university of malaya. the authors express their gratitude for the comments of the reviewer and the assistance of the editors to expedite the copy editing of the paper. the authors are jointly responsible for any remaining errors. the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 139 references akdogan, h. (1992). “behaviour of systematic risk in a regionally integrated model for stock prices”, economics letters, vol 39: 213-216. bekaert, k.g., harvey, c.r. and ng, a. (2005). “market integration and contagion”, journal of business, vol 78: 39-69. bollerslev, t. (1986). “generalized autoregressive conditional heteroskedasticity”, journal of econometrics, vol 31: 307-327. bollerslev, t. and wooldridge, j.m., (1992). “quasi-maximum likelihood estimation and inference in dynamic models with time varying covariances”, econometric reviews, vol 11: 143-172. brown, r.l., durbin, j. and evans, j. m., (1975). “techniques for testing the constancy of regression relationships over time”, journal of the royal statistical society, series b, vol 37: 149-192. campbell, j.y. and hamao, y., (1992). “predictable stock returns in the united states and japan: a study of long-term capital market integration”, journal of finance, vol 47: 43-69. chen g.m., firth, m. and rui, o.m., (2002). “stock market linkages: evidence from latin america”, journal of banking and finance, vol 26: 11131141. davidson, s., faff, r. and hillier, d., (2003). “gold factor exposures in international asset pricing”, journal of international financial markets, institutions & money, vol 13: 271-289. engle, r.f. and ng, v.k., (1993). “measuring and testing the impact of news on volatility”, journal of finance, vol 48: 1749-1778. frankel, j.a. and wei, s.j., (1998). “open regionalism in a world of continental trade blocs”, imf working paper, no. wp/98/10. frankel, j.a., stein, e. and wei, s.j., (1995). “trading blocs and the americas: the natural, the unnatural, and the super-natural”, journal of development economics, vol 47 (no 1): 61-96. fratzscher, m., (2002). “financial market integration in europe: on the effects of emu on stock markets”, international journal of finance and economics, vol 7: 165-193. goh, k.l., wong, y.c. and kok, k.l., (2005). “financial crisis and intertemporal linkages across the asean-5 stock markets”, review of quantitative finance and accounting, vol 24: 359-377. johnson, r. and soenen, l., (1993). “stock market reaction to ec economic and monetary intention”, european management journal, vol 11 (no 1): 85-92. johnson, r. and soenen, l., (2003). “economic integration and stock market comovement in the americas”, journal of multinational financial management, vol 13: 85-100. johnson, r., lindvall, j. and soenen, l., (1994). “ec economic and monetary integration: implications for european equity investors”, european management journal, vol 12 (no 1): 94-101. 140 the international journal of banking and finance, 2008/09 vol. 6. number 1: 2008: 117-140 lessard, d.r., (1973). “international portfolio diversification: a multivariate analysis for a group of latin american countries”, journal of finance, vol 28: 619-633. rillo, a.d., (2004). “trade and financial integration: is there a link?”, asean one, november 7-8. ripley, d.m., (1973). “systematic elements in the relationships of national stock market indices”, review of economics and statistics, vol 55: 356-361. soydemir, g., (2000). “international transmission mechanism of stock market movements: evidence from emerging equity markets”, journal of forecasting, vol 19: 149-176. international journal of banking and finance 9-1-2008 international asset pricing models: the case of asean stock markets chee-wooi hooy kim-leng goh recommended citation cetak a4 ijbf6(2)09.indd the international journal of banking and finance, 2008/09 vol. 6. number 2: 2009: 61-75 61 ijbf monetary integration in east asia: why does it take so long? mariusz k. krawczyk ryukoku university, japan abstract the launch of the economic and monetary union in europe and the 1997 fi nancial crisis that underscored the disadvantages of currently employed exchange rate regimes raised questions about the feasibility of a similar monetary unifi cation project for east asia. being one of the most dynamically growing regions in the world, east asia has the potential for a successful implementation of a monetary union. the paper examines why, despite substantial political emphasis being placed on the issue of monetary integration, the progress to date has been slower than could be expected. the major fi nding is that, although east asia may actually benefi t from establishing its monetary union in the long run, a specifi c political culture that prevails in the region and misconceptions about the sequencing of the process prevent the east asian monetary union from materialising. possible short and mid-term policy solutions follow. keywords: monetary union, east asia, sequencing monetary reforms, monetary unifi cation jel classifi cations: e42, e52 1. introduction east asia, one of the most dynamically growing regions in the world, has long been thought to be a potential candidate for a regional monetary union. amid the successful launch of the european monetary union, the debate on the issue of regional monetary integration (or at least close monetary cooperation) in east asia heated up in the aftermath of the fi nancial crisis that hit the region a decade ago in july 1997.1 1 for instance williamson (2000), berben et al. (2002), kawai and takagi (2002), ogawa (2002), kawai and motonishi (2004), ogawa and kawasaki (2006), ito and park (2004), zhang et al. (2004), ogawa and ito (2002), kwack (2004), chan-lee (2005), kim (2004) and others. ht tp :// ijb f.u um .e du .m y 62 monetary integration in east asia: why does it take so long?: 61-75 the crisis underscored the danger of running fi xed exchange rates (or managed exchange rates), with independent monetary policy and open capital accounts, and resulted in a growing awareness of the importance of regional monetary cooperation in east asia. there is now a strong perception that, had an effective regional monetary cooperative institutional framework been implemented, the crisis itself might have been prevented or at least mitigated and that the benefi ts of the intra-regional cooperation could not be fully accrued without a well-established monetary cooperative framework. accordingly, a group of east asian countries2 engaged in a series of appropriate initiatives that aimed at enhancing the region’s monetary cooperation. yet in spite of high priority given to the issue of monetary unifi cation at least during high-level meetings of regional bodies, east asia’s progress towards its common currency seems to be falling short of expectations. furthermore, there is evidence showing that, after the immediate post-crisis initial switching to fl oating rates, the east asian countries have gradually returned to the de facto dollar peg.3 looking for answers about the reasons for slower than expected progress towards monetary unifi cation in east asia, this paper examines three possible explanations. first, east asia may really not need the monetary union after all. second, there may be the lack of political will to implement the agreed policies. third, east asia may be following its original path to monetary unifi cation that need not necessarily be identical to the approaches adopted in europe or in the persian gulf area. the rest of the paper is organized as follows. the second section presents arguments in favour and against alternative exchange rate regimes for east asia. next, the following section questions the desirability of the region’s monetary union using a basic optimum currency area framework. the main political initiatives towards the monetary union in east asia are presented in the fourth section. section fi ve argues that, it is mainly asia’s original path to monetary unifi cation that is to blame for the slower than expected progress. finally, the concluding section, after having answered the paper’s main question, provides a set of necessary, in this author’s opinion, steps that may lead to achieving monetary unifi cation in the region. 2. alternative exchange rate regimes for east asia the debate whether the exchange rate should be fi xed or fl exible has remained unresolved ever since milton friedman began championing the cause for exchange rate fl exibility. advocates of both corner solutions can be found on either end of the economic theory spectrum. a keynesian economist would argue for instance that a fl exible exchange rate could absorb external and internal shocks without affecting domestic price and wage levels, while his support for 2 for the needs of this paper, i refer to the ten asean countries together with china, japan and (the republic of) korea (known as asean+3) and hong kong and taiwan as east asia. 3 for instance fukuda (2006), ogawa (2002), ogawa and kawasaki (2006). ht tp :// ijb f.u um .e du .m y monetary integration in east asia: why does it take so long?: 61-75 63 a fi xed exchange rate regime would be based on the belief that it provides a necessary protection from speculative attacks (that result from imperfect foreign exchange markets): the later being a defence for his devising and supporting the bretton woods. on the other hand, a market economy believer would argue that fl exible exchange rates are a necessary condition for retaining independence of domestic monetary policies under free capital movement (that is indispensable for effi cient functioning of a market). yet fi xed exchange rates could, in his reasoning, provide a nominal anchor necessary for stabilising fl exible domestic prices and wages. the debate is far from being resolved and in practice a wide range of intermediate regimes has been in place. the asian currency crisis constitutes a milestone in the debate on the optimal exchange rate regimes for east asia. most of the region had adopted a de facto dollar peg before the 1997 crisis (despite offi cial claim of managed fl oat or basket peg). undoubtedly, the fi xed exchange rates helped to encourage capital infl ows that contributed to the excellent performance of east asian economies before 1997. yet even then the vulnerabilities of the system were quite obvious; the high interest rates necessary for stabilising rapidly growing economies were encouraging more portfolio investment that in turn contributed to the overheating in the region (the impossibility to maintain fi xed a exchange rate, capital mobility and independent monetary policy at the same time). after 1997, while leaving the debate on two corner solutions still largely unresolved, the focus of the debate here has shifted to costs and benefi ts of three alternative exchange rate regimes, a return to the dollar peg, a currency basket regime and a monetary union for east asia. aside from the return to the de facto dollar peg after the crisis, some analysts have argued in favour of explicit adoption of a dollar peg (hong kong chose to do so ahead of the uncharted days post-merger with china).4 for reasons ranging from transaction costs, direction of trade fl ows to historical conditions at least some countries of east asia could be considered as a part of a “greater dollar zone” (with canada, mexico and parts of latin america being a “natural dollar zone”). as fukuda (2006) convincingly shows, for reasons of history and expectation, invoicing in dollars dominated in thai (85.7 percent in 2001, down from 92.0 percent in 1997) and korean (87.43 percent in 2001 down from 89.21 percent in 1997) transactions in spite of the fact that their trade with the united states amounted to only 20.3 and 20.9 percent respectively (against 49 percent of east asia share in thailand and 44.3 percent share of east asia in korea’s trade). history, lower transaction costs, and a nominal anchor for stabilisation policies constitute the main arguments in favour of a dollar peg. on the other hand, the arguments on the loss of independent monetary policy and the possibility of asymmetric shocks still hold (table 1). 4 for instance mckinnon (2004) or barro (2004). ht tp :// ijb f.u um .e du .m y 64 monetary integration in east asia: why does it take so long?: 61-75 table 1: the dollar peg in east asia benefi ts costs an effi cient medium of exchange (lower transaction costs) if a country does not satisfy the oca criteria, the economy would suffer from asymmetric shocks a nominal anchor (low infl ation) it can be carried out unilaterally by a small economy. regional cooperation is possible if all economies peg to the us dollar the entire region becomes dependent on the third party (frb) monetary policy decisions currency attack would be weaker and contagion would be contained if japan joined (china is in) the entire region would be unifi ed optimally source: from loosely based on wyplosz (2002) and ito and park (2004). although being the second best solution, the de facto dollar peg has returned in practice as the fi rst choice of an exchange rate regime in east asia. but much more attention has been paid to a currency basket peg and to a currency union as optimal forms of monetary unifi cation in the region. many academicians would prefer the former while the policy makers in the regions apparently favour the latter one. one of main arguments in favour of a basket regime is its ability to stabilise real effective exchange rates (as such a basket would contain currencies of major trading partners). regarding quite high trade weight with japan and the european union, the argument that a basket, apart from dollar, containing the euro and japanese yen could mitigate the 1997 crisis looks fairly plausible. another argument in favour of a currency basket peg is its moderating infl uence on capital fl ows. comparing to a single currency peg, a basket peg contains a higher exchange rate risk and therefore has a depressing infl uence on capital fl ows. a basket peg provides an economy with some stabilising anchor (compared to a fl exible exchange rate) and with more fl exibility than would happen under a fi xed exchange rate regime.5 there are, however, also some diffi culties associated with a currency basket peg. first, complicated basket calculations would certainly result in a loss of transparency. given diffi culties in immediate evaluating the actions the monetary authorities undertake, the credibility of the basket peg may be adversely affected. needless to say, interventions within the currency basket 5 while the idea of the currency basket for east asia that comprises three major world currencies has been promoted mainly in japan, it does not seem to have been given a lot of consideration in other countries of the region. instead, an alternative of a basket that would be composed of regional currencies has been suggested. i am grateful to woosik moon of seoul national university for pointing this issue in an earlier version of the paper. ht tp :// ijb f.u um .e du .m y monetary integration in east asia: why does it take so long?: 61-75 65 regime may be technically much more complex than those that take place under a single currency peg. second, since each country’s optimal basket is different, adopting a currency basket peg may result in diffi culties in policy coordination in the region.6 table 2 summarises the main costs and benefi ts of a currency basket regime. table 2: a currency basket regime benefi ts costs fewer fl uctuations in trade balance no currency crisis prevention moderate capital fl ows loss of transparency (due to complicated basket calculations) some fl exibility in managing external shocks loss of fl exibility (compared to a free fl oat) gives some anchor to prevent misalignment (compared to a free fl oat) loss of nominal anchor (compared to a fi xed exchange rate), hence higher risk premium allows for joint appreciation and depreciation in the region policy coordination problems source: from loosely based on wyplosz (2002), ito and park (2004), and ogawa et al. (2004). the successful launch of the economic and monetary union in europe has provided an additional argument for those who support a similar project in east asia. some, especially those involved in policy making, seem to take for granted that east asia will have its own version of a common currency introduced in the (relatively) near future.7 the advantages and drawbacks of a monetary union are relatively well known (table 3). a monetary union arrangement provides exchange rate stability necessary for promotion of trade and investment within the region while, at the same time, avoiding dependency on a third party monetary policy (one of the drawbacks of the dollar peg). here the monetary policy would be conducted by an independent (regional) central bank. a monetary union also provides a fi rm anchor both in a monetary and institutional sense (peer pressure for meeting policy coordination goals if a framework such as the europe’s growth and stability pact has been established). one can also expect that, once in place, a monetary union may deepen the economic correlation within the region. the loss of an independent monetary policy for each member country has been the most pronounced cost of the monetary union. however, long before the debate on currency union emerged, the proponents of currency board arrangement had argued that the loss of an independent monetary policy measures associated with establishing a currency board was actually overstated. it happens because the growing interdependence of the world economy does not allow for a completely independent monetary 6 ogawa and ito (2002) show that the probability of a sub-optimal outcome is not negligible under such circumstances. 7 for instance chino (2004) or asahi shinbun (2005). ht tp :// ijb f.u um .e du .m y 66 monetary integration in east asia: why does it take so long?: 61-75 policy and, in addition, even the independence of a monetary policy does not warrant its optimality.8 moreover, as wyplosz (2002) argues, there is a possibility that a common monetary policy performed by a regional central bank may not vary signifi cantly from each member country’s optimal monetary policy. needless to say, in order to implement a monetary union successfully, the region needs to meet the optimum currency area criteria. table 3: a monetary union benefi ts costs exchange rate stability (trade and investment promotion) loss of independent monetary policy a (regional) central bank independence peer pressure for macroeconomic policy coordination asymmetric shocks may cause problems without suffi cient capital mobility and fi scal transfers source: from loosely based on wyplosz (2002), and ito and park (2004). most of research done on the issue of monetary integration in east asia seems to be quite in favour of establishing a monetary union in the region. even critical papers (chan-lee, 2005 or kim, 2004) predict that a monetary unifi cation of some sort will eventually emerge in the region over a longer time horizon. yet in spite of the favourable comments and various political initiatives, the monetary integration in east asia still seems to be decades away. the following sections will attempt to search for reasons behind the slower than expected progress. 3. is a monetary union desirable for east asia? the benefi ts of monetary union outweigh its costs when the union’s partners are trading intensively with each other, when their economies business cycles are correlated and they suffer from symmetric shocks, when there is high degree of policy coordination in place, production factors mobility between member states is high, fi nancial markets well developed and so on. in other words, when the entire region constitutes an optimum currency area (oca). this section attempts to address briefl y this issue. most empirical evidence seems to suggest that the region does indeed satisfy the oca conditions, more or less. trade openness indices for east asian economies (defi ned as a share of total trade in a country’s gdp) show that, with the notable exceptions of myanmar and japan, the economies of the region are as open as the eu-15 and more open than the eu economies were in 1990 (before the maastricht treaty, table 4). 8 for instance hanke and schuler (1993). ht tp :// ijb f.u um .e du .m y monetary integration in east asia: why does it take so long?: 61-75 67 east asian economies do similarly well on account of the intra-regional trade. for instance kawai and motonishi (2005) as well as yano and van anh (2006) show that the intra-regional trade in east asia has already reached the level comparable to the pre-maastricht european union (fig.1). the caveat is that the asean economies do not trade much between themselves; in 2003 only 24 percent of their trade was reaching other asean countries. the same applies to the asian new industrialised economies (nies; hong kong, korea, singapore and taiwan); only 16,1 percent of their trade was directed to the region in 2003. but the trade between the asean, the nies and china already accounted for 44.1 percent of their total trade in 2003. when the trade with japan is included the intra-regional trade amounts to 54 percent of the region’s total trade. figure 1 (based on kawai and motonishi, 2005) shows that this is more than the intra-regional trade within nafta and it is comparable to the intensity of the intra-regional trade in the european union in 1990 (before the maastricht treaty). even more support on the oca criteria comes from the estimated correlations of real sector macroeconomic variables (real gdp, real personal consumption, and real gross fi xed capital formation) as well as fi nancial variables (real money supply, real stock price, and real effective exchange rates) and price variables (gdp defl ator, consumer price index, and wholesale price index). kawai and motonishi (2005) estimations for the period 1983-2000 as well as yano and van anh (2006) results for 1980-2002, show that east asian economies do on this account no worse than european countries. a similar conclusion emerges from estimations of demand and supply shock correlations for east asia, eu15 and nafta. the results9 show a similar level of correlation for all groups of countries. given the specifi c character of the criteria in question (i.e. once a monetary union has been formed, economic integration will deepen and the degree of asymmetry of shocks will decline) the results support the plausibility of an east asian monetary union. some problems arise when the labour mobility criterion is tested. although limited availability of data constitutes a serious diffi culty, yano and van anh (2006) conclude that the labour mobility in the region is rather low. for major economies (japan, korea, taiwan, and thailand), the approximate stock of migrant workers in 2000 did not exceed 1-2 percent of the labour force in spite of enormous income disparities between labour exporting and labour importing countries (ppp adjusted income per capita in japan, singapore, and hong kong is 8-9 times higher than that in indonesia or philippines). only hong kong (5-7 percent), malaysia (10-15 percent), and singapore (25-27 percent) have shares of migrant workers higher than average. in the presence of high capital mobility (except for malaysia and china) one may argue that, similarly to europe, labour mobility has largely been replaced with trade and capital fl ows. the quality of institutional framework constitutes another serious diffi culty for east asia as shown in chan-lee (2005). in spite of a relatively high level of capital account openness, the quality of the fi nancial sector prevents capital fl ows 9 kawai and motonishi (2005), kwack (2004), yano and van anh (2006) and zhang et al. (2004) ht tp :// ijb f.u um .e du .m y 68 monetary integration in east asia: why does it take so long?: 61-75 from being effi ciently utilised. table 5 (based on data from chan-lee, 2005) presents, apart from the capital account openness, the assessment of the quality of fi nancial systems (scale 1 to 10) and the perception of business environment (here the reversed price waterhouse and coopers’ “business opacity” index). while some of the east asian countries (hong kong, japan, singapore or taiwan) managed to establish relatively sound fi nancial systems, much is to be done to match the level of the eu “southern periphery” (greece, portugal, spain) not to mention the united states. the problems are aggravated by phenomena common for most of emerging market economies such as insuffi cient property rights protection and contract enforcement, poor transparency, low level of corporate governance and widespread corruption. table 4: trade openness in east asia and the european union (total trade as a percentage of gdp) east asia 1990 2002 eu-15 1990 2002 brunei china cambodia hong kong indonesia japan (rep. of) korea laos malaysia myanmar philippines singapore taiwan thailand vietnam 88.5 24.2 17.4 218.3 42.2 19.8 51.4 30.5 124.1 3.1 46.0 300.8 74.6 61.4 54.1 113.4a) 49.1 101.8 248.8 53.6 21.1 65.2 40.9 177.2 1.0a) 87.7 273.7 83.5 98.1 115.0 austria belgium denmark finland france germany greece ireland italy luxembourg netherlands portugal spain sweden united kingdom 56.4 111.1b) 50.3 39.1 37.0 50.0 33.6 93.7 31.9 129.4 87.2 60.2 29.0 46.8 41.2 69.7 169.1 61.0 59.3 43.1 55.4 31.1 114.2 41.8 95.4 99.1 52.4 43.6 61.1 39.1 a) 2001. b) 1993. source: kawai and motonishi (2005). figure 1: intra-regional trade (as percent of total trade) ht tp :// ijb f.u um .e du .m y monetary integration in east asia: why does it take so long?: 61-75 69 table 5: quality of financial markets quality of fi nancial systems (1-10), 2003 business environment (1-10), 2001 capital account openness (1-4), 1999 china hong kong indonesia japan (rep. of) korea malaysia philippines singapore taiwan thailand 2.7 6.9 1.9 6.9 5.4 5.7 4.3 7.1 6.6 3.9 1.3 5.5 2.5 4.0 2.7 n. a. n. a. 7.1 3.9 3.3 1.5 4.0 3.5 3.5 3.5 3.0 3.0 4.0 2.5 2.5 greece portugal spain 3.8 4.6 6.2 4.3 n. a. n. a. 3.0 3.0 3.0 united states 9.1 6.4 4.0 n. a. – data not available. source: adapted from chan-lee (2005). except for hong kong and singapore, the perception of the business environment in east asia is not encouraging. poor condition of the region’s fi nancial markets damages east asia’s image as an optimum currency area. to summarise, on most of accounts east asia (or at least its parts) seems to be satisfying the oca criteria. even if not matching the level of the eu monetary union member countries, east asia is close to the levels the eu member states had reached before they started the process of monetary unifi cation (i.e. before the maastricht treaty). however, the disparities between the east asian countries are much bigger than in europe and much is to be done in the area of institution building. nevertheless the answer to the question whether east asia needs its monetary union seems to be largely positive (at least for parts of the region). 4. a lack of political will? this section attempts to address the issue of whether it is the lack of political will to implement a monetary integration that is responsible for the slow progress in achieving the goal. there may be no straightforward answer to this problem. on the one hand, there is abundance of political initiative towards deepening integration in the region. asean+3 heads of state have met annually since 1997. even before their fi rst meeting took place, the manila framework group, a regional surveillance body had been established in november, 1997. other initiatives include a bilateral swap arrangement (chiang mai initiative) in may, 2000, establishing an inter-governmental east asia study group or the non-government east asia vision group (both agreed at the november, 2000 ht tp :// ijb f.u um .e du .m y 70 monetary integration in east asia: why does it take so long?: 61-75 meeting in singapore), the asian bond initiative of 2003 and so on. a series of bilateral negotiations has produced a handful of bilateral free trade agreements between countries of the region. despite being often highly publicised, many of those initiatives never develop beyond the stage of a study group or a political statement: see table 6. even the most successful one, the chiang mai initiative still does not seem to be fulfi lling its entire potential. verbal support and statements of good will do not change the fact that, in contrast to europe, east asian governments seem to be very reluctant to delegate parts of their prerogatives to supranational institutions. table 6: asean+3 declarations on economic and monetary cooperation in east asia venue and date agenda kuala lumpur, december 1997 exchange of views on regional currency problems amid the asian currency crisis hanoi, december 1998 deepening an economic cooperation aimed at economic recovery in the region manila, november 1999 the fi rst offi cial joint statement by the asean+3 summit singapore, november 2000 the assessment of the chiang mai initiative agreement. establishing the non-government east asia vision group (eavg) and its intergovernmental counterpart, the east asian study group (easg) for working-out guiding principles of economic and monetary cooperation in the region brunei, november 2001 accepting the eavg report. korea’s president kim dae jung announces his country’s readiness to establish an east asian free trade area pnom-penh, november 2002 accepting the easg report on establishing a free trade area in east asia bali, october 2003 accepting a strategy for establishment of the asean community by 2020 (the declaration of the 2nd asean accord) vientiane, november 2004 establishing an offi cial bureau of the asean+3 in the asean structures (asean+3 unit) kuala lumpur, december 2005 renewed commitment towards narrowing development gap between member states cebu, january 2007 commitment towards accelerating the establishment of the asean community by 2015 source: compilation based on the asean internet home page (www.asean.org actual as of may 2007). ht tp :// ijb f.u um .e du .m y monetary integration in east asia: why does it take so long?: 61-75 71 europeans’ willingness to do so has been the result of their historical experience of two world wars that generated a desire to establish a political and economic institutional framework, and able to prevent a recurrence of confl ict. there is apparently a lack of such consensus among east asian nations. the lack of shared beliefs in political democracy and the presence of unresolved historical disputes result more often in emotional confrontations9 rather than in compromises based on principles of solidarity and confl ict resolution. the resultant distrust translates then into a characteristic pattern of many low profi le initiatives. on the other hand, after the bitter controversy surrounding the imf standby agreements with east asian economies in the aftermath of the 1997 crisis no self-respecting asian leader would ignore the issue of deepening the regional integration here. east asia, one of the fastest growing regions of the world economy, is highly under-represented in various international bodies (the imf, g-8, united nations and so on). there is a growing frustration over the disparity between the region’s own growing economic might and its lack of political clout. awareness of one’s own helplessness can also be an important driving force behind cooperative regional efforts. considering the number of undertaken integration initiatives, not necessarily the lack of the real political will, but rather the specifi c political culture of distrust and lack of commonly shared cultural and political set of values contributes to slower than expected progress of the integration process. the procrastinated negotiations on the free trade agreement between japan and the republic of korea may illustrate the point. 5. is east asia following its own specifi c path towards monetary union? even if east asia is indeed on its own path towards monetary union, the path is very different from the one the european monetary union has followed. after having achieved its original goal of establishing a customs union the european countries undertook series of attempts to stabilise the intra-european exchange rate (dating back to the pre-ems days). then, the leading monetary body (germany’s monetary authority) took over setting the pace of the region’s monetary policy and, after relaxing capital movement controls (that did not happen without diffi culties) europe’s common currency was fi nally established. east asia’s path towards monetary integration differs signifi cantly from the european one. first, east asia as a region has not pursued any formal trade liberalisation. instead the region’s trade has been regulated on the basis of bilateral liberalisation agreements (notably agreements on free trade between the largest countries of the region, japan, korea, taiwan, and china have yet to be 9 the latest examples could include angry demonstrations and violent reactions to the statements of public fi gures regarding historical territorial suzerainty between korea and china or thailand and cambodia, as well as public outrage in china and korea against japanese prime ministers’ visits to the controversial yasukuni shrine, not to mention the real still unresolved territorial disputes between japan, korea, and china. ht tp :// ijb f.u um .e du .m y 72 monetary integration in east asia: why does it take so long?: 61-75 achieved). primarily for that reason the monetary integration in east asia stands for a policy goal itself, while in the case of europe it has been a mere economic policy tool used for further deepening of economic and political integration. a second important difference is the fact that, contrary to the european experience, east asian countries (with the notable exception of china) had liberalised their capital accounts even before their fi nancial markets were well developed. although the capital controls relaxation was arguably necessary for achieving high growth rate at the end of the last century,10 capital mobility makes it diffi cult to sustain a fi xed exchange rate regime (let it be either dollar or currency basket peg) as a precondition for establishing a monetary union in the region. third, there appears to be no obvious candidate to assume the role germany has performed in the european union at the end of 1980s and in the 1990s. the position of japan, once undisputed candidate for a regional leadership, has been seriously damaged by its prolonged economic depression. as for china, with its low level of economic development, internal discrepancies.11 under developed fi nancial systems, not convertible currency, and capital account restrictions, it is hard to imagine it taking over a regional economic leadership in the near future. fourth, the east asian political establishment seems to have taken it for granted that the economic integration goes precisely along the balassa (1962) fi ve-consecutive-steps pattern. it requires fi rst establishing a free trade area (fta) in order to reach a customs union level. the latter can later develop into a common market and a full-fl edged economic union can follow. finally, the integrating nations can establish a complete political and economic union.12 revolutionary as it was half a century ago, the theory neglects the fact that it requires a common policy to reach level two of customs union while the fta does not require any common policy at all (simply, the abolishing of customs should suffi ce for the fta to work). the european union has never been a free trade association. already in 1957 it aimed at entering the balassa (1962) pattern at stage two. accordingly, given the fact it was not the european free trade association (efta) that produced europe’s common currency, forming the network of ftas around the region will not eventually lead to a monetary union as that would require a common policy, something the fta actually misses. 6. conclusion: what is wrong with east asia? the east asian path to monetary unifi cation suffers from the adverse effects of its reversed order (capital account liberalisation ahead of monetary integration) 10 another way of reasoning would be that they had to liberalise their capital accounts precisely because their capital markets were not developed enough to stem capital infl ows necessary for high growth rates. 11 it is a good question whether we can really think about a single unifi ed china in economic terms; systemic and income differences between the regions (hong kong, shanghai, beijing, the island of hainan, manchuria, tibet, inner mongolia, western provinces etc) are so big that one can think about the country as a group of economic entities (forcibly) unifi ed in political terms as for instance the british empire once used to be. 12 it is precisely as the line of argument runs in nakashima (2006). ht tp :// ijb f.u um .e du .m y monetary integration in east asia: why does it take so long?: 61-75 73 that creates a confl ict between monetary cooperation (understood by the region’s opinion makers as a capital movement liberalisation) and exchange rate coordination (ranging between a single currency or currency basket peg and currency union altogether). also the misconception about the sequence of economic integration (i.e. taking the balassa, 1962 theory prima facie) has taken its toll on the speed of monetary integration in the region (or may take in the near future). together with the lack of regional leadership (the role germany has played in european processes), specifi c political environment that precludes bold political initiatives, and, to put it straightforward, lack of political vision for the region’s future (i.e. political union) probably rules out any monetary union in east asia in the near future. however that does not mean nothing can be done in order to deepen monetary integration in the region. in my opinion, the following measures could be helpful for establishing the monetary union in east asia in the midand long run. first, political energy should switch from concluding bilateral fta towards establishing at least a customs union in the region.13 then, building sound fi nancial markets should follow. the character of the optimum currency area theory allows for almost discretional interpretation of results. but, let aside mundell’s original criteria, it is the soundness of the fi nancial system that spells the ability to withstand the adverse shocks that are likely to accompany the monetary unifi cation project. at their current condition east asian fi nancial markets are as weak as they were at the onset of the 1997 currency crisis and probably could not withstand any similar challenge. third, through strengthening the chiang mai initiative, the joint surveillance of economic policies and coordination of exchange rate policies could ultimately lead to establishing a decision making framework (or institutions) based on the principle of mutual consent. only from that point can one realistically think about the east asian monetary unifi cation process. besides, there is the problem of what to do with china. if the country is going to assume leadership in the region, then a phased opening of its capital markets has to take place at some time. but, analogously to the russia’s relations with the eu, there is a well-grounded doubt about the wisdom of creating such an unbalanced integration body. the monetary unifi cation in east asia is not a political illusion. however, it is not an immediate necessity either. as the economic and monetary union in europe has taken half a century to materialise, also here, after having addressed its specifi c problems, fi fty years from now, east asia may be well on its way to a monetary (and economic) union. author statement: i am grateful to sadayoshi takaya of kansai university, junji yano of hiroshima university and other participants of the japan society for monetary economics international finance division meeting held on april 7, 2007 at kansai university in osaka, as well as to thomas wright of ryukoku university for their helpful comments. the usual disclaimer applies. 13 or parts of it if political considerations allow for excluding the weak elements of the system ht tp :// ijb f.u um .e du .m y 74 monetary integration in east asia: why does it take so long?: 61-75 references asahi, s. (2005, november 5). asian currency unit sets stage for integration. balassa, b. (1962). the theory of economic integration. london: george allen & unwin ltd. berben, r., jean, m.b., ekniti, n., kanit, s., pisit, p., and piyaporn s. (2002). requirements for successful currency regimes: the dutch and thai experiences. report on the study group on exchange rate regimes for asia (kobe research project). chan l., and james, h. (2005). the illusive quest for an asian common currency: economic mirage or realpolitik? cnaec research series no. 05-05. korea institute for international economic policy. chino, t. (2004, june). consider a single asian currency. asian wall street journal. fukuda, s. (2006, july 6). on the determinants of exporters’ currency pricing: history vs. expectations. paper presented at the 81st annual conference of the western economic association international, san diego, usa. hanke, s., and kurt, s. (1993). currency boards and currency convertibility. cato journal, 12(3). ito, t., and yung, c. p. (2004). exchange rate regimes in east asia. in monetary and fi nancial integration in east asia. new york, palgrave: asian development bank. kawai, m., and taizo, m. (2005, february 18). is east asia an optimum currency area? paper presented at the rokko forum on international finance, kobe university, japan. kawai, m., and shinji, t. (2002, may 29). towards regional monetary cooperation in east asia. paper presented at the hamburg institute of international economics conference on east asian monetary and financial cooperation. kim, j.s. (2004, december 2-3). is the pegged regime a feasible alternative in east asia. paper presented at the conference on financial interdependence and exchange rate regimes at japan ministry of finance policy research institute. kwack, s. y. (2004). an optimum currency area in east asia: feasibility, coordination and leadership. journal of asian economics, 15(1), 153169. montiel, p. j. (2004). an overview of monetary and fi nancial integration in east asia. in monetary and fi nancial integration in east asia. new york, palgrave: asian development bank. murase, t. (2007). currency and monetary cooperation in east asia (東アジ アの通過・金融協力－higashi ajia no tsuuka-kinyuu kyouryoku). tokyo: keiso shobo (in japanese). nakashima, t. (2006). the necessity of east asian community (東アジア共同 体の「必要性」－higashi ajia kyoudoutai no hitsuyousei), erina discussion paper no. 0605 (in japanese). economic research institute for northeast asia. ht tp :// ijb f.u um .e du .m y monetary integration in east asia: why does it take so long?: 61-75 75 ogawa, e. (2002). economic interdependence and international coordination in east asia. in report on the study group on exchange rate regimes for asia (kobe research project). ogawa, e., and takatoshi, i. (2002). on the desirability of a regional basket currency arrangement. journal of the japanese and international economies, 16(3), 317-334. ogawa, e., takatoshi, i., and yuri, n. s. (2004). costs, benefi ts, and constraints of the currency basket regime for east asia. in monetary and fi nancial integration in east asia. new york, palgrave: asian development bank. ogawa, e., and kentaro k. (2006, july 6). adopting a common currency basket arrangement into the “asean plus three”. paper presented at the 81st annual conference of the western economic association international, san diego, usa. williamson, j. (2000). exchange rate regimes for emerging markets: reviving the intermediate option. policy analyses in international economics, 60. institute for international economics. dc: washington. wyplosz, c. (2002). fiscal policy: institutions vs. rules. (cepr discussion paper no. 3238). yano, j., and tran, t. v. a. (2006). on asian monetary union. hiroshima university, mimeo. zhang, z., kiyotaka, s., and michael, mc. (2004). is a monetary union feasible for east asia? applied economics, 36(10), 1031-1043. ht tp :// ijb f.u um .e du .m y ht tp :// ijb f.u um .e du .m y how to cite this article: kuah, y. c., zakaria, z., & choong, c. k. (2020). the impact of manager characteristic on capital structure in malaysian manufacturing smes. international journal of banking and finance, 15(1), 21-37. https://doi.org/10.32890/ijbf2020.15.1.9930 the impact of managerial characteristics on capital structure in malaysian manufacturing smes 1kuah yoke chin 2zuriawati zakaria 3choong chee keong faculty of business and finance, universiti tunku abdul rahman, malaysia 1corresponding author: kuahyc@utar.edu.my; 2zuriawatiz@utar.edu.my; 3choongck@utar.edu.my a r t i c l e i n f o article history: received 19 august 2019 revised 14 november 2019 accepted 26 november 2019 published 31 march 2020 keywords: managerial characteristics, manufacturing, sme, financing. jel code: d24 a b s t r a c t access to capital is a critical factor in stimulating small business creation and growth especially in developing companies. the failure of small business entities in securing the needed capital would entail them remaining small and limit their ability to create goods, services, and innovations in the marketplace, including jobs. this study focuses on manufacturing smes by examining the influence of managerial characteristics (age, gender, work experience and level of education) on capital structure towards technology improvement. based on a survey of 219 respondents, the results indicated that male managers, managers with a lot of work experience and educated managers preferred internal and external financing. furthermore, internal and external financing also showed positive relationships to improve manufacturing technology the international journal of banking and finance, vol. 15. number 1, 2020: 21-37 21 22 the international journal of banking and finance, vol. 15, no 1, 2020 : 21-37 1. introduction business discontinuation is a crucial aspect of economic dynamics, and entries and exits of businesses are closely connected. given that a large number of smes have been unsuccessful, it is expedient to investigate the causes of poor performance and failure of these firms (arasti, 2011). the reasons for business failure are many and varied, such as limited sources of financing, low productivity and lack of business experience. the majority of studies on business failures focused on the business itself rather than on the owner of the business. it cannot be denied that entrepreneurs and related factors are critical factors that contribute to the failure of smes (arasti, zandi, & talebi, 2012). according to baldwin, bian, dupuy and gellatly (2000), managerial experience and financial management were vital issues that determined the success or failure of a firm. based on survey results, almost half of the canadian insolvent firms became bankrupt due to internal problems caused by managerial inexperience and lack of knowledge in sustaining the business. based on the findings in arasti et al. (2012), the failed business owners and managers were mostly in the middle-aged group and they usually did not succeed in the establishment phase of the business. their businesses did not survive more than three years because of their personal characteristics. management characteristics have contributed largely towards bankruptcy (ooghe & de prijcker, 2008). insufficient and inappropriate skills of management account for the failure of unsuccessful start-up companies. additionally, some managers have experience in management but are without industry-related knowledge and display very authoritarian behaviour that increases the rigidity of the company. apart from management qualities and skills, many personal characteristics strongly affect the performance of a company. an increasing number of scholars (zabri, 2012; borgia & newman, 2012) have begun to question the role played by owners’ or managers’ characteristics and attitude as critical factors in smes. owners or managers have substantial influence on financing decisions in a company (borgia & newman, 2012). performance. this study provides and adds new knowledge to corporate managers to serve as benchmarks in making decisions on company performance. it could also enhance company ability to deal with competitive environments. the impact of managerial characteristics on capital structure in malaysian manufacturing smes : 21-37 23 companies in the same industry have management whose characteristics differ. they employ different capital structure models resulting in diverse outcomes (kyenze, 2014). many researches have discussed the relationship between managers’ characteristics and debt financing, but little has been done in the malaysian context, especially in smes. thus, this study attempts to bridge the gap, given that an analysis of managers’ characteristics is crucial to understanding the capital structure of firms. the purpose of this study is to examine if there is any significant managerial characteristic(s) such as age, gender, level of education or background experience which influences the composition of a firm’s capital structure. next, this study will investigate the effect of internal and external financing on technology improvement. 2. theoretical model the theories suggest that firms decide on a capital structure based on attributes that determine the various costs and benefits associated with debt and equity financing (abor & biekpe, 2009). after the work of modigliani and miller (1958), several theories of capital structure have been developed to explain the optimal capital structure. pecking order theory, trade-off theory and agency theory are among the most popular theories on capital structure. the modigliani and miller theory (m&m) forms the foundation of capital structure theories (kumar & rao, 2015). the assumption of the modigliani and miller theory (1958) is that firms have a particular set of expected cash flows. after selection of a certain proportion of debt and equity to finance a firm’s assets, the cash flow would be allotted among its investors. investors and firms are assumed to have equal access to financial markets, which allows for homemade leverage. therefore, the leverage of a firm has no effect on the market value of the firm (luigi & sorin, 2009). m&m assumes that the market is perfect and has no influence on the value of a firm; investors have homogeneous expectations; it is a tax free economy and there are no transaction costs. however, this theory is inapplicable as it is based on restrictive assumptions and is inconsistent with the real world. investors prefer to buy undervalued shares and sell at higher prices to earn income (salim & yadav, 2012). hence, modigliani and miller (1963) revised their position by incorporating tax benefits as determinants of capital structure. to strengthen this argument, m&m explained that a firm that honors its tax obligations benefits from partially offsetting interest called tax shield in the form of payment of lower taxes. thus, firms are able to maximize their value by employing debt due to tax shield benefits associated with debt use (ahmad, abdullah & roslan, 2012). m&m was criticized due to some weaknesses and 24 the international journal of banking and finance, vol. 15, no 1, 2020 : 21-37 inapplicable assumptions. nevertheless, it provided the foundation for other theories put forth that took into consideration market imperfections. the m&m has been expanded into the pecking order theory and trade-off theory. the pecking order theory (pot) propounded by myers and majluf (1984) argues that firms prioritize their sources of financing and that they usually prefer internal financing. if external financing is needed then debt is chosen and equity would be the last resort. to avoid the underinvestment problem, managers seek to finance new projects using a security that is not undervalued by the market, such as internal funds and riskless debt. this is because a firm has more information than outsiders and information asymmetry investors demand more premiums for information-sensitive securities (e.g. equity). this implies that firms with high levels of profit have a lower leverage ratio as they are capable of financing investment needs through internal sources, and therefore external sources are not necessary. however, when the company is unprofitable, they tend to use debt as external financing when their cash flow is not enough to fulfill investment needs (dwaikat et al., 2014). the second theory is trade-off theory (tot) which implies that leverage has a positive relationship which is contrary to the pecking order theory. it assumes the presence of an optimal capital structure that maximizes shareholder’s wealth and simultaneously minimizes external claims to wealth. it considers the trade-off between benefits of interest tax shield of debt and cost of financial distress (kumar & rao, 2015). the advantage of borrowing allows companies to attain tax shield, that is, a company pays lower tax when it incurs more debt. companies are most likely to use debt financing up to a certain level until the cost of financial distress starts to surface (saarani & shahadan, 2013). a highly leveraged firm with a high debt ratio is always associated with the need for high returns. this is because the firm is exposed to bankruptcy risks if not managed well. this explains why highly leveraged firms require high returns to compensate for the risk. incidentally, interest payment for debt is tax-deductible. an optimal capital structure could therefore be achieved by the firm to enjoy maximum tax benefits. to conclude, trade-off theory gathers the pros and cons of the usage of debt. the usage of debt in capital structure gives a reduction in tax burden; however, an increase in the level of debt comes with the risk of bankruptcy, since the likelihood of bankruptcy rises with the degree of the company’s indebtedness (dwaikat et al., 2014). the last theory is agency cost provided by jensen and meckling (1976) which discusses the conflict of interest between principals (shareholders) and decision makers (agents) of firms (managers, board members, etc.). this conflict stems from the differences in behaviour or decisions by pointing out that the parties (agents and shareholders) often have different goals, and different tolerances towards risk. for instance, some managers tend to guide the firm the impact of managerial characteristics on capital structure in malaysian manufacturing smes : 21-37 25 towards achieving their personal goals rather than maximizing benefits for the shareholders. hence, the main conflict that shareholders face is ensuring that managers (agents) do not invest free cash flow in unprofitable projects (nassar, 2016). in companies with high cash flow and profitability, increase in debts can be used as a toll for reducing the scope of the manager in order that resources of the company may not be wasted as a result of their activities (negasa, 2016). there are some researchers who argue that leverage ratio is high when managerial ownership is low, as agency cost is also low. figure 1. theoretical model (sources: developed for research) 3. manager's characteristics and financing preferences the subject of capital structure has been extensively examined. despite the enormous amount of literature that analyzed the capital structure of companies, studies conducted on this subject related to managerial characteristics are limited (dwaikat et al., 2014). most of the theoretical and empirical studies that addressed these topics concentrated on the large number of listed firms that dispersed ownership structure. it is doubtful that these findings could be applied to firms under the control of large stockholders, especially those owned by families; in particular, small and medium enterprises (smes) which have specific goals to concentrate on their survival in the long run. it is argued that smes have more aversion to risk than listed companies. to prevent the entrance of new stockholders, small companies will employ debt in the capital as leverage. 4 board members, etc.). this conflict stems from the differences in behaviour or decisions by pointing out that the parties (agents and shareholders) often have different goals, and different tolerances towards risk. for instance, some managers tend to guide the firm towards achieving their personal goals rather than maximizing benefits for the shareholders. hence, the main conflict that shareholders face is ensuring that managers (agents) do not invest free cash flow in unprofitable projects (nassar, 2016). in companies with high cash flow and profitability, increase in debts can be used as a toll for reducing the scope of the manager in order that resources of the company may not be wasted as a result of their activities (negasa, 2016). there are some researchers who argue that leverage ratio is high when managerial ownership is low, as agency cost is also low. figure 1. theoretical model sources: developed for research 3. literature review the subject of capital structure has been extensively examined. despite the enormous amount of literature that analyzed the capital structure of companies, studies conducted on this subject related to managerial characteristics are limited (dwaikat et al., 2014). most of the theoretical and empirical studies that addressed these topics concentrated on the large number of listed firms that dispersed ownership structure. it is doubtful that these findings could be applied to firms under the control of large stockholders, especially those owned by families; in particular, small and medium enterprises (smes) which have specific goals to concentrate on their survival in the long run. it is argued that smes have more aversion to risk than listed companies. to prevent the entrance of new stockholders, small companies will employ debt in the capital as leverage. age gender level of education experience capital structure (internal and external financing) technology improvement 26 the international journal of banking and finance, vol. 15, no 1, 2020 : 21-37 3.1 age ceos need to be of a mature age as they are tasked with making important decisions, especially related to external financing (kyenze, 2014). aterido, beck and lacovone (2011) reported that older individuals are more likely to use formal banking services compared to younger people. however, the findings of uddin (2014) and zabri and lean (2014) revealed that the problem faced in raising debt capital does not differ with the age of entrepreneurs. this is contradictory as earlier research by coleman and carsky, (1999) suggested that young entrepreneurs were noted to be more likely to use external credit. based on data taken from the 1993 national survey of small business finances, the authors found that a young entrepreneur sees growth potential in his firm as compared to a mature manager and thus requires more external capital to fund its growth. older managers tend to be more risk averse whereas younger managers are willing to undertake risky innovative growth strategies (ting, azizan, & kweh, 2014). from the survey in the us and uk on sme financing, vos, yeh, carter and tagg (2007) reported that young entrepreneurs of smes utilized external financing more actively while older entrepreneurs relied on a great extent on retained profits. bogdana (2009) attempted to summarize and compare the findings on the impact of the age of ceos on capital structure. he concluded that a ceo’s age is negatively, statistically and significantly related to corporate capital structure, based on fixed and random effects in panel data. this means that the older the ceos, the more conservative they become, and consequently, the less they borrow. h1: there is a positive relationship between managers’ age and leverage in malaysian manufacturing smes. 3.2 gender there is less evidence of actual discrimination by financial institutions against female sme owners in terms of the average time taken to approve a loan, the tenure (term) of the loan or the interest rate charged (watson, newby & mahuka, 2009). the findings failed to uncover any perceptions of discrimination by financial institutions against female sme owners. therefore, there is no difference in bank loan rejection rates by gender. the problem faced in raising debt capital does not vary with the gender of entrepreneurs (cassar, 2004; uddin, 2014; zabri & lean, 2014; kyenze, 2014) however, fourati and affes (2013) found that men are more likely to obtain external funding than women. male business owners make the most use of external sources of finance, notably bank overdrafts, bank loans and suppliers’ credit (vos et al., 2007). the impact of managerial characteristics on capital structure in malaysian manufacturing smes : 21-37 27 interestingly, constantinidis, cornet and asandei (2006) found that women entrepreneurs in the industry sector resorted to bank loans to expand the operational size of their firms. irwin (2007) pointed out that men are more likely to use family and friends’ advice to access finance for smes in the uk, whilst women seek external advice, such as from business links and enterprise agencies. in 2007, a majority of female-owned firms sought external funding just as the majority of male-owned firms (jung, 2010). yet, female managers are more risk-averse and borrow less on behalf of the company (bogdana, 2009). the decision on the amount of borrowings is not dependent on the gender of the ceo, as it is collectively made during executive board meetings. women experienced more problems in raising external funds than their male counterparts (wellalage, dupati, & fauzi, 2013; kapunda, 2016). the female entrepreneurs in the african countries have to overcome greater legal barriers compared to their male peers (aterido et al., 2011). firms with female cfos face tighter credit availability and are more often required to provide collateral despite their approved loans. furthermore, banks prefer to claim mortgage collaterals when lending to companies under the control of female cfos, while they are more inclined to require guarantee collaterals when lending to companies under the control of male cfos. this suggests that the collateral clauses are more restrictive to female than to male cfos. thus, it could be surmised that female cfos are less likely to obtain external financing from financial institutions (xu, li, & chang, 2016). h2: male managers are preferred in debt financing than female managers. 3.3 education highly educated entrepreneurs are able to manage their businesses better than those who are less educated. in addition, education is a strong predicator of the use of banking services, with linear increase in its use in most countries (aterido et al., 2011). highly educated entrepreneurs are usually successful in managing smes (chong & mahmoud, 2013) and ceos who have mbas or law degrees tend to have greater leverage (frank & goyal, 2007), and therefore resort to banks and external financing (fourati & affes, 2013). this could be attributed to the fact that they have sufficient financial knowledge and are aware of the varied sources of financing. a manager’s level of education is significantly related to debt financing (zabri & lean, 2014) and firm performance (kyenze, 2014). constantinidis et al. (2006) studied financing patterns of enterprises owned by women in belgium and found that women owners with low educational levels were more likely to encounter problems with regard to access to financing 28 the international journal of banking and finance, vol. 15, no 1, 2020 : 21-37 than those who were highly educated. indeed, women with higher educational qualifications and additional training are in an advantageous position in terms of securing financing. the less educated sme owners tend to actively seek external financing to avert the fear of loan denials (vos et al., 2007). however, zabri, ahmad and lean (2015) noted that the more knowledgeable the ownermanagers, the lesser their preference for external equity financing. whereas, borgia and newman (2012) showed that the education level of owners or managers in chinese smes did not significantly impact the amount of debt used in their capital structure. this was consistent with earlier findings by cassar (2004) who affirmed that educational levels of entrepreneurs did not impact financing preferences. h3: there is a positive relationship between managers’ education and leverage in malaysian manufacturing smes. 3.4 experience a manager’s experience could be considered as a measure of reputation and private entrepreneurs who are experienced are more likely to choose formal financing. managers with a greater level of business experience are found to take advantage of bank financing (zabri et al., 2015). this is because owner’s or manager’s experience plays an important role in reducing asymmetric information flow between the firm and external investors. borgia and newman (2012) observed that the owner’s or manager’s experience is significantly and positively related to the level of a firm’s leverage. fourati and affes (2013) stated that high industrial experience increases the probability of using external equity in financing venture activities. hence, high capital requirements for large firms discourage inexperienced and undercapitalized entrepreneurs from undertaking large ventures (parsa, rest, smith, parsa and bujisic, 2015). constantinidis et al. (2006) noted that women did not encounter problems of access to external financing if they had business experience, that is, they had worked as employees or were owners in the family business. however, according to cassar (2004), bogdana (2009), zabri and lean (2014), and klacmer calopa et al. (2014), the diverse experiences of entrepreneurs did not appear to impact on financing preferences. on the contrary, scott and irwin (2009) found that inexperienced owner-managers faced difficulties raising debt. they did not have enough credit scoring to obtain loans compared to the more experienced entrepreneurs. h4: there is a positive relationship between managers’ experience and leverage in malaysian manufacturing smes. the impact of managerial characteristics on capital structure in malaysian manufacturing smes : 21-37 29 4. methodology the data was collected using structured questionnaire survey. prior to using the questionnaire for data collection, a detailed pre-pilot test was conducted by sending the questionnaires to 30 malaysian manufacturing smes to increase its validity. the pilot study helped to dispel doubts about the validity and reliability of the instrument. thus, the final questionnaire incorporated suggestions obtained from the pilot study and necessary amendments were made. the developed questionnaire consisted of two sections. section one captured the demographics of the respondents (managers or owners) such as age, gender, level of education and experience. section two, required the respondents to indicate their preferences on capital structure, sources of funding and their technology performance. census method is the best method to collect data from smes listed in the sme corp. as it incorporates all companies in the manufacturing sector from the fmm directory 2018. out of the 1365 questionnaires distributed, 235 sample responses were returned. however, only a total of 225 of the questionnaires were usable. 10 questionnaires were not included in the data analysis due to incomplete response. a total of 225 respondents represented approximately 15.6 percent covering a broad range in the malaysian manufacturing sector. this response rate of 15.6 percent is a common response rate within the context of research in malaysian manufacturing companies (olusegun, hasbullah, & nordin, 2014). however, there were six sets of questionnaires which fell out of the range and were considered as outliers. henceforth, the data set in this study was 219 after removing the undesirable outliers. 5. results 5.1 descriptive analysis this study employed a sample size of 219 respondents with the following descriptive characteristics: 45.2 percent were male and 54.8 percent were female. most of them were chinese from the age of 31 to 40 years old. 81 percent of the respondents had more than 5 years to 10 years of work experience. the majority of the respondents were owners/managers who had work experience in local private firms. 30 the international journal of banking and finance, vol. 15, no 1, 2020 : 21-37 table 1. demographic profile of respondents frequency percentage (%) gender male 99 45.2 female 120 54.8 age below 20 0 0 21–30 49 22.4 31–40 92 42.0 41–50 63 28.8 51 and above 15 6.8 ethnicity malay 4 1.8 chinese 211 96.3 indian 4 1.8 level of education no formal education 0 0 primary school 7 3.2 pmr/ srp/ lce 0 0 spm/ mce 18 8.2 diploma 121 55.3 advance diploma 121 55.3 degree 10 4.6 master’s degree 0 0 doctor of philosophy/ dba 0 0 professional certification 2 0.9 current position owner 47 21.5 owner/manager 70 32.0 finance manager 71 32.4 accounting manager 31 14.2 (continued) the impact of managerial characteristics on capital structure in malaysian manufacturing smes : 21-37 31 frequency percentage (%) length of service less than 3 years 55 25.1 more than 3 years to 5 years 23 10.5 more than 5 years to 10 years 141 64.4 more than 10 years 0 0.0 work experience yes 179 81.7 no 40 18.3 work experience as owner 27 12.3 shareholder 49 22.4 employee 129 58.9 others 14 6.4 work experience in local private firm 123 56.2 government service 7 3.2 multinational corporation (mnc) 36 16.4 government linked company (glc) 2 0.9 others 51 23.3 sources: developed for research 5.2 findings and discussion all collected data were analyzed using statistical package for the social sciences (spss) version 23 and partial least squares-structural equation model (pls-sem). the structural model was analyzed using smartpls 3.2 to perform pls-sem to test the hypothesized relationships between independent variables (such as age, gender, level of education and experience of respondents) with dependent variable as leverage in malaysian manufacturing companies as shown in figure 2. 32 the international journal of banking and finance, vol. 15, no 1, 2020 : 21-37 figure 2. diagram of second-order constructs. there are several steps for formative measurement. the first step is to conduct a collinearity assessment of indicators. collinearity is assessed by computing the tolerance which represents the amount of variance of one formative indicator that is not explained by other indicators in the same block. variance inflation factor (vif) is a reciprocal of the tolerance and it is a related measure of collinearity. a vif value of 5 or higher shows the potential presence of a collinearity problem in pls-sem (hair, ringle, & sarstedt, 2011). table 2 shows that all the vifs of the indicators are below the threshold of 5, suggesting that these indicators are distinct and that there is no collinearity problem. the next step in accessing the formative measurement model is by assessing the significance and relevance of the formative indicators if there is no collinearity issue in the formative measurement. this procedure is carried out by undergoing the bootstrapping procedure. table 2 summarizes the results for the formative measure of internal and external financing by showing the original outer weight estimates, t-values, and significance levels. the findings from table 2 shows that gender significantly influenced decisions to obtain internal and external funding at 5% level of significance. gender showed a negative association with external funding decision. it suggested that male ceos in the manufacturing sector had less reliance on the use of external financing for company growth. as compared to women entrepreneurs in the industrial sector who relied more on bank loans as a source of financing to expand business operations (constantinidis et al., 2006;, irwin, 2007; jung, 2010). the age of ceos showed a positive but insignificant relationship with company decision to use or not to use external financing. 10 figure 2. diagram of second-order constructs. there are several steps for formative measurement. the first step is to conduct a collinearity assessment of indicators. collinearity is assessed by computing the tolerance which represents the amount of variance of one formative indicator that is not explained by other indicators in the same block. variance inflation factor (vif) is a reciprocal of the tolerance and it is a related measure of collinearity. a vif value of 5 or higher shows the potential presence of a collinearity problem in pls-sem (hair, ringle, & sarstedt, 2011). table 2 shows that all the vifs of the indicators are below the threshold of 5, suggesting that these indicators are distinct and that there is no collinearity problem. the next step in accessing the formative measurement model is by assessing the significance and relevance of the formative indicators if there is no collinearity issue in the formative measurement. this procedure is carried out by undergoing the bootstrapping procedure. table 2 summarizes the results for the formative measure of internal and external financing by showing the original outer weight estimates, t-values, and significance levels. table 2. formative measure, variance inflation factor (vif) and outer weight significance test results dependent variable independent variable vif outer weights t statistics significance internal and external financing age 1.000 1.000 0.044 no gender 1.000 1.000 1.565** yes level of education 1.000 1.000 1.596** yes work experience 1.000 1.000 1.528* yes note: *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent the impact of managerial characteristics on capital structure in malaysian manufacturing smes : 21-37 33 table 2. formative measure, variance inflation factor (vif) and outer weight significance test results dependent variable independent variable vif outer weights t statistics significance internal and external financing age 1.000 1.000 0.044 no gender 1.000 1.000 1.565** yes level of education 1.000 1.000 1.596** yes work experience 1.000 1.000 1.528* yes note: *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent nevertheless, problems faced by companies in raising debt financing were equally the same regardless of whether the manager was younger or older (zabri & lean, 2014). ting, azizan and kweh (2014) reported that younger managers preferred to take risks to increase company growth and relied on external financing compared to older managers. work experience showed a positive relationship with company source of capital at 10% significant level. experienced managers in the manufacturing sector focused more on internal or external financing to increase their company performance. with their vast experience, these managers know the consequences they may need to face when using external financing. therefore these managers are more cautious when choosing debt capital. this result concurred with previous studies by borgia and newman (2012), fourati and affes (2013) and zabri et al. (2015) who found that managers with high levels of business experience chose bank financing to reduce asymmetric information between company and investors. the more highly educated the managers, the higher the probability of them choosing external financing. the results showed a strong positive relationship between the level of education and source financing at 5% significant level. furthermore, internal and external financing also shows a positive relationship at 1% significant level in improving manufacturing technology performance in table 3. 34 the international journal of banking and finance, vol. 15, no 1, 2020 : 21-37 table 3. relationship of internal and external financing on technology improvement dependent variable t statistics independent variable technology improvement 2.338*** internal and external financing note: *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent 6. conclusion a total of 219 responses were collected from a survey to examine the impact of managerial characteristics namely age, gender, work experience and level of education on source of capital financing of smes (in the manufacturing sector). based on pls sem bootstrapping, the results revealed that only managers’ age showed an insignificant influence on company leverage decision. the level of education and work experience strongly influenced managers when making decisions to use internal or external financing. managers who were more educated with work experience preferred to use external financing. the findings were in line with the trade-off theory where more knowledgeable managers know the benefits of lower tax payments when incurring debt, however, the bankruptcy clause is increased if they are unable to generate higher returns to compensate risks. therefore, these managers may prefer to use internal financing to finance the growth of the company. managers’ gender also showed a negative and significant relationship to leverage. thus, financial decisions by management are crucial in ensuring that a firm’s capital structure is optimal. different managements have different leverage privileges with managers trying to attain optimal capital structure. variations in capital structure affect the costs of equity or debt and thereby, have an impact on the capital costs of firms. therefore the choice of type of debt in the capital structure is important for all companies. references ahmad, z., abdullah, n. m. h., & roslan, s. (2012). capital structure effect in firm performance: focusing on consumers and industrials sectors in malaysian firms. international review of business research papers, 8(5), 137–155. arasti, z. (2011). an empirical study on the causes of business failure in the iranian context. african journal of business management, 5(17), 7488– 7498. the impact of managerial characteristics on capital structure in malaysian manufacturing smes : 21-37 35 arasti, z., zandi, f., & talebi, k. (2012). exploring the effect of individual factors on business failure in iranian newly established small businesses. international business research, 5(4), 2. aterido, r., beck, t., & lacovone, l. (2011). gender and finance in sub-saharan africa: are women disadvantaged? world bank policy research working paper series. baldwin, j. r., bian, l., dupuy, r., & gellatly, g. (2000). failure rates for new canadian firms: new perspectives on entry and exit. failure rates for new canadian firms: new bogdana, d. (2009). corporate capital structure choice: does managers’ gender matter. working thesis, kyiv school of economics. borgia, d., & newman, a. (2012). the influence of managerial factors on the capital structure of small and medium-sized enterprises in emerging economies: evidence from china. journal of chinese entrepreneurship, 4(3), 180–205. cassar, g. (2004). the financing of business start-ups. journal of business venturing, 19(2), 261–283. chong, k. g., & mahmoud, k. a. (2013). the determinants of sme succession in malaysia, from entrepreneurship perspective. journal of advanced social research, 3(12), 350–361. constantinidis, c., cornet, a., & asandei, s. (2006). financing of womenowned ventures: the impact of gender and other owner-and firm-related variables. venture capital, 8(2), 133–157. coleman, s., & carsk, m. (1999). sources of capital for small family-owned businesses: evidence from the national survey of small business finances. family business review, 12(1), 73–84. dwaikat, n. k., queiri, a. r., & aziz, m. n. (2014). capital structure of family companies. international conference on business, law and corporate social responsibility (icblcsr’14). frank, m. z., & goyal, v. k. (2007). corporate leverage: how much do managers really matter? available at ssrn 971082. fourati, h., & affes, h. (2013) the capital structure of business start-up: is there a pecking order theory or a reversed pecking order? -evidence from the panel study of entrepreneurial dynamics. technology and investment, 4(4), 244. hair, j. f., ringle, c. m., & sarstedt, m. (2011). pls-sem: indeed a silver bullet. journal of marketing theory and practice, 19(2), 139–151. irwin, d. (2007). how sme owner’s characteristics influence external advice and access to finance. aston university. jung, o. (2010). small business financing profiles: women entrepreneurs, sme financing data initiative. 36 the international journal of banking and finance, vol. 15, no 1, 2020 : 21-37 kapunda, s. m. (2016). sme finance, development and trade in botswana: a gender perspective. business management review, 11(1). 30–54. klacmer calopa, m., horvat, j., & lalic, m. (2014). analysis of financing sources for start-up companies. management: journal of contemporary management issues, 19(2), 19–44. kyenze, a. m. (2014). the effect of managers’ characteristics on the capital structure of firms listed at the nairobi securities exchange. doctoral dissertation, university of nairobi. luigi, p., & sorin, v. (2009). a review of the capital structure theories. annals of faculty of economics, 3(1), 315–320. ooghe, h., & de prijcker, s. (2008). failure processes and causes of company bankruptcy: a typology. management decision, 46(2), 223–242. parsa, h. g., rest, j. p. i, van der, smith, s. r, parsam, r. a., & bujisic, m. (2015). why restaurants fail? part iv the relationship between restaurant failures and demographic factors. cornell hospitality quarterly, 56(1), 80–90. saarani, a., & shahadan, f. (2013). the determinants of capital structure of smes in malaysia: evidence from enterprise 50 (e50) smes. asian social sciences, 9(6), 64. salim, m., & yadac, r. (2012). capital structure and firm performance: evidence from malaysian listed companies. proceedings-social and behavioral sciences, 65, 156–166. scott, j. m., & irwin, d. (2009). discouraged advisees? the influence of gender, ethnicity and education in the use of advice and finance by uk smes. environment and planning c: government and policy, 27(2), 230–245. ting, i. w. k., azizan, n. a., & kweh, q. l. (2014). human governance and firm’s leverage decision: evidence from malaysian listed companies. social sciences & humanities, 22(3), 879–890. uddin, m. a. (2014). problems faced by micro, small and medium enterprises in raising debt capital. proceedings of the social sciences icssr. vos, e., yeh, a. j. y., carter, s., & tagg, s. (2007). the happy story of small business financing. journal of banking & finance, 31(9), 2648–2672. watson, j., newby, r., & mahuka, a. (2009). gender and the sme “finance gap”. international journal of gender and entrepreneurship, 1(1), 42– 56. wellalage, n., dupati, g., & fauzi, f. (2013). institutional factors create additional glass ceilings: evidence from female owned south-asian smes’ access to external financing. asian institute of management. xu, x., li, y., & chang, m. (2016). female cfos and loan contracting: financial conservatism or gender discrimination? an empirical test based on collateral clauses. china journal of accounting research, 9(2), 153– 173. the impact of managerial characteristics on capital structure in malaysian manufacturing smes : 21-37 37 zabri, s. m. (2012). the determinants of capital structure among smes in malaysia. in proceedings of international conference of technology management, business and entrepreneurship. zabri, s. m., & lean, j. (2014). sme managers’ financing preferences: the case of successful smes in malaysia. in proceedings of 5th asia pacific business research conference. zabri, s. m, ahmad, k., & lean, j. (2014). understanding owner-managers’ preferences towards different sources of financing: the case of successful smes in malaysia. advanced sciences letter, 21(5), 1435–1438. how to cite this article: sultana, i., & rahman, m. m. (2020). determinants of bank cost efficiency: empirical evidence from bangladesh. international journal of banking and finance, 15(1), 39-71. https://doi.org/10.32890/ijbf2020.15.1.9931 determinants of bank cost efficiency: empirical evidence from bangladesh 1israth sultana 2mohammad morshedur rahman university of chittagong, chattogram, bangladesh 1israthctg@gmail.com 2corresponding author: mmrais@cu.ac.bd; a r t i c l e i n f o article history: received 8 october 2019 revised 3 december 2019 accepted 13 january 2020 published 31 march 2020 keywords: cost efficiency, risk, capital regulation, banks, panel data. jel code: c23, e22, g18, g21, g32, h21 a b s t r a c t due to the trust of depositors, banks should be responsible for efficient utilization of resources to achieve cost efficiency (ce) which in turn contributes to raising income. previous studies found that the average ce of banks in bangladesh was around 80%. this study aims to find the determinants of ce in bangladesh from a sample of 33 banks during a period from 2009 to 2016. stochastic frontier approach (sfa) was used to measure ce in the first stage. in the second stage, different types of regression estimations were used like pooled ordinary least square, fixed effect or random effect panel regression, system-dynamic panel data estimation and arellano-bond dynamic panel data estimation for comparison. the results showed that generalized method of moments (gmm) specifically the arellanobond dynamic panel data estimation was best suited for problems of the international journal of banking and finance, vol. 15. number 1, 2020: 39-71 39 40 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 1. introduction in developing countries like bangladesh, where people are interested in safe savings, the importance of commercial banks is immense. the trend of saving among people and likewise, the number of commercial banks have been increasing since independence. the banking industry in bangladesh started with six nationalized commercial banks, three state owned specialized banks and nine foreign banks after independence. in the 1980s, the banking industry achieved significant expansion with the entry of private banks. now there are 58 scheduled banks in bangladesh which operate under the full control and supervision of the bangladesh bank, and recently three new banks have obtained permission to conduct their business. as there is intense rivalry among existing and upcoming banks, cost efficiency (ce) is an important concept to secure competitive advantage. commercial banks play an important role in maintaining stability and progress of the economy by properly mobilising financial resources across the country. banks can increase economic growth and financial stability through efficient intermediation. on the other hand, if banks were inefficient then it would induce an economic crisis. banna et al. (2017) argued that efficient banks can contribute not only to the productivity of the country but also lend more endogeneity, serial correlation, heteroskedasticity and cross sectional dependence in data the results revealed that regulatory capital, risk measured by non-performing loan ratio, liquidity measured by total loans to total deposits, and level of operating costs had a significant negative impact on ce. in contrast, lagged cost efficiency, profitability, years of operation, net interest income had a significant positive impact on ce. to attain competitive advantage by performing with higher ce, policymakers should focus on capital regulation measured by capital adequacy ratio, risk level, profit earning capacity, aggressiveness of banks, bank size, years of operation and level of operating costs. determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 41 strength to the economy to fight against negative and external shocks. therefore it can be said that efficiency in the banking sector of bangladesh is crucial for economic growth and for the stability of the financial sector which leads to the stability of the whole economy. nowadays, ce is getting huge attention in improving bank performance including proper investment allocation. it is the way to save money by providing quality services in a better way. according to ariff and luc (2008) “ce refers to a firm's minimization of costs reflected by how close its cost is to what a best-practice firm would be for producing the same amount of outputs which are sold at certain prices with the given prices of inputs.” in the context of banking, the measurement of efficiency is considered as a tool for taking management decisions regarding the improvement of bank performance and it provides bank specific information related to efficiency gain (tecles & tabak, 2010). however, ab-rahim et al. (2012) argued that banks, within a competitive environment, need to examine their performance for survival which is dependent on their productive efficiencies. blejer (2006) claimed that the capability of emerging markets for adoption of optimal stabilization policies is impeded due to inefficiencies, financial instability and financial market imperfection. the study also added that financial efficiency is the basis for the success of any economy; hence an emerging economy should emphasize financial efficiency which would lead to financial stability. schumpeter (1982) concluded that the economy would be better if it has a more efficient financial system. therefore, a better allocation of financial resources is embodied in the improvement of banking performance. as a result, the economy will attract more private investments that lead to economic growth. in recent years, the income of the people of bangladesh has been increasing and since most of the people of this country are risk averse, banking operations should continue to grow. in addition, banks also need to show their performance in a better way to attain the confidence of depositors. although the economy is largely dependent on the financial system, bangladesh has been facing challenges in stimulating private investments for a long period of time. the banking industry in bangladesh is undergoing a critical time due to some deregulation and non-performing loans. the government has earmarked the financial sector as a major area to be developed for the development of the whole economy. some banks are striving hard to remain competitive. in order to thrive in the modern competitive business world, these banks need to focus attention on their performance. in this regard, the performance measurement of banks is important. evaluation of banking efficiency is valuable from both the microeconomic and macroeconomic perspective (berger & mester, 1997). from the microeconomic view point , inefficiency in the banking system is vital because competition is stiff and there is a huge development in banking institutions, regulations and supervision. on the other hand, from the macroeconomic view 42 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 point, the soundness of the financial market and the cost of intermediation are being influenced by efficiency of the banking sector (ab-rahim et al., 2012). the existing literature is divided into three categories: event studies, operating performance studies through ratio analysis and frontier analysis studies. event studies are basically conducted to show the impact of an event, for example, merger to the value of shareholders. operating performance studies are conducted based on ratio analysis which allows for comparison of different sized banks. both of these approaches have been criticized due to their limitations (koher et al., 2000; nail & parisi, 2005). to overcome these limitations, this study uses stochastic frontier analysis (sfa) to measure banking ce. in addition, when banks are less cost efficienct, the risk of failure will be higher, which may in turn, lead the economy of the country to depression (podpiera & podpiera, 2005). as a small and developing country, if a bank fails, the government of bangladesh has little capacity to support the banking sector. therefore, the difference between successful and inefficient banks should be identified (hassan & hassan, 2018). moreover, the central bank of bangladesh emphasized the need for an efficient banking sector. thus, the authorities urged that more research be conducted on the banking sector of bangladesh (khanam & nghiem, 2006). there are very few studies focused on ce in bangladeshi banks and those studies found that the average ce in bangladesh was around 80% (hassan & hassan, 2018: khanam & nghiem, 2006). they argued that the ce of banks should be improved. it is not sufficient to ascertain efficient and inefficient banks without knowing what factors are working behind a bank’s ce. to fulfill that need, determinants of a bank’s ce should be identified. moreover, basel i, ii, iii capital regulations have been implemented in bangladesh to mitigate the risk taking behavior of banks in bangladesh. with the introduction of capital regulation and based on the risk taking behavior of banks in bangladesh, the impact of this capital regulation along with the risk on ce of banks in bangladesh need to be examined. there are several studies conducted on ce of commercial banks in the context of developed countries, but there are not many studies conducted in the context of developing or least developed countries. based on our knowledge, there is no empirical study in bangladesh on this issue. therefore our objective is to contribute to the existing literature by examining the determinants of ce of 33 bangladeshi commercial banks spanning the period from 2009 to 2016. specifically, this study looks into the impact of capital regulation, risk-taking behavior and relevant factors that influence a bank’s ce. moreover, as tasks in banks are repetitive in nature, the learning curve and experience curve can be applied in bangladesh for reducing and managing operating cost in banks thereby increasing ce (fioretti, 2007; teplitz, 1991; wright, 1936). bush (2015) suggested that experience is related to ce gains in the banking sector. boot and determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 43 thakor (1997) found that efficiency achieved from experience may reduce labor cost in cases of similar types of loan monitoring. hence, the study also explores the impact of experience and learning on ce. the rest of the paper is organized as follows: section 2 reviews the literature and provides testable hypotheses. section 3 introduces the research methodology. section 4 presents the empirical results and reports on the robustness of testing, and section 5 concludes the study with recommendations. 2. literature review and hypotheses development there is a large number of studies both on single countries as well as on cross countries that focused on banking efficiency. hereafter, we briefly report previous studies that covered the ce of banks and the factors of ce in both developed and developing countries. financial instability, inefficiencies, and financial market imperfections hamper the adoption of stabilization policies in emerging markets. it is argued that the financial regulatory paradigm, embodied in basel, may improve financial stability but reinforces cyclicality (blejer, 2006). the study concluded that emphasis should be given on financial efficiency because it will enhance financial stability without increasing cyclicality. the literature on banks’ ce can be divided into two groups. in the first group, we report on a number of studies that focused on the identification of the impact of various specific measures of transition like capital structure, capital measure, risk measure, ownership structure, banking reform, economic reform, financial liberalization and financial crisis, etc., on ce in a number of countries (spulbăr & niţoi, 2014). in the second group, we report on studies that focused on the determinants of ce. 2.1 literature on impact of various measures of transition on ce among the previous studies in this group, abdul-majid et al. (2011a) found that the east asian financial crisis caused a short-term increase in efficiency. the study also concluded that bank efficiency declined because conventional banks were operating under islamic banking windows. their findings were also supported by a study from hammim et al. (2006). in another study, abdul-majid et al. (2011b) showed the impact of islamic banking on ce and found that higher inputs were required for islamic banks. the study also concluded that merged banks used higher input and the financial crisis showed a negative impact on bank performance in the increasing volume of nonperforming loans. alam (2012) focused on the impact of regulatory and supervisory structures on efficiency and found that regulations, strict monitoring 44 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 of banking operations and higher supervisory power of the authorities increased the efficiency of islamic banks but decreased the efficiency of conventional banks. the results also supported the findings of chortareas et al. (2012). the existence of competition-efficiency hypothesis was confirmed in studies by andrieş and căpraru (2014) and kyj and isik (2008). hall and simper (2013) found evidence that perfect competition prevailed in korean banking which offered a basis for merger policy where there were scale economies to be obtained and reduction in average costs that enhanced efficiency. by working on chinese banks, ariff and luc (2008) suggested that joint-stock banks were more ce than state-owned banks. banking reforms also played a vital role in changing banking efficiency. ataullah and le (2006) focused on banking reforms in indian banking and suggested that banking efficiency increased due to economic reforms in india. the study also found that fiscal deficit negatively influenced bank efficiency. beyond the above studies, avkiran (2011) suggested that financial ratios can be predicted by regressing ratios on efficiency estimates. the study also reported that data envelopment analysis (dea) can also be used to objectively identify benchmarks for ratio analysis based on actual observed data collected from peers where an input–output specification consisted of key financial ratios. the heterogeneity of environmental conditions calls for the consideration of efficiency measurement (battaglia et al., 2010). the study showed that environmental conditions substantially influenced efficiency estimates. in addition, battese et al. (2000), das and ghosh (2009) focused on efficiency in the deregulation of the swedish and indian banking industry, respectively. the study found significant technical inefficiency of labor used in swedish banks. on the other hand, environmental conditions of a country play an important role in the definition and specification of the common frontier in different countries (dietsch & lozano-vivas, 2000). the study suggested that environmental variables significantly affected the difference in efficiency for french and spanish banking industries. in another study, esho (2001) argued that estimates of ce and efficiency ranking would be biased if subsidies were ignored. despite the above, hasan and marton (2003) analyzed the experiences and developments of the banking sector during the transitional process. the study suggested that flexible techniques to privatization, easy policy for foreign banks and early reorganization initiatives induced the development of a comparatively stable banking system thus raising efficiency. the study also concluded that foreign banks and banks with higher foreign bank ownership showed lower inefficiency. the finding was also supported by findings from kasman and yildirim (2006). in addition, hauner (2005) found that state-owned banks in germany and austria were more cost efficient. determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 45 2.2 literature on determinants of ce a number of other studies have focused on the determinants of bank ce which included bank-level variables like capital ratio, risk level, bank size, liquidity, performance, years of operation, loan loss provision, non-performing loans, etc., and macroeconomic variables such as growth in gross domestic product, inflation, banking sector development, etc. with regard to bank level factors, a number of studies were conducted on both developing and developed countries. those studies found that government ownership, population density, demand density, credit risk, liquidity ratio, solvency ratio, bank size, equity to total assets and market concentration resulted in positive impact on different measures of efficiency and also ce (ab-rahim et al., 2012; chan et al., 2014; girardone et al., 2004). on the other hand, off balance sheet exposure, interest sensitivity, merger, macroeconomic conditions, capitalization, credit risk, asset quality and management quality, and problem loans resulted in a negative impact on different measures of efficiency and also ce (ab-rahim et al., 2012; berger & deyoung, 1997; chan et al., 2014; chang & chiu, 2006; chiu et al., 2008; drake & hall, 2003; garza-garcia, 2012; girardone et al., 2004). in addition, mixed results were also found regarding the ce determinants of banks. for example, chang and chiu (2006) argued that risk factors impacted bank efficiency. the study also found no significant difference with bank efficiency index when it only considered market or credit risk. finally, the study concluded that different risk based bank efficiency were influenced by different factors. other than the studies mentioned, chan et al. (2014) found that bank insolvency risk (as measured by z-scores) was positively related to efficiency. in a study, chiu et al. (2008) showed that capital adequacy was an influential factor in evaluating the efficiency of banks and highly capitalized banks had average efficiency scores which were significantly higher than those of low capitalized banks.in a study, christopoulos et al. (2002) showed that smaller banks were more efficient than larger ones. the study also found that bank loans, investment and economic performance were positively related to the ce of the greek commercial banking sector. in another study, das and ghosh (2009) reflected that inefficiency was important on the revenue side of banking activity. the study also revealed that allocative inefficiency took a larger portion of outlay cost. the study also suggested that other than the factors mentioned, product diversity and prudential indicators were important variables that drove efficiency. eisazadeh and shaeri (2012) also found other factors that affected efficiency. these were: financial development, the degree of market competition, macroeconomic stability, legal rights and contract laws, political stability and better governance. the study recommended to build stronger institutions, improve governance and promote more competition. in addition, age, average deposit size and interest rate spreads also contributed to ce (esho, 2001). 46 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 in another study, fries and taci (2005) revealed that there was a non-linear relationship between progress in banking reform and ce. the study focused on the transformation of banking in the post-communist transition in european countries and found that banks with large foreign shares had lower costs and state-owned banks were less efficient than private banks. private banks with foreign ownership were more ce (garza-garcía, 2012). in contrast, gardener et al. (2011) found that state-owned banks were more efficient than those of private banks. in terms of macroeconomic variable, a positive association was found between gdp and ce (chan et al., 2014; garza-garcía, 2012). in terms of efficiency, a number of studies used x-efficiency as a proxy for efficiency, for example, altunbaş et al. (2001), altunbas et al. (2000), isik and hassan (2003), kenjegalieva et al. (2009), kraft and tırtıroğlu (1998) and kyj and isik (2008). based on the literature, we selected the following variables as the potential factors of banks’ ce: capital ratio, risk, performance, years of operation, level of operating costs, size of bank, lagged ce, number of bank branches, ownership structure, etc. and the following alternative hypotheses were developed: h1 : more capital makes the bank more cost efficient. (capital has a positive relationship with bank ce.) h2 : risky banks are less cost efficient. (bank ce is negatively related to risk profile.) h3 : profitable banks are cost efficient. (the relationship between bank cost efficiency and bank profitability is positive.) h4 : experienced banks are more cost efficient. (age of bank is positively related to bank cost efficiency.) h5 : banks with high operating cost are less cost efficient. (there is a negative correlation between operating cost and cost efficiency.). h6 : large banks (in terms of assets) are most cost efficient. (bank size is positively correlated with cost efficiency.) h7 : efficient banks may remain efficient. (previous year efficiency is positively related to current efficiency.) 2.3 literature on ce, risk, capital regulation and bank performance a significant number of studies were focused on efficiency along with risk, capital regulation and performance, for example, altunbas et al. (2007), altunbas et al. (2000), bitar et al. (2018), chan et al. (2014), deelchand and padgett (2009b), faccio et al. (2016), färe et al. (2004), fethi and pasiouras (2010), fiordelisi and mare (2013), fiordelisi et al. (2011), koetter (2008), kwan and eisenbeis (1997), zhang et al. (2012) and sarkar et al. (2019). determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 47 2.4 literature on ce in developed countries in the context of developed countries, altunbaş et al. (2001), altunbas et al. (2007), andrieş and căpraru (2014), chortareas et al. (2012), and fiordelisi et al. (2011) investigated european banks, altunbas et al. (2000), deelchand and padgett (2009b), drake and hall (2003) examined japanese banks, carvallo and kasman (2005) researched on north american and caribbean banks, chortareas et al. (2009), christopoulos et al. (2002), chortareas et al. (2009) investigated greek banks, čiković and cvetkoska (2018) studied the macedonian banking sector, kumbhakar and tsionas (2008), färe et al. (2004) investigated the us banking industry, carvallo and kasman (2005) researched on banks in latin america, dietsch and lozano-vivas (2000) examined french and spanish banks, esho (2001) investigated australian banks, fiordelisi and mare (2013) on italian banks, garza-garcía (2012) researched on banks in mexico and eisazadeh and shaeri (2012) studied banks in african countries. 2.5 literature on ce in asian countries in asian countries, abdul-majid et al. (2011a), abdul-majid et al. (2011b), abrahim et al. (2012) researched on malaysian banks, anwar (2015) investigated indonesian banks, ariff and luc (2008), avkiran (2011), berger et al. (2009), dong et al. (2014) examined chinese banks, chiu et al. (2008), chang and chiu (2006), chiu et al. (2011) studied banks in taiwan, ataullah and le (2006) examined indian banks, bitar et al. (2018) researched on banks in oecd countries and hall and simper (2013) investigated korean banks. while studies by aliyu and yusof (2016), bitar et al. (2018), chan et al. (2014), chortareas et al. (2012), and gardener et al. (2011) were based on cross countries. among the studies conducted in bangladesh, zheng et al. (2018) assessed the effect of human capital efficiency on bank risk and capital regulation. the study found a significant positive association between risk and human capital efficiency; in contrast, the relationship was negative with capital. miah and sharmeen (2015) examined the relationship between capital, risk and efficiency when comparing islamic banks with traditional banks. their study showed that conventional banks were more efficient in managing costs. 2.6 literature on research methods used in ce with regards to the calculation of efficiency, some studies used dea for example, ab-rahim et al. (2012), anwar (2015), ataullah and le (2006), avkiran (2006), avkiran (2011), chang and chiu (2006), chiu et al. (2011), chortareas et al. (2009), chortareas et al. (2012), christopoulos et al. (2002), čiković and cvetkoska (2018), das and ghosh (2009), dong et al. (2014), fethi 48 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 and pasiouras (2010) and gardener et al. (2011). in some other studies, sfa was used, for example, abdul-majid et al. (2011b), anwar (2015), battaglia et al. (2010), dong et al. (2014), esho (2001), miah and sharmeen (2015), zheng et al. (2018). in the second stage regression analysis, a majority of the studies used tobit regression, berger and deyoung (1997) used granger causality technique, deelchand and padgett (2009b), and garza-garcía (2012) used two stage least square regression, miah and sharmeen (2015) used seemingly unrelated regression technique. only a few studies used gmm, for example, han et al. (2012), ataullah and le (2006), and zheng et al. (2018). anwar (2015) suggested that dea should be used for estimating technical efficiency and sfa be used for cost efficiency. avkiran (2006) argued that “dea identifies the inefficiency in a particular decision making unit (dmu) by comparing it to similar dmus regarded as efficient, rather than trying to associate a dmu’s performance with statistical averages that may not be applicable to that dmu.” in accordance with the studies of anwar (2015) and avkiran (2006), our study has contributed to the existing literature by integrating determinants of banking efficiency into the areas of sfa methodology in the 1st stage and gmm in the 2nd stage regression analysis in the context of bangladeshi banking system across individual domestic banks. we applied dynamic panel gmm technique (doytch & uctum, 2011) to correct potential endogeneity, heteroscedasticity, cross section dependence and autocorrelation problems. in addition, banks can learn from experience and can increase ce through learning and experience. however, this is not evident in the literature. hence, we have also contributed to the existing literature by focusing on the learning and experience theory which is rarely focused in the literature. 3. methodology 3.1 data and sources on 31st december 2016, there were 58 commercial banks operating their businesses in bangladesh. among them six were state owned commercial banks, three were state-owned specialized banks, 40 were private commercial banks and nine were foreign commercial banks (annual report of bangladesh bank). initially the sample consisted of 58 banks. three specialized banks were excluded because of their non-involvement in commercial activities. due to the lack of available data, we also excluded nine foreign commercial banks. a total of 13 commercial banks were also excluded from the sample because they were either new or that data were not available. in fine-tuning, the final sample consisted of 33 bangladeshi commercial banks for the period from 2009 to 2016 including determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 49 four state owned commercial banks. the study was based on secondary data. these data were drawn from the annual published reports of respective banks and from their websites. 3.2 measurement of variables 3.2.1 dependent variable cost technical efficiency is the dependent variable in this study. sfa was used to calculate ce. total cost is the dependent variable in the sfa model and a number of inputs and outputs were used to calculate ce. the sfa used to calculate the efficiency of each bank in this study was based on the stochastic frontier production methodology which was originated by aigner et al. (1977). the stochastic cost frontier model was developed (kwan & eisenbeis, 1997; schmidt & lovell, 1979) based on the production frontier model. according to this methodology, due to inefficiency and random noise in the observed cost of a bank, the following is formulated to deviate from cost-efficient frontier (deelchand & padgett, 2009a). for the nth bank, ln tcn = f (ln qi, ln pj) + εn where, tcn represents total operating cost including financial costs, qi indicates outputs, pj stands for input, εn shows the deviation of actual total cost of a bank from the cost-efficient frontier. table 1 shows the list of input and output variables used in the calculation of ce. table 1. description of input and output variables used in the calculation of ce (1) variable definition source dependent variable , total cost (tc) operating cost including interest expenses used in all previous studies loans(q1) total loan and advances dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) total deposits(q2) total deposits xiaoqing maggie and heffernan (2007) other earning assets(q3) total assets net off total loans and fixed assets dong et al. (2014) (continued) 50 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 error term has two disturbance terms given as follows: where, vn is the random error term and it is also known as the twosided noise component and we assume that this is independent and identically distributed vn refers to the stochastic nature of the cost function. un represents the nonnegative disturbance which indicates deviation from the efficient cost frontier of individual banks and serves as a proxy for technical efficiency and assumed to be distributed independently of vn and follows a halfnormal distribution or truncated normal distribution i.e. n (0, ). the notation n (0, ) and n (0, ) mean normally distributed with zero mean and variance. by using intermediation approach (sealey & lindley, 1977) and by following deelchand and padgett (2009a), we developed the following multiproduct trans log cost function to specify the cost function: according to jondrow et al. (1982), the expected value of un, on conditional εn, represents the cost-inefficiency of bank n (which is defined as cn). variable definition source outputs non-interest income(q4) income other than interest on loan and advances dong et al. (2014), xiaoqing maggie and heffernan (2007) off-balance sheet items (q5) off-balance sheet commitments and contingencies manlagñit (2011) inputs price of labor (p1) ratio of staff expenses and number of employees berger and hannan (1998), dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) price of deposits(p2) ratio of interest expenses and total customer deposits dong et al. (2014), xiaoqing maggie and heffernan (2007) , manlagñit (2011) price of physical capital(p3) ratio of other noninterest expenses and physical capital. dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) 12 inputs price of labor (p1) ratio of staff expenses and number of employees berger and hannan (1998), dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) price of deposits(p2) ratio of interest expenses and total customer deposits dong et al. (2014), xiaoqing maggie and heffernan (2007) , manlagñit (2011) price of physical capital(p3) ratio of other non-interest expenses and physical capital. dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) error term has two disturbance terms given as follows: n = un + vn where, vn is the random error term and it is also known as the twosided noise component and we assume that this is independent and identically distributed n (0, ). vn refers to the stochastic nature of the cost function. un represents the nonnegative disturbance which indicates deviation from the efficient cost frontier of individual banks and serves as a proxy for technical efficiency and assumed to be distributed independently of vn and follows a half-normal distribution or truncated normal distribution i.e. n (0, ). the notation n (0, ) and n (0, ) mean normally distributed with zero mean and variance. by using intermediation approach (sealey & lindley, 1977) and by following deelchand and padgett (2009a), we developed the following multiproduct trans log cost function to specify the cost function: + + ½ + ½ (2) according to jondrow et al. (1982), the expected value of un, on n, represents the cost-inefficiency of bank n (which is defined as cn). = =[ ][ (3) n to standard deviation of vn is the standard normal density function. cn can be estimated by using equation (3). we used computer software called frontier version 4.1 (3) 12 inputs price of labor (p1) ratio of staff expenses and number of employees berger and hannan (1998), dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) price of deposits(p2) ratio of interest expenses and total customer deposits dong et al. (2014), xiaoqing maggie and heffernan (2007) , manlagñit (2011) price of physical capital(p3) ratio of other non-interest expenses and physical capital. dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) error term has two disturbance terms given as follows: n = un + vn where, vn is the random error term and it is also known as the twosided noise component and we assume that this is independent and identically distributed n (0, ). vn refers to the stochastic nature of the cost function. un represents the nonnegative disturbance which indicates deviation from the efficient cost frontier of individual banks and serves as a proxy for technical efficiency and assumed to be distributed independently of vn and follows a half-normal distribution or truncated normal distribution i.e. n (0, ). the notation n (0, ) and n (0, ) mean normally distributed with zero mean and variance. by using intermediation approach (sealey & lindley, 1977) and by following deelchand and padgett (2009a), we developed the following multiproduct trans log cost function to specify the cost function: + + ½ + ½ (2) according to jondrow et al. (1982), the expected value of un, on n, represents the cost-inefficiency of bank n (which is defined as cn). = =[ ][ (3) n to standard deviation of vn is the standard normal density function. cn can be estimated by using equation (3). we used computer software called frontier version 4.1 12 inputs price of labor (p1) ratio of staff expenses and number of employees berger and hannan (1998), dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) price of deposits(p2) ratio of interest expenses and total customer deposits dong et al. (2014), xiaoqing maggie and heffernan (2007) , manlagñit (2011) price of physical capital(p3) ratio of other non-interest expenses and physical capital. dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) error term has two disturbance terms given as follows: n = un + vn where, vn is the random error term and it is also known as the twosided noise component and we assume that this is independent and identically distributed n (0, ). vn refers to the stochastic nature of the cost function. un represents the nonnegative disturbance which indicates deviation from the efficient cost frontier of individual banks and serves as a proxy for technical efficiency and assumed to be distributed independently of vn and follows a half-normal distribution or truncated normal distribution i.e. n (0, ). the notation n (0, ) and n (0, ) mean normally distributed with zero mean and variance. by using intermediation approach (sealey & lindley, 1977) and by following deelchand and padgett (2009a), we developed the following multiproduct trans log cost function to specify the cost function: + + ½ + ½ (2) according to jondrow et al. (1982), the expected value of un, on n, represents the cost-inefficiency of bank n (which is defined as cn). = =[ ][ (3) n to standard deviation of vn is the standard normal density function. cn can be estimated by using equation (3). we used computer software called frontier version 4.1 12 inputs price of labor (p1) ratio of staff expenses and number of employees berger and hannan (1998), dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) price of deposits(p2) ratio of interest expenses and total customer deposits dong et al. (2014), xiaoqing maggie and heffernan (2007) , manlagñit (2011) price of physical capital(p3) ratio of other non-interest expenses and physical capital. dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) error term has two disturbance terms given as follows: n = un + vn where, vn is the random error term and it is also known as the twosided noise component and we assume that this is independent and identically distributed n (0, ). vn refers to the stochastic nature of the cost function. un represents the nonnegative disturbance which indicates deviation from the efficient cost frontier of individual banks and serves as a proxy for technical efficiency and assumed to be distributed independently of vn and follows a half-normal distribution or truncated normal distribution i.e. n (0, ). the notation n (0, ) and n (0, ) mean normally distributed with zero mean and variance. by using intermediation approach (sealey & lindley, 1977) and by following deelchand and padgett (2009a), we developed the following multiproduct trans log cost function to specify the cost function: + + ½ + ½ (2) according to jondrow et al. (1982), the expected value of un, on n, represents the cost-inefficiency of bank n (which is defined as cn). = =[ ][ (3) n to standard deviation of vn is the standard normal density function. cn can be estimated by using equation (3). we used computer software called frontier version 4.1 12 inputs price of labor (p1) ratio of staff expenses and number of employees berger and hannan (1998), dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) price of deposits(p2) ratio of interest expenses and total customer deposits dong et al. (2014), xiaoqing maggie and heffernan (2007) , manlagñit (2011) price of physical capital(p3) ratio of other non-interest expenses and physical capital. dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) error term has two disturbance terms given as follows: n = un + vn where, vn is the random error term and it is also known as the twosided noise component and we assume that this is independent and identically distributed n (0, ). vn refers to the stochastic nature of the cost function. un represents the nonnegative disturbance which indicates deviation from the efficient cost frontier of individual banks and serves as a proxy for technical efficiency and assumed to be distributed independently of vn and follows a half-normal distribution or truncated normal distribution i.e. n (0, ). the notation n (0, ) and n (0, ) mean normally distributed with zero mean and variance. by using intermediation approach (sealey & lindley, 1977) and by following deelchand and padgett (2009a), we developed the following multiproduct trans log cost function to specify the cost function: + + ½ + ½ (2) according to jondrow et al. (1982), the expected value of un, on n, represents the cost-inefficiency of bank n (which is defined as cn). = =[ ][ (3) n to standard deviation of vn is the standard normal density function. cn can be estimated by using equation (3). we used computer software called frontier version 4.1 12 inputs price of labor (p1) ratio of staff expenses and number of employees berger and hannan (1998), dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) price of deposits(p2) ratio of interest expenses and total customer deposits dong et al. (2014), xiaoqing maggie and heffernan (2007) , manlagñit (2011) price of physical capital(p3) ratio of other non-interest expenses and physical capital. dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) error term has two disturbance terms given as follows: n = un + vn where, vn is the random error term and it is also known as the twosided noise component and we assume that this is independent and identically distributed n (0, ). vn refers to the stochastic nature of the cost function. un represents the nonnegative disturbance which indicates deviation from the efficient cost frontier of individual banks and serves as a proxy for technical efficiency and assumed to be distributed independently of vn and follows a half-normal distribution or truncated normal distribution i.e. n (0, ). the notation n (0, ) and n (0, ) mean normally distributed with zero mean and variance. by using intermediation approach (sealey & lindley, 1977) and by following deelchand and padgett (2009a), we developed the following multiproduct trans log cost function to specify the cost function: + + ½ + ½ (2) according to jondrow et al. (1982), the expected value of un, on n, represents the cost-inefficiency of bank n (which is defined as cn). = =[ ][ (3) n to standard deviation of vn is the standard normal density function. cn can be estimated by using equation (3). we used computer software called frontier version 4.1 9 determinants of bank cost efficiency: empirical evidence from bangladesh:29-54 and lovell, 1979) based on the production frontier. according to this methodology, due to inefficiency and random noise in the observed cost of a bank, the following is formulated to deviate from cost-efficient frontier (deelchand & padgett, 2009a). for the nth bank, ln tcn = f (ln qi, ln pj) + εn (1) where, tcn represents total operating cost including financial costs, qi indicates outputs, pj stands for input, εn shows the deviation of actual total cost of a bank from the cost-efficient frontier. table 1 shows the list of input and output variables used in the calculation of ce. table 1. description of input and output variables used in the calculation of ce variables definition sources dependent variable , total cost (tc) operating cost including interest expenses used in all the previous studies outputs loans(q1) total loan and advances dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) total deposits(q2) total deposits xiaoqing maggie and heffernan (2007) other earning assets(q3) total assets net off total loans and fixed assets dong et al. (2014) non-interest income(q4) income other than interest on loan and advances dong et al. (2014), xiaoqing maggie and heffernan (2007) off-balance sheet items (q5) off-balance sheet commitments and contingencies manlagñit (2011) inputs price of labor (p1) ratio of staff expenses and number of employees berger and hannan (1998), dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) price of deposits(p2) ratio of interest expenses and total customer deposits dong et al. (2014), xiaoqing maggie and heffernan (2007) , manlagñit (2011) price of physical capital(p3) ratio of other non-interest expenses and physical capital. dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) error term has two disturbance terms given as below: εn = un + vn where, vn is the random error term and it is also known as the two-sided noise component and we assume that this is independent and identically distributed n (0, ). vn refers the stochastic nature of the cost function. un represents the nonnegative disturbance which indicates deviation from the efficient cost frontier of individual bank and serves as a proxy for technical efficiency and assumed to be distributed independently of vn and follows a half-normal distribution or truncated normal distribution i.e. n (0, ). the notation n (0, ) and n (0, ) mean 9 determinants of bank cost efficiency: empirical evidence from bangladesh:29-54 and lovell, 1979) based on the production frontier. according to this methodology, due to inefficiency and random noise in the observed cost of a bank, the following is formulated to deviate from cost-efficient frontier (deelchand & padgett, 2009a). for the nth bank, ln tcn = f (ln qi, ln pj) + εn (1) where, tcn represents total operating cost including financial costs, qi indicates outputs, pj stands for input, εn shows the deviation of actual total cost of a bank from the cost-efficient frontier. table 1 shows the list of input and output variables used in the calculation of ce. table 1. description of input and output variables used in the calculation of ce variables definition sources dependent variable , total cost (tc) operating cost including interest expenses used in all the previous studies outputs loans(q1) total loan and advances dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) total deposits(q2) total deposits xiaoqing maggie and heffernan (2007) other earning assets(q3) total assets net off total loans and fixed assets dong et al. (2014) non-interest income(q4) income other than interest on loan and advances dong et al. (2014), xiaoqing maggie and heffernan (2007) off-balance sheet items (q5) off-balance sheet commitments and contingencies manlagñit (2011) inputs price of labor (p1) ratio of staff expenses and number of employees berger and hannan (1998), dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) price of deposits(p2) ratio of interest expenses and total customer deposits dong et al. (2014), xiaoqing maggie and heffernan (2007) , manlagñit (2011) price of physical capital(p3) ratio of other non-interest expenses and physical capital. dong et al. (2014), xiaoqing maggie and heffernan (2007), manlagñit (2011) error term has two disturbance terms given as below: εn = un + vn where, vn is the random error term and it is also known as the two-sided noise component and we assume that this is independent and identically distributed n (0, ). vn refers the stochastic nature of the cost function. un represents the nonnegative disturbance which indicates deviation from the efficient cost frontier of individual bank and serves as a proxy for technical efficiency and assumed to be distributed independently of vn and follows a half-normal distribution or truncated normal distribution i.e. n (0, ). the notation n (0, ) and n (0, ) mean (2) determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 51 where, λ is the ratio of the standard deviation of un to standard deviation of vn, φ is the cumulative standard normal density function, and ϕ is the standard normal density function. cn can be estimated by using equation (3). we used computer software called frontier version 4.1 developed by coelli (1996) for stochastic frontier production and cost function was estimated by the method of maximum likelihood. 3.2.2 independent variables: determinants of ce the potential determinants of bank cost efficiency are enumerated in this section. table 2 represents the circumstantial variables used in the study. we considered only the bank specific variables in our study. as mentioned in the literature part, based on previous studies as well as theoretical background we selected a number of bank specific variables which included capital ratio, risk level, profitability, bank size, total loans to total deposits, ownership dummy, operating expenses, non-interest income, number of branches and years of operation as potential determinants of ce. capital ratio was measured by using capital adequacy ratio (car) as required by basel capital regulation, it is noted that, commercial banks in bangladesh have to maintain a capital of tk 4,000 million or 10% on risk weighted assets, whichever is higher. secondly, to see the impact of risk profile (risk) we considered the ratio of non-performing loans to total loans (npltl) as a proxy for risk. in a third world country like bangladesh, banks which provide more loans, are expected to have high credit risk. it is expected that npltl has a negative effect on ce. in accordance with previous studies, ownership structure represented by ownd was taken as the independent variable to see its impact on ce. if the bank is a private commercial bank, then the ownd equals 1 and otherwise, 0. in some studies it was found that state owned commercial banks were more efficient whereas in other studies, it was the private commercial banks which were more efficient. therefore, there is no clear opinion on this variable. to determine whether more profitable banks were more cost efficient, return on assets (roa) was used to see the impact of profitability. nowadays, banks provide fee-based services. we used non interest income to total assets (niita) as a proxy to see the impact of this fee-based service on ce. in bangladesh, this type of banking activities is still not very much popular. since generating income from this new service line may be costly, that is why, in some cases its correlation with bank ce can be mixed. to see the impact of lending propensity on ce we used the ratio of total loans to total deposits (tltd) as a proxy for lending propensity. the bank which lends more can generate more income in return. so, it is expected to have a positive effect on bank ce. it is expected that as banks grow they can perform efficiently. to examine the impact of experience of banks in ce,we used number of years in operation (age) as a proxy to measure experience and we expect that banks with more 52 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 years in operation are more ce. to see the impact of bank size we considered natural logarithm of total assets (size). some authors found that small banks were more cost inefficient than large banks. some other authors found that both small and large banks were cost efficient. it is expected that bank size has a positive effect on ce. it was found in the literature that banks with higher levels of operating costs were less cost efficient. hence, we considered the level of operating cost (oc) to observe its impact on ce. it is expected to have a negative association between them. finally, to see whether banks with a high number of branches are cost efficient or not, we considered a number of branches (nb) as a potential determinant of ce. to see whether ce showed persistence results we considered ce in the last year or lagged ce which is proxied as ce(-1) as a potential determinant of ce. we expected a positive relationship between lagged ce and ce. table 2. description of variables used in the study variable definition source cost efficiency (ce) calculated from sfa hafez (2018); zheng et al. (2018); tecles and tabak (2010) capital (car) ratio of capital to risk-weighted assets (tier 1 capital + tier 2 capital) / risky weighted assets hafez (2018); avkiran (2011); hughes and mester (2008);chang and chiu (2006) risk (npltl) non performing loan to total loan carvallo and kasman (2005); lee et al. (2014); zheng et al. (2018) profitability (roa) return on total asset sun and chang (2011); vennet (2002); ariff and luc (2008); maudos et al. (2002); ray and das (2010) size (size) natural logarithm of total asset tecles and tabak (2010); wezel (2010); ariff and luc (2008); girardone et al. (2004); carvallo and kasman (2005); staikouras et al. (2008) total loan to total deposit (tltd) the ratio of total loan to total deposit anwar (2015) ownership (ownd) ownd equals 1 if the bank is a private commercial bank and 0 otherwise ariff and luc (2008), xiaoqing maggie and heffernan (2007); xiaogang et al. (2005); schulz et al. (2009) operating expenses (oc) total opertaing expenses dietsch and lozano-vivas (2000) (continued) determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 53 variable definition source non-interest income ratio (niita) non interest income to total assets claessens et al. (2001); vennet (2002) nb number of branches authors’ idea experience of banks (age) no. of years commencing from operation authors’ idea. 3.3 econometric model there has been a longstanding debate in measuring efficiency of a cost frontier. the methodologies can be classified broadly into two models, parametric and non-parametric. in the parametric model sfa, the thick frontier approach (tfa) and distribution free approach (dfa) are used. in non-parametric approaches linear programming techniques are applied such as dea and free disposal hull analysis (fdha) (eisazadeh & shaeri, 2012). however, until today, researchers have reached no consensus about the best appropriate estimation methodology. in some studies, dea method is applied because it is a non-parametric and deterministic method and, hence, does not need an anterior assumption about the analytical cost function. with dea, multiple inputs and outputs of a complex cost or production function can be explained with a single efficiency indicator. in our study, we used sfa because it has some advantages over dea. first of all, the research output of decision-making units (dmu) are not fixed but affected by many random variables. such random measurement errors are taken into account under sfa. secondly, in sfa, the regression parameters and efficiency values are unbiasedly estimated in single step (wang & schmidt, 2002). thirdly, sfa mitigates the problem of explicit mitigation of a functional connection, because under this approach it is possible to test what production function fits the data best (ehlers, 2011). last but not least, hierarchically performance model development is possible by combining multi-level models with sfa. for potential endogeneity problem, two step dynamic panel data approach was applied which was suggested by arellano and bover (1995) and blundell and bond (2000). we also applied dynamic panel gmm technique (doytch & uctum, 2011) to correct potential heteroscedasticity, endogeneity and autocorrelation problems. the standard gmm estimator is known to be a consistent estimator of as n (the size of the units) approaches infinity. however, this standard gmm estimator has been found to have poor finite sample properties (bias) in the case in which the series is highly persistent (blundell & bond, 1998). in these circumstances, the lagged levels of the series are only weakly correlated with subsequent first differences, thus leading to weak instruments for the firstdifferenced equations. arellano and bover (1995) and blundell and bond (1998) 54 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 demonstrated that the system-gmm approachby adding additional moment restrictions permits lagged first differences to be used as instruments in the levels equations, and this corrects any bias that could emerge using the standard gmm estimator. out of several gmm approaches, the two step feasible gmm is more popular, particularly for the system gmm. before applying gmm, the variables were tested for their correlation, other tests like serial correlation test, test for heteroskedasticity, test for cross sectional dependence, test for endogeinity were conducted. diagnosis tests like sargan test, arellano-bond order 1 (2) test for first (second) order correlation for gmm estimator were also performed. on the basis of literature review and our chosen variables, we developed the following econometric model to investigate the impact of potential determinants on ce of commercial banks in bangladesh. ceit= α0+ α1carit+ α2riskit+ α3 ceit-1+ α4roait + α5sizeit +α6tltdit+ α7owndit +α8 nbit+ α9 ocit + α10 niita it + α11ageit +eit (4) here, subscript i represents the cross-sectional dimension across the bank and subscript t represents the time dimension. error term is represented by eit. 3.4 model efficiency test we conducted robustness checks to validate the empirical results of gmm estimators namely the arellano-bond dynamic panel data estimation. for testing the robustness, we made three changes in our independent variables. we used equity to total assets (eqtta) instead of car as a proxy for capital regulation, return on equity (roe) instead of roa as a proxy for profitability measure, loan loss provisions to non-performing loans (llpnpl) instead of npltl as a proxy for risk measure. we used these three changes because these three variables are very crucial for the banking sector in bangladesh. for instance, banks need to maintain minimum regulatory capital according to basel iii, and this capital regulation is introduced to control the risk taking behavior of banks. finally, banks manager are responsible for the profit earning capacity of banks since their aim is to maximize the wealth of shareholders. 4. empirical findings results are derived from the econometric model in this section. in measuring ce, we used sfa model. to determine the degree of multicollinearity among the independent variables and the level of correlation between ce and its determinants, pearson’s correlation coefficient was used. in our further analysis, we used four different regression estimators—ordinary least square (ols), panel random/fixed effect (re/fe) model, gmm specifically system dynamic panel data estimation and arellano-bond dynamic panel data estimation. we determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 55 also used hausman test to see whether panel fe/re model was appropriate for our panel data (aliyu & yusof, 2016; hausman, 1978). in addition, to test the existence of serial correlation, wooldridge test for autocorrelation (wooldridge, 1990) was applied. breusch-pagan/cookweisberg test of heteroscedasticity was also used to support the re requirement for balanced panel observation (breusch & pagan, 1979; cook & weisberg, 1982). to see cross sectional dependence of variables, we ran pesaran’s test (pesaran, 2004). in addition, we performed a hausman test of endogeneity in our measurement to find out whether there was endogeneity (hausman & taylor, 1981). finally, sargan’s test was carried out to show that the tools were valid (sargan, 1958). this section is divided into three sub-sections. section 1 reports on descriptive statistics, section 2 presents the correlation analysis and finally section 3 reports on the regression analysis. 4.1 descriptive statistics descriptive statistics of all variables, used in this study are reported in table 3. the average value of efficiency was 1.108 or 110.8%. the maximum value was 1.55 or 155% and the minimum value was 1.01 or 101%. a value of more than 100% was referred to as cost inefficiency of banks in bangladesh. the standard deviation (sd) of efficiency was 10.98% which indicated that efficiency varied very slightly from bank to bank. table 3. descriptive statistics variable mean std. dev. minimum maximum observation ce 1.108132 0.109836 1.01146 1.54623 264 car 13.63954 64.55062 -108.49 102.500 264 npltl 0.085236 0.155629 0.0073 1.00000 264 tltd 0.857377 0.508061 0.083489 0.88289 264 roa 0.009769 0.015658 -0.109182 0.05099 264 niita 0.028712 0.012360 0.002694 0.10110 264 age 22.01515 9.892952 8.00000 45.0000 264 oc 4267.204 3225.439 455.3226 19579.57 264 roe 0.121273 0.224974 -2.74082 1.40864 264 size 11.92754 0.763578 9.413911 13.9992 264 source: authors’ calculations. 56 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 the average car was 13.64% which was higher than the minimum requirement of capital in bangladesh i.e. 10% whereas the maximum value was 102.00% and the minimum value was –108.49%. the average credit risk (npltl) was 9% and it ranged from a minimum 0.07% to a maximum 100%. the average total liabilities to total deposits was 85.73%, whereas the maximum value was 88% and the minimum value was 83%. the average roa and roe, which indicated profitability, was 0.97% and 12.12%, respectively, which ranged from -10.91% to 5.09% and -274% to 140%, respectively. both of these varied slightly from bank to bank (sd was 0.016 and 0.22, respectively). the mean niita was 2.9% and it ranged from a minimum of 0.03% to a maximum of 10.11%. the average age of bank was 22.01 years, whereas the maximum age was 45 years and the minimum age was 8 years. from the table it can also be seen that the average operating cost was 4267.20 million bangladeshi taka, with a maximum value of 19579.57 million bangladeshi taka and a minimum value of 455.32 million bangladeshi taka. the table also shows that the mean size of banks (natural logarithm of total assets) was 11.927 and it ranged from a minimum of 9.415 to a maximum of 13.99. 4.2 correlation the degree of correlation between variables used in the study is reported in table 4. we used pearson’s correlation coefficient here. from the matrix, it can be seen that the correlations between the majority of the variables did not support the severe problem of multicollinearity, only ownd and nb variables showed the problem of multicollinearity, hence we did not report them in the correlation matrix. it is noted that, multicollinearity exists when the correlation between two variables exceeds 0.80 (kennedy, 2008). it can be seen from the table that the correlation coefficient between the independent variables were less than 0.80. therefore, it can be concluded that there is no multicollinearity problem in our case. table 4. correlation matrix for the dependent and independent variables ce 1 car 2 npltl 3 roa 4 tltd 5 age 6 niita 7 oc 8 size 9 1 1.000 2 -0.162*** 1.000 3 0.453*** -0.114* 1.000 4 -0.403*** 0.116* -0.750*** 1.000 (continued) determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 57 ce 1 car 2 npltl 3 roa 4 tltd 5 age 6 niita 7 oc 8 size 9 5 0.005 -0.057 -0.055 0.031 1.000 6 0.093 0.091 0.279*** -0.195*** -0.116* 1.000 7 -0.048 0.091 -0.223*** 0.428*** -0.014 0.008 1.000 8 0.137** 0.318*** 0.019 -0.029 -0.106 0.615*** 0.025 1.000 9 -0.255*** 0.268*** -0.281*** 0.201*** -0.086* 0.553*** 0.019 0.083*** 1.000 note: the table shows the results of pearson’s correlation coefficient. ***,** and * indicates significance at 1%, 5% and 10%, respectively. in addition, to examine the existence of multicollinearity, we used the variance inflation factor (vif) measures. the results are reported in table 5 as follows. vif is equal to 1/tolerance and it is always less than or equal to 1. there is no formal vif value for determining the presence of multicollinearity. values of vif that exceed 10 are often regarded as indicating multicollinearity (o’brien, 2007). in the vif measure (table 5), the vif of nb variable was found to be at 11.70 which indicated there was existence of collinearity. based on the correlation matrix and vif figures, the nb and ownd variables were omitted in the regression analysis. in the case of other variables, the issue of multicollinearity did not subvert our results. table 5. variance inflation factor variable vif 1/vif nb 11.70 0.085489 ownd 7.83 0.127687 size 5.75 0.173947 oc 5.19 0.192623 npltl 3.40 0.293878 roa 2.85 0.351361 age 2.82 0.354103 niita 1.34 0.748760 car 1.17 0.856657 tltd 1.04 0.963078 mean vif 4.31 58 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 4.3 regression analysis table 6 demonstrates the outcome of gmm estimator where bank ce is the dependent variable. in the first case, we ran the regression using ols regression where we assumed that all the assumptions of ols were valid. in the second case, we applied hausman test for panel fe or re. in our hausman test for panel fe or re, we found that p value was greater than 0.05, so we could not reject that the null hypothesis of re was appropriate. then we ran the regression by using panel re model. the results are reported in table 6. in the third case, we determined whether the data had the problem of endogeneity, heteroscedasticity, autocorrelation and/or cross sectional dependency. the p-value from hausman test for endogeneity was 0.000 which suggested that there was a problem of endogeneity where car and npltl were endogenous variables in the equation of ce. as a result, the ols and re panel model were not appropriate in our case. p values from the serial correlation test was 0.000 which indicated that the null hypothesis of no serial correlation was rejected. this meant that the error terms were serially correlated. in the breuschpagan/cook-weisberg test for heteroscedasticity, the p-value was below an appropriate edge (i.e. p<0.05), therefore the null hypothesis of homoscedasticity was rejected. for testing the presence of cross-sectional independence we took on pesaran's test. here, the p value was 0.000, which indicated there was cross sectional dependency among variables. based on the above test, the gmm data estimator was appropriate for our case. for the dynamic nature of data we used both system-dynamic panel data estimation and arellano-bond dynamic panel data estimation and the results are reported in table 6. it can be seen from table 6 that the arellano-bond dynamic panel data estimation was appropriate for our case, because it showed better results than any other estimator. hence, the outcome from the arellano-bond dynamic panel data estimation were presented in the next paras. from table 6, it can be seen that, the coefficient of capital was negative and it was statistically significant at 1% level of significance. this implied that capitalized banks are not as cost efficient as undercapitalized banks. in a developing country like bangladesh, the cost of capital is more and therefore it is expensive to maintain more capital. hence banks with high capital is less efficient. this finding is inconsistent with our hypothesis h1 and contrasts with findings by hafez (2018). the coefficient of risk (npltl) was negative and statistically significant at 1% level of significance. this states that banks with high risk are less cost efficient than banks having low risk. the negative sign also indicates that riskier banks are more cost inefficient. bad management hypothesis becomes evident in that case. this outcome is unswerving to our preliminary hypothesis (h2) and is supported by carvallo and kasman (2005). turning to the impact of bank profitability on ce, it is evident that profitable banks are more cost efficient. they can organize their costs well which may in turn make them more profitable. this complies with our initial hypothesis (h3). determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 59 table 6. determinants of ce variable ols panel re system-dynamic panel data estimation arellano-bond dynamic panel data estimation constant 2.4543***(14.94) 1.168***(20.34) -0.04575***(-4.91) -0.0499***(-21.45) ce(-1) 1.0327***(1616.48) 1.0549***(748.73) car -0.0003***(-4.10) -0.0004***(-2.66) -3.39e-06***(-9.19) -1.08e-06***(-5.96) npltl 0.05619(1.09) -0.0296**(-2.43) -0.00138***(-20.56) -0.0012***(-9.09) roa -0.8819*(-1.86) 0.2994***(2.82) 0.02845***(15.88) 0.000762***(2.83) tltd -0.0010(-0.12) -0.00024(-0.14) -0.00005(-1.18) -0.0255***(-12.43) age 0.0023***(3.17) 0.0079***(8.65) 0.00022***(11.24) 0.01662***(6.16) niita 0.4703(1.13) 0.08821(071) 0.03702***(14.69) 0.000207***(6.08) oc 0.0036***(12.04) -4.8e-06*** (-5.52) -5.6e-07***(-15.04) -1.23e-07***(-5.88) size -0.139***(-10.05) -0.036***(-6.38) -0.0034***(-88.52) -0.00074***(-6.07) xttest0 (p-value) 0.0000 hausman test for panel fe/re (p-value) 0.5273 serial correlation test (wooldridge test for autocorrelation, p-value) 0.000 0.000 breusch-pagan/ cook-weisberg test for heteroscedasticity (p-value) 0.000 0.000 pesaran’s test of cross sectional independence (p-value) 0.000 0.000 hausman test for endogeneity f (p-value) 16.82 (0.000) 17.38 (0.000) sargan’s test (p-value) 0.999 0.989 ar(1) 0.26 0.24 ar(2) 0.38 0.36 observations 264 264 264 264 number of banks 33 33 33 33 note: the table shows the empirical results from ols, panel re, system gmm and arellano-bond estimation respectively. the dependent variable is cost efficiency (ce) in each case. ***,** and * indicates significance at 1%, 5% and 10%, respectively. t-statistics are shown in parentheses. ce(-1) indicates cost efficiency in lag one. 60 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 the coefficient of tltd was negative which indicated that more aggressive banks were involved in lending activities as compared to deposits and therefore were less cost efficient. in the case of developed countries, the scenario may be different (ariff & luc, 2008). however, in the case of developing countries like bangladesh, where people have a tendency to default on loans, a higher tltd may hinder efficiency because it increases the cost related to cash collection. as expected, experience (age) had a positive and statistically significant impact on cost efficiency. the findings here supports our initial hypothesis (h4). it demonstrates that as the experience of banks increases, banks can manage their cost more efficiently. experience theory and learning theory are effective in the case of banks in bangladesh. the coefficient between niita was positive and significant which indicated that banks which had more fee income, were more cost efficient. this may happen because this source of income is more stable and riskless than that of interest income. it could also be seen from the results that the coefficient of operating cost was negative and statistically significant at 1% level of significance which implied that banks with high operating cost were less cost efficient. this outcome also supports our previous literature and our hypothesis (h5). in respect of bank size, it was found that size had a negative and statistically significant impact on banks’ ce which suggested that large banks were less cost efficient than those of small banks. one possible reason may be the inclusion of state owned banks in the data. in bangladesh, state owned banks are large and hence found to be inefficient. many control mechanisms help balance the interests of owners and managers in private commercial banks and improve their quality that is missing from state-owned commercial banks (e.g.: potentially higher exposure of private businesses to takeovers and subsequent job losses, etc.) (fama, 1980; perera et al.; 2007). moreover, managers of stateowned companies may be inclined to follow their own interests or the interests of lobbyists to the detriment of those of the company (di patti & hardy, 2005; perera et al., 2007). our outcomes supported ariff and luc (2008) and berger et al. (2009), but contradicted that of xiaogang et al. (2005). this finding does not support our hypothesis (h6). finally, the coefficient of ce(-1) was positive and significant at 1% level of significance which meant that cost efficiency of the current year was also the result of previous year’s cost efficiency. it meant that banks’ cost efficiency showed persistent results which suggested that banks with high efficiency in the previous year(s) were also efficient in the current year. this result supports our hypothesis (h7). diagnostic tests of gmm estimator in all cases showed that the models were accurately specified. for instance, the p-values of sargan’s tests were determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 61 insignificant, indicating that the null hypothesis that instruments which were not exogeneous were not rejected and confirmed that the instruments used were valid. similarly, the insignificant ar(2) p-values confirmed that there was no second-order serial correlation in residuals. 4.4 model efficiency and robustness test we conducted robustness checks to validate the empirical results of gmm estimators and the estimated results were reported in the following table 7. after replacing those three variables as mentioned in the methodology section, we tested our data again for the existence of multicollinearity, serial correlation, heteroscedasticity and cross section dependence and endogeneity problems. the results showed that there was no problem of multicollinearity. null hypotheses of no serial correlation, homoscedasticity in error terms, no endogeneity and cross sectional independence were rejected. therefore, we applied only gmm estimation. the model showed similar results and was consistent with table 6 in its association with the determinants of ce. in terms of the level of significance of all factors, we also found similarities with some exceptions for the t-value. similar to table 6, we found that all the variables were highly significantly related with ce in bangladesh. hence, we can conclude that our model for determinants of ce is robust for different measures of variables. table 7. determinants of cost efficiency (arellano-bond dynamic panel data estimation) variable coefficient constant -0.3189***(-18.51) ce(-1) 0.9849***(81.29) eqtta -1.07e-05***(-6.18) llpnpl -0.0035***(-8.07) roe 0.000892***(3.38) tltd -0.02093***(-13.58) age 0.01776***(7.19) niita 0.000309***(7.06) oc -1.54e-06***(-6.76) size -0.00087***(-7.04) serial correlation test (wooldridge test for autocorrelation, p-value) 0.000 (continued) 62 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 variable coefficient breusch-pagan/cook-weisberg test for heteroscedasticity (p-value) 0.000 pesaran’s test of cross sectional independence (p-value) 0.000 hausman test for endogeneity f (p-value) 21.56 (0.000) sargan’s test (p-value) 0.765 ar(1) 0.28 ar(2) 0.50 observations 264 number of banks 33 note: the table shows the empirical results from arellano-bond estimation respectively. the dependent variable is cost efficiency (ce). ***,** and * indicates significance at 1%, 5% and 10%, respectively. t-statistics are shown in parentheses. ce(-1) indicates cost efficiency in lag one. 5. conclusion and recommendation this study empirically examined the factors affecting ce for the latest data set of bangladeshi banks for the period from 2009 to 2016. empirical results showed that among the pooled ols, fe or re panel regression, system-dynamic panel data estimation and arellano-bond dynamic panel data estimation, the arellano-bond dynamic panel data estimation was the most suited for our data which had problems of endogeneity, heteroscedasticity, serial correlation and cross sectional dependence. the empirical results showed that capitalized banks were not as cost efficient as undercapitalized banks. in a developing country like bangladesh, it is expensive to maintain more capital. hence banks with high capital is less efficient. it is also evident that banks with high risk are less cost efficient than banks having low risk. as a result, bad management hypothesis becomes evident in this case. the results also showed that profitable banks were more cost efficient as they could organize their cost well which would in turn make them more profitable. while more aggressive banks which were involved in more lending activities as compared to deposits were less cost efficient. in bangladesh, people have a tendency to default on loans, thus a higher tltd may hinder efficiency because it increases the loss and cost. it was also observed from the results that experienced banks were more cost efficient which supported the experience theory and learning theory. results also stated that banks with more non-interest income were more cost efficient determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 63 than those with lower non-interest income. this is because this source of income is more stable and riskless than that of interest income. it can also be seen from the results that banks with high operating cost were less cost efficient. in respect of bank size, it was found that large banks were less cost efficient than those of small banks. one possible reason may be the inclusion of state owned banks in the data. in bangladesh, state owned banks are large and hence found to be inefficient. finally, ce of the current year is also the result of previous year’s ce which indicates the persistency in ce of banks in bangladesh. the empirical results from the current study provides considerable policy applications. policymakers should exercise caution in designing regulatory capital as this could decrease the efficiency of banks and they also focus on aggressiveness of banks in providing loans so that bad loans. the bank manager should diversify banking activities to generate non-interest income from commissions, charges, etc. policymakers and regulators should also focus on monitoring the operating efficiency of banks and to consider all potential technologies to reduce operating costs of bangladeshi banks. in terms of the management of large banks, their operations should be scaled down to become more efficient. in addition, the government should focus specifically on stateowned banks and that inexperienced banks can be merged with experienced ones to benefit from the experience and learning effect. finally, policymakers should also concentrate on efficient banks so that they can continuously maintain their efficiency. it would be difficult to raise more revenue in a small economy with 58 banks in bangladesh, thus they must be efficient in managing their resources and costs. one needs to be cautious, in comparing the results of previous studies that evaluates the determinants of banks’ ce in developed countries as this literature is still in its infancy. further areas of research should seek to evaluate the consistency of our findings by way of application to a more representative and contemporary sample of banks in developed countries. the approach could also be expanded to examine the consistency of results by using alternative measures of efficiency like dea and by using alternative accounting measures of the variables as well as by including some other determinants. references ab-rahim, r., md-nor, n. g., ramlee, s., & ubaidillah, n. z. (2012). determinants of cost efficiency in malaysian banking. international journal of business & society, 13(3). abdul-majid, m., saal, d. s., & battisti, g. (2011a). efficiency and total factor productivity change of malaysian commercial banks. the service industries journal, 31(13), 2117–2143. 64 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 abdul-majid, m., saal, d. s., & battisti, g. (2011b). the impact of islamic banking on the cost efficiency and productivity change of malaysian commercial banks. applied economics, 43(16), 2033–2054. aigner, d., lovell, c. a. k., & schmidt, p. (1977). formulation and estimation of stochastic frontier production function models. journal of econometrics, 6(1), 21–37. alam, n. (2012). the impact of regulatory and supervisory structures on bank risk and efficiency: evidence from dual banking system. asian journal of finance & accounting, 4(1), 216–244. aliyu, s., & yusof, r. m. (2016). profitability and cost efficiency of islamic banks: a panel analysis of some selected countries. international journal of economics and financial issues, 6(4), 1736–1743. altunbas, y., carbo, s., gardener, e. p., & molyneux, p. (2007). examining the relationships between capital, risk and efficiency in european banking. european financial management, 13(1), 49–70. altunbaş, y., gardener, e. p., molyneux, p., & moore, b. (2001). efficiency in european banking. european economic review, 45(10), 1931–1955. altunbas, y., liu, m.-h., molyneux, p., & seth, r. (2000). efficiency and risk in japanese banking. journal of banking & finance, 24(10), 1605–1628. andrieş, a. m., & căpraru, b. (2014). the nexus between competition and efficiency: the european banking industries experience. international business review, 23(3), 566–579. anwar, m. (2015). bank efficiency and lending propensity: evidence from commercial banks in indonesia. doctoral dissertation, school of management. arellano, m., & bover, o. (1995). another look at the instrumental variable estimation of error-components models. journal of econometrics, 68(1), 29–51. ariff, m., & luc, c. (2008). cost and profit efficiency of chinese banks: a nonparametric analysis. china economic review, 19(2), 260–273. ataullah, a., & le, h. (2006). economic reforms and bank efficiency in developing countries: the case of the indian banking industry. applied financial economics, 16(9), 653–663. avkiran, n. k. (2006). developing foreign bank efficiency models for dea grounded in finance theory. socio-economic planning sciences, 40(4), 275–296. avkiran, n. k. (2011). association of dea super-efficiency estimates with financial ratios: investigating the case for chinese banks. omega, 39(3), 323–334. banna, h., ahmad, r., & koh, e. h. (2017). determinants of commercial banks’ efficiency in bangladesh: does crisis matter? the journal of asian finance, economics and business (jafeb), 4(3), 19–26. determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 65 battaglia, f., farina, v., fiordelisi, f., & ricci, o. (2010). the efficiency of cooperative banks: the impact of environmental economic conditions. applied financial economics, 20(17), 1363–1376. battese, g. e., heshmati, a., & hjalmarsson, l. (2000). efficiency of labour use in the swedish banking industry: a stochastic frontier approach. empirical economics, 25(4), 623–640. berger, a. n., & deyoung, r. (1997). problem loans and cost efficiency in commercial banks. journal of banking & finance, 21(6), 849–870. berger, a. n., & hannan, t. h. (1998). the efficiency cost of market power in the banking industry: a test of the “quiet life” and related hypotheses. review of economics and statistics, 80(3), 454–465. berger, a. n., hasan, i., & zhou, m. (2009). bank ownership and efficiency in china: what will happen in the world’s largest nation? journal of banking & finance, 33(1), 113–130. berger, a. n., & mester, l. j. (1997). inside the black box: what explains differences in the efficiencies of financial institutions? journal of banking & finance, 21(7), 895–947. bitar, m., pukthuanthong, k., & walker, t. (2018). the effect of capital ratios on the risk, efficiency and profitability of banks: evidence from oecd countries. journal of international financial markets, institutions and money, 53, 227–262. blejer, m. i. (2006). economic growth and the stability and efficiency of the financial sector. journal of banking & finance, 30(12), 3429–3432. blundell, r., & bond, s. (1998). initial conditions and moment restrictions in dynamic panel data models. journal of econometrics, 87(1), 115–143. blundell, r., & bond, s. (2000). gmm estimation with persistent panel data: an application to production functions. econometric reviews, 19(3), 321–340. boot, a. w., & thakor, a. v. (1997). banking scope and financial innovation. the review of financial studies, 10(4), 1099–1131. breusch, t. s., & pagan, a. r. (1979). a simple test for heteroscedasticity and random coefficient variation. econometrica: journal of the econometric society, 1287–1294. bush, g. (2015). learning by banking: testing for experience effects in the financial sector. available at ssrn 2688578. carvallo, o., & kasman, a. (2005). cost efficiency in the latin american and caribbean banking systems. journal of international financial markets, institutions and money, 15(1), 55–72. chan, s. g., karim, m. z. a., burton, b., & aktan, b. (2014). efficiency and risk in commercial banking: empirical evidence from east asian countries. the european journal of finance, 20(12), 1114–1132. 66 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 chang, t. c., & chiu, y. h. (2006). affecting factors on risk-adjusted efficiency in taiwan's banking industry. contemporary economic policy, 24(4), 634–648. chiu, y.-h., chen, y.-c., & bai, x.-j. (2011). efficiency and risk in taiwan banking: sbm super-dea estimation. applied economics, 43(5), 587– 602. chiu, y.-h., jan, c., shen, d.-b., & wang, p.-c. (2008). efficiency and capital adequacy in taiwan banking: bcc and super-dea estimation. the service industries journal, 28(4), 479–496. chortareas, g. e., girardone, c., & ventouri, a. (2009). efficiency and productivity of greek banks in the emu era. applied financial economics, 19(16), 1317–1328. chortareas, g. e., girardone, c., & ventouri, a. (2012). bank supervision, regulation, and efficiency: evidence from the european union. journal of financial stability, 8(4), 292–302. christopoulos, d. k., lolos, s. e., & tsionas, e. g. (2002). efficiency of the greek banking system in view of the emu: a heteroscedastic stochastic frontier approach. journal of policy modeling, 24(9), 813–829. čiković, k. f., & cvetkoska, v. (2018). efficiency of the macedonian banking sector: a parametric approach. cea journal of economics, 12(2), 17– 26. claessens, s., demirgüç-kunt, a., & huizinga, h. (2001). how does foreign entry affect domestic banking markets? journal of banking & finance, 25(5), 891–911. coelli, t. j. (1996). a guide to frontier version 4.1 : a computer program for stochastic frontier production and cost function estimation.7, 1-33, cepa working papers. cook, r. d., & weisberg, s. (1982). residuals and influence in regression. new york: chapman and hall. das, a., & ghosh, s. (2009). financial deregulation and profit efficiency: a nonparametric analysis of indian banks. journal of economics and business, 61(6), 509–528. deelchand, t., & padgett, c. (2009a). the relationship between risk, capital and efficiency: evidence from japanese cooperative banks. icma centre discussion papers in finance, dp2009-12. deelchand, t., & padgett, c. (2009b). the relationship between risk, capital and efficiency: evidence from japanese cooperative banks. capital and efficiency: evidence from japanese cooperative banks (december 18, 2009). di patti, e. b., & hardy, d. c. (2005). financial sector liberalization, bank privatization, and efficiency: evidence from pakistan. journal of banking & finance, 29(8–9), 2381–2406. determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 67 dietsch, m., & lozano-vivas, a. (2000). how the environment determines banking efficiency: a comparison between french and spanish industries. journal of banking & finance, 24(6), 985–1004. dong, y., hamilton, r., & tippett, m. (2014). cost efficiency of the chinese banking sector: a comparison of stochastic frontier analysis and data envelopment analysis. economic modelling, 36, 298–308. doytch, n., & uctum, m. (2011). does the worldwide shift of fdi from manufacturing to services accelerate economic growth? a gmm estimation study. journal of international money and finance, 30(3), 410–427. drake, l., & hall, m. j. (2003). efficiency in japanese banking: an empirical analysis. journal of banking & finance, 27(5), 891–917. ehlers, r. s. (2011). comparison of bayesian models for production efficiency. journal of applied statistics, 38(11), 2433–2443. eisazadeh, s., & shaeri, z. (2012). an analysis of bank efficiency in the middle east and north africa. the international journal of banking and finance, 9(4), 28–47. esho, n. (2001). the determinants of cost efficiency in cooperative financial institutions: australian evidence. journal of banking & finance, 25(5), 941–964. faccio, m., marchica, m.-t., & mura, r. (2016). ceo gender, corporate risktaking, and the efficiency of capital allocation. journal of corporate finance, 39, 193–209. fama, e. f. (1980). agency problems and the theory of the firm. journal of political economy, 88(2), 288–307. färe, r., grosskopf, s., & weber*, w. l. (2004). the effect of risk-based capital requirements on profit efficiency in banking. applied economics, 36(15), 1731–1743. fethi, m. d., & pasiouras, f. (2010). assessing bank efficiency and performance with operational research and artificial intelligence techniques: a survey. european journal of operational research, 204(2), 189–198. fiordelisi, f., & mare, d. s. (2013). probability of default and efficiency in cooperative banking. journal of international financial markets, institutions and money, 26, 30–45. fiordelisi, f., marques-ibanez, d., & molyneux, p. (2011). efficiency and risk in european banking. journal of banking & finance, 35(5), 1315–1326. fioretti, g. (2007). the organizational learning curve. european journal of operational research, 177(3), 1375–1384. fries, s., & taci, a. (2005). cost efficiency of banks in transition: evidence from 289 banks in 15 post-communist countries. journal of banking & finance, 29(1), 55–81. 68 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 gardener, e., molyneux, p., & nguyen-linh, h. (2011). determinants of efficiency in south east asian banking. the service industries journal, 31(16), 2693–2719. garza-garcía, j. g. (2012). determinants of bank efficiency in mexico: a twostage analysis. applied economics letters, 19(17), 1679–1682. girardone, c., molyneux, p., & gardener, e. p. (2004). analysing the determinants of bank efficiency: the case of italian banks. applied economics, 36(3), 215–227. hafez, h. m. m. (2018). examining the relationship between efficiency and capital adequacy ratio: islamic versus conventional banks—an empirical evidence on egyptian banks. accounting and finance research, 7(2), 232. hall, m. j., & simper, r. (2013). efficiency and competition in korean banking. applied financial economics, 23(10), 881–890. hammim, m., abdullah, n., & al-habshi, s. (2006). efficiency of islamic banks in malaysia: a stochastic frontier approach. journal of economic cooperation among islamic countries, 27(2), 37–70. han, y., kim, m. h., & kim, w.-j. (2012). determinants of profit efficiency: evidence from korean savings banks. applied financial economics, 22(12), 1003–1016. hasan, i., & marton, k. (2003). development and efficiency of the banking sector in a transitional economy: hungarian experience. journal of banking & finance, 27(12), 2249–2271. hassan, m., & hassan, m. (2018). cost efficiency of banking sector of bangladesh: evidence using the stochastic frontier analysis. asian journal of empirical research, 8(6), 208–224. hauner, d. (2005). explaining efficiency differences among large german and austrian banks. applied economics, 37(9), 969–980. hausman, j. a. (1978). specification tests in econometrics. econometrica: journal of the econometric society, 1251–1271. hausman, j. a., & taylor, w. e. (1981). a generalized specification test. economics letters, 8(3), 239–245. hughes, j. p., & mester, l. j. (2008). efficiency in banking: theory, practice, and evidence. isik, i., & hassan, m. k. (2003). efficiency, ownership and market structure, corporate control and governance in the turkish banking industry. journal of business finance & accounting, 30(9–10), 1363–1421. jondrow, j., knox lovell, c. a., materov, i. s., & schmidt, p. (1982). on the estimation of technical inefficiency in the stochastic frontier production function model. journal of econometrics, 19(2–3), 233–238. kasman, a., & yildirim, c. (2006). cost and profit efficiencies in transition banking: the case of new eu members. applied economics, 38(9), 1079–1090. determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 69 kenjegalieva, k. a., simper, r., & weyman-jones, t. g. (2009). efficiency of transition banks: inter-country banking industry trends. applied financial economics, 19(19), 1531–1546. kennedy, p. (2008). a guide to econometrics. malden, mass: blackwell publishing. khanam, d., & nghiem, h. s. (2006). efficiency of banks in bangladesh: a nonparametric approach. regional studies-islamabad, 25(1), 93. koetter, m. (2008). the stability of bank efficiency rankings when risk preferences and objectives are different. european journal of finance, 14(2), 115–135. kohers, t., huang, m.-h., & kohers, n. (2000). market perception of efficiency in bank holding company mergers: the roles of the dea and sfa models in capturing merger potential. review of financial economics, 9(2), 101– 120. kraft, e., & tırtıroğlu, d. (1998). bank efficiency in croatia: a stochasticfrontier analysis. journal of comparative economics, 26(2), 282–300. kumbhakar, s. c., & tsionas, e. g. (2008). scale and efficiency measurement using a semiparametric stochastic frontier model: evidence from the us commercial banks. empirical economics, 34(3), 585–602. kwan, s., & eisenbeis, r. a. (1997). bank risk, capitalization, and operating efficiency. journal of financial services research, 12(2-3), 117–131. kyj, l., & isik, i. (2008). bank x-efficiency in ukraine: an analysis of service characteristics and ownership. journal of economics and business, 60(4), 369–393. lee, c.-c., hsieh, m.-f., & yang, s.-j. (2014). the relationship between revenue diversification and bank performance: do financial structures and financial reforms matter? japan and the world economy, 29, 18–35. manlagñit, m. c. v. (2011). cost efficiency, determinants, and risk preferences in banking: a case of stochastic frontier analysis in the philippines. journal of asian economics, 22(1), 23–35. maudos, j., pastor, j. m., pérez, f., & quesada, j. (2002). cost and profit efficiency in european banks. journal of international financial markets, institutions and money, 12(1), 33–58. miah, m. d., & sharmeen, k. (2015). relationship between capital, risk and efficiency: a comparative study between islamic and conventional banks of bangladesh. international journal of islamic and middle eastern finance and management, 8(2), 203–221. nail, l., & parisi, f. (2005). bank mergers and their impact: a survey of academic studies: there seems to be no clear or universal silver bullet for postmerger success in banking as presented in empirical studies. bank accounting & finance, 18(4), 3–11. 70 the international journal of banking and finance, vol. 15, no 1, 2020 : 39-71 o’brien, r. m. (2007). a caution regarding rules of thumb for variance inflation factors. quality & quantity, 41(5), 673–690. perera, s., skully, m., & wickramanayake, j. (2007). cost efficiency in south asian banking: the impact of bank size, state ownership and stock exchange listings. international review of finance, 7(1–2), 35–60. pesaran, m. h. (2004). general diagnostic tests for cross section dependence in panels. podpiera, a., & podpiera, j. (2005). deteriorating cost efficiency in commercial banks signals an increasing risk of failure: czech national bank. ray, s. c., & das, a. (2010). distribution of cost and profit efficiency: evidence from indian banking. european journal of operational research, 201(1), 297–307. sargan, j. d. (1958). the estimation of economic relationships using instrumental variables. econometrica: journal of the econometric society, 393-415. sarkar, s., sensarma, r., & sharma, d. (2019). the relationship between risk, capital and efficiency in indian banking: does ownership matter? journal of financial economic policy, 11(2), 218–231. schmidt, p., & lovell, c. k. (1979). estimating technical and allocative inefficiency relative to stochastic production and cost frontiers. journal of econometrics, 9(3), 343–366. schulz, k. m., molenda-figueira, h. a., & sisk, c. l. (2009). back to the future: the organizational–activational hypothesis adapted to puberty and adolescence. hormones and behavior, 55(5), 597–604. schumpeter, j. a. (1982). the theory of economic development: an inquiry into profits, capital, credit, interest, and the business cycle (1912/1934). transaction publishers.1982, january, 1, 244. sealey, c. w., & lindley, j. t. (1977). inputs, outputs, and a theory of production and cost at depository financial institutions. the journal of finance, 32(4), 1251–1266. spulbăr, c., & niţoi, m. (2014). determinants of bank cost efficiency in transition economies: evidence for latin america, central and eastern europe and south-east asia. applied economics, 46(16), 1940–1952. staikouras, c., mamatzakis, e., & koutsomanoli-filippaki, a. (2008). cost efficiency of the banking industry in the south eastern european region. journal of international financial markets, institutions and money, 18(5), 483–497. sun, l., & chang, t.-p. (2011). a comprehensive analysis of the effects of risk measures on bank efficiency: evidence from emerging asian countries. journal of banking & finance, 35(7), 1727–1735. tecles, p. l., & tabak, b. m. (2010). determinants of bank efficiency: the case of brazil. european journal of operational research, 207(3), 1587– 1598. determinants of banks cost efficiency: empirical evidence from bangladesh : 39-71 71 teplitz, c. j. (1991). the learning curve deskbook: a reference guide to theory, calculations, and applications: quorum books. vennet, r. v. (2002). cost and profit efficiency of financial conglomerates and universal banks in europe. journal of money, credit and banking, 254– 282. wang, h.-j., & schmidt, p. (2002). one-step and two-step estimation of the effects of exogenous variables on technical efficiency levels. journal of productivity analysis, 18(2), 129–144. wezel, t. (2010). bank efficiency amid foreign entry: evidence from the central american region: international monetary fund. wooldridge, j. m. (1990). a note on the lagrange multiplier and f-statistics for two stage least squares regressions. economics letters, 34(2), 151–155. wright, t. p. (1936). factors affecting the cost of airplanes. journal of the aeronautical sciences, 3(4), 122–128. xiaogang, c., skully, m., & brown, k. (2005). banking efficiency in china: application of dea to pre-and post-deregulation eras: 1993–2000. china economic review, 16(3), 229–245. xiaoqing maggie, f. u., & heffernan, s. (2007). cost x-efficiency in china's banking sector. china economic review, 18(1), 35–53. zhang, j., wang, p., & qu, b. (2012). bank risk taking, efficiency, and law enforcement: evidence from chinese city commercial banks. china economic review, 23(2), 284–295. zheng, c., gupta, a. d., & moudud-ul-huq, s. (2018). effect of human capital efficiency on bank risk-taking behavior and capital regulation: empirical evidence from a developing country. asian economic and financial review, 8(2), 231–247. 81 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: hanafi, a. h. a., md-rus, r., & taufil mohd, k. n. (2021). predicting financial distress in malaysia and its effect on stock returns. international journal of banking and finance, 16(2), 81-110. https://doi.org/10.32890/ ijbf2021.16.2.4 predicting financial distress in malaysia and its effect on stock returns 1ahmad harith ashrofie hanafi, 2rohani md-rus & 3kamarun nisham taufil mohd 1faculty of business and finance universiti tunku abdul rahman, malaysia 2&3 school of economics, finance and banking universiti utara malaysia, malaysia corresponding author: harith@utar.edu.my received: 17/1/2021 revised: 9/3/2021 accepted: 23/3/2021 published: 15/6/2021 abstract unstable economic conditions have an adverse impact on the financial performance of firms, leading to financial distress, which is an unfavourable situation for investors as it may affect their investment returns. thus, this study attempted to predict financial distress and to examine the effect of financial distress on stock returns by using firms listed on bursa malaysia from 1990 to 2020. this study used the logit model to find the probability of bankruptcy and also as a proxy for financial distress risk in the asset pricing model. from this study, financial distress risk was found to be insignificant in pricing stock returns in all tested models. this finding demonstrates that 1 82 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 financial distress risk does not affect stock returns since this risk may be eliminated through diversification. keywords: predicting financial distress, financial distress risk, stock returns. jel classification: g12, g17, g33. introduction uncertainties in economic conditions can affect the financial performance of firms and in some cases, lead to financial distress. this situation not only affects small firms but also listed firms. based on the records of bursa malaysia, or the stock exchange of malaysia, the number of firms classified as practice note 17 (pn17) firms increased from 21 firms in 2017 to 24 firms in 2018. these numbers may seem small compared to the total number of firms in the market, but firms’ bankruptcy risk level and shareholder return may be affected if this increasing trend continues. thus, a better understanding of financial distress and its effect on returns is imperative. financial distress can be defined as the inability of an entity to repay its debt. it is also known as insolvency (sathye et al., 2003). further, financial distress risk can be defined as the probability where a firm is unable to meet its obligations. therefore, a firm is considered financially distressed if it cannot meet its current obligations and the value of its assets is less than the value of its liabilities. this definition is in accordance with studies conducted by altman (1968) and ohlson (1980), which are considered to be among the earliest studies on predicting financial distress. since the studies by altman (1968) and ohlson (1980), other researchers such as pindado et al. (2008), bhunia and sarkar (2011), thai et al. (2014), and ming and akhtar (2014) have conducted numerous studies on predicting financial distress in many different countries. however, a model developed in one country may not be suitable to be used in another country. according to md-zeni and ameer (2010), prediction models used in developed countries may not be suitable for developing countries such as malaysia. they explained further that the models for developed countries are based on useful quality 83 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 data, backed by strong research fundamentals in predicting financial distress, robust financial distress laws that help to identify financial distress clearly, and with minimal intervention by the government. these aspects support the financial distress prediction accuracy. it is therefore crucial to have a high accuracy financial distress prediction model for stakeholders such as corporations, investors, creditors, and regulators in making financial decisions and managing the market. thus, this study intends to develop a financial distress prediction model using financial ratios based on the logit model to predict financial distress for the malaysian market. this is to ensure that the model developed can accurately measure the financial distress risk based on the malaysian market. it is essential to have a highly accurate model for measuring financial distress risk, as it may affect stock returns. this is because stakeholder decisions can be affected especially investors who make investment decisions based on stock risk and return level. due to this situation, previous studies such as by shumway (1996), dichev (1998), and sabbaghi (2015) incorporated financial distress risk into stock return pricing models to determine the relationship between financial distress risk and returns. however, the results obtained were inconclusive, and in addition most of the studies were conducted in developed markets (md-rus, 2011; opler & titman, 1994; sabbaghi, 2015; shumway, 1996; simlai, 2014). on the other hand, studies in emerging markets like malaysia are relatively scarce. thus, this study attempts to incorporate financial distress risk as one of the factors to explain stock returns in malaysia and to understand the relationship between financial distress risk and stock returns. to the researcher’s knowledge, there is a limited number of studies measuring the effects of the probability of financial distress on stock returns in malaysia. thus, the results of this study is expected to be significant for market players in malaysia since it will provide a clearer picture of financial distress risk and its effect on stock returns. literature review financial distress prediction model altman (1968) and beaver (1966) were the pioneers who contributed to the introduction of models for financial distress prediction. beaver 84 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 (1966) used univariate analysis to predict failures among 79 firms by using financial ratios and found that the analysis could better predict failed firms in comparison to random prediction. altman (1968), on the other hand, employed multivariate discriminant analysis (mda) to predict financial distress. based on the results of that study, altman proposed the z-score model to predict financial distress by using ratios such as profitability, liquidity, solvency, and cash flow. however, ohlson (1980) highlighted specific problems arising from the mda, namely that it was found to violate the assumptions of normality and group dispersion which could result in bias in the test of significance and estimated error rates. due to the highlighted problems, ohlson (1980) introduced the logit model that does not have the same assumptions as the mda. this model usually uses average data and is considered as a single period model. a total of 105 failed and 2,058 non-failed firms within the period between 1970 and 1976 were used to predict firm failure by using the logit model. the results demonstrated the suitability of the logit model for predicting firm failure. based on the results by ohlson (1980), many researchers have used the logit model to predict financial distress (abdul manab et al., 2015; ming & akhtar, 2014; platt & platt, 2008; yap et al., 2012). zaki et al. (2011) shifted the research focus from non-financial firms to financial firms as the sample of their study, which included commercial and islamic banks in the united arab emirates (uae). their study used the logit model, probit model, and log-logistic (loglog) link function. the results showed that the logit model could be used to predict financial distress among banks in the uae. in a study conducted by ong et al. (2011), they used the logit model to predict corporate failure among public listed firms in malaysia. the results obtained highlighted the importance of the ratios used in the model, whereby the model used in the study had an excellent accuracy of 91.5 percent when only five main financial ratios were used. this accuracy rate was much higher than the one recorded in a previous study by low et al. (2001). nevertheless, a financial distress prediction model does not depend solely on the model used, as predictor variables also play an important 85 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 role in predicting financial distress. previous studies by abdul manab et al. (2015), chiaramonte and casu (2016), etemadi et al. (2009), and md-zeni and ameer (2010) used financial ratios to predict financial distress. other researchers used current ratio (cr) in developing their financial distress prediction models as it directly measures the ability of a firm to meet its short-term obligations using only current assets (abdullah, 2006; daily & dalton, 1994; elloumi & gueyié, 2001; ganesalingam & kumar, 2001; youn & gu, 2010). cr can be defined as current assets divided by current liabilities (elloumi & gueyié, 2001; foster & zurada, 2013; parker et al., 2002; wang & li, 2007). md-zeni and ameer (2010) used cr as one of the predictors in predicting the turnover of financially distressed firms in malaysia. they found that liquidity, represented by the cr, was significant in predicting the turnover of distressed firms. ugurlu and aksoy (2006) used sales to working capital (swc) and found a positive relationship between swc and the probability of a firm going bankrupt. however, their results contradicted findings by yap et al. (2012) as the swc was found to be ineffective in predicting financial distress. like other ratios, the activity ratio has also been used by researchers in predicting financial distress (abdul manab et al., 2015; tan & dihardjo, 2001; tirapat & nittayagasetwat, 1999; wang & li, 2007). parker et al. (2002) and ong et al. (2011) included days’ sales in accounts receivable (dsc) in predicting financial distress. this ratio was found to be significant since it reflected the ability of a firm to collect payments for its credit sales. ong et al. (2011) explained that the faster the firm was able to collect payments for its credit sales, the lower the probability of the firm becoming a financially distressed one. earnings before interest and tax to sales (ebits) is one of the commonly used ratios that represents profitability in the prediction of financial distress (abdul manab et al., 2015; parker et al., 2002; tan & dihardjo, 2001; thai et al., 2014; ugurlu & aksoy, 2006). parker et al. (2002) used this ratio as a proxy for return on assets, which also represents the ability of a firm to recover from financial distress. parker et al. (2002) claimed that ebits should have a negative relationship with the probability of financial distress. some researchers used the net profit margin (npm) to represent the profitability ratio (pindado et al., 2008; wang & li, 2007). yap et al. (2012) found a negative 86 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 relationship between npm and firm financial failure. indeed, the profitability ratios are important in predicting financial distress, as these ratios are statistically significant in distinguishing firm financial performance (md-zeni & ameer, 2010). many ratios have been used to represent firm leverage, and one of them is debt ratio (dr). lee and yeh (2004) defined dr as total debt divided by total assets, and they found this ratio to be significant in predicting financial distress. meanwhile, ugurlu and aksoy (2006) used long-term debt to total debt (ldtd) ratio to represent firm leverage and found it to be significantly helpful in decreasing the probability of financial distress. abdullah and ahmad (2005) used shareholders’ funds to total liabilities (sftd) as it represents a firm’s capital structure, and they found the ratio to be significant in predicting financial distress. thus, if a firm relies too much on liabilities, it has a high probability of getting itself into financial distress. chen et al. (2013), fich and slezak (2008), and youn and gu (2010) considered financial cost elements such as interest coverage ratio (icr) as one of their variables. youn and gu (2010) explained that this ratio contained a wide range of information related to earning, productivity, ability to pay interest, and indebtedness of a firm. they found that the icr had a negative coefficient, indicating that a firm with a high-interest coverage ratio would have a lower probability of being in financial distress. financial distress risk and stock return previous studies showed diverse results when analysing the relationship between financial distress risk and returns. denis and denis (1995) found a positive relationship between financial distress risk and stock returns. their result was supported by shumway (1996), who suggested that the risk of default is significant to affect stock returns. shumway (1996) explained that the average returns is strongly and positively related to distress risk and has a weak correlation to the size of a firm. the study conducted by sabbaghi (2015) indirectly examined the relationship between systematic risk and financial distress risk by investigating the relationship between the aggregate volatility of market return and financial distress risk. sabbaghi’s study revealed a positive correlation between aggregate 87 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 volatility and the momentum used to represent financial distress risk. the results indirectly showed that stock return has a positive relationship with financial distress risk. boubaker et al. (2018) examined whether financial distress risk affects stock return by using 12 portfolios sorted by size, book-tomarket, and leverage and a portfolio of distressed firms covering an 18-year period. the main goal of this study was to identify the risk factors that best captured the default risk in the french context. the results demonstrated that the risk premium for the relative distress factor was positively significant only for the distressed firm portfolio but insignificant for the non-financial distress firm portfolio. this is because the level of financial distress risk for non-financial distress firms is too small whereby it does not give any significant effect on the stock return of non-financial distress firms. mselmi et al. (2019) examined financial distress, liquidity, and valueat-risk effect on stock returns for the french stock market. financial distress was found to be consistently and positively significant to the pricing of stock returns for the financial distress portfolio in all models used in the study, which was somewhat similar to the results of boubaker et al. (2018). mselmi et al. (2019) also found that financial distress was significant to the pricing of stock returns only in the absence of size and book-to-market factors. this indicated that the existence of size and book-to-market factors had already captured the effects of financial distress on stock returns. mselmi et al. (2019) further explained that the absence of size and book-to-market factors led to financial distress risk becoming significant but left the proportion of return unexplained, as the model could not explain the return well. chhapra et al. (2020) investigated the relationship between default risk to represent financial distress and a cross-section of stock returns based on pakistan’s stock market using monthly returns between 2001 and 2016. they found that the stock of firms that were significantly exposed to non-diversified default risk yielded higher returns. this indicated that financial distress represented by default risk was positively significant in affecting stock returns. the positively significant results obtained from these previous studies were in accordance with the risk-return trade-off theory which stated that stocks with high level risk (including financial distress risk) should generate high returns. 88 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 however, there were also studies that obtained negatively significant results such as a study by opler and titman (1994) which found that leveraged firms within distressed industries obtained lower stock returns, which did not portray the correct relationship between risk and stock returns. the authors explained that this result was due to the pure leverage effect therefore indicating that distress risk had a negative effect on stock returns. instead of developing a prediction model like in the previous studies for example, shumway (1996), dichev (1998) used the z-score and o-score to develop a financial distress portfolio based on the probability of financial distress generated by both models. the results demonstrated that firms with high levels of financial distress risk earned lower average returns compared to firms with low levels of financial distress risk. thus, the author concluded that financial distress has a negative relationship with stock returns. all these negative results contradicted the risk-return trade-off theory which proposed that high risk led to high returns. this situation could be due to a firm’s leverage effect (opler & titman, 1994). this is because high leverage effect could negatively affect a firm’s income and also investor’s income or dividend which would directly result in a negative effect on returns. lastly, there were also previous studies that obtained insignificant results such as a study by idrees and qayyum (2018) which investigated the relationship between financial distress risk and equity returns of financially distressed firms listed on the pakistan stock exchange (psx). they found that distress risk had a negative coefficient but proved statistically insignificant in determining stock returns. hence, they concluded that there was no relationship between expected stock returns and bankruptcy risk. using manufacturing companies listed on the indonesia stock exchange from 2015 to 2017, sudirgo et al. (2019) later examined the effects of financial distress, financial performance, and liquidity on stock returns. the z-score was used to calculate the financial distress risk in the study. the results demonstrated that the financial distress variable had an insignificant effect on stock returns. these insignificant results might be due to the financial distress risk that could be reduced or eliminated through diversification since financial distress risk is a part of unsystematic risk. thus, the study concluded that there was no significant effect of financial distress risk on stock returns. 89 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 methodology data this study focused on firms listed on the malaysian stock market from 1990 to 2020. the estimated sample consisted of data from 1057 firms that were used to develop the prediction model. due to the highly volatile ratios that were heavily affected by the economic conditions and the slightly different interpretation of the ratios, firms under the financial and properties industries were excluded from the sample of study (md-rus & abdullah, 2005). this study focused on firms listed on the malaysian stock market because listed firms are significant players in this market. the selected firms were later classified into two categories, namely (1) financially distressed firms and (2) nonfinancially distressed firms. thus, this study defined a financially distressed firm as a firm that fulfils one of the following conditions: (1) the firm is classified by bursa malaysia as a financially distressed firm under pn4, pn17 or/and amended pn 17, and (2) there is a deficit in the adjusted shareholders’ equity on a consolidated basis. firms that did not fulfil the criteria were classified as non-financially distressed firms. unlike ong et al. (2011), this study did not match the number of non-financially distressed firms to the number of financially distressed firms in order to avoid sample bias, as highlighted by sori et al. (2001). annual reports for both groups of firms were collected from the bloomberg terminal. financial information was used to calculate each firm’s financial ratios for each year. these ratios, consisting of liquidity, profitability, leverage, and efficiency ratios, formed the independent variables or predictors in developing the financial distress prediction model. the stock price for each firm was used to calculate the return/excess return for each firm to identify the related risk factors. this study used firm size, value, and financial distress probability as the variables representing the risk factors for returns. firm returns were calculated based on monthly returns starting from six months after the fiscal year-end. this was to ensure the availability of accounting data for measuring financial distress probability. 90 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 to investigate the risk factors of returns, this study first ranked all firms from the lowest to the highest financial distress risk based on the probabilities generated from the earlier part of this study to develop a set of 10 portfolios. thus, portfolio 1 consisted of firms with the lowest financial distress risk, while portfolio 10 consisted of firms with the highest financial distress risk. the average monthly return for each portfolio was calculated. the average size, value, and financial distress probability were calculated to represent all the independent variables. lastly, this study regressed portfolio returns with all independent variables using the fama-macbeth regression which is similar to the procedure used by dichev (1998) in investigating risk factors of stock returns. data analysis according to shumway (2001) the logistic model (logit model) is a statistical technique that is appropriate when the independent variables are metric variables and the dependent variable is a nonmetric (categorical nominal) variable. the logistic analysis aims to identify the best-fitting model to explain the relationship between the predictors or explanatory variables and the dependent variable with the most parsimonious yet reasonable model based on statistics. thus, this study adopted the logit prediction model from ohlson (1980) and abdullah and ahmad (2005). zi = β’ xi + ui (1) where: zi = xi = financial ratios of companies ui = error term non-distressed companies’ probability and likelihood function can be simply defined as follows: pi = (2) 7 according to shumway (2001) the logistic model (logit model) is a statistical technique that is appropriate when the independent variables are metric variables and the dependent variable is a non-metric (categorical nominal) variable. the logistic analysis aims to identify the best-fitting model to explain the relationship between the predictors or explanatory variables and the dependent variable with the most parsimonious yet reasonable model based on statistics. thus, this study adopted the logit prediction model from ohlson (1980) and abdullah and ahmad (2005). zi = β’ xi + ui (1) where: zi = { 0 ⋯ non − financially distressed firms 1 ⋯ financially distressed firms xi = financial ratios of companies ui = error term non-distressed companies’ probability and likelihood function can be simply defined as follows: pi = 1 1+𝑒𝑒−𝑧𝑧𝑖𝑖 (2) where: zi = β’ xi + ui in detail, it is written as follows: pi = 1 1+𝑒𝑒−(𝛽𝛽’ 𝑥𝑥𝑖𝑖 + 𝑢𝑢𝑖𝑖 ) (3) equation (3) represents the cumulative logistic distribution function. equation (1) is used to estimate the weight for each financial ratio of the selected firms. based on equation (3), if pi represents the probability of non-distress, then (1 − pi) represents the probability of distress. thus, 1pi = 1 1+𝑒𝑒(𝛽𝛽’ 𝑥𝑥𝑖𝑖 + 𝑢𝑢𝑖𝑖 ) (4) the optimal weight, β, can be estimated where the likelihood value is maximised. to obtain the probability of distress, β is substituted into the cumulative probability function (equation 4). a firm with a probability of less than 0.5 is classified as a non-distressed firm, whereas a firm with a probability of more than 0.5 is classified as a distressed firm based on the calculated probabilities from the logit model. to obtain the most appropriate predictors for the maximum likelihood estimation, this study used the stepwise procedure to exclude the insignificant independent variables and only took significant variables to be included in the model. it is crucial to identify the significant predictors based on the financial ratios that can differentiate between financially distressed and non-financially distressed firms. thus, the stepwise procedure helps in identifying the most appropriate independent variables. this procedure excludes insignificant independent variables and only includes variables that are significant in the model. this method is similar to the method by nam and taehong (2000) and md-rus and abdullah (2005). the financial ratios to be included in the 7 according to shumway (2001) the logistic model (logit model) is a statistical technique that is appropriate when the independent variables are metric variables and the dependent variable is a non-metric (categorical nominal) variable. the logistic analysis aims to identify the best-fitting model to explain the relationship between the predictors or explanatory variables and the dependent variable with the most parsimonious yet reasonable model based on statistics. thus, this study adopted the logit prediction model from ohlson (1980) and abdullah and ahmad (2005). zi = β’ xi + ui (1) where: zi = { 0 ⋯ non − financially distressed firms 1 ⋯ financially distressed firms xi = financial ratios of companies ui = error term non-distressed companies’ probability and likelihood function can be simply defined as follows: pi = 1 1+𝑒𝑒−𝑧𝑧𝑖𝑖 (2) where: zi = β’ xi + ui in detail, it is written as follows: pi = 1 1+𝑒𝑒−(𝛽𝛽’ 𝑥𝑥𝑖𝑖 + 𝑢𝑢𝑖𝑖 ) (3) equation (3) represents the cumulative logistic distribution function. equation (1) is used to estimate the weight for each financial ratio of the selected firms. based on equation (3), if pi represents the probability of non-distress, then (1 − pi) represents the probability of distress. thus, 1pi = 1 1+𝑒𝑒(𝛽𝛽’ 𝑥𝑥𝑖𝑖 + 𝑢𝑢𝑖𝑖 ) (4) the optimal weight, β, can be estimated where the likelihood value is maximised. to obtain the probability of distress, β is substituted into the cumulative probability function (equation 4). a firm with a probability of less than 0.5 is classified as a non-distressed firm, whereas a firm with a probability of more than 0.5 is classified as a distressed firm based on the calculated probabilities from the logit model. to obtain the most appropriate predictors for the maximum likelihood estimation, this study used the stepwise procedure to exclude the insignificant independent variables and only took significant variables to be included in the model. it is crucial to identify the significant predictors based on the financial ratios that can differentiate between financially distressed and non-financially distressed firms. thus, the stepwise procedure helps in identifying the most appropriate independent variables. this procedure excludes insignificant independent variables and only includes variables that are significant in the model. this method is similar to the method by nam and taehong (2000) and md-rus and abdullah (2005). the financial ratios to be included in the 91 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 where: zi = β’ xi + ui in detail, it is written as follows: (3) equation (3) represents the cumulative logistic distribution function. equation (1) is used to estimate the weight for each financial ratio of the selected firms. based on equation (3), if pi represents the probability of non-distress, then (1 − pi) represents the probability of distress. thus, (4) the optimal weight, β, can be estimated where the likelihood value is maximised. to obtain the probability of distress, β is substituted into the cumulative probability function (equation 4). a firm with a probability of less than 0.5 is classified as a non-distressed firm, whereas a firm with a probability of more than 0.5 is classified as a distressed firm based on the calculated probabilities from the logit model. to obtain the most appropriate predictors for the maximum likelihood estimation, this study used the stepwise procedure to exclude the insignificant independent variables and only took significant variables to be included in the model. it is crucial to identify the significant predictors based on the financial ratios that can differentiate between financially distressed and non-financially distressed firms. thus, the stepwise procedure helps in identifying the most appropriate independent variables. this procedure excludes insignificant independent variables and only includes variables that are significant in the model. this method is similar to the method by nam and taehong (2000) and md-rus and abdullah (2005). the financial ratios to be included in the prediction model must be statistically significant based on the p-values generated by the model. to confirm the significance of the independent variables, this study also tested the variables’ overall significance using the likelihood ratio. the independent variables in this model consisted of the financial ratios, namely, liquidity, activity, profitability, and leverage ratios. in 7 according to shumway (2001) the logistic model (logit model) is a statistical technique that is appropriate when the independent variables are metric variables and the dependent variable is a non-metric (categorical nominal) variable. the logistic analysis aims to identify the best-fitting model to explain the relationship between the predictors or explanatory variables and the dependent variable with the most parsimonious yet reasonable model based on statistics. thus, this study adopted the logit prediction model from ohlson (1980) and abdullah and ahmad (2005). zi = β’ xi + ui (1) where: zi = { 0 ⋯ non − financially distressed firms 1 ⋯ financially distressed firms xi = financial ratios of companies ui = error term non-distressed companies’ probability and likelihood function can be simply defined as follows: pi = 1 1+𝑒𝑒−𝑧𝑧𝑖𝑖 (2) where: zi = β’ xi + ui in detail, it is written as follows: pi = 1 1+𝑒𝑒−(𝛽𝛽’ 𝑥𝑥𝑖𝑖 + 𝑢𝑢𝑖𝑖 ) (3) equation (3) represents the cumulative logistic distribution function. equation (1) is used to estimate the weight for each financial ratio of the selected firms. based on equation (3), if pi represents the probability of non-distress, then (1 − pi) represents the probability of distress. thus, 1pi = 1 1+𝑒𝑒(𝛽𝛽’ 𝑥𝑥𝑖𝑖 + 𝑢𝑢𝑖𝑖 ) (4) the optimal weight, β, can be estimated where the likelihood value is maximised. to obtain the probability of distress, β is substituted into the cumulative probability function (equation 4). a firm with a probability of less than 0.5 is classified as a non-distressed firm, whereas a firm with a probability of more than 0.5 is classified as a distressed firm based on the calculated probabilities from the logit model. to obtain the most appropriate predictors for the maximum likelihood estimation, this study used the stepwise procedure to exclude the insignificant independent variables and only took significant variables to be included in the model. it is crucial to identify the significant predictors based on the financial ratios that can differentiate between financially distressed and non-financially distressed firms. thus, the stepwise procedure helps in identifying the most appropriate independent variables. this procedure excludes insignificant independent variables and only includes variables that are significant in the model. this method is similar to the method by nam and taehong (2000) and md-rus and abdullah (2005). the financial ratios to be included in the 7 according to shumway (2001) the logistic model (logit model) is a statistical technique that is appropriate when the independent variables are metric variables and the dependent variable is a non-metric (categorical nominal) variable. the logistic analysis aims to identify the best-fitting model to explain the relationship between the predictors or explanatory variables and the dependent variable with the most parsimonious yet reasonable model based on statistics. thus, this study adopted the logit prediction model from ohlson (1980) and abdullah and ahmad (2005). zi = β’ xi + ui (1) where: zi = { 0 ⋯ non − financially distressed firms 1 ⋯ financially distressed firms xi = financial ratios of companies ui = error term non-distressed companies’ probability and likelihood function can be simply defined as follows: pi = 1 1+𝑒𝑒−𝑧𝑧𝑖𝑖 (2) where: zi = β’ xi + ui in detail, it is written as follows: pi = 1 1+𝑒𝑒−(𝛽𝛽’ 𝑥𝑥𝑖𝑖 + 𝑢𝑢𝑖𝑖 ) (3) equation (3) represents the cumulative logistic distribution function. equation (1) is used to estimate the weight for each financial ratio of the selected firms. based on equation (3), if pi represents the probability of non-distress, then (1 − pi) represents the probability of distress. thus, 1pi = 1 1+𝑒𝑒(𝛽𝛽’ 𝑥𝑥𝑖𝑖 + 𝑢𝑢𝑖𝑖 ) (4) the optimal weight, β, can be estimated where the likelihood value is maximised. to obtain the probability of distress, β is substituted into the cumulative probability function (equation 4). a firm with a probability of less than 0.5 is classified as a non-distressed firm, whereas a firm with a probability of more than 0.5 is classified as a distressed firm based on the calculated probabilities from the logit model. to obtain the most appropriate predictors for the maximum likelihood estimation, this study used the stepwise procedure to exclude the insignificant independent variables and only took significant variables to be included in the model. it is crucial to identify the significant predictors based on the financial ratios that can differentiate between financially distressed and non-financially distressed firms. thus, the stepwise procedure helps in identifying the most appropriate independent variables. this procedure excludes insignificant independent variables and only includes variables that are significant in the model. this method is similar to the method by nam and taehong (2000) and md-rus and abdullah (2005). the financial ratios to be included in the 92 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 this study, the liquidity ratios were represented by the current ratio (cr) and sales to working capital (swc). days’ sales in accounts receivable (dsc) was used to represent the activity ratios. furthermore, this study used earnings before interest and tax to sales (ebits) and net profit margin (npm) to represent profitability in predicting financial distress. debt ratio (dr), long-term debt to total debt ratio (ldtd), shareholders’ funds to total debt (sftd), and interest coverage ratio (icr) were used to represent leverage ratios. the subsequent analysis was to find the determinants of return. before determining the stock returns, this study first sorted all the selected stocks based on the level of financial distress, where portfolio 1 consisted of firms with the lowest financial distress probabilities whereas portfolio 10 consisted of firms with the highest financial distress probabilities to analyse the mean returns for the decile portfolios based on stock monthly returns from june 1991 to june 2020. next, the average values of financial distress probability, monthly stock returns, firm size, and firm value were calculated for the decile portfolios. this study compared the lower financial distress decile portfolios (portfolios 1 to 5) against the higher financial distress decile portfolios (portfolios 6 to 10). the results would provide a picture on the characteristics of the stocks based on financial distress level, firm size, and firm value. next, this study conducted further analysis to determine stock returns. this study adopted a method similar to the one used by dichev (1998), which was based on famamacbeth regression introduced by fama and macbeth (1973) with the inclusion of financial distress risk variables. dichev (1998) included two major factors to represent the risk factors related to returns, namely size and value. the model developed in this study is as follows: rit = α + β1 fdit + β2 sizeit + β3 valueit + εit (5) where rit is the monthly portfolio return, α is the intercept, fdit is the financial distress risk, sizeit is the market value, valueit is the book-tomarket value, and βn is the coefficient for the risk factor, and ε is the error term. this study used size and value as independent variables since both variables are the most common variables used as risk factors for returns. 93 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 this study also included financial distress risk factors as independent variable. although there are other risk factors to determine returns such as momentum, this study only focused on these three variables since these variables are more suitable to be used in fama-macbeth (1973) regression and does not involve complex calculation approach. in this study, size or market value was calculated based on the log of the product of fiscal year-end price and the number of shares outstanding. the book-to-market value was calculated as the common equity value divided by the market value of the firm. the measurements for both size and value were similar to those used in previous studies (dichev, 1998; zaretky & zumwalt, 2007). as for financial distress risk, the probability of financial distress was generated using the prediction model developed in the early stage. this approach is similar to dichev (1998) and zaretzky and zumwalt (2007), which used scores generated from financial distress prediction model. the simple average for all variables in each portfolio was calculated for each observation year. all variables were regressed using fama-macbeth (1973) regression to obtain the coefficients for the hypothesis testing of this study. results and analysis the results in table 1 shows the mean difference analysis based on the financial condition of the firms. firms were divided into two groups: 1) distressed firms and 2) non-financial distressed firms. the mean, standard error, and mean difference values between them were calculated and analysed for each variable. the results showed that all the variables had significant mean differences between both groups. the results also showed that financial distress firms tended to have negative swc, ebits, npm, and icr. in contrast, non-financial distressed firms recorded positive values for all selected variables. subsequently, this study continued with a correlation analysis for all independent variables to predict financial distress. the results are shown in table 2. the outcome clearly showed that correlations between cr and sftd, dr and sftd, and npm and ebits were quite strong, which could lead to a multicollinearity problem. thus, this study conducted a variance inflation factor (vif) analysis to detect the multicollinearity problem. 94 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 ta bl e 1 m ea n d iff er en t a na ly si s d is tr es se d fi rm s n on -f in an ci al d is tr es se d fi rm s m ea n st d. e rr or . m ea n st d. e rr or m ea n d if f. p -v al ue c r 1. 08 76 0. 07 03 2. 21 01 0. 08 85 1. 12 25 0. 00 00 ** * sw c -1 9. 98 33 21 .4 97 2 14 .9 69 9 8. 48 30 34 .9 53 1 0. 04 14 ** d sc 15 1. 28 27 10 .7 23 5 10 3. 88 1 2. 98 62 -4 7. 40 18 0. 00 00 ** * e b it s -0 .0 77 7 0. 01 67 0. 08 39 0. 00 62 0. 16 16 0. 00 00 ** * n pm -0 .3 06 3 0. 03 59 0. 04 18 0. 00 65 0. 34 82 0. 00 00 ** * d r 0. 72 75 0. 02 46 0. 43 07 0. 00 87 -0 .2 96 8 0. 00 00 ** * l d t d 0. 20 61 0. 02 09 0. 24 31 0. 00 83 0. 03 42 0. 04 17 ** sf t d 0. 75 44 0. 26 39 1. 95 95 0. 10 58 1. 20 51 0. 00 00 ** * ic r -0 .3 67 2 0. 30 88 0. 19 26 0. 07 14 0. 55 99 0. 00 40 ** * n ot e: c r r ep re se nt s cu rr en t r at io , s w c r ep re se nt s sa le s to w or ki ng c ap ita l, d sc r ep re se nt s da ys ’ s al es in a cc ou nt s re ce iv ab le , e b it s re pr es en ts ea rn in gs b ef or e in te re st a nd t ax t o sa le s, n pm r ep re se nt s ne t pr ofi t m ar gi n, d r r ep re se nt s de bt r at io , l d t d r ep re se nt s lo ng -t er m d eb t to t ot al de bt , s ft d r ep re se nt s sh ar eh ol de rs ’ f un ds to to ta l d eb t a nd i c r r ep re se nt s in te re st c ov er ag e ra tio . t he a st er is k sy m bo l ( *) f or p -v al ue r ep re se nt s st at is tic al ly s ig ni fic an t a t d iff er en t l ev el s: * ** s ta tis tic al ly s ig ni fic an t a t 1 % , * * st at is tic al ly s ig ni fic an t a t 5 % , a nd * s ta tis tic al ly s ig ni fic an t a t 1 0% . 95 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 ta bl e 2 c or re la tio n a na ly si s c r s w c d sc e b it s n pm d r l d t d sf t d i c r c r 1. 00 00 sw c -0 .0 19 7 1. 00 00 d sc -0 .0 13 9 -0 .0 02 3 1. 00 00 e b it s 0. 13 48 0. 02 62 -0 .2 68 1 1. 00 00 n pm 0. 23 26 0. 02 47 -0 .2 22 9 0. 71 41 1. 00 00 d r -0 .6 41 8 -0 .0 00 3 0. 16 01 -0 .2 41 6 -0 .4 41 6 1. 00 00 l d t d 0. 05 75 -0 .0 83 -0 .1 96 4 0. 25 30 0. 01 49 0. 06 48 1. 00 00 sf t d 0. 79 33 -0 .0 10 7 -0 .0 64 6 0. 00 75 0. 11 65 -0 .6 89 4 -0 .0 68 2 1. 00 00 ic r 0. 03 88 0. 00 36 -0 .0 33 9 0. 13 47 0. 10 33 -0 .0 63 6 -0 .0 63 6 0. 02 30 1. 00 00 n ot e: c r r ep re se nt s cu rr en t r at io , s w c r ep re se nt s sa le s to w or ki ng c ap ita l, d sc r ep re se nt s da ys ’ s al es in a cc ou nt s re ce iv ab le , e b it s re pr es en ts ea rn in gs b ef or e in te re st a nd ta x to s al es , n pm re pr es en ts n et p ro fit m ar gi n, d r re pr es en ts d eb t r at io , l d t d re pr es en ts lo ng -t er m d eb t t o to ta l d eb t, sf t d re pr es en ts s ha re ho ld er s’ fu nd s to to ta l d eb t a nd ic r re pr es en ts in te re st c ov er ag e ra tio . t he s ta r s ym bo l ( *) fo r p -v al ue re pr es en ts s ta tis tic al ly si gn ifi ca nt a t d if fe re nt le ve ls : * ** s ta tis tic al ly s ig ni fic an t a t 1 % , * * st at is tic al ly s ig ni fic an t a t 5 % , a nd * s ta tis tic al ly s ig ni fic an t a t 1 0% . 96 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 the results of the vif analysis are shown in table 3. based on the outcome, all variables generated a vif value of less than 10. the results indicated that none of the selected variables suffered from a severe multicollinearity problem. thus, all variables could be used in developing a financial distress prediction model later. table 3 collinearity diagnostics analysis vif cr 3.10 swc 1.00 dsc 1.15 ebits 2.38 npm 2.66 dr 2.63 ldtd 1.22 sftd 3.64 icr 1.00 mean vif 2.09 note: cr represents current ratio, swc represents sales to working capital, dsc represents days’ sales in accounts receivable, ebits represents earnings before interest and tax to sales, npm represents net profit margin, dr represents debt ratio, ldtd represents long-term debt to total debt, sftd represents shareholders’ funds to total debt and icr represents interest coverage ratio. this study continued with the development of a financial distress prediction model using the logit model. the results in table 4 showed that liquidity ratios represented by the current ratio (cr) and sales to working capital (swc) were negatively significant in predicting financial distress. these results were similar to the findings of previous studies, such as parker et al. (2002), juniarti (2013), and chiaramonte and casu (2016). the results imply that as firms’ liquidity increases, the ability to meet their short-term obligations with current assets will increase, leading to a decrease in the probability of financial distress. the activity ratios are represented by days’ sales in accounts receivable (dsc) in table 4. this variable was positively significant in predicting 97 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 financial distress; similar to the results of ong et al. (2011) and parker et al. (2002). this is because the slower a firm collects payments for its credit sales the higher will be the probability of financial distress due to the lower revenue collections within the period, therefore impairing the ability of the firm to settle its debts. table 4 financial distress prediction model coefficient std. error z p-value const -6.8162 0.7177 -9.497 0.000*** cr -0.3278 0.1789 -1.832 0.067* swc -0.0005 0.0002 -1.786 0.074* dsc 0.0032 0.0012 2.816 0.005*** ebits −6.5610 1.8673 −3.514 0.000*** npm −3.6935 0.8706 −4.242 0.000*** dr 9.3561 0.9779 9.567 0.000*** icr 0.0045 0.0024 1.841 0.065* mcfadden r-squared 0.6275 number of observations 1057 note: this stepwise logit regression analysis used firm’s average cross-sectional data for each variable. cr represents current ratio, swc represents sales to working capital, dsc represents days’ sales in accounts receivable, ebits represents earnings before interest and tax to sales, npm represents net profit margin, dr represents debt ratio, and icr represents interest coverage ratio. the star symbol (*) for p-value represents statistically significant at different levels: *** statistically significant at 1%, ** statistically significant at 5%, and * statistically significant at 10%. table 4 also shows the results for profitability ratios represented by earnings before interest and tax to sales (ebits) and net profit margin (npm). the results showed that both profitability ratios were negatively significant in predicting financial distress. these results were in accordance with studies by ong et al. (2011), juniarti (2013), polemis and gounopoulos (2012) and thai et al. (2014). furthermore, these results imply that a high ability to generate profit will help a firm to reduce its financial distress risk. 98 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 table 4 also shows the results for debt ratio (dr), which is a crucial variable for predicting financial distress. this is because the variable’s coefficient recorded the highest value among all the variables and was found to be statistically significant. previous studies by lee and yeh (2004), sori et al. (2001) and alifiah and tahir (2018) obtained similar results. they stated that as a firm increased the proportion of its debt to finance its assets but defaulted on repaying its debt, the financial distress risk would increase. another significant leverage ratio in predicting financial distress is interest coverage ratio (icr), which represents the ability of a firm to meet financing cost on its debt. the results contradicted the results of previous studies by fich and slezak (2008) and youn and gu (2010). this is because high interest coverage ratio also indicates that the firm has a low debt level and at the same time ignores opportunities to grow through leverage. thus, it is difficult for a firm to predict its future earnings, cover its future debt payment, and to increase the probability of financial distress to occur. although increasing leverage could help the firm to grow and reduce financial distress, the firm still needs to control the level of its debt in order to avoid financial distress. table 5 prediction model accuracy analysis main sample predicted 0 1 accuracy rate actual 0 819 43 95.01% 1 64 131 67.17% overall accuracy 89.9% based on table 5, the study found that the model could correctly predict 131 financial distress cases out of a total of 195 cases in the sample. thus, the model recorded an accuracy of 67.18 percent in predicting financial distress. this accuracy rate is considered moderate, indicating that the model is reliable in predicting financial distress. this model could also predict 819 non-financial distress cases out of a total of 862 cases in the sample. hence, 95.01 percent 99 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 of non-financial distress cases could be predicted correctly by using the model. thus, the overall accuracy of this model was 89.9 percent which is considered high, making it a reliable model. after developing the financial distress prediction model, this study used the model to generate the probability of financial distress (fd). the generated fd value was used to represent financial distress risk and also as one of the variables to determine stock returns. before determining the stock returns, this study conducted a descriptive analysis. the results of the descriptive analysis are shown in table 6. table 6 descriptive analysis mean std. dev. min max return 0.0057 0.1909 -0.98 14.875 fd 0.1344 0.2639 0 1 size 19.0903 1.6621 12.2234 24.8228 value 1.7732 11.4493 -521.72 332.722 note: fd represents financial distress risk generated from the prediction model, return represents monthly portfolio return starting from six months after the fiscal year-end, size represents firm’s size based on market capitalisation, and value represents firm’s value based on book-to-market value. based on table 6, the results showed that the firm’s return was between 14.875 and -0.98 with the average return of 0.0057 or 0.57 percent. although the range of firm return value was considered to be large, the variation was small since the standard deviation was only 0.1909. thus, this clearly showed that only a few firms managed to generate higher than average returns. the results also showed that the average financial distress risk represented by financial distress probability was 0.1344 with the standard deviation of 0.2639. this indicated that only a few firms could be categorised as financially distressed. as for size that is based on log of market value, the data showed that firm size was within the range of 12.2234 to 24.8228 with the average size of 19.0903 and standard deviation of 1.6621. lastly, the value presented by book-to-market value recorded the average value of 1.7732 which showed that on average most of the firms had a higher book value 100 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 compared to the market value. however, the variation between the observations was considered large since the standard deviation was 11.4493 and the range of observations was between -521.72 to 332.722. table 7 correlation analysis among independent variables fd size value fd 1.0000 size -0.1577 1.0000 value -0.0517 -0.1081 1.0000 note: fd represents financial distress risk generated from the prediction model, return represents monthly portfolio return starting from six months after the fiscal year-end, size represents firm’s size based on market capitalisation, and value represents firm’s value based on book-to-market value. this analysis was followed by correlation analysis in order to understand the correlation between selected independent variables where the results are summarized in table 7. based on the results, the correlations between all variables were negative, and the values were greater than negative 0.2. the results indicated that the selected variables were weakly correlated between each other. next, this study continued the analysis by creating 10 decile portfolios based on firms’ financial distress probabilities generated from the model developed earlier. the results in table 8 showed that the range of average monthly returns for lower financial distress portfolios (portfolios 1 to 5) was from 0.14 to 1.42 per cent. in comparison, the range of average monthly returns for higher financial distress portfolios (portfolios 6 to 10) was from 0.33 to 0.66 percent. thus, the lower financial distress risk portfolios tended to have higher returns compared to the higher financial distress risk portfolios. the table also shows that as financial distress risk increases, the average return decreases. this result contradicted the risk-return trade-off theory whereby high-risk state investments should obtain higher returns compared to low-risk investments. in addition, on average the lower financial distress risk portfolios consisted of more large firms and value firms compared to the higher financial distress risk portfolios. 101 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 table 8 average portfolio analysis portfolio fd return size value 1 0.0001 0.0067 19.4563 1.8115 2 0.0007 0.0014 19.2096 2.7371 3 0.0013 0.0142 19.7713 3.3521 4 0.0030 0.0063 19.3556 1.7619 5 0.0079 0.0077 19.2523 1.6385 6 0.0201 0.0066 19.2400 1.5267 7 0.0518 0.0045 19.0783 1.9819 8 0.1307 0.0046 19.0048 1.4928 9 0.3412 0.0054 19.9124 1.3713 10 0.8059 0.0033 19.5029 0.2875 total 0.1344 0.0057 19.0868 1.8132 note: fd represents financial distress risk generated from the prediction model, return represents monthly portfolio return starting from six months after the fiscal year-end, size represents firm’s size based on market capitalisation, and value represents firm’s value based on book-to-market value. the results also showed that on average, large firms and value firms could generate higher average returns compared to small firms and growth firms. the results for low distress portfolios consisting of large size firms and generating high returns was similar to dichev (1998) while the results for high distress portfolios consisting of growth firms and generating lower returns was similar to griffin and lemmon (2002). thus, these findings could be early indicators before conducting fama-macbeth regression to determine the relationship between financial distress risk and stock returns. next, this study examined the relationship between the risk factors, including financial distress risk, and return based on fama-macbeth regression, and the results are shown in table 9. first, all the selected risk factors, namely financial distress, size, and value, were combined into model 1 and the results showed size as being positively significant and value as negatively significant. in contrast, financial distress was found to have an insignificant effect on return. in model 2, this 102 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 study combined the risk factors, namely size and financial distress, in affecting stock returns. the results showed size as being positively significant in affecting stock returns while financial distress was insignificant with a negative sign. these results were consistent with the results obtained in model 1. as for model 3, value was found to be negatively significant, which was consistent with the results of model 1. however, financial distress showed a negative coefficient and an insignificant result. thus, for this model, financial distress risk did not have a significant effect on stock returns. model 4 consisted of only value and size. the results showed a significant value with a negative sign, indicating a significant negative effect on stock returns. meanwhile, size was found to be positively significant in affecting stock returns. lastly, model 5 was the model for univariate analysis, which investigated the effect of financial distress risk on stock returns. however, the result for this variable was found to be statistically insignificant to affect returns since the p-value for the variable’s t-test showed values of more than 0.10. based on the results obtained in table 9 using fama-macbeth regression analysis, all models demonstrated that a firm’s value had a significant and negative coefficient in affecting stock returns. as a firm’s value increases, the firm’s returns will decrease (bauer & agarwal, 2014; kothari & shanken, 1997; taneja, 2010). this result was similar to the study by reilly and brown (2011), who asserted that growth stocks should obtain higher returns compared to value stocks due to growth stocks’ higher risk levels. fama and french (1995) explained that a high book-to-market ratio indicated poor firm performance in earnings and profitability compared to a low book-tomarket ratio stock. thus, to obtain a high return in terms of large price appreciation and dividend from a firm’s profit, investors will focus their investment on low value, or high growth stocks. the results for all models also showed that size was positively significant in affecting returns. these results indicated that a large firm’s stock earned a higher return compared to a small firm’s stock (hassan & javed, 2011; shoaib & siddiqu, 2016). mobarek and mollah (2005) explained that the significant positive results were due to the assumptions that all investors had a homogenous expectation, 103 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 table 9 fama-macbeth regression result constant fd size value r-square model 1 -0.0084 -0.0282 0.0040 -0.0057 0.0739 [0.0083] [0.0268] [0.0009] [0.0023] (-1.10) (-1.05) (4.54***) (-2.46**) model 2 -0.0135 -0.0270 0.0045 0.0658 [0.0078] [0.0269] [0.0009] (-1.72*) (-1.00) (4.99***) model 3 0.0130 -0.0306 -0.0065 0.0551 [0.0053] [0.0267] [0.0024] (2.46**) (-1.14) (-2.76***) model 4 -0.0096 0.0041 -0.0057 0.0303 [0.0085] [0.0009] [0.0021] (-1.13) (4.56***) (-2.73***) model 5 0.0094 -0.0284 0.0471 [0.0049] [0.0267] (1.93*) (-1.06) number of observations 141425 note: fd represents financial distress risk generated from the prediction model, size represents firm’s size based on market capitalisation, and value represents firm’s value based on book-to-market value. the value inside [ ] represents standard error value, while value in ( ) represents t-statistic value. the star symbol (*) for p-value represents statistically significant at different levels: *** statistically significant at 1%, ** statistically significant at 5%, and * statistically significant at 10%. were risk-averse and preferred investing in a large firm that was stable and with a lower risk compared to small firms. institutional investors and foreign and local mutual funds preferred investing in a large firm that could generate high returns with higher dividend payout ratios (gompers et al., 2010; rashid & abbas, 2011; zhu, 2010). 104 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 lastly, all models showed that financial distress risk was not significant in affecting stock returns. this result contradicted shumway (1996), dichev (1998), and sabbaghi (2015) but was similar to the study by md-rus (2011). this result was consistent with the risk categories highlighted by reilly and brown (2011) who categorised financial distress as a financial risk which was a part of unsystematic risk. thus, it could be fully diversified via portfolio diversification and would not affect return. conclusion this study focused on predicting financial distress based on the malaysian stock market and investigated the effects of financial distress risk on stock returns. this study began by developing financial distress prediction models using the stepwise logit regression model. this model was used to generate the probabilities of financial distress representing financial distress risk. firms’ financial distress risk, size, and value were used as independent variables to determine stock returns based on fama-macbeth regression. the results of the financial distress prediction model based on the stepwise logit model showed that current ratio, sales to working capital; days’ sales in accounts receivable, earnings before interest and taxes to sales, net profit margin, debt ratio, and interest coverage ratio were significant in predicting financial distress. meanwhile, long-term debt to total debt ratio and shareholders’ fund to total debt were dropped based on the results of the stepwise logit model. subsequently, this study used the probabilities generated from the prediction model developed to represent financial distress risk together with size and value as the independent variables in determining stock returns. as for the models that included financial distress risk, size, and value, only size and value were found to be significant. meanwhile, financial distress risk was found to be insignificant in all the models—both multivariate and univariate—in determining stock returns. the results indicated that financial distress risk does not affect stock returns, which implied that it could be eliminated through diversification. based on the results obtained, corporate management should carefully manage the financial aspects of liquidity, efficiency, profitability and leverage as they will impact a firm’s financial distress risk level. 105 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 meanwhile, the results also indicate that creditors should consider a firm’s liquidity, efficiency, profitability and leverage aspects in making financial decisions to avoid lending capital to financially distressed firms. as for investors, the results show that investment in financial distress risk firm does not affect returns. thus, investors should not be concerned about a firm’s financial distress risk when they develop their portfolios since it can be eliminated through diversification. however, investors can still use the results from the developed prediction model as a guide to measure a firm’s level of financial distress risk. the result also provides current and future researchers with an enhanced understanding in predicting financial distress and the effect of financial distress on returns in the malaysian stock market. in addition, this study suggests that a more comprehensive model which includes more predictors such as cash flow ratios, market ratios, corporate governance variables, and macroeconomic variables should be developed. future studies could also try to compare as many models as possible in order to develop the best model that can be used to measure and predict financial distress risk accurately. lastly, future studies could also incorporate the financial distress probabilities generated from the developed model into fama and french threefactor model, and fama and french four-factor model, specifically to study the emerging markets including the malaysian market to gain a better understanding of the relationship between financial distress and stock returns. acknowledgment the authors gratefully acknowledge the permission to use the services and facilities of perpustakaan sultanah bahiyah, universiti utara malaysia. this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. references abdul manab, n., yen, n., & md-rus, r. (2015). the determinants of credit risk in malaysia. procedia social and behavioral sciences, 172, 301–308. 106 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 abdullah, n. a., & ahmad, a. h. h. (2005). detecting financial distress. international journal of management studies, 12(1), 77–95. abdullah, s. n. (2006). board structure and ownership in malaysia: the case of distressed listed companies. corporate governance, 6(5), 582–594. alifiah, m., & tahir, m. (2018). predicting financial distress companies in the manufacturing and non-manufacturing sectors in malaysia using macroeconomic variables. management science letters, 8(6), 593–604. altman, e. i. (1968). financial ratios, discriminant analysis and the prediction of corporate bankruptcy. the journal of finance, xxiii(4), 589–609. bauer, j., & agarwal, v. (2014). are hazard models superior to traditional bankruptcy prediction approaches? a comprehensive test. journal of banking & finance, 40, 432–442. beaver, w. h. (1966). financial ratios as predictors of failure. journal of accounting research, 71–111. bhunia, a., & sarkar, r. (2011). a study of financial distress based on mda. journal of management research, 3(2), 1-11. boubaker, s., hamza, t., & vidal-garcía, j. (2018). financial distress and equity returns: a leverage-augmented three-factor model. research in international business and finance, 46, 1–15. chen, y., zhang, l., & zhang, l. (2013). financial distress prediction for chinese listed manufacturing companies. procedia computer science, 17, 678–686. chhapra, i. u., zehra, i., kashif, m., & rehan, r. (2020). is bankruptcy risk a systematic risk ? evidence from pakistan stock exchange. etikonomi, 19(1), 51–62. chiaramonte, l., & casu, b. (2016). capital and liquidity ratios and financial distress. evidence from the european banking industry. the british accounting review. doi:10.1016/j.bar.2016.04.001 daily, c. m., & dalton, d. r. (1994). bankruptcy and corporate governance: the impact of board composition and structure. academy of management journal, 37(6), 1603–1617. denis, d. j., & denis, d. k. (1995). causes of financial distress following leveraged recapitalizations. journal of financial economics, 37(2), 129–157. 107 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 dichev, i. d. (1998). is the risk of bankruptcy a systematic risk? the journal of finance, 53(3), 1131–1147. elloumi, f., & gueyié, j.-p. (2001). financial distress and corporate governance: an empirical analysis. corporate governance, 1(1), 15–23. etemadi, h., anvary rostamy, a. a., & dehkordi, h. f. (2009). a genetic programming model for bankruptcy prediction: empirical evidence from iran. expert systems with applications, 36(2 part 2), 3199–3207. fama, e. f., & french, k. r. (1995). size and book to market factors in earnings and returns. journal of finance, 50(1), 131–155. fama, e. f., & macbeth, j. d. (1973). risk, return, and equilibrium: empirical tests. the journal of political economy, 81(3), 607– 636. fich, e. m., & slezak, s. l. (2008). can corporate governance save distressed firms from bankruptcy? an empirical analysis. review of quantitative finance and accounting, 30(2), 225– 251. foster, b. p., & zurada, j. (2013). loan defaults and hazard models for bankruptcy prediction. managerial auditing journal, 28(6), 516–541. ganesalingam, s., & kumar, k. (2001). detection of financial distress via multivariate statistical analysis. managerial finance, 27(4), 45–55. gompers, p., ishii, j., & metrick, a. (2010). extreme governance: an analysis of dual-class firms in the united states. review of financial studies, 23(3), 1051–1088. griffin, j. m., & lemmon, m. l. (2002). book–to–market equity, distress risk, and stock returns. the journal of finance, lvii(5), 2317–2336. hassan, a., & javed, m. t. (2011). size and value premium in pakistani equity market. african journal of business management, 5(16), 6747. idrees, s., & qayyum, a. (2018). the impact of financial distress risk on equity returns : a case study of non-financial firms of pakistan stock exchange. munich personal repec archive, (85346), 1–22. juniarti, j. (2013). good corporate governance and predicting financial distress using logistic and probit regression model. jurnal akuntansi dan keuangan, 15(1), 43–50. 108 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 kothari, s. p., & shanken, j. (1997). book-to-market, dividend yield, and expected market returns: a time-series analysis. journal of financial economics, 44(2), 169–203. lee, t., & yeh, y. (2004). corporate governance and financial distress: evidence from taiwan. corporate governance, 12(3), 378–388. low, s. w., nor, f. m., & yatim, p. (2001). predicting corporate financial distress using the logit model: the case of malaysia. asian academy of management journal, 6(1), 49–61. md-rus, r. (2011). an empirical investigation on the relationship between risk of bankruptcy and stock return. international journal of management studies, 18(1), 117-134. md-rus, r., & abdullah, n. a. h. (2005). empirical investigation of failing companies in an emerging capital market. in 16th asian finance association conference, malaysia. md-zeni, s., & ameer, r. (2010). turnaround prediction of distressed companies: evidence from malaysia. journal of financial reporting and accounting, 8(2), 143–159. ming, x., & akhtar, h. (2014). impact of corporate governance on financial distress: an evidence from shanghai stock market an analysis based on a panel data and logit model. interdisciplinary journal of contemporary research in business, 6(6), 1–11. mobarek, a., & mollah, s. (2005). the general determinants of share returns: an empirical investigation on the dhaka stock exchange. review of pacific basin financial markets and policies (rpbfmp), 08(04), 593–612. mselmi, n., hamza, t., lahiani, a., & shahbaz, m. (2019). pricing corporate financial distress : empirical evidence from the french stock market. journal of international money and finance, 96, 13–27. nam, j.-h., & taehong, j. (2000). bankruptcy prediction: evidence from korean listed companies during the imf crisis. journal of international financial management & accounting, 11(3). ohlson, j. a. (1980). financial ratios and the probabilistic prediction of bankruptcy. journal of accounting research, 18(1), 109– 131. ong, s.-w., yap, v. c., & khong, r. w. l. (2011). corporate failure prediction: a study of public listed companies in malaysia. managerial finance, 37(6), 553–564. 109 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 opler, t. c., & titman, s. (1994). financial distress and corporate performance. journal of finance, 49(3), 1015–1040. parker, s., peters, g. f., & turetsky, h. f. (2002). corporate governance and corporate failure: a survival analysis. corporate governance, 2(2), 4–12. pindado, j., rodrigues, l., & de la torre, c. (2008). estimating financial distress likelihood. journal of business research, 61(9), 995–1003. platt, h. d., & platt, m. b. (2008). financial distress comparison across three global regions. journal of risk and financial management, 1(1), 129–162. polemis, d., & gounopoulos, d. (2012). prediction of distress and identification of potential m&as targets in uk. managerial finance, 38(11), 1085–1104. rashid, a., & abbas, q. (2011). predicting bankruptcy in pakistan. theoretical & applied economics, 18(9), 103-128. reilly, f. k., & brown, k. c. (2011). investment analysis and portfolio management. cengage learning. sabbaghi, o. (2015). volatility, distress risk, and the cross-section of portfolio returns. review of accounting and finance, 14(2), 149–171. sathye, m., bartle, j., vincent, m., & boffey, r. (2003). credit analysis and lending management. john wiley & sons. shoaib, a., & siddiqu, m. a. (2016). quantile methods for testing the applicability of capm and ff-model in pakistan. in gai international academic conferences proceedings (p. 78). shumway, t. (1996). size, overreaction, and book-to-market effects as default premia. http://www-personal.umich.edu/~shumway/ papers.dir/defprm1.pdf, shumway, t. (2001). forecasting bankruptcy mode accurately: a simple hazard model. the journal of business, 74(1), 101–124. simlai, p. (2014). firm characteristics, distress risk and average stock returns. accounting research journal, 27(2), 101–123. sori, z. m., hamid, m. a. a., nassir, a. m., & mohamad, z. a. (2001). forecasting corporate failure in malaysian industrial sector firms. asian academy of management journal, 6(1), 1-16. sudirgo, t., yuniarwati., & bangun, n. (2019). pengaruh financial distress, financial perfomance dan likuiditas terhadap stock return. business management journal, 15(2), 77–92. tan, c. n. w., & dihardjo, h. (2001). a study of using artificial neural networks to develop an early warning predictor for credit 110 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 union financial distress with comparison to the probit model. managerial finance, 27(4), 56–77. taneja, y. p. (2010). revisiting fama french three-factor model in indian stock market. vision: the journal of business perspective, 14(4), 267–274. thai, s. b., goh, h. h., teh, b. h., wong, j. c., & ong, t. s. (2014). a revisited of altman zscore model for companies listed in bursa malaysia. international journal of business and social science, 5(12), 197–207. tirapat, s., & nittayagasetwat, a. (1999). an investigation of thai listed firms’ financial distress using macro and micro variables. multinational finance journal, 3(2), 103–125. ugurlu, m., & aksoy, h. (2006). prediction of corporate financial distress in an emerging market: the case of turkey. cross cultural management: an international journal, 13(4), 277– 295. wang, z., & li, h. (2007). financial distress prediction of chinese listed companies: a rough set methodology. chinese management studies, 1(2), 93–110. yap, b. c. f., munuswamy, s., & mohamed, z. (2012). evaluating company failure in malaysia using financial ratios and logistic regression. asian journal of finance & accounting, 4(1), 330– 344. youn, h., & gu, z. (2010). predicting korean lodging firm failures: an artificial neural network model along with a logistic regression model. international journal of hospitality management, 29(1), 120–127. zaki, e., bah, r., & rao, a. (2011). assessing probabilities of financial distress of banks in uae. international journal of managerial finance, 7(3), 304–320. zaretzky, k., & zumwalt, j. k. (2007). relation between distress risk, book-to-market ratio and return premium. managerial finance, 33(10), 788–797. zhu, t. (2010). do domestic and foreign institutional investors have similar portfolio preferences? in 2010 international conference on management and service science (pp. 1–4). the roles of health awareness and knowledge in medical takaful purchase intention muhammad zarunnaim haji wahab* asma mohd tajuddin islamic business school, universiti utara malaysia, malaysia *corresponding author: zarunnaim@yahoo.com ________________________________________________________________ a r t i c l e i n f o _____________________________ article history: received 18 april 2019 revised form 24 october 2019 accepted 3 november 2019 published 6 january 2020 ____________________________ keywords: medical takaful, customer’s intention, health awareness, knowledge, public sector officers. jel code: m 3, m 1, i1 a b s t r a c t _________________________________ the present study aims to examine the factors of health awareness and knowledge that influence customers’ intention to use medical takaful coverage among the public sector officers in northern malaysia. a quantitative research methodology was adopted to answer the objectives progressively. in the current research, data from 313 respondents were collected via questionnaires from three different organisations, namely the police department, the fire and rescue firefighter department, and the education department. moreover, the data analysis was performed using spss version 22. the analysis revealed a significant and positive relationship between health awareness and knowledge and the customers’ intention to purchase medical takaful. apart from that, the results also indicated that only health awareness has a significant influence on the intention with the strongest impact. on the other hand, the knowledge variable had no significant influence on customers’ intention to use medical the international journal of banking and finance, vol. 14, 2018-2019 : 95-116 95 96 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 takaful coverage among the respondents. a number of important limitations need to be considered in the present study which will be further explained in the conclusion section. furthermore, the current research provides some major implications. first, on behalf of the takaful industry, the findings of the present study encourage the takaful operators, marketing teams, staffs, and agents to improve their performance by increasing their efficiency, creativity, innovation, and knowledge of product features in developing a good impression among potential customers. second, on behalf of the public sector officers, the results of the study show the important need to enhance the customers’ intention to use medical takaful coverage. third, the suitability of factors such as health awareness, knowledge and customers’ intention in the previous theory and literature on takaful industry has been proven in the current research. this study helps the marketing team of the takaful operators to formulate marketing strategies that can attract potential customers to participate in the medical takaful coverage, especially among the public sector officers. 1. introduction the basic concept of takaful is to protect humans from risks, while also promoting the element of brotherhood, solidarity, and mutual assistance among the members of a society (marhanum et al., 2013). the application of takaful is based on the shari’ah-compliant product which is in line with the al-quran and as-sunnah. more importantly, several verses from the quran such as verse 2 of surah al-ma’idah and verse 201 of surah al-baqarah acknowledge the validity of the elements applied in takaful. the roles of health awareness and knowledge in medical takaful purchase intention: 95-116 97 according to hamid et al. (2011) the concepts of takaful and conventional insurance are different. takaful is a concept developed based on three principles, namely mutual responsibility, cooperation with each other, and protecting one another from any kind of difficulties, disasters, and other misfortunes. regarding this matter, it should be understood that the contribution fund is based on the concept of tabarru’. most islamic jurists agree that the operation of the conventional insurance contradicts the rules and requirements of the shari’ah. a clear explanation for this matter is that the operational activities of conventional insurance evidently involve the elements of uncertainty (gharar), gambling (maysir), and interests (riba) (hamid et al., 2011; munawar, 2013). in the takaful industry, the assessment of the customers’ intention is important for the purpose of enticing them to participate in takaful products. accordingly, it is necessary to examine customers’ purchasing attitude and behavior in order to fulfil customers’ intention and satisfaction. regarding this matter, gillmore et al. (2002) and abubakar et al. (2017) define intention as a relationship between behaviors and attitudes within a human action. hence, intention can be utilized to predict how individuals behave based on their preexisting attitudes and behavioral intentions in purchasing an item. as a result, customers’ satisfaction can be achieved if the decision is accurately made within their intention. a factor that influences customers’ intention is health awareness. a study by benjamin et al. (2014) highlighted the importance of health protection for individuals who have to bear the high cost of medical expenses in the future. indeed, kumanyika et al. (2000) found that only a few succeeded in maintaining a healthy lifestyle even though they initially had the intention of doing so. webb and sheeran (2006) indicated that good intentions influence the actual behavior to pursue healthy behavior in an attempt to ensure the best actual goal-oriented behavior for public health awareness. in malaysia, public health care services have been experiencing establishment pressures due to the increasing demand and limited available resources. current services have almost provided full insurance to the overall population and all civil servants. subsequently, this causes health care services and patients to suffer from overcrowding, understaffing, lack of quality and waste of time. meanwhile, a study by dyah and rizal (2006) stated that nearly 60% of malaysians intended to seek private primary care but 73.2% could not afford the cost. therefore, the findings of the study revealed the strong influence of health awareness towards the intention of seeking private healthcare among malaysians. the second factor refers to the knowledge of the public sector officers on takaful as well as how it affects their intention to use medical takaful coverage. past research by rahman et al. (2008) was offered by the industry to fulfil the customers’ needs in managing the risk which is in line with the shari’ah ways. the study examined the purchasing behavior and knowledge of malaysian undergraduate students with regard to the differences between takaful and insurance products. in particular, the study emphasized the following four 98 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 main factors that reflect how knowledge can affect customers’ intention: services delivery, services provision reputation, electronic services, and convenience. the findings showed that undergraduate students have the intention to use insurance products because they perceived them to be better than the takaful products in all the aforementioned factors. as previously mentioned, takaful products are different from conventional insurance. the knowledge of the differences of both products is important because it allows customers to compare the better alternative prior to opting for a protection policy. research by saif (2012) stated that around 80% of the existing customers changed to takaful when a yemeni conventional insurance company introduced takaful products in 2008. however, it was also argued that the insurance sector realized the importance of takaful products in terms of attracting the intention of yemeni customers and investors because the content and policy of takaful products are transparent compared to any non-shari’ah elements. furthermore, aziz et al. (2011) conducted a research to study government servants based on four main explanatory variables, namely product knowledge, awareness, advertising, and benefit of the product. the results indicated that the variables have a significant influence on the intention of malaysian government employees towards the level of acceptance of motor takaful products. subsequently, this showed that knowledge about the products is important to influence customers’ intention. moreover, sarwar and qureshi (2013) stated that the lack of insurance knowledge is one of the major barriers in the intention of purchasing health and life insurance. in the case of the present study, it is crucial to note that public sector officers may not be aware of their health condition due to their work pressure. as a result, they may suffer from an illness in the future without having the chance to take any prevention. a study by ellis (1995) reported that occupational stress can be inferred from the problems faced by the employees or from environmental interaction. more importantly, it should be noted that the consequences of this stress may manifest in terms of health problems, emotional reactions and organizational outcomes. apart from that, another issue of health awareness is the reduction of healthcare subsidies for public sector officers. the economist (2014) stated that healthcare subsidization by the government is no longer sustainable. according to the world health organization (who), the malaysian government continues with the subsidy-rationalization program in order to reduce the overall government expenditure whereby an outlay as large as healthcare tends to look increasingly unsustainable (11th april 2014). in other words, the healthcare provision for public sector officers is in a precarious situation due to the reduction of subsidies by the government from year to year. therefore, it will be unfortunate for the public sector officers in the future because they need to pay using their own money to receive healthcare services. the present study focuses on three groups of public sector officers, namely the police, firefighters, and teachers due to their risky and uncertain the roles of health awareness and knowledge in medical takaful purchase intention: 95-116 99 job conditions. in this regard, the organization for economic co-operation and development (oecd) pointed out that some countries have repeatedly cut the fees for hospitals as well as the subsidies for medication and equipment (oecd, 2009). moreover, it should be noted that cutting the subsidies will increase the cost of healthcare; hence, the advent of sophisticated treatment will burden the people in covering the rest of the medical cost (eiji & jun, 2012). therefore, it is important for the public sector officers to obtain sufficient knowledge about the healthcare services provided by the takaful industry such as the medical takaful to prevent themselves from bearing high medical costs due to any uncertainty in the future. a search of the literature showed that previous studies focused more on customer intention to participate in takaful from various factors which included mass media, word of mouth, and subjective norms (maizaitulaidawati et al., 2016; farhat et al., 2019), attitude, subjective norm, and perceived behavior (maizaitulaidawati & asmak, 2013), perception of takaful operators’ performance (haji wahab, 2018), and customer beliefs (bhatti & md husin, 2019). however, a limited number of available studies focused on the influence of health awareness and knowledge factors on customers’ intention. apart from that, the literature also showed that none of the studies focused on medical takaful coverage, especially among public sector officers in malaysia. therefore, this paper attempted to examine the relationship between health awareness and knowledge towards the intention to choose medical takaful coverage among public sector officers in the northern region of malaysia. other than that, this paper also aimed to fill the gaps in understanding the concept of medical takaful as well as the factors that influence its demand. the organization of this paper is as follows: section 1 begins with a brief introduction of takaful, followed by section 2 which presents the literature. next, section 3 provides the methodology of this research, while section 4 details the results and discussion of the study. finally, section 5 focuses on the conclusion and limitations of the research. 2. literature review 2.1 health awareness health is described as the level of functional or metabolic efficiency of a living organism. in humans, health is the ability of individuals or communities to adapt and self-manage when they are faced with physical, mental, or social challenges (huber et al., 2011). in 1948 the who constitution, defined health in a broader sense as a state of complete physical, mental and social well-being rather than merely the absence of disease or infirmity. on another note, awareness can be defined in several ways. kjeldskov and skov (2004) defined awareness as the ability to adapt to the changing 100 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 circumstances as well as respond according to the context of use. meanwhile, neaton et al. (1993) defined awareness as the knowledge of being hypertensive which is often obtained through a healthy worker. gregory et al. (1996) described awareness as a factor that influences a healthy lifestyle, including the effect of publicity on diet-related health subjects. according to the study, asians are described as individuals who are less likely to be aware of the importance of beneficial change in health awareness, especially with regard to the diet content provided by public health campaigns, media publicity and advertising program. more importantly, this health awareness contributes to the intention of enrolling in medical takaful scheme. zuriah and rosylin (2008) mentioned that intention enables people to be aware of medical takaful in the attempt to compensate one another after sustaining an accidental loss. a person in distress or experiencing loss may not be brought about by animosity between humans which is usually observed during unexpected events such as natural death, health adversity, unemployment and sickness. a possible explanation for this matter may be because insurance in its essence is an important device created to solve many social obligations. mohamed (2008) mentioned that the intention of takaful is to obligate funds to be distributed for the purpose of strengthening the health care system by aggressively targeting the poorest population and rural areas. on another note, it should be understood that the general community health is imperative. a study by piviki (2002) encouraged community participation in health planning which includes the effort to enhance community awareness on health issues as well as the opportunity to increase networking between health care service providers and community members. the study showed that health awareness influences the intention of the community to join and obtain incentive values from the health planning program. naail and norsham (2014) also highlighted that the awareness of shari’ah compliance product is important in muslims’ daily life. moreover, this awareness may act as a response towards individuals’ intention in preparing for unexpected events in future. nevertheless, a high number of individuals are still not taking any steps towards the preparation despite its importance. drechsler and jütting (2007) through who (2006) determined the urgent need for public health awareness as a sustainable tool for health financing in order to reduce the high amount of the out-of-pocket payments as well as the incidence of catastrophic health shocks in the developing world. on the contrary, studies by fletcher and hastings (1984), kunreuther and pauly (2005) and oluyemi (1995) highlighted that budget constraint is one of the reasons that affect customers’ intention to not participate in an insurance scheme. overall, the public provided the reason that they are confronted with various expenses which leave them with no expendable income to pay for the insurance premium. the literature discussed in this study has concluded that many disadvantages may occur due to the lack of awareness regarding the importance the roles of health awareness and knowledge in medical takaful purchase intention: 95-116 101 of practicing good health care. referring to the public sector officers, the berita harian newspaper in 2012 showed the statement of the ministry of health where the increasing number in claims for takaful coverage was a result of the recent increase in the number of people who suffered from several diseases. therefore, this study aimed to measure whether health awareness influences the intention of the public sector officers to participate in medical takaful coverage. thus, the first hypothesis as follows: hypothesis 1: there is a significant relationship between health awareness and customers’ intention in using medical takaful coverage among public sector officers in northern malaysia. 2.2 knowledge knowledge can be interpreted in several ways. ackoff (1999) defines knowledge as conveying instructions and answering how-to questions. in addition, he describes knowledge as a combination of information and understanding with the aim of increasing efficiency. however, brucks (1985) explains knowledge in the context of a product by defining it as customers’ intention towards specific information which concerns the primary source of the product knowledge of a given product. other than that, it also includes customers’ experience with the actual product and advertisement with the aim of influencing the customer’s decision to select a product (rao & monroe, 1988). furthermore, burton et al. (2009) determine the consistent ranking of product knowledge as the most important factor that influences the intention of a customer with regard to their purchase behavior. nevertheless, alnemer (2015) states that the lack of knowledge and intention about the basics and technicality of the products and services can lead to an obvious deficiency in the customer-motivated reasons to have insurance products and services. regarding the level of knowledge, most researches seem to agree that customers with a higher level of education tend to have higher intention to own insurance coverage (beck & webb, 2003; browne & kim, 1993; li et al., 2007). more importantly, empirical evidence reveals that knowledge of conventional and islamic finance is higher for the highly educated customer (bley & kuehn, 2004). a considerable number of studies indicates that the intention to purchase is slightly affected by knowledge (yin et al., 2010). jeddi and zaiem (2010) found that the level of knowledge provides some impact on customers’ intention. hence, it can be concluded that a high level of knowledge is related to the ability to understand the benefits of risk management and long-term saving. consequently, this leads to high precision in risk assessments that are positively correlated to the intention to demand any product (beck & webb, 2003; lo et al., 2011; sherrick et al., 2004). 102 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 on a similar note, other researchers have also highlighted the importance of knowledge. hamid and othman (2009) emphasized that the level of knowledge and understanding among muslims in malaysia may contribute to enhancing the intention of customers to purchase takaful products. furthermore, naail and norsham (2014) mentioned that customers’ buying intention is based on knowledge of takaful that can be obtained through reading, the internet, and promotions given by takaful agents. a study by lin and chen (2006) indicated that customers are easily influenced by the product and service knowledge during the buying intention and buying decision. accordingly, the study implied that customers’ purchase intention and purchase decision under different product involvement tend to have a positive effect on product knowledge. moreover, a study by sharif et al. (2011) revealed that about 47% of the respondents wished to know more about takaful terms, followed by 26.9% of the respondents who aimed to know more about the claiming process, 21.6% about the product, and 1.5% about related information. hence, knowledge of takaful is important for customers because it can affect their buying intention. accordingly, it is crucial to gain relevant insights into the buyer decision process or strategies in buying because it is one of the useful factors in understanding consumer behaviour. in most cases, consumers usually build a certain set of cognition or pieces of information and knowledge about the product offering prior to deciding whether to buy or not to buy the product. in other words, the lack of knowledge on takaful may affect customers’ intention and decision to either buy the product or not. therefore, ghazali et al. (2011) and ahmad et al. (2010) stressed that the poor knowledge and understanding of insurance and takaful products in malaysia, further affect customers’ intention to use the services. consequently, the present study aimed to investigate the effect of knowledge on customers’ intention among public sector officers. thus, the second hypothesis is as follows: hypothesis 2: there is a significant relationship between knowledge and customers’ intention in using medical takaful coverage among public sector officers in northern malaysia. 2.3 customers’ intention in the present time, the theory of reasoned action (tra) model developed by martin fishbein and icek ajzen in 1975 has been widely and frequently used by classic persuasion models in psychology for the purpose of understanding the relationship between behavior, attitude, intention and belief within human action (md-taib et al., 2008). moreover, gillmore et al. (2002) explained that this theory has been used to predict and create an assumption on how individuals behave based on their pre-existing attitudes and behavioral intentions. in short, it can be understood that an individual will make the decision to do something if he or she has the intention to perform a certain behavior. the roles of health awareness and knowledge in medical takaful purchase intention: 95-116 103 on another note, several reasons have been identified as to why this model was adopted by previous studies. first, the tra model is useful in determining or predicting individual intention. apart from that, this theory is also flexible and can be easily adjusted in different contexts of research fields. moreover, this fact is supported by published studies that investigated islamic credit coverage (muhammad ali et al., 2017; hanudin amin, 2013), islamic bank adoption (sulaiman et al., 2016), and zakah (islamic tax) compliance (ram & roszaini, 2014). therefore, this study chose to adopt this model for the purpose of predicting the behavioral intention to use medical takaful coverage. data from published studies have shown that there is a positive and significant relationship between health awareness and customers’ intention. moreover, this statement is supported by various studies including research by montano et al. (1997) who demonstrated that health awareness is positively correlated to intention, while elmanan (2005) found that health treatment has a positive correlation with customers’ intention. therefore, those who are aware of health takaful are likely to have a high intention to get takaful coverage (zuriah & rosylin, 2008). furthermore, piviki (2002) stated that health planning is an effort to instil health awareness which is positively correlated to customers’ intention. a number of past studies found a significant positive relationship between knowledge and customers’ intention. this statement is consistent with previous research by burton et al. (2009) who found that product knowledge is correlated to the intention of customer purchasing behaviour. meanwhile, alnemer (2015) identified that a lack of knowledge has a positive correlation with intention because the customers are reluctant to participate in insurance products and services. in addition, dettmann and dimitri (2007) reported that there is a positive association between the level of education and intention in regard to customers’ interest. more importantly, a study by hamid and othman (2009) managed to find a positive statistical relationship between knowledge and intention to purchase takaful products and services. therefore, based on the evidence above, the current research adopted the theory of reasoned action (tra) to determine customers’ intention. thus, the third hypothesis is as follows: hypothesis 3: there is a significant influence between health awareness, knowledge and customers’ intention in using medical takaful coverage among public sector officers in northern malaysia. 3. methodology 3.1 theoretical framework figure 1 shows the theoretical framework of the present study as well as the variables that need to be examined according to the objective of the study. 104 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 figure 1. the theoretical framework of the study. 3.2 instrument and measurement items the present study adopted a survey-based questionnaire which consisted of 20 questions, including the demographic questions that were designed to answer the proposed hypothesis: the factors that affect customers’ intention to use medical takaful coverage among public sector officers in northern malaysia. the questionnaire was further divided into three parts but excluded demography. moreover, it should be noted that all items intended to measure the constructs in the present study were adopted the previously validated instruments. first, cooper and sutherland’s (1987) questions were adapted to test health awareness, followed by the second part from amim (2000) which was compiled and added to the questionnaire in order to test knowledge. the third part was adopted from fishbein and ajzen (1975) in order to test customers’ intention to use medical takaful coverage. more importantly, some of the items were modified to suit the target respondents. the questionnaires employed a six-point likert scale from ‘extremely disagree’ to ‘extremely agree’ for data collection purposes. 3.3 sample the population was drawn using the disproportionate stratified random sampling from three public sector officers in northern malaysia, namely the royal malaysia police department, the fire and rescue department, and the state education department. overall, there were 7,304 officers from the three government agencies. following the suggestion of krejcie and morgan (1970), 367 officers were selected as the sample size for the present study (as cited in sekaran & bougie, 2012). the allocation of the distributed questionnaires was as follows: 300 questionnaires for teachers, 40 for policemen, and 27 for firefighters. however, only 313 questionnaires were returned and valid for data analysis. the high response rate was attributed to the presence of the researcher at the organisations in ensuring the completeness of the questionnaires. 9 determinants of attitude towards zakat on employment income in nigeria: 29-48 statement is supported by various studies including research by montano et al. (1997) who demonstrated that health awareness is positively correlated to intention, while elmanan (2005) found that health treatment has a positive correlation with customers’ intention. therefore, those who are aware of health takaful are likely to have a high intention to get takaful coverage (zuriah & rosylin, 2008). furthermore, piviki (2002) stated that health planning is an effort to instil health awareness which is positively correlated to customers’ intention. a number of past studies found a significant positive relationship between knowledge and customers’ intention. this statement is consistent with previous research by burton et al. (2009) who found that product knowledge is correlated to the intention of customer purchasing behaviour. meanwhile, alnemer (2015) identified that a lack of knowledge has a positive correlation with intention because the customers are reluctant to participate in insurance products and services. in addition, dettmann and dimitri (2007) reported that there is a positive association between the level of education and intention in regard to customers’ interest. more importantly, a study by hamid and othman (2009) managed to find a positive statistical relationship between knowledge and intention to purchase takaful products and services. therefore, based on the evidence above, the current research adopted the theory of reasoned action (tra) to determine customers’ intention. thus, the third hypothesis is as follows: hypothesis 3: there is a significant influence between health awareness, knowledge and customers’ intention in using medical takaful coverage among public sector officers in northern malaysia. 3. methodology 3.1 theoretical framework figure 1 shows the theoretical framework of the present study as well as the variables that need to be examined according to the objective of the study. independent variables dependent variable figure 1. the theoretical framework of the study 3.2 instrument and measurement items the present study adopted a survey-based questionnaire which consisted of 20 questions, including the demographic questions that were designed to answer the health awareness (cooper & sutherland, 1987). customers intention (fishbein & ajzen, 1975) knowledge (amim, 2000) the roles of health awareness and knowledge in medical takaful purchase intention: 95-116 105 3.4 data analysis this study used descriptive and inferential analyses. moreover, several statistical tools and methods were used from the spss software version 22.0 for the purpose of data analysis and hypothesis testing. the data analysis included pilot test, descriptive statistic, pearson correlation, and multiple regression analysis. 3.5 pilot test a total of fifty pre-respondents from the public sector officers were selected to participate in the pilot test prior to the distribution of the final questionnaire. the purpose of the pilot test is to ensure the clarity of the questionnaire’s statements to the respondents as well as to detect other possible weaknesses in the questionnaire. therefore, the pilot study was aimed to provide some feedback that could help the researcher to improve the questionnaire. in particular, some of the improvements involved the restructuring of the statements in simpler words in order to ensure the ease of comprehension. more importantly, the pilot test was also used to examine the question flow as well as to determine whether or not to remove or improve several questions prior to the large-scale implementation. moreover, this idea is supported by bryman and bell (2007) who stated that it could assist in obtaining extra information and accurate responses from the respondents. table 1. reliability test variable no. of items cronbach’s alpha pilot test customers intention health awareness knowledge 5 5 5 0.977 0.742 0.760 table 1 shows the result interpreted using the cronbach’s alpha value obtained from the pilot test. the outcome obtained was in the range of 0.742 to 0.977. according to sekaran and bougie (2010), the items are considered good if the value is in the range of 0.70 to 0.9, while the items are considered excellent if the value is more than 0.90. in the case of this study, the items were reliable to be tested since the value was more than 0.70 (a > 0.7). 4. results and discussions 4.1 respondent’s demographic profile frequency distribution was employed in the present study for the purpose of finding the frequency characteristics of the respondent profile. in the case of 106 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 the current research, 367 questionnaires were distributed to the respondents. however, only 313 questionnaires were returned and valid for data analysis. hence, demographic information such as gender, marital status, age, type of public sector and job category was obtained from the respondents. the demographic profile of the respondents for the present study is represented in table 2. table 2. result of the demographic profile demographic items no. of respondents percent gender male 105 33.5 female 208 66.5 marital status single 47 15 married 254 81.2 divorce 12 3.8 age less than 25 years 17 5.4 26-35 years 93 29.7 36-50 years 166 53 51 and above 37 11.8 type of public sector officers teachers 252 80.5 fire fighters 23 7.3 police 38 12.1 job category professional and managerial 241 77 supportive groups 72 23 the majority of the respondents in the study were female with a total of 208 respondents (66.5 %), whereas the remaining 33.5% represented 105 male respondents. regarding the marital status, 47 respondents (15.0%) were single, 254 respondents (81.2%) were married, and 12 respondents (3.8%) were divorced. meanwhile, most respondents were between the ages of 36 and 50 years old (166 respondents or 53.0%), followed by 93 respondents (29.7%) in the age range of 26 and 35 years old, 37 respondents (11.8%) who were above 51 years old, and only 17 respondents (5.4%) were younger than 25 years old. table ii also shows the type of public sector officers as follows: 252 respondents (80.5%) were teachers, 38 respondents (12.1%) were police, and 23 respondents (7.3%) were fire fighters. finally, it should be noted that 241 respondents (77.0%) were from professional and managerial groups, while the remaining 72 respondents (23.0%) were from supportive groups. the roles of health awareness and knowledge in medical takaful purchase intention: 95-116 107 4.2 relationship between customers’ intention and its determinants the pearson correlation analysis was utilised in the study for the purpose of determining the relationship between health awareness and knowledge with regard to customers’ intention to use medical takaful coverage among public sector officers. table 3 illustrates a significant positive relationship (r = 0.256) between health awareness and customers’ intention to use medical takaful coverage. therefore, hypothesis 1 is accepted. table 3. correlation between health awareness and customers’ intention correlations customers intention health awareness .256** ** correlation is significant at the 0.01 level (2-tailed). table 4 shows that there is a significant positive relationship (r = 0.151) between knowledge and customers’ intention to use medical takaful coverage. therefore, hypothesis 2 is accepted. table 4. correlation between knowledge and customer’s intention correlations customers intention knowledge .151** ** correlation is significant at the 0.01 level (2-tailed). 4.3 the influence of determinants on customers’ intention multiple regression analysis was conducted to determine the significant influence of health awareness and knowledge on customers’ intention to use medical takaful coverage. the result indicated that the r2 value was 0.069 or 6.9%. hence, the present study indicated that 6.9% of the variance in customers’ intention to use medical takaful coverage was significantly explained by the two independent variables of the study, namely health awareness and knowledge. on the other hand, 93% of the variance was explained by other untested factors. moreover, the unstandardized β-coefficient for health awareness had a significant value of 0.302, while knowledge it was 0.062 with a significant value of 0.280. therefore, it is clear that only health awareness significantly influenced customers’ intention to use medical takaful coverage, whereas the knowledge variable was insignificant due to the value exceeding 0.05. therefore, hypothesis 3 is rejected. 108 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 table 5. multiple regression analysis of the coefficient unstandardized coefficients standardized coefficients t sig. b std. error beta (constant) 2.836 0.308 9.214 0.000 health awareness 0.302 0.077 0.232 3.923 0.000 knowledge 0.062 0.058 0.064 1.082 0.280 dependent variable = customers intention r² = 0.069, f–value = 11.512, significant = 0.000 the following equation represents the multiple regression analysis: γ = α + β1x1 + β2x2 where γ = intention to use medical takaful coverage α = intercepts (constant value) β = unstandardized beta coefficient x1 = health awareness (ha) x2 = knowledge (kn) therefore, the equation for the multiple regression analysis is shown below: γ = 2.836 + 0.302ha + 0.062kn the formula above explains that customers’ intention to use medical takaful coverage will increase by 0.302 units when one unit of health awareness is increased. meanwhile, an increase of one unit of knowledge reflects an increase in customers’ intention to use medical takaful coverage by 0.062 units. 4.4 discussion several published studies by tsai et al. (2013), jayawardhena et al. (2009), munnukka (2007), and porter and donthu (2006) found that the sociodemographic variables which include age, education, gender, and others can influence the intention of consumers when making decisions. moreover, many researchers considered gender and age as important factors that affect customers’ intention (zhang, 2005; riquelme & rios, 2010; jun et al., 2008; dewan et al., 2009; singh, 2014). meanwhile, a study by maizaitulaidawati and asmak (2013) revealed that demographic variables are crucial in determining customers’ intention to participate in a family takaful scheme. the roles of health awareness and knowledge in medical takaful purchase intention: 95-116 109 accordingly, the demographic analysis showed that most of the public sector officers who participated as respondents in the present study were mature people (36-50 years old). moreover, another large segment of the customers was in the age range of 26 to 35 years old (young people). furthermore, the results revealed that the majority of the respondents possessed professional and managerial positions (241 respondents). therefore, it is more relevant for takaful operators to provide diverse and unique medical takaful schemes that suit the preference and satisfaction of both groups. in other words, takaful operators ought to consider tailoring their products and services to suit the characteristics of the customers. the outcomes revealed that there was a significant positive relationship among all the independent variables (health awareness and knowledge) and the dependent variable (customers’ intention). in the case of the present study, the positive value (r = 0.256) signified a weak relationship between health awareness and customers’ intention. the result is in line with previous studies which stated that health awareness has a positive relationship with customers’ intention (montano et al., 1997; elmanan, 2005; zuriah & rosylin, 2008; piviki, 2002). similarly, the present study revealed a positive relationship between knowledge and customers’ intention with a value of 0.007 (p < 0.01, sig.2tailed). the value (r = 0.151) signified a weak relationship between knowledge and customers’ intention. moreover, the result is consistent with prior research (dettmann & dimitri, 2007; burton et al., 2009; hamid & othman, 2009; alnemer, 2015). regarding the last objective, the result indicated that only health awareness significantly influenced customers’ intention. however, knowledge had no significant influence because the value of 0.280 exceeded the significant value of 0.05. apart from that, the result showed the β-value of health awareness was 0.302, while for knowledge it was 0.062. hence, it can be concluded that health awareness provides the strongest impact on the dependent variable as well as influences customers’ intention to use medical takaful coverage among public sector officers in the northern region of malaysia. more importantly, all the measured items can be used to evaluate the practice of relationship marketing, particularly for the takaful industry. 5. implication, limitation and conclusion 5.1 implication the results of the present study have provided a number of useful implications for certain groups. on behalf of the takaful industry, this study acknowledged the important need for takaful operators, marketing teams, staffs, and agents to improve their performance by increasing their efficiency, creativity, innovation, and knowledge of product features in developing a good impression among potential customers. hence, the findings of the present study are hoped to be useful by allowing the marketing teams to formulate marketing strategies. 110 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 the results offered several significant methods that can be adopted to pique the intention of potential customers towards the medical takaful coverage facility. first, takaful operators should keep afloat customers’ needs and preferences to ensure a creative and innovative marketing strategy. more importantly, this will help the customers to have more awareness of takaful products, especially medical takaful coverage. accordingly, a survey should be carried out in real-time because the need to wait for a long period of time may limit the precision and usefulness of the information in the decision-making process. therefore, comparisons should be made amidst the fast adaptation of marketing strategies. second, it is very important for takaful operators to educate their staffs and agents in ensuring that they have a good understanding of the product features. in other words, it is necessary to enhance staff knowledge for the purpose of providing a clear picture of the takaful product to potential customers. moreover, the knowledge is aimed to attract potential customers who are concerned about the necessity of the medical takaful coverage scheme. accordingly, it is highly suggested that public sector officers have adequate education and awareness regarding the importance of applying for medical takaful coverage for their health. overall, this is to ensure that the customers are covered from unfortunate events. meanwhile, promotional tools can be a useful approach to explain the benefits of medical takaful coverage products, which is likely to result in greater intention among potential customers. regarding the enhancement of knowledge and expertise among takaful staffs and agents, a series of seminars and workshops should be held to discuss the current knowledge related to principles, practices and product features for the purpose of building competence. on the other hand, the seminars are expected to provide good opportunities for senior managers to improve the concepts and principles of takaful. meanwhile, workshops are believed to be helpful to officers who are directly involved with the structuring and transactions of takaful products because they are expected to further develop their competence and expertise. the results of this study have provided a great amount of useful information to the public sector institutions. more importantly, the findings emphasised the need to increase the intention to use medical takaful coverage among public sector officers. in addition, forums and conferences are important avenues to create health awareness, knowledge and intention to participate in takaful protection. therefore, it is important to ensure that the officers are always in good health to enable them to receive productive and efficient services. previous theory and literature related to health awareness, knowledge and customers’ intention have proven suitable to be adopted even in the takaful industry research. the present study further contributes to the takaful industry field by further investigating the possibility of a relationship between health awareness and knowledge among public sector officers in regard to their intention to use takaful protection, especially the medical takaful coverage. accordingly, the current research successfully expanded the existing knowledge of takaful products and services in malaysia and other countries. the roles of health awareness and knowledge in medical takaful purchase intention: 95-116 111 1.2 limitation there are some limitations in this study. first, the location and the target respondents were limited because this study only focused on three public sectors in the northern region of malaysia. hence, it is recommended for future research to further investigate various types of public sectors and locations in order to obtain a wide range of responses. second, the sample size of the present study is considered small. in terms of directions for future research, further work could increase the sample size by including other segments such as private sectors to obtain a different view on customers’ intention. third, the present study only focused on the factors of health awareness and knowledge towards customers’ intention. therefore, another possible area of future research would be to include other factors that influence customers’ intention. 6. conclusions the present study aimed to achieve three main objectives. first, the current research attempted to identify the significant relationship between health awareness and customers’ intention. the results illustrate a significant positive relationship between both factors. next, the second objective was to examine the significant relationship between knowledge and customer intention. the results revealed that both independent variables, namely health awareness and knowledge had significant positive relationships with the dependent variable (customers’ intention). finally, the third objective was to determine the significance of health awareness, knowledge and customers’ intention in using medical takaful coverage. the findings revealed that only health awareness significantly influenced the customers’ intention of public sector officers in northern malaysia. references abubakar u. farouk, kamil md idris & ram al jaffri saad. (2017). determinants of attitude towards zakat on employment income in nigeria, the international journal of banking and finance, 13(1), 29-48. ackoff, r. l. (1999). ackoff ’s best. john wiley & sons. ahmad, m. i., masood, t., & khan, m. s. (2010) problems and prospects of islamic banking: a case study of takaful, mpra paper. ajzen, i. (1991). the theory of planned behavior: organizational behaviour and human decision processes, 50, 179-211. 112 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 alnemer, h. (2015). participants’ perceptions about takaful operators disclosures system: an empirical study on saudi arabia. islamic finance, performance and efficiency. v (3). gerlach press. germany. amim, m. o. (2000). factors influencing the perception of government servantstowards takaful as an alternative to conventional insurance. thesis (m.o.m). international islamic university malaysia (iium). gombak.selangor. aziz, w. a. b. w. a, mat, a. b. c., & zin, e. a. m. e. w. (2011). a study of contributing factors in islamic motor insurance. gaziantep university journal of social science. vol. 10 no. 1, pp. 1-20. beck, t., & webb, i. (2003). economic, demographic and institutional determinants of life insurance consumption across countries. the world bank economic review, vol. 17, no. 1, pp. 51-88. benjamin, n. w. j., fong l. h., goon l. t., tan x. y., & tan y. c. (2014). the determinants on health and life insurance demand among malaysians. faculty of business and finance, department of finance. universiti tunku abdul rahman. berita harian (bh) online. insurans rm1 b. retrieved from:www. takafulinsurance.com.my/wp-content/uploads/2012/06/beritaharianinsurans1-billion.jpg bhatti, t., & md husin, m. (2019). an investigation of the effect of customer beliefs on the intention to participates in family takaful schemes, journal of islamic marketing. bley, j., & kuehn, k. (2004). conventional versus islamic finance: student knowledge and perception in the united arab emirates. international journal of islamic financia in services, 5, 17-30. browne, m. j., & kim, k. (1993). an international analysis of life insurance demand. journal of risk and insurance, 60, 616-634. brucks, m. (1985). the effect of product class knowledge on information search behavior. journal of consumer research, 12(1), 1-16. bryman, a., & bell, e. (2007). business research method (2nd ed.). oxford university press inc. burton, s. e., howlett & tangari, a. h. (2009). food for thought: how will the nutrition labeling of quick service restaurant menu items influence consumers’ product evaluations, purchase intentions, and choices?. journal of retailing, 85(3), 258-273. colman, a. (2015). theory of reasoned action. a dictionary of psychology. cooper, c. l., & sutherland, v. j. (1987). job stress, mental health, and accidents among offshore workers in the oil and gas extraction industries. journal of occupational medicine, 29, 119-125. dettmann, r., & dimitri, c. (2007). who’s buying organic vegetables? demographic characteristics of us consumers. journal of food distribution research, 49-62. dewan, s., low, g., & land, l. (2009). consumer choice model of mobile banking. proceedings of the 20th australasian conference on information systems, paper no. 101, monash university, melbourne, 2-4 december. the roles of health awareness and knowledge in medical takaful purchase intention: 95-116 113 drechsler, d., & jütting, j. (2007). different countries, different needs: the role of private health insurance in developing countries. journal of health politics, policy and law, 32(3), 497-534. dyah, p. s., & rizal, a. m. (2006). patients’ satisfaction in antenatal clinic hospital universiti kebangsaan malaysia. jurnal kesihatan masyarakat (malaysia), 12(1), 09-18. eiji, t., & jun, k. (2012). the roles of public and private insurance for the health care reform of japan. hitotsubashi university, national institute of population and social security research, tokyo. policy research institute, ministry of finance, japan, public policy review, 8(2), july 2012. ellis, k. (1995). the organizational approach to stress management. psychological health in the workplace: understanding and managing occupational stress. p cotton (ed.) australian psychological society, 1995. elmanan, n. a. (2005). evaluation of contribution rate in health insurance system khartoum state”, faculty of economic and social studies. department of econometrics. university of khartoum. khartoum. farhat, k., aslam, w., & sany sanuri, b. (2019). predicting the intention of generation m to choose family takaful and the role of halal certification. journal of islamic marketing. fishbein, m., & ajzen, i. (1975). belief, attitude, intention and behavior: an introduction to theory and research. addison-wesley, reading, ma. fletcher, k. p., & hastings, w. j. (1984). consumer choice: a study of insurance buying intention, attitudes and beliefs. the service industries journal, 4(2), 174-188. ghazali, p. l. b., mohd, i. b., mamat, m. b., & ahmad, w. m. a. w. (2011). mathematical modelling in family takaful. journal of applied science, 11(19), 3381-3388. gillmore, r. m., matthew a., morrison, d., wilsdon, a., elizabeth, w., hoppe, m., nahom, d., murowchick, e. (2002). teen sexual behavior: applicability of the theory of reasoned action. journal of marriage and family, 64(4), 885-897. gregory, y. h. l., chris, l., maria, mc c., ifran., m. & gareth, b. (1996). ethnic differences in public health awareness, health perceptions and physical exercise: implications for heart disease prevention. ethnicity & health, 1(1), 47-53. haji wahab, m. (2018). perception of the takaful operators’ performance towards customers’ intention in using medical takaful card among public sector. journal of islamic marketing, 9(3), 527-543. hamid, m. a., & rahman, n. m. n. a. (2011). commitment and performance: a case of takaful (islamic insurance) representatives in malaysia. australian journal of basic and applied sciences, 5(10), 777-785. hamid, m. a., & othman, m. s. (2009). a study on the level of knowledge and understanding among muslims towards the concepts, arabic and 114 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 shariah terms in islamic insurance (takaful). european journal of social sciences, 10(3), 468-478. hanudin amin.(2013). factors influencing malaysian bank customers to choose islamic credit coverage: empirical evidence from the tra model.journal of islamic marketing, 4(3), 245-263. huber m, knottnerus ja, green l, horst h, jadad ar, kromhout d et al. (2011). how should we define health?. bmj 2011;343: d4163. jayawardhena, c., kuckertz, a., karjaluoto, h., & kautonen, t. (2009). antecedents to permission based mobile marketing: an initial examination. european journal of marketing, 43(3/4), 473-499. jeddi, n., & zaiem, i. (2010). the impact of label perception on the consumer’s purchase intention: an application on food products. ibima business review, 1-14. jun, b., pil, k., jae, w., & byung, g. (2008). adoption of on-line banking service considering the moderate effects of on-line banking service type. proceedings of the 2008 second international conference on future generation communication and networking symposia, 77-80. kjeldskov, j., & skov, m. (2004). supporting work activities in healthcare by mobile electronic patient records. in proceedings of the 6th asia pacific conference on human computer interaction, apchi 2004, rotorva, new zealand, 2004. krejcie, r. v., & morgan, d. w. (1970). determining sample size for research activities. educational and psychological measurement, 30, 607-610. kumanyika, s. k., van horn, l., bowen, d., perri, m. g., rolls, b. j., czajkowski, s. m., et al. (2000). maintenance of dietary behavior change. health psychology, 19(suppl. 1), 42–56. kunreuther, h. & pauly, m. (2005). insurance decision-making and market behavior. foundations and trends in microeconomics, 1(2), 63-127. li, d., moshirian, f., nguyen, p., & wee, t. (2007). the demand for life insurance in oecd countries. the journal of risk and insurance, 74(3), 637-652. lin & chen. (2006). the influence of the country-of-origin image, product knowledge and product involvement on consumer purchase decisions: an empirical study of insurance and catering services in taiwan. journal of consumer marketing, 23(5), 248-265. lo, a. s., cheung, c., & law, r. (2011). hong kong residents’ adoption of risk reduction strategies in leisure travel. journal of travel and tourism marketing, 28, 240-260. md-taib, f. m., ramayah, t., & abdul-razak, d. (2008), factor influencing intention to use diminishing partnership home financing, international journal of islamic and middle eastern finance and management, 1(3), 235-248. maizaitulaidawati md husin., & asmak ab rahman (2013). what drives consumers to participate into family takaful schemes? a literature review. journal of islamic marketing, 4(3), 264-280. maizaitulaidawati md husin., noraini ismail., & asmak ab rahman (2016). the roles of mass media, word of mouth and subjective norm in family the roles of health awareness and knowledge in medical takaful purchase intention: 95-116 115 takaful purchase intention. journal of islamic marketing, 7(1), 5973, https://doi.org/10.1108/jima03-2015-0020 marhanum, c., nurdianawati, i. a., & s. salwani, r. (2013). measuring takaful agents’ understanding towards the objectives and concepts of takaful. journal of islamic finance, 2(1), 020-030. iium institute of islamic banking and finance. mohamed, g. k. (2008). financing health care in sudan: is it a time for the abolishing of user charges?. sudanese journal of public health, 2(1), 38-47. montano, d. e., thompson, b., victoria, m. t., & janice, m. (1997). understanding mammography intention and utilization among women in an inner city public hospital clinic. university of washington, seattle, washington. preventive medicine: 26, 817–824 (1997). article no.: pm970215 muhammad ali., syed ali raza., & chin-hong puah (2017). factors affecting to select islamic credit coverages in pakistan: the tra model. journal of islamic marketing, 8(3), 330-344. munawar iqbal. (2013). islamic finance: an attractive new way of financial intermediation. the international journal of banking and finance, 10(2), 1-24. munnukka, j. (2007). characteristics of early adopters in mobile communications markets. marketing intelligence & planning, 25(7), 719-731. naail, m. k., & norsham, m. n. (2014). factors influencing the choice of takaful over conventional insurance: the case of malaysia. journal of islamic finance, 3(2), 001 – 014. iium institute of islamic banking and finance. neaton, j. d. et al. (1993). treatment of mild hypertension study. final results. jama 1993; 270, 713–724. oecd. (2009). economic survey of japan 2009. oecd publishing. oluyemi, s. a. (1995). recent development in the nigerian banking system and insured banks’ assets portfolio behaviour: an empirical study, nigerian deposit insurance corporation (ndic). quarterly review, 5(4), 32-44. piviki, j. r. (2002). practical strategies for facilitating meaningful citizen involvement in health planning. ottawa, canada: university of ottawa. porter, c. e., & donthu, n. (2006). using the technology acceptance model to explain how attitudes determine internet usage: the role of perceived access barriers and demographics. journal of business research, 59(9), pp. 999-1007. ram al jaffri saad., & roszaini haniffa. (2014). determinants of zakah (islamic tax) compliance behavior. journal of islamic accounting and business research, 5(2), 182-193. rao, a. r., & monroe k. b. (1988). the moderating effect of prior knowledge on cue utilization in product evaluations. journal of consumer research, 15 (september), 253-264. rahman, a., ali, n., che-seman, a., & wan-ahmad, w. m. (2008). comparative study between auto takaful and auto insurance purchasing behaviour among malaysian undergraduates. jurnal syariah, 16(1), 75-88. 116 the international journal of banking and finance, vol. 14, 2018-2019: 95-116 riquelme, h. e., & rios, r. e. (2010). the moderating effect of gender in the adoption of mobile banking. the international journal of bank marketing, 28(5), 328-341. saif, m. (2012). islamic insurance: a competitive business in yemen. yemen times news letter 1545 (9th february). retrieved on december 12, 2013, available online at: http://www.yementimes.com/en/1545/business/360/ islamic insurance-a-competitive-business-in yemen.htm. sarwar, a., & qureshi, h. a. (2013). awareness and willingness to buy private health insurance and a look into its future prospects in pakistan. european journal of business and social sciences (ejbss), 2(1), 69-81. sekaran, u., bougie, r. (2010). research method for business: a skill building approach (4th ed.) new york, john and sons publication. sekaran, u., bougie, r. (2012). research method for business: a skill building approach (5th ed). new york, john and sons publication. sharif, m. b., hafizah, n. h. m., & jamil, m. a. a. (2011). consumer perceptions on takaful business in brunei darussalam. international conference on management (icm 2011) proceeding, sultan sharif ali islamic university and legal advisor at abrahams davidson & co., brunei. sherrick, b. j., barry, p. j., ellinger, p. n., & schnitkey, g. d. (2004). factors influencing farmers’ crop insurance decisions. american journal of agricultural economics, vol. 86, no. 1, pp. 103-114. singh, s. (2014). customer perception of mobile banking: an empirical study in national capital region delhi. journal of internet banking and commerce, 19(3), 2-23. sulaiman lujja., mustafa omar mohammad., & rusni hassan. (2016). modelling public behavioral intention to adopt islamic banking in uganda: the theory of reasoned action. international journal of islamic and middle eastern finance and management, 9(4), 583-600. the economist. (2014). how sustainable is malaysian healthcare? retrieved from: http://www.eiu.com/. tsai, c., zhu, d., & jang, y. (2013). a study on the consumer adoption behaviors of internet bank. journal of networks, 8(12), 2956-2963. webb, t. l., & sheeran, p. (2006). does changing behavioral intentions engender behavioral change? a meta-analysis of the experimental evidence. psychological bulletin, 132, 249–268. world health organization (who). (2006). world health report-working together for health. geneva: who. yin, s. w. l., du, l., & chen, m. (2010). consumers’ purchase intention of organic food in china. journal of the science of food and agriculture, 90(8), 1361-1367. zhang, y. (2005). age, gender, and internet attitudes among employees in the business world. computers in human behavior, 21(1), 1-10. zuriah, a. r., & rosylin, m. y. (2008). family takaful: its role in social economic development and as a savings and investment instrument in malaysia. shari’a journal, 16(1), 89-105, international islamic university malaysia (iium). malaysia. analysts earnings forecasts distribution 37 the international journal of banking and finance, volume 8 (number 3), 2011: pages 37-53 analysts earnings forecasts distribution henry leung the university of sydney, australia ____________________________________________________________ abstract consensus measures on earnings forecasts such as the ibes mean and median are point estimates of sample distributions of analyst earnings forecasts. often, these consensus measures serve as informational proxies for investors in their decision making process. this study utilises the australian ibes earnings forecast data from 1988 through 2008 to show that analyst earnings forecast distributions are non-normal across the 20-year period. these results suggest the possibility of a more accurate surrogate consensus than the simple ibes mean and median. this, in turn, may have some bearing on those who generally employ analysts’ consensus earnings forecasts for stock valuation and modelling purposes. keywords: analysts forecasts, earnings, forecast distribution, ibes, non-normality of forecasts, mean, median jel classification: g43 _____________________________________________ 1. introduction the literature on security analyst earnings forecasts often utilises consensus measures (mean or median) published by the institutional brokers estimate system (ibes). often, these consensus measures serve as informational proxies (brown and kim, 1991) used by investment advisory services in activities such as stock valuation and in gauging stock returns predictability. thus, earnings forecast accuracy (low earnings forecast error) is desired in the making of investment decisions. the ibes consensus measures are point estimates of analyst earnings forecast distributions. the usage of these consensus earnings forecasts within the investment community implies that analyst earnings estimates distributions are presumed to be intrinsically normal because statistical theory has shown that the unweighted mean or median are the best estimators of a normal distribution. however, if analyst earnings estimates distributions are shown to be non-normal, then a more accurate surrogate consensus may be leung: analysts earnings forecasts distribution 38 needed. if so, this result will have practical implications for those who employ the ibes consensus earnings forecasts such as the different stakeholders of capital markets described in kothari (2001). specifically, a more accurate surrogate consensus may be used to derive more accurate stock valuations and allow more reliable recommendations to be made by those who provide investment advice. some studies, as a by-product of other research, have suggested that analyst earnings forecast distributions may possibly be non-normal (ali, klein and rosenfeld, 1992; collins and hopwood, 1980) but no extensive empirical tests have been carried out on analyst earnings estimates distributions to examine this claim. this paper tests this hypothesis using ibes earnings forecast data and non-parametric tests of best fit to a normal distribution. the results, based on contemporaneous monthly ibes analyst earnings estimates from 1988 through 2008, are largely consistent with the hypothesis that analyst earnings estimates distributions are non-normal. the significance of the deviation of skewness and kurtosis distributional characteristics are also examined. they too corroborate distributional nonnormality. the rest of the paper is organised as follows. section 2 develops the hypotheses in the context of the ibes literature. section 3 discusses the data sample and the selection criteria. section 4 describes the research design and methodology followed by the analyses of results in section 5. finally, section 6 provides a summary of the main findings in relation to the underlying theoretical framework with a short discussion on possible future research directions. 2. hypothesis one germane capital markets research area is the investigation of the relationship between financial information and capital markets. kothari (2001) indicates that this is driven by practical needs within the investment advisory services industry to seek a better understanding of fundamental analysis and valuation techniques for the added value of financial reporting. there is a constant demand for correct investment decisions in the industry and in turn a demand is placed on more accurate analysts’ earnings forecasts and the consensus earnings forecast. in many instances, the ibes consensus is used to represent the underlying per period distribution of analyst earnings estimates. it is assumed with this usage that analyst earnings forecasts are a priori normally distributed because in statistical theory the mean or the median international journal of banking and finance, vol. 8, iss. 3 [2011], art. 2 39 is the best estimate of a normal distribution (kreyszig, 1993, p. 1219). however, it is expected in this study that empirical evidence will show otherwise. one reason for the expected non-normal distributions is that analyst earnings forecasts contain lower or upper bounds. if so, then skewed distributions will result. data that have a lower bound are often skewed right as are security prices; while data that have an upper bound are often skewed left. in the case of earnings forecasts, the lower bound is zero earnings per share whilst the upper bound is constrained by prior growth rate of actual announced earnings stream. 1 hence there exists the possibility of either left or right skewness in any sample data set. skewness can also result from start-up effects. for example, initial analyst earnings forecasts for a firm may contain a large number of upper outliers (failures) causing left skewness in the earnings forecast distribution. as the forecast horizon 2 shortens, earnings forecast accuracy improves and forecasts tend to cluster around its mean which moves towards the actual earnings result. collins and hopwood (1980) point out this improvement in forecast accuracy over 4 quarters prior to the announcement date. thus, the following hypothesis is testable: hypothesis 1. hnull: contemporaneous ibes analyst earnings forecast distributions are normal. halternative: contemporaneous ibes analyst earnings forecast distributions are non-normal. assuming that hypothesis one holds true, the general shape of the distribution will be investigated using higher ordered moment measures such as skewness and excess kurtosis and goodness of fit testing against known distribution types. next, the ibes data are described and the sample data selection methodology is discussed. our aim in this section is to develop a robust methodology to identify measures and derive robust results on this distributional characteristics of earnings forecasts. 1 this is based on the assumption of stable earnings streams of large-cap firms. for small-cap firms, earnings streams will be more volatile and bounds may not exist. 2 see appendix 1 for definition. leung: analysts earnings forecasts distribution 40 3. data and sample selection the analysts’ earnings forecasts data are sourced from the institutional brokers estimate system (ibes). the ibes data set (financial, 2000) contains earnings per share forecasts before extraordinary items, adjusted for stock splits, stock dividends, and other capitalisation changes. analyst earnings estimates made for all australian firms spanning from 1 july 1988 through 30 june 2008 are selected. actual earnings figures are also sourced from ibes in order to minimise idiosyncratic noise in the forecast metrics because ibes applies the same adjustment process to both its earnings forecasts and actual earnings figures. the observation sample meets the selection criteria: actual earnings figures are announced annually; analyst earnings forecasts of annual actual earnings are available for all firms; analyst earnings forecasts are shifted to the next closest ibes consensus publication date occurring on the third thursday of every month; 3 and a forecast horizon of 11 months prior to the actual earnings announcement. 4 these criteria yield 574,438 monthly earnings estimates and 264,926 monthly ibes consensus figures. finally, ibes uses its analyst earnings forecasts to compute its monthly consensus. ibes’s computational process is governed by internal rules determining whether analyst’s estimates made in certain statistical periods are carried forward into subsequent periods. first, it is predicted that an analyst estimate carry forward process is employed by ibes in its consensus computation because, in cases where an earnings estimate is published by an analyst and this estimate in the subsequent statistical period, it is neither replaced by peer analysts of the same brokerage firm nor the analyst himself/herself, then the implicit translational effect of an analyst estimate’s information content into subsequent months needs to be captured. additionally, events necessitating the exclusion 5 or stoppage 6 of estimates also 3 this criterion implies the assumption of equal informational content within all analyst estimates announced between two monthly ibes publication dates. 4 this ensures the study focuses on analyst earnings forecast distributional characteristics results which are current in relation to the subsequent actual earnings. the rationale underlying the omission of earnings estimates 12 months prior to actual earnings figures is due to the possible distortional effect the information content of a firm’s previous actual earnings figures may have on analyst earnings estimates made in that same month but which contribute to the actual earnings of the subsequent period. (brown, lee, taylor and walter, 2004) 5 excluded estimates denote those estimates made in a statistical period but not included in the ibes consensus calculation. these forecasts have deviated from accepted standard as defined by the majority of analysts covering a particular issue. ibes contacts the analysts making these forecasts for confirmation regarding the exclusion and queries the methodology behind the estimates. 6 stopped estimates indicate when a brokerage firm’s analysts removed his/ her earnings forecast from the ibes database due to conflict of interest between the brokerage firm and the company for which earnings are estimated. examples of causes of earnings publication stoppage include a brokerage firm underwriting a firm’s equity issues or when the investment banking international journal of banking and finance, vol. 8, iss. 3 [2011], art. 2 41 need to be incorporated. for example, earnings estimates made in a statistical period needs to be carried forward into subsequent periods until a new estimate is made by the same analyst or when an estimate publication stoppage is being placed on the brokerage firm for which the analyst is employed. these stoppages may be due to a variety of reasons including underwriting agreements between the brokerage firm and the firm for which earnings estimates are made. an understanding of the different ibes processing rules ensures the distributions of analysts’ earnings forecasts undergoing tests of normality are consistent with the ones ibes use to generate its consensus. employment of the ibes processing rules to the filtered subset yields 303,016 monthly earnings estimates and 42,730 monthly ibes consensus observations for the time period 1 july, 1988 through 30 june, 2008. table 1 shows the change in sample size as a result of the filtering process. table 1. elimination of firms and estimates due to data cleansing and employment of ibes internal rules criteria number of ibes analysts' annual earnings forecasts number of ibes consensus (mean and median counted as 1 consensus) (1) original non-us source data set 9,144,148 7,569,139 (2) post sample data selection and cleansing process followed by extraction of australian data 574,438 264,926 (4) carry forward estimates and application of ibes internal rules. 1,212,064 251,354 (5) select data from years 1988 through 2008 303,016 42,730 the next section discusses the research design of this study. 4. research design 4.1 forecast horizon and observation interval a monthly contemporaneous analysis of the 11 months prior actual earnings forecasts is applied across a 20-year interval from 1 july 1988 through 30 june 2008. contemporaneous macroeconomic factors which may impact earnings forecasts are controlled in the sample that arm of a brokerage firm is involved with the mergers or acquisition activity of the client firm. leung: analysts earnings forecasts distribution 42 spans over four economic cycles. 7 possible macroeconomic factors that are considered are fluctuations in interest rates, exchange rates, business spending, consumer sentiment, and fiscal policy changes. 4.2 forecast error metric – analyst earnings relative forecast bias prior to the discussion of the statistical techniques pertaining to the testing of the hypothesis, it is important to refine measures of forecast bias that are used in the past forecast literature to obtain an appropriate benchmark to facilitate the study of analyst earnings forecast distribution characteristics. the main motivation of this study lies in our interest to evaluate the significance of non-normality in cross-sectional analyst earnings forecast distributions. in analyst earnings forecasts research, consensus point estimates of analyst earnings forecasts sample distributions are used in the calculation of two types of forecast error measures. these are (i) firm-based, which measures the deviation of the consensus estimate earnings away from the actual earnings, after dividing it by a deflator to achieve scaling and (ii) analyst-based, which measures the difference between an analyst’s forecast error and the per period consensus of all analysts’ forecast errors, again scaled by a deflator. this is termed the analyst’s relative forecast error measure (following brown, 1999). the latter error measure is altered in this study. an analyst earnings relative forecast (aerf) bias measure may is defined as: _ _ _ _ _ analyst earnings forecast analyst earnings consensus aerf bias deflator   (1) the use of deflators in earnings forecast research has been scarce. when attempted, they are often not sufficiently detailed to yield a generalisation of result nor robustness of result. thus, for completeness and for robustness of results, this study employs four deflators in relation (1). they are: (i) a firm’s share price at 11 months prior actual earnings are reported pricet=-11; (ii) a firm’s share price at each statistical period pricet; (iii) the period’s consensus analysts’ earnings forecast cons; and (iv) a modified version of the mean absolute percentage error (mape) as suggested by makridakis (1993). the fourth measure, which is denoted by mpe hereafter, operates best in a majority of situations and at the same time satisfies theoretical and practical concerns. it is able to 7 artis et al. (1997) found average economic cycles to be 51 months in duration across a study of 12 countries (g7 and 5 european countries) from years 1961 through 1993. thus, the data sample in this study covers 20 years x 12 months/ 51 months=4.7 economic cycles. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 2 43 mitigate scaling problems, is robust from one data set to another, and is protected from the influence of outliers. the mpe measure improves upon the mape measure by avoiding scaling problems by dividing the error (actual-forecast) by the average of the actual and forecast, that is, (a+f)/2. hence the mathematical form of the error measure mpe becomes: ( ) / 2 a f mpe a f    (2) where a is the actual reported earnings and f is the analyst earnings forecast. for the purpose of this study, the per period analyst earnings forecast consensus item (cons) replaces the actual reported earnings (a) to form the analyst earnings relative forecast (aerf) bias as included in equation (1). 4.3 goodness of fit tests hypothesis 1 is tested with three goodness of fit tests using aerf_bias (relation 1) together with the deflators mpe, pricet=-11, pricet and cons. the goodness-of-fit tests are the kolmogorov-smirnov (k-s) test, the andersondarling (a-d) test and the cramér-von mises (cvm) test (siegel and castellan, 1988). all were developed to examine whether a given distribution fits a specified distribution (a normal distribution in this study). the a-d test has been modified to overcome limitations 8 of the ks test but it is only available for a limited number of distribution types. 5. results 5.1 effect of ibes internal rules on ibes consensus mean and analyst estimates' mean table 2 provides descriptive statistics on the effects of the ibes internal rules on the percentage matches between analyst estimates’ mean and the ibes consensus mean from the 1 july 1988 through 30 june 2008. results suggest that the application of the ibes internal rules is necessary for specific analyst earnings forecasts to be identified and is, therefore, included in the sample distribution for goodness of fit tests to be carried out. 8 the skewness in sensitivity towards the centre of the distribution and the need for the theoretical distribution to be fully specified. leung: analysts earnings forecasts distribution 44 panel a of table 2 reports the mean, median and standard deviation of the cross sectional per year end period percentage matches of the analyst estimates’ mean and the ibes consensus mean both prior to and posterior to the application of ibes internal rules. within each year end period, only those firms with analyst estimates and ibes consensus persisting from 11 statistical periods through reporting date are selected to control for sample size effects. for example, for the year end period 1994-1995, there were 2,160 (2,987) analysts’ forecasts for firms which published estimates for at least 1 statistical period prior each firms’ reporting date for prior (posterior) to the employment of the ibes internal rules. this was reduced significantly to 957 (2,508) when firms with at least 11 statistical periods prior reporting date were imposed for selection. in the case of the posterior application of the ibes internal rules, the highest number of matches occurred in the financial fiscal year 1994-1995 (94.06 percent) and the lowest number of matches occurred in the financial fiscal year 1990-1991 (18.11 percent). another result is the low matching percentage (<50 percent) in the earlier year end periods (19881993) and the subsequent increase in matching percentage to a high of greater than 50 percent in the later year end periods (1993-2008). a possible explanation for both the high range value (94.51 percent-18.11 percent=76.4 percent) and increase in percentage of matches from below to over 50 percent is due to the changes to internal rules used by ibes, including the amendments, removal or addition of internal rules which existed in the earlier periods but not in the later periods or vice versa. further, these changes in internal rules may be attributable to regulatory changes or the way multiple earnings forecasts are chosen and provided by brokerage firms to ibes. 9 for example, regulatory changes include changes to the number of stoppage months placed upon a brokerage firm’s estimates from when a stop signal is initiated by ibes. the greater number of statistical periods of the posterior as opposed to the prior application of ibes internal rules across all year end periods is attributable to the carry forward of analyst estimates into subsequent periods which initially did not have any existing estimate(s) made by the same analyst. this implies that some firms in the data set prior to being treated with the ibes internal rules will have some statistical periods without any estimates. since the sample selection chooses only those firms with estimates from 11 9 using the sample data selection process detailed in section 3, it is found that there are multiple earnings forecasts supplied by different analysts of the same brokerage firm in the same statistical period, only the most recent estimate is used by ibes in its consensus calculation. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 2 45 statistical periods through reporting date, firm selection is carried out after the ibes internal rules are applied. table 2. effect of ibes internal rules on ibes consensus mean and analyst estimates' mean panel a: comparison of ibes consensus mean and analyst estimates' mean before and after the application of ibes internal rules year end period number of statistical periods match_prior_ibes a (%) number of statistical periods match_post_ibes b (%) 30/06/1988 1/07/1989 572 2.27 1,507 19.38 30/06/1989 1/07/1990 627 4.31 1,507 22.16 30/06/1990 1/07/1991 616 4.22 1,375 18.11 30/06/1991 1/07/1992 594 4.88 1,540 35.13 30/06/1992 1/07/1993 759 4.74 1,529 46.96 30/06/1993 1/07/1994 902 6.32 1,837 73.43 30/06/1994 1/07/1995 957 7.31 2,508 94.06 30/06/1995 1/07/1996 858 6.76 2,794 91.30 30/06/1996 1/07/1997 979 5.11 2,717 84.62 30/06/1997 1/07/1998 1,078 7.79 2,860 87.20 30/06/1998 1/07/1999 1,221 6.55 3,267 71.93 30/06/1999 1/07/2000 1,012 8.10 3,036 78.69 30/06/2000 1/07/2001 990 7.17 3,707 73.37 30/06/2001 1/07/2002 792 9.60 3,839 85.60 30/06/2002 1/07/2003 864 8.94 3,294 83.23 30/06/2003 1/07/2004 973 6.42 3,185 92.42 30/06/2004 1/07/2005 987 7.82 3,832 93.11 30/06/2005 1/07/2006 1023 8.92 2,484 95.10 30/06/2006 1/07/2007 1102 7.24 3,172 94.51 30/06/2007 1/07/2008 1059 6.63 3,491 91.72 30/06/1988 1/07/2008 17,965 6.82 53,481 78.92 mean 898 6.56 2674 71.60 median 965 6.70 2827 83.93 standard deviation 185 1.84 847 27.18 panel b: pair-wise tests for differences between before and after the application of ibes internal rules on matching of ibes consensus and individual estimates pair-wise tests statistic p-value student's t (t-statistic) 9.23 (<.0001) sign (m-statistic) 8.38 (0.0001) wilcoxon signed rank (s-statistic) 67.93 (0.0001) _____________________________ a the match percentage of ibes consensus and individual estimates' mean prior application of ibes internal rules of each year end period is defined as: match_prior_ibes= number of firm specific ibes consensus mean equal to mean of corresponding individual analyst estimates in the same statistical period within the year end period prior application of  ibes internal rules number of statistical periods within the year end period b the match percentage of ibes consensus and individual estimates' mean posterior application of ibes internal rules of each year end period is defined as match_post_ibes= number of firm specific ibes consensus mean equal to mean of corresponding individual analyst estimates in the same statistical period within the year end period posterior application o f ibes internal rules) (number of statistical periods within the year end period) leung: analysts earnings forecasts distribution 46 panel b of table 2 reports the significance of the difference between two conditions, before and after the application of ibes internal rules. it is expected that the latter condition is significantly and positively different from the prior condition due to the expected improvement due to the application of the ibes internal rules will have on the comparison of the ibes consensus to the individual estimates’ mean. the significant and positive test statistics (a p-value less < 0.05) from all three pair-wise tests, namely the student’s t test (p<0.0001), the sign test (p=0.0001) and the wilcoxon signed rank test (p=0.0001) confirm this conjecture. 5.2 descriptive statistics to maintain consistency and continuity in the analyses, the set of analyst detailed estimates data which have undergone the treatment of ibes internal rules in the previous section is now used to generate the set of contemporaneous distributional properties from 11 months prior through actual earnings reporting on a per month (statistical period) basis. table 3 provides the descriptive statistics of the analyst earnings forecast bias (aerf_bias) distributions using the four different deflators pricet=-11, pricet, cons, and mpe, for all australian ibes firms for the 20 test years. for example, the median of the fifth month prior reporting for aerf_bias_cons is calculated as follows. the median of each firm-specific distribution of analyst earnings forecasts at five months prior to the firm’s actual reporting date is computed for all firms. the aerf_bias_cons of each median is then determined, followed by averaging all median-aerf_bias_cons with the resulting mean value being the required contemporaneous median relative forecast value. by taking the averages of forecasts summarised per period, bias values are generated across all firms. for instance, the bias of the distribution of analysts’ earnings forecasts was first computed for each asx firm at 11 months prior to actual earnings announcement. then the mean of the bias across all these asx firms was generated to produce a summary bias value at 11 months prior to actual earnings reporting. distortions to aerf_bias caused by zero deflator values 10 need to be addressed. forecasts with zero deflators are removed to prevent the introduction of infinite values. for example, the deflator mpe when used with analysts’ earnings relative forecast bias, is by 10 results for non-removal of zero deflators were found to be consistent with the case when zero denominators were removed. to avoid division by zero errors for the non-removal case, the relevant analyst estimates were assigned large relative forecast bias values of 100,000. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 2 47 definition the analyst estimate plus period’s consensus. if the forecast earnings value is equal to the negative of the period consensus, then the resulting aerf_bias_mpe value will be infinite. table 3: descriptive statistics on contemporaneous analysts' relative forecast bias (aerf_bias) distributions using deflators pricet=-11, pricet, cons and mpe year end period 1/7/1988 through 30/6/2008 months prior to reporting of actual (forecast indicator = 1) a aerf_bias deflated by firm’s share price at statistical period t=-11 (aerf_bias_pricet=-11) aerf_bias deflated by firm’s share price at statistical period t (aerf_bias_pricet) median (%) 1st moment mean 2nd moment standard deviation 3rd moment skewness b 4th moment excess kurtosis c median (%) 1st moment mean 2nd moment standard deviation 3rd moment skewness 4th moment excess kurtosis 11 -0.20 0.00 0.021 0.27 0.79 -0.20 0.00 0.023 0.33 0.72 10 -0.23 0.00 0.020 0.31 0.79 -0.22 0.00 0.024 0.26 0.83 9 -0.28 0.00 0.029 0.30 0.89 -0.24 0.00 0.025 0.31 0.81 8 -0.26 0.00 0.023 0.31 0.85 -0.25 0.00 0.030 0.32 0.88 7 -0.24 0.00 0.024 0.29 0.93 -0.24 0.00 0.034 0.34 0.90 6 -0.16 0.00 0.027 0.32 0.99 -0.18 0.00 0.025 0.30 1.02 5 -0.23 0.00 0.022 0.31 1.08 -0.22 0.00 0.030 0.28 1.10 4 -0.20 0.00 0.026 0.34 1.20 -0.21 0.00 0.035 0.35 1.14 3 -0.17 0.00 0.020 0.34 1.24 -0.23 0.00 0.036 0.34 1.19 2 -0.21 0.00 0.024 0.33 1.22 -0.24 0.00 0.029 0.35 1.23 1 -0.22 0.00 0.021 0.28 0.95 -0.21 0.00 0.032 0.35 1.01 mean -0.22 0.00 0.023 0.31 0.99 -0.22 0.00 0.029 0.32 0.99 months prior to reporting of actual (forecast indicator = 1) aerf_bias deflated by period consensus (aerf_bias_cons) aerf_bias deflated by average of period consensus and forecast (aerf_bias_mpe) median (%) 1st moment mean 2nd moment standard deviation 3rd moment skewness 4th moment excess kurtosis median (%) 1st moment mean 2nd moment standard deviation 3rd moment skewness 4th moment excess kurtosis 11 -2.19 0.00 0.365 0.34 0.69 -0.19 0.00 0.439 0.12 0.64 10 -1.24 0.00 0.261 0.29 0.78 -1.00 0.00 0.378 0.04 0.68 9 -1.55 0.00 0.252 0.31 0.78 -0.79 0.00 0.385 0.10 0.71 8 -3.11 0.00 0.352 0.27 0.81 -0.27 0.00 0.526 0.12 0.75 7 -1.08 0.00 0.276 0.30 0.86 -0.22 0.00 0.404 0.13 0.85 6 -1.57 0.00 0.319 0.29 0.91 4.70 0.00 0.722 0.10 0.93 5 1.65 0.00 0.493 0.33 1.01 -0.09 0.00 0.516 0.05 0.95 4 3.13 0.00 0.486 0.32 1.09 -0.88 0.00 1.070 0.12 1.07 3 -0.22 0.00 0.465 0.34 1.12 -0.23 0.00 0.500 0.07 1.13 2 1.41 0.00 0.474 0.35 1.28 0.23 0.00 0.462 0.15 1.16 1 -1.76 0.00 0.419 0.35 1.20 -0.01 0.00 0.323 0.10 1.13 mean -0.59 0.00 0.378 0.32 0.96 0.11 0.00 0.520 0.10 0.91 ___________________________________ a all firms' cross-sectional analyst forecast bias distribution parameters such as median and the first four moments (per statistical period) were grouped by the number of statistical periods up to 11 months prior the corresponding firm's reporting date. in this way the timing differences between the earnings reporting dates of different firms and their corresponding forecast bias distributions are controlled. this was employed to all firms with year end dates 1 st july 1988 through 30 th june 2008. b skewness is equal to 0 for a standard normal distribution and values may range between negative infinity and positive infinity. c excess kurtosis is used rather than kurtosis. kurtosis omits the subtraction of 3 so that for a standard normal distribution, the excess kurtosis is equal to 0 and kurtosis is equal to -3. the excess kurtosis must lie between -2 and infinity. it is termed leptokurtic when it is greater than 0 and platykurtic when it is less than 0. leung: analysts earnings forecasts distribution 48 the results suggest a positive bias (median less than mean) in general, consistent with prior reported empirical evidence of optimistic analyst forecast bias (e.g., brown, 1998 and mande, wohar and ortman, 2003). the median, on average, is consistently negative across all error deflator types except for aerf_bias_mpe. the mean is consistent with the expectation that when distributions of analysts’ earnings estimates distributions are normalised to a mean of zero, the average across all firms will also be zero. the analysts optimism bias is corroborated by the other two distributional characteristics, i.e, skewness and kurtosis. these show that, on average, aerf_bias distributions are positively skewed and is leptokurtic in nature 11 months or less from the day of actual earnings release, suggesting that analyst earnings estimates distributions are fairly asymmetric. this result is exhibited by the positive mean skewness and excess kurtosis across the 4 deflators. 5.3 non-normality of contemporaneous distributions table 4 reports the results for the goodness-of-fit tests and skewness/kurtosis location tests of cross-sectional aerf_bias distributions in each of the 11 months prior reporting for firm years. panel a reports the statistics and p-values of the kolmogorov-smirnov test, the anderson-darling test and the cramér-von mises test while panel b reports the statistics and p-values of the student's t test, the sign test and the wilcoxon signed rank test. in panel a, aerf_bias distributions reveal a rejection of fit to a normal distribution, with the d-statistic (<0.01), a-sq-statistic (<0.005) and the w-sq-statistic (<0.005) significantly positive for all relative forecast metrics in all statistical periods prior to the releases of actual earnings. panel b reports the significance of analyst forecast bias distributions’ skewness and kurtosis from zero, the value expected from a normal distribution. for succinctness, only nonsignificant results are shown. they are relatively sparse given that there are 33 non-significant results of the 264 tests run (3 statistical tests for skewness and excess kurtosis each spread over 11 months and 4 deflators). moreover, the 3 non-significant results for bias metrics aerf_bias_pricet=-11, aerf_bias_pricet and aerf_bias_cons may be attributable to the decreasing power-efficiency of the sign test for large samples, with the wilcoxon signed rank test and the student’s t test been more power-efficient in large sample cases exhibiting opposing results. likewise, the overwhelmingly non-significant results (30/66*100=45%) for aerf_bias_mpe may have been caused by a metric dependent factor because the other 3 bias metrics have overwhelmingly reported polarised results. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 2 49 table 4: contemporaneous relative forecast distribution normality panel a: analyst relative forecast bias distribution goodness-of-fit to normal distribution months prior to reporting of actual (forecast indicator = 1) b analyst earnings relative forecast deflated by firm’s share price at statistical period t=-11 (aerf_pricet=-11) sample size per statistical period = 3232 analyst earnings relative forecast deflated by firm’s share price at statistical period t (aerf_pricet) sample size per statistical period = 3555 kolmogorovsmirnov d-statistic andersondarling a-sq-statistic cramér-von mises w-sq-statistic kolmogorovsmirnov d-statistic andersondarling a-sq-statistic cramér-von mises w-sq-statistic 11 0.57 1210 261 0.60 1203 261 (<0.0100) a (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 10 0.57 1200 275 0.51 1245 257 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 9 0.51 1190 272 0.57 1239 265 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 8 0.56 1230 266 0.55 1245 274 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 7 0.54 1272 270 0.60 1224 266 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 6 0.53 1202 259 0.52 1269 260 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 5 0.51 1254 264 0.60 1215 266 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 4 0.58 1222 260 0.52 1234 268 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 3 0.55 1270 258 0.59 1270 259 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 2 0.58 1206 269 0.58 1235 269 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 1 0.57 1263 257 0.51 1244 258 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) months prior to reporting of actual (forecast indicator = 1) b analyst earnings relative forecast deflated by period consensus (aerf_cons) sample size per statistical period = 3550 analyst earnings relative forecast deflated by average of period consensus and forecast (aerf_mpe) sample size per statistical period = 3550 kolmogorovsmirnov d-statistic andersondarling a-sq-statistic cramér-von mises w-sq-statistic kolmogorovsmirnov d-statistic andersondarling a-sq-statistic cramér-von mises w-sq-statistic 11 0.52 1189 258 0.52 1025 220 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 10 0.52 1192 264 0.57 1026 227 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 9 0.54 1197 259 0.48 1121 229 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 8 0.56 1246 273 0.51 1111 226 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 7 0.58 1201 256 0.50 1085 228 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 6 0.52 1215 272 0.54 1071 226 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 5 0.58 1227 261 0.56 1117 234 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 4 0.59 1185 266 0.48 1045 230 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 3 0.58 1232 259 0.56 1075 221 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 2 0.57 1233 270 0.48 1028 226 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) 1 0.53 1236 271 0.49 1013 224 (<0.0100) (<0.0050) (<0.0050) (<0.0100) (<0.0050) (<0.0050) leung: analysts earnings forecasts distribution 50 panel b: significance of skewness and kurtosis from zero c months prior to reporting of actual (forecast indicator = 1) analyst earnings relative forecast deflated by firm’s share price at statistical period t=-11 (aerf_pricet=-11) sample size per statistical period = 3232 analyst earnings relative forecast deflated by firm’s share price at statistical period t (aerf_pricet) sample size per statistical period = 3555 student's t t-statistic sign m-statistic wilcoxon signed rank s-statistic student's t t-statistic sign m-statistic wilcoxon signed rank s-statistic skewness excess kurtosis skewness excess kurtosis skewness excess kurtosis skewness excess kurtosis skewness excess kurtosis skewnes s excess kurtosis 11 39 34 (0.1732) (0.1243) 10 9 8 … 1 months prior to reporting of actual (forecast indicator = 1) analyst earnings relative forecast deflated by period consensus (aerf_cons) sample size per statistical period = 3550 analyst earnings relative forecast deflated by average of period consensus and forecast (aerf_mpe) sample size per statistical period = 3550 student's t t-statistic sign m-statistic wilcoxon signed rank s-statistic student's t t-statistic sign m-statistic wilcoxon signed rank s-statistic skewness excess kurtosis skewness excess kurtosis skewness excess kurtosis skewness excess kurtosis skewness excess kurtosis skewness excess kurtosis 11 37 -0.60 -19.48 -15171 (0.1872) (0.6576) (0.4603) (0.7462) 10 -1.63 -8.73 -27837 (0.1739) (0.7809) (0.5597) 9 -0.77 20.91 -5827 (0.4787) (0.4443) (0.9806) 8 -0.29 24.73 11829 (0.9115) (0.3575) (0.8015) 7 1.14 32.42 60383 (0.4326) (0.2701) (0.2757) 6 -0.71 -1.07 -7282 (0.637) (1.0277) (0.9489) 5 0.32 14.77 29254 (0.8348) (0.7140) (0.3468) 4 0.53 5.97 27892 (0.7356) (1.0247) (0.5360) 3 0.43 -13.21 23755 (0.8343) (0.7381) (0.6057) 2 0.68 15.26 36728 (0.6450) (0.4919) (0.3409) 1 ____________________________________ a parenthetical number are p-values. b all firms' cross-sectional analyst relative forecast bias distribution parameters such as median and the first four moments (per statistical period) were grouped by the number of statistical periods up to 11 months prior the corresponding firm's reporting date. in this way the timing differences between the earnings reporting dates of different firms and their corresponding relative forecast bias distributions are controlled. this was employed to all firms with year end dates 1 st july 1988 through 30 th june 2008. c only non-significant test statistics and probabilities are shown due to the scarcity of such evidence, indicating that it may have arisen due to reasons not related to the underlying characteristics of the distribution but rather on the method of analysis such as the sensitivity of the particular significance test used. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 2 51 as such, the earnings forecast dependency of the mpe deflator may have introduced skewness into the underlying bias distribution. the other relative forecast bias metrics aerf_bias_pricet=-11, aerf_bias_pricet and aerf_bias_cons depends on the constant deflators share price and actual reported earnings. finally, those results not shown have significantly positive statistics (p-values<0.0001) suggesting that the skewness and excess kurtosis values are not of a normal distribution. this evidence is consistent with those previously presented. 6. conclusions and future research prior research suggests that many facets of the analyst earnings forecast literature base their findings on the assumption that analyst earnings forecast distributions are normal. the use of the ibes consensus as a proxy for expected earnings figures in stock valuation or in the provision of investment recommendations suffers from this assumption. since the ibes consensus data consist of the mean and median as the point estimates representing the underlying analyst earnings estimates and that the mean and median are the best estimates of a normal distribution, then the distribution type is implied to be normal. rather, this study shows that contemporaneous ibes analyst earnings forecast distributions are significantly non-normal, thus providing evidence that the assumption is not valid. once timing and informational inconsistencies are alleviated through data selection processes and implementation of the ibes internal rules, goodness of fit tests are carried out to determine whether analyst earnings estimates distributions are non-normal. the results obtained in this paper are largely consistent with the hypothesis of nonnormality. significant non-normality is shown to be uniformally present across all periods of contemporaneous monthly analyst relative forecast distributions over 20 years (1988-2008) for up to 11 months prior to the actual earnings reporting date on a per firm in each reporting year end basis for all four metric deflators: (i) by firm’s share price at each period, (ii) by firm’s share price at the beginning of the 11 month forecast horizon, (iii) by period consensus, and (iv) by the average of analyst forecast and period consensus. subsequent analyses of skewness and kurtosis distributional characteristics for the same data set also reveal results consistent with nonnormality. leung: analysts earnings forecasts distribution 52 this strong evidence pointing towards analyst earnings distributions to be non-normal suggests that a need for possible development of an improved surrogate consensus based alternate distribution types, which may be of interest to those investors who employ the consensus analysts’ earnings forecasts for stock valuation and modelling purposes. author information: dr henry leung is a staff member in the discipline of finance, the university of sydney, nsw 2006, australia; e-mail: henry.leung@sydney.edu.au. appendix 1: glossary of ibes data terms term description analyst code this is a unique identifier for each analyst. broker code this is a unique identifier for each brokerage firm. estimation date this is the date an estimate was reported by an analyst to ibes. excluded date this denotes the period when analyst’s forecasts which have been excluded from ibes consensus calculation. these forecasts have deviated from accepted standard as defined by the majority of analysts covering a particular issue. ibes contacts the analysts making these forecasts for confirmation or to query the methodology behind the estimates. fiscal period this is the frequency by which a company performance measure is reported. forecast horizon this is the period of time from the date the earnings forecast is made to the next earnings announcement date. ibes ticker this permanently and uniquely identifies an estimate made at a certain point in time. reporting date this is the date actual earnings figures are announced. companies may not report earnings to the marketplace ranging up to 6 months after the fiscal period end date. statistical period this is the date range (approximately a month in duration) between two subsequent ibes summary consensus statistics publication. this occurs after the ibes monthly production run which occurs on the evening of the third thursday of every month. it executes snapshots of analysts’ estimates and calculates the consensus level data. stop date this indicates when a brokerage firm’s analysts removed his/ her earnings forecast from the ibes database due to conflict of interest between the brokerage firm and the company for which earnings are estimated. examples of causes of earnings publication stoppage include a brokerage firm underwriting a firm’s equity issues or when the investment banking arm of a brokerage firm is involved with the mergers or acquisition activity of the client firm. international journal of banking and finance, vol. 8, iss. 3 [2011], art. 2 53 references ali, a., klein, a., and rosenfeld, j., (1992). analysts' use of information about permanent and transitory earnings components in forecasting annual eps. accounting review, 67: 183– 198. artis, m. j., kontolemis, z. g., and osborn, d. r., (1997). business cycles for g7 and european countries. journal of business, 70(2): 249-279. brown, l. d., (1998). managerial behaviour and the bias in analysts' earnings forecasts. school of accountancy, georgia state university. brown, l. d., (1999). predicting individual analyst earnings forecast accuracy. school of accountancy, georgia state university. brown, l. d., and kim, k. j., (1991). timely aggregate analyst forecasts as better proxies for market earnings expectations. journal of accounting research, 29(2): 382-385. brown, p., lee, p. j., taylor, s. l., and walter, t., (2004). are security analysts "walked down" to beatable forecasts?. school of accounting, university of new south wales. collins, w. a., and hopwood, w. s., (1980). a multivariate analysis of annual earnings forecasts generated from quarterly forecasts of financial analysts and univariate time-series models. journal of accounting research, 18(2): 390-406. financial, f. c. t., (2000). i/b/e/s summary history v2.0 (1 st ed.), first call thomson financial. kothari, s. p., (2001). capital markets research in accounting. journal of accounting & economics, 31(1-3): 105-230. kreyszig, e., (1993). advanced engineering mathematics (7 th ed.), john wiley & sons, inc. makridakis, s., (1993). accuracy measures: theoretical and practical concerns. international journal of forecasting, 9(4): 527-529. mande, v., wohar, m. e., and ortman, r. f. (2003). an investigation of asymmetric earnings forecasts of japanese financial analysts. multinational business review, 11(1): 13. siegel, s., and castellan, j. n. j., (1988). nonparametric statistics for the behavioral sciences (2 nd ed.), mcgraw-hill. leung: analysts earnings forecasts distribution 1 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: bakri, m. a., abd jalil, m. i., & hassan, z. (2021). dividend policy in malaysia: a comparison of determinants pre and post malaysian code on corporate governance. international journal of banking and finance, 16(2), 1-22. https://doi.org/10.32890/ ijbf2021.16.2.1 dividend policy in malaysia: a comparison of determinants pre and post malaysian code on corporate governance mohd ashari bakri, mohamad isa abd jalil & zakiah hassan labuan faculty of international finance universiti malaysia sabah corresponding author: mohd.ashari@ums.edu.my received: 2/11/2020 revised: 2/1/2021 accepted: 24/1/2021 published: 15/6/2021 abstract this study was aimed at examining the differences between dividend policy determinants preand post-malaysian code on corporate governance (mccg) 2012. several factors, including profitability, lagged dividend, free cash flow, debt, firm size, investment opportunities and market risk were tested. the study investigated a total number of 631 non-financial firms in malaysia that covered 7830 firm-year observations from 2005 to 2011 (pre-mccg) and from 2013 to 2019 (post-mccg). the study used pooled ordinary least square (ols) and random and fixed effect, with a robust standard error. the results demonstrated that from seven factors tested only four factors were found to be significant in determining dividend policy in pre-mccg, and five factors in post mccg. the pre-mccg test 2 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 revealed that before the revised mccg 2012, the factors determining dividend policy were as follows: profitability, lagged dividend, debt, and firm size. however, there were slight changes in the range of determinants affecting dividend policy, post-mccg 2012. the post mccg test revealed that profitability, lagged of dividend, and firm size consistently determined firm dividend policy; however, debt was no longer a significant determinant of dividend policy post mccg. additionally, investment opportunity and market risk were found to be significant determinants of dividend policy post-mccg in 2012. keywords: dividend policy, corporate governance, emerging markets, malaysia. jel classification: g1, g3, g35. introduction dividend policy has been listed as the top ten unresolved issues in the world of corporate finance (brealey & myers, 2005). according to black (1976), the complexity of the dividend could be visualised as a piece of the puzzle that was hard to fit each other. one of the main reasons the dividend was classified as a puzzle was that a firm’s decision to pay or not pay, increase, or decrease the dividend would substantially influence the firm’s performance and financial position. for example, if assuming a firm has decided to pay a dividend; in this case, the firm would have less financial capability to finance its investment opportunities. conversely, if the firm has decided not to pay a dividend, it would affect the firm’s valuation and damage its reputation. the interconnected link between a firm’s dividend policy and its other policies indicates how important it is to achieve an optimum dividend level. several past empirical studies have revealed that a dividend policy could be affected by profitability, debt level, firm size, investment opportunity and cash flow. although a substantial number of studies have been conducted regarding dividend policy determinants, there has been minimal attention given to corporate governance policy and its impact from the determinants of dividend policy. according to bakri et al. (2019), the dividend could be used as a substitute for weak 3 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 governance to maintain a good relationship with shareholders. this implied that the governance or corporate governance could influence a firm’s dividend in policy to a certain degree. for example, strong governance has been posited as significantly influencing a firms’ intention to pay more dividends (mitton 2004; petrasek, 2012). the strong influence of governance quality over the dividend policy and the minimal attention given to the determinants of dividend especially in the emerging markets creates a room to be filled in. motivated by this concern, the study was aimed at determining whether the changes in the malaysian code on corporate governance in 2012 might have influenced a firm’s dividend policy. furthermore, changes in the malaysian corporate governance code have made it an imperative to investigate these issues because the country’s market has experienced substantial growth in recent years. malaysia is considered one of the most developed capital markets among emerging markets (yusof & ismail, 2016). the malaysian capital markets have continuously become the backbone of financial sources, with rm 90 billion raised via the primary market for the fourth consecutive year (security malaysia, 2016). during this financial year, rm 86 billion was raised via bond issuance and rm4 billion raised via initial public offering (ipo) (security malaysia, 2016). in 2017, the malaysian capital market grew by 12.6 percent from the previous year, achieving a sum of rm 3.2 trillion with the total capital raised amounting to rm 146.6 billion (security malaysia, 2018). the exponential growth of malaysia capital markets and the changes in the policy on the malaysia code on corporate governance has motivated the present study to investigate whether the factors that have determined dividend policy in malaysia have changed over the past few years, after the changes introduced in the mccg in 2012. the study would contribute to the existing body of knowledge on dividend policy by expanding the existing empirical evidence in several ways. first, the present study sought to examine the determinants of dividend policy using the corporate governance changes as the basis of comparison. this have not been tested or has been overlooked in past studies. second, the study might be very helpful to the practitioner, especially to the management team, in determining its dividend policy or to prioritise on optimising dividend policy; strategies which could 4 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 maximise their shareholders’ wealth, and at the same time without jeopardising the firm’s performance. the rest of the paper is structured as follows; the following sections discuss the study’s theoretical framework, literature review related to dividend policy determinants, and how past theories and empirical evidence have been used to derive the present study’s hypotheses. next, are the sections on the methodology, findings, and discussions. finally, the paper concludes by discussing the implications and limitations of the study, and suggestions for future research. literature review in general, at the end of each financial year, the firm with a profit will decide on whether to distribute its profit or not. if so, how much will this profit be distributed in the form of a dividend? for more than a decade, the dividend issues remained a mystery among researchers and practitioners. for instance, the firm is required to decide its optimum dividend policy, and this is because, the dividend will affect other the firm’s other policies, such as firm investment and capital structure, which will be directly linked to firm-specific characteristics (smith & watts, 1992; gaver & gaver, 1993; braclay et al., 1995). the complexity of the dividend often makes it being visualised as a puzzle which is difficult to piece together (black, 1976). the puzzle of the dividend remains as potential issues that require further research. there are issues such as the regulation, policy, and economic changes that significantly influence how a firm determines its dividend policy throughout a year. thus, to identify what has changed over the past year, research to update the body of knowledge in this aspect is crucial. this will enable stakeholders to understand whether the determinants of dividend policy has changed over the years, and whether these have impacted the firm’s dividend policy. one of the earliest studies with regards to dividend determinants was conducted by lintner (1956). this study revealed that a firm’s profitability and past year dividend had a major influence over the firm’s dividend policy. the study posited that the reasons for this influence were that the investors preferred a dividend with a stable 5 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 payment rate. the finding was also supported by much empirical evidence from other later studies, which also showed that the dividend was significantly influenced by profits and past year dividend (pruitt & gitman, 1991). since the classic study by lintner (1956), more studies have been conducted to examine the factors that may influence the dividend policy. one of the earliest theories relating to dividend policy is the signalling theory. according to miller and modigliani (1961), a dividend announcement would convey some information about a firm’s performance, making the shareholders and investors likely to react to the announcement. bhattacharya (1979) also supported this theory, when they showed that there was a profitability correlation with dividend policy. furthermore, the dividend announcement reflected the stability of a firm’s financial performance in the future (kale & noe, 1990). the dividend reflected a firm’s financial performance stability because a firm generally would only announce a dividend payment or any dividend increase. consider the scenario in which a firm has decided on a different dividend policy, for example, a dividend omission or decrease. in this instance, it has conveyed information to the shareholders and investors that the firm has become less profitable or could suffer a potential loss. as a result, the market would punish the firm that has made dividend omission, or has decreased their dividend by mitigating the demand for their stocks, or share sell by those firms, resulting in a low firm valuation. besides signalling theory, agency cost theory has also been related to dividend policy. the theory is based on the principal and agent relationship, in which a manager act as an agent and the owner or shareholder act as a principal. the problem arises when the manager’s action is not aligned with the best interest of the shareholder. agency cost theory suggests that the dividend can be used as a tool for the mitigated agency problem. easterbrook (1984) and jensen (1986) in their study indicated that the dividend could be used as a mechanism to divert the attention of insiders from using the excess cash to invest in an unprofitable project, or for personal use. agency cost theory also suggested that the dividend can be used to mitigate the agency’s problem by reducing agency cost related to free cash flow, debt, firm growth, investment opportunities, firm size and risk (jensen & meckling, 1976; rozeff, 1982; jensen, 1986; utami & inanga, 2011). 6 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 the theory indirectly indicates that these factors can potentially influence the dividend policy (yusof & ismail, 2016). to investigate free cash effect based on agency cost theory, the study used “fcf” as a proxy for free cash flow. in general, the higher the profitability, the more cash available, and the more dividend paid to mitigate agency cost. in this regard, agency cost theory, pecking order theory, and signalling theory might help to explain the relationship between dividend and profitability (fama & french, 2002; yarram & dollery, 2015). kuzucu (2015) discovered that dividend and profitability had a negative relationship. in contrast, al-malkawi (2007) bokpin (2011), patra et al. (2012) identified profitability as a key determinant which has a positive association with dividend policy. to investigate this effect on dividend policy, present study has used roa as a proxy for profitability and fcf to proxy for free cash flow. to examine the impact of profitability on a firm’s dividend policy, and following fang et al. (2014) and jiang et al. (2017), the present study used “roa” which is a return on asset as a proxy for profitability. whereas to examines the cash availability on dividend determinants, the study used fcf following yusof and ismail (2016). in the context of the debt and dividend policy relationship, a firm with more debt requires excess cash to settle its debt obligations to prevent from default. it will then lead to a reduction, in terms of available cash to sustain the operating expenses. as a result, the management team may reduce the funds available to shareholders and reduce the term of dividend pay outs. in other words, a debt and a dividend have a negative relationship. supporting this idea, several empirical studies have been conducted in the past years. for example, al-malkawi (2007) discovered a negative association between debt financing and dividend policy when examining the determinants of dividend policy in jordan. similarly, the same findings were also found in yusof and ismail (2016); al-shubiri (2011) and ramli (2010). however, not all previous studies showed a negative association between debt level and dividend policy; for instance, appannan and sim (2011) found a positive association between debt level and dividend pay outs. in contrast, singla and samanta (2019) documented an insignificant relationship between leverage and dividend. to investigate the firm’s debt-level effect on dividend policy, the present study used “debt” to refer to the total liabilities to total asset ratio as a proxy for debt level. 7 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 generally, when the firm size increases, it will pay more dividend to mitigate the agency problem, which is likely to increase. the reasons for the increasing agency problem among larger firms are that a large firm has more widespread ownership and therefore, difficulty in controlling its internal and external financing activities, relative to small firms (yusof & ismail, 2016). the positive relationship between firm size and dividend policy has been well documented in numerous past empirical researches such as in ramli (2010), mehrani et al. (2011), hashemi and zadeh (2012), and yusof and ismail (2016). besides the finding on the positive association between dividend policy and firm size, past studies also recorded a negative association between firm size and dividend policy, such as in the studies by ahmed and javid (2009) and huda and farah (2011). to investigate firm size effect on dividend policy, the present study has used “size” to refer to a natural logarithm of a total asset to proxy for the firm size, following past studies such as those by yusof and ismail (2016); and dewasiri et al. (2019). in the context of investment opportunity and its relationship with dividend policy, agency cost theory has been able to provide a rational explanation. according to this theory, a firm with no or less growth and investment opportunities would be prone to being exposed to agency costs related to free cash flow (yusof & ismail, 2016). according to jensen (1986), to reduce firm agency cost, a firm with less investment and growth opportunities would pay more dividend. in contrast, a firm with greater investment and growth opportunities would be paying less or no dividend because they would have required more cash to finance its investment opportunity. the negative association between investment and growth opportunities, and dividend policy were recorded in many previous studies, such as those by rozeff (1982), jensen et al. (1992), al-kuwari (2010), and yusof and ismail (2016). to investigate this effect, the present study used “inv” to refer to the ratio of retained earnings over total asset, following yusof and ismail (2016). the factor of risk was also often associated with dividend policy in many past studies. a greater need for external sources of financing demonstrates a massive fluctuation in firm cash flow, which also indicates a higher firm risk. thus, to mitigate the concerns regarding external sources of financing, a firm would often pay less dividend 8 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 (rozeff, 1982). the negative relationship between risk and dividend policy were recorded in several past studies such as in ramli (2010) and al-shubiri (2011), which discovered a negative relationship between dividend policy and firm risk. in contrast, al-shabibi and ramesh (2011) found a positive relationship between a firm’s risk and the firm’s decision to pay a dividend. in contrast, not all research discovered a negative association between risk and dividend payment. for example, franc-dabrowska et al. (2020) found an insignificant relationship between risk and dividend policies. to examine the influence of risk on a firm’s dividend policy, the present study used “risk” to refer to one year of market beta, following yusof and ismail (2016). based on the postulates of signalling theory, agency cost theory, the lintner model and empirical evidence discovered in previous studies, the hypotheses for the malaysian context of dividend policy were developed as follows: h1 : there is a positive relationship between profitability and firm dividend policy preand post-mccg 2012. h2 : there is a positive relationship between lagged of dividend and firm dividend policy pre-and post-mccg 2012. h3 : there is a positive relationship between free cash flow and firm dividend policy preand post-mccg 2012. h4 : there is a negative relationship between debt level and firm dividend policy preand post-mccg 2012. h5 : there is a positive relationship between firm size and firm dividend policy preand post-mccg 2012. h6 : there is a negative relationship between investment opportunities and firm dividend policy pre-and postmccg 2012. h7 : there is a negative relationship between market risk and firm dividend policy preand post-mccg 2012. methodology sample selection and data collection the present study used a non-financial firm sample which was made available in datastream within malaysia. the reason the present study has elected to exclude a firm within banks and in non-financial sectors 9 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 was because of the issue of the high leverage and industry regulations (dewasiri et al., 2016). this study also included a one year lagged of dividend per share to mitigate the concerns of serial correlation. as a result, the study lost some firm-year observations. there were two periods covered in this study; the first is the period between 2005 to 2011 (pre-mccg 2012), and the second is the period between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-andpost mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: where, = dividend per share = return on asset = lagged of dividend = free cash flow = total liabilities over total asset = natural logarithm of total asset between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets 10 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 = investment opportunity = market risk = industries fixed effect = year fixed effect = error terms table 1 variables definitions variables represented by definitions dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/total assets firm size size natural logarithm of total assets investment opportunity inv retained earnings/total assets market beta risk 1 year of market beta findings and discussions descriptive statistics the descriptive statistics presented in table 2 showed the mean, standard deviation, and minimum and maximum value for each variable tested in this study. as demonstrated in table 2, the average dividend per share and return on assets were 0.05 and 3.57, respectively. the average lagged of dividend, free cash flow, debt, size, investment opportunities, and market risk were 0.05, 0.02, 0.37, 12.70, 0.08, and 1.05 respectively. before examining the main analysis, the study conducted several diagnostic tests. these tests were conducted to ensure the robustness of the result. they included tests on normality, heteroscedasticity, an autocorrelation test and lastly a multi-collinearity test. to mitigate the concern about any outlier, the study winsorizing the data at 1 and 99 percentiles. next, the study examined the potential heteroscedasticity of the data using the white heteroscedasticity test. to reduce this concern, as indicated by the test, the study used the robust standard errors calculation. between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets between 2013 to 2019 (post-mccg 2012). the main justification for selecting these two time frames for this study was the aim of examining the pre-and-post mccg 2012 impact on dividend policy determinants. data analysis the data were analysed using stata (version 13). more specifically, the regression method was conducted using pooled ols, random as well as fixed-effect analysis. before analysing the data, the study also carried out several diagnostic tests to ensure the data were free from any multi-collinearity and heteroscedasticity issue. the diagnostic test to identify multi-collinearity issues included the correlation matrix analysis and variance inflation factors analysis. furthermore, the study also used the robust standard errors calculation analysis to ensure that the data were free from heteroscedasticity issues. the dependent variable used in this study was the dividend per share. whereas the independent variables used in this study included the following: profitability, lagged of dividend, free cash flow, debt level, firm size, investment opportunities and market risk. the study also added industries and year fixed effect which used dummies to control for industry and year effect. the model used in this study is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽0 + 𝛽𝛽1𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽3𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐷𝐷𝑆𝑆𝑆𝑆𝐷𝐷𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽7𝑅𝑅𝑆𝑆𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 where, dpsi,t = dividend per share roai,t = return on asset dpsi,t−1 = lagged of dividend fcfi,t = free cash flow debti,t = total liabilities over total asset sizei,t = natural logarithm of total asset invi,t = investment opportunity riski,t = market risk δi,t = industries fixed effect μi,t = year fixed effect εi,t = error terms table 1 definition and proxy of variables constructs represent by proxy variables dividend dps dividend per share profitability roa return on asset lagged of dividend ldps dividend per share free cash flow fcf free cash flow debt debt total liabilities/ total assets firm size size natural logarithm of total assets 11 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 table 2 descriptive statistics of the variables variable obs mean std. dev. min max dps 7830 0.0415 0.0917 0 0.59 roa 7830 3.5717 9.5172 -35.41 31.96 ldps 7830 0.0406 0.0897 0 0.59 fcf 7830 0.0114 0.1519 -0.617 0.569 debt 7830 0.3736 0.2007 0.0235 0.9308 size 7830 12.7062 1.6258 5.7930 19.0014 inv 7830 0.0889 0.4020 -1.8470 0.7086 risk 7830 1.0470 0.6926 -0.678 3.268 the study also used the lag of dependent variable to solve the potential autocorrelation in the data, as indicated by the breusch pagan lagrange multiplier (lm) test. finally, the study also conducted a multi-collinearity test. following hair et al. (2010), any value higher than 4.00 in vif score and 0.60 in the pearson correlation matrix was considered as having high multi-collinearity. as shown in table 3 and 4, no variables, excluding lagged dependent variables, had a value higher than 4.00 and 0.60. this indicated that there was no risk of multi-collinearity. table 3 variation inflation factor analysis of determinants of dividend policy variable vif 1/vif inv 1.71 0.5850 size 1.66 0.6031 roa 1.42 0.7059 debt 1.37 0.7302 ldps 1.3 0.7711 risk 1.09 0.9161 fcf 1.06 0.9431 mean vif 1.37 12 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 ta bl e 4 p ea rs on c or re la tio n m at ri x r es ul ts v ar ia bl es d ps r o a l d ps fc f d eb t si ze in v r is k d ps 1 r o a 0. 34 53 ** * 1 l d ps 0. 93 21 ** * 0. 31 50 ** * 1 fc f 0. 09 58 ** * 0. 21 81 ** * 0. 07 38 ** * 1 d eb t -0 .0 28 9* * -0 .1 11 7* ** -0 .0 24 6* * -0 .1 07 3* ** 1 si ze 0. 37 48 ** * 0. 24 64 ** * 0. 36 94 ** * 0. 07 04 ** * 0. 29 03 ** * 1 in v 0. 27 17 ** * 0. 48 42 ** * 0. 26 78 ** * 0. 14 66 ** * -0 .2 82 0* ** 0. 36 37 ** * 1 r is k -0 .1 68 9* ** -0 .1 22 5* ** -0 .1 69 1* ** -0 .0 24 8* * 0. 09 25 ** * 0. 10 78 ** * -0 .1 18 4* ** 1 n ot e. * , * * an d * de no te s ta tis tic al s ig ni fic an ce a t 1 % , 5 % a nd 1 0% le ve l o f s ig ni fic an ce re sp ec tiv el y. 13 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 regression results the main analysis was as presented in table 5. the hausman test indicated that the fixed effect was the most appropriate choice, which referred to model iii pre-mccg and model vi post-mccg 2012. from the seven variables tested in this study, four variables were found to be significant pre-mccg 2012 and five variables in post-mccg 2012. referring to model iii, roa, ldps, debt, and size were found to be significant. from among the four significant variables pre-mccg 2012, roa, ldps and size, were significant at a 1 percent level, while debt was significant at 10 percent. from among the four significant variables, roa, ldps, and size showed a positive relationship, whereas debt was negatively related. the post-mccg 2012 analysis results presented in table 6 showed that there were slight changes in the range of dividend policy determinants; it was found that from the original seven factors tested post-mccg 2012, only five variables were found to be significant including roa, ldps, size, inv and risk. the five factors found to be significant were the following: roa and ldps were significant at 1 percent level, size and inv were significant at 5 percent level and risk was significant at 10 percent. among the five factors, three variables, namely roa, ldps and size, were positively related to dividend whereas, two factors, namely debt and risk, were found to be negatively correlated. hence pre-mccg 2012, h1 (profitability), h2 (lagged of dividend), h4 (debt), h5 (size) were supported, and the remaining variables were not supported. on comparison, post-mccg 2012, h1(profitability), h2 (lagged of dividend), h5 (size), h6 (inv) and h7 (risk) were supported, while the remaining variables were not supported. the positive and significant relationship between profitability and dividend preand post-mccg 2012 indicated that as a firm’s profitability increases, it would increase its dividend payment, which indirectly supported the postulates of signalling theory. as the firm’s performance became better, it could reward its shareholders in the form of a dividend payment. the results in this study were consistent with many other previous research studies, such as those by al-malkawi (2007), ahmed and javid (2009), al-kuwari (2010), ramli (2010), mehrani et al. (2011), al-shabibi and ramesh (2011), yusof and ismail (2016), and dewasiri et al. (2016). however, the 14 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 findings regarding dividend and profitability were not corroborated in the results found in anil and kapoor (2008) and appannan and sim (2011) where the researchers discovered a negative association between dividend and profitability. the documented positive relationship between dividend and lagged of dividend pre-and post-mccg 2012 were consistent with those in lintner (1956) and yusof and ismail (2016). lintner (1956) had suggested that firm profitability and the past year dividend (lagged dividend) were major determinants of firm dividend policy. according to yusof and ismail (2016), the influence of past year dividend over the current year dividend was the result of the investor preferring a much stable dividend rate. as a result, a firm tended to make the dividend decision based on the past year dividend policy. the negative association between debt level and dividend pre-mccg 2012 was consistent with past studies such as those by yusof and ismail (2016), al-shubiri (2011), ramli (2010), and al-malkawi (2007). this result also indicated that the debt level provided incentives to the firm to limit its dividend payment to fulfil its debt obligation. the finding also indirectly suggested that the debt level was influenced by the bank covenant, which might restrict the amount of dividend that could be paid by the firms. however, the result was inconsistent with several past studies such as in appannan and sim (2011), and singla and samanta (2019), which had demonstrated a negative and insignificant relationship. the significant positive relationship between firm size and dividend policy preand post-mccg 2012, indicated that the agency cost theory was supported. according to this theory, the larger the firm size, the greater the percentage of widespread ownership. this most likely had created a difficult to control internal and external financing sources (yusof & ismail, 2016). thus, to control agency costs associated with the greater widespread ownership found in a large firm, the large firm tended to increase dividend payment. the findings from this study with regards to firm size and dividend policy had been consistent with those in previous studies such as those in al-malkawi (2007), al-kuwari (2010), ramli (2010), yusof and ismail (2016) and dewasiri et al. (2019). the negative association between investment opportunities and dividend policy post-mccg 2012 was consistent with agency cost 15 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 theory. according to this theory, a firm with little to no ability to demonstrate a high potential growth rate would be highly exposed to agency cost, especially regarding free cash flow (yusof & ismail, 2016). therefore, to mitigate those concerns, a firm with a greater investment opportunity represented by inv in the present study would most likely pay less dividend. the reasons were due to the firm requiring the cash to finance its investment opportunity. the negative relationship between dividend and risk has found support in past empirical evidence presented in al-shubiri (2011) and ramli (2010). the results indicated that a firm with greater market risk had more cash flow issues or greater fluctuation with regards to the firm cash flow. as a result, the firm might reduce its dividend payment to ensure better cash flow to meet its obligation and finance its investment opportunity. however, the results were found to be not consistent with other past empirical studies such as those in francdabrowska et al. (2020) which had found an insignificant relationship between dividend and risk. table 5 panel ordinary least square, random effect and fixed effects (with robust standard errors) model model i: pooled ordinary least square (robust standard errors) model ii: random effect (robust standard errors) model iii: fixed effect (robust standard errors) regressors regression coefficient t-statistics regression coefficient z-statistics regression coefficient t-statistics constant roai,t ldpsi,t fcfi,t debti,t sizei,t invi,t riski,t -0.0364 0.0004 0.8896 0.0091 -0.0134 0.0045 -0.0030 -0.0031 -3.39** 4.71*** 28.44*** 1.43 -3.22** 4.97*** -2.41* -3.02** -0.0344 0.0005 0.9138 0.0088 -0.0128 0.0037 -0.0031 -0.0033 -4.99*** 6.14*** 32.41*** 1.60 -3.49** 5.22*** -2.38* -3.83*** -0.1031 0.0005 0.3994 -0.0087 -0.0208 0.0107 -0.0050 0.0006 -3.62** 5.36*** 5.00*** -1.10 -2.57* 4.57*** -1.91 0.44 industries year r-squared yes yes 0.8321 no no 0.8267 no no 0.7543 note. *, ** and * denote statistical significance at 1%, 5% and 10% level of significance respectively. 16 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 table 6 panel ordinary least square, random effect and fixed effects (with robust standard errors model model iv: pooled ordinary least square (robust standard errors) model v: random effect (robust standard errors) model vi: fixed effect (robust standard errors) regressors regression coefficient t-statistics regression coefficient z-statistics regression coefficient t-statistics constant roai,t ldpsi,t fcfi,t debti,t sizei,t invi,t riski,t -0.0267 0.0005 0.9075 0.0091 -0.0031 0.0020 -0.0050 -0.0022 -3.43** 6.24*** 46.28*** 1.43 -1.04 3.45** -3.78*** -3.56*** -0.0176 0.0005 0.9217 0.0094 -0.0034 0.0018 -0.0053 -0.0016 -2.64** 6.34*** 51.09*** 1.52 -1.35 2.86** -4.22*** -2.59* -0.0399 0.0004 0.4343 -0.0076 -0.0098 0.0054 -0.0077 -0.0032 -2.01* 4.28*** 8.20*** -1.11 -1.85 3.32** -3.48** -2.36* industries year r-squared yes yes 0.9078 no no 0.9060 no no 0.8876 note. *, ** and * denote statistical significance at 1%, 5% and 10% level of significance respectively. additional test the results presented in table 5 and 6 might be influenced by endogeneity with regard to omitted variables bias. according to jiang, ma and shi (2017), the firm fixed effect could reduce the concern of endogeneity, especially regarding omitted variables bias. thus, the pesent study re-examined the model using the firm fixed effect to mitigate omitted variables bias following jiang et al. (2017) and bakri et al. (2020), the results are as presented in table 7. the results revealed that dividend policy determinants remained persistent even after controlling for endogeneity concerns regarding omitted variables bias both in pre-and-post mccg 2012, as represented via model vii and model viii. additionally, fcf was also found to be significant post-mccg. 17 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 table 7 robustness test-endogeneity of omitted variables bias model model vii: firm fixed effects (pre-mccg 2012) model viii: firm fixed effects (post-mccg 2012) regressors regression coefficients t-statistics regression coefficients t-statistics constant -0.1031 -3.39 -0.0399 -2.23 roai,t 0.0005 5.09 0.0004 5.58 ldpsi,t 0.3995 22.18 0.4343 30.00 fcfi,t -0.0087 -1.82 -0.0076 -2.15 debti,t -0.0208 -2.92 -0.0098 -1.79 sizei,t 0.0107 4.27 0.0054 3.78 invi,t -0.0050 -1.42 -0.0077 -2.81 riski,t 0.0006 0.43 -0.0032 -3.34 industries yes yes year yes yes r-squared 0.7543 0.8876 note. *, ** and * denote statistical significance at 1%, 5% and 10% level of significance respectively. conclusion this study investigated the factors that might influence firm dividend policy in malaysia. examining a total of 631 non-financial firms listed in malaysia, the study discovered that the range of determinants of dividend policy demonstrated slight changes pre-and-post mccg 2012. pre-mccg results revealed that profitability, lagged of dividend, firm size, and debt significantly determined firm dividend policy in malaysia. post-mccg 2012, the range of factors had changed; profitability, lagged of dividend, size, investment opportunity and market risk were found to be significant determinants of firm dividend policy. specifically, the result was robust, even after controlling for endogeneity concerns, especially with regard to omitted variables bias. the present study has shown originality in its approach and this is important as it has provided empirical evidence on the range 18 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 of dividend policy determinants that could change as a result of comparing two different timelines. compared to previous studies in the field, this study has made a significant contribution by comparing pre-and-post mccg 2012, which was neglected in the past. the study has also contributed to the growing body of knowledge based on empirical evidence obtained from studies of dividend policy. the present study has also added a corporate governance perspective within the malaysian context. however, the study has some limitations within the context of the selected timeframe on the latest mccg changes. for example, the study only examined pre-and-post mccg 2012. future research may want to include important recent changes on the mccg, which happened in 2017. such an updated analysis would be very beneficial for management to settle on a reasonable or ideal dividend strategy that would support shareholder capital without neglecting corporate governance regulation, which might jeopardise firm evaluation and reputation. acknowledgment this research received no specific grant from any funding agency. references ahmed, h., & javid, a. y. (2009). the determinants of dividend policy in pakistan. international research journal of finance and economics, 29, 110-125. anil, k., & kapoor, s. (2008). determinants of dividend payout ratios – a study of indian information technology sector. international research journal of finance and economics, 15, 63-71. appannan, s., & sim, l. w. (2011). a study on leading determinants of dividend policy in malaysia listed companies for food industry under consumer product sector. paper presented at 2nd international conference on business and economic research proceeding. bakri, m. a., bany-ariffin, a. n, matemilola, b. t., & lau, w.t. (2020). moderating role of financial market development on the relationship between stock liquidity and dividend. asian 19 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 academy of management journal of accounting and finance, 16(2), 77-99. bakri, m. a., noordin, b. a. a., tunde, m. b., & theng, l. w. (2019). moderating effects of governance quality on the relationship between stock liquidity and dividend in emerging market countries. jurnal pengurusan (ukm journal of management), 57. http://ejournal.ukm.my/pengurusan/article/view/36855/10112. barclay, m. j., smith, c. w., & watts, r. l. (1995). the determinants of corporate leverage and dividend policies. journal of applied corporate finance, 7(4), 4-19. bhattacharya, s. (1979). imperfect information, dividend policy, and “the bird in the hand” fallacy. the bell journal of economics, 10(1), 259-270. black, f. (1976). the pricing of commodity contracts. journal of financial economics, 3, 167–179. bokpin, g. a. (2011). ownership structure, corporate governance and dividend performance on the ghana stock exchange. journal of applied accounting research, 12(1), 61-73. brealey, r., & myres, s. (2005). principle of corporate finance (8th ed.). mcgraw-hill. dewasiri, n. j., koralalage, w. b. y., azeez, a. a., jayarathne, p. g. s. a., kuruppuarachchi, d., & weerasinghe, v. a. (2019). determinants of dividend policy: evidence from an emerging and developing market. managerial finance, 45(3), 413-429. easterbrook, f. h. (1984). two agency-cost explanations of dividends. the american economic review, 74(4), 650-659. fama, e., & french, k. (2002). testing trade off and pecking order predictions about dividends and debt? the review of financial studies, 15(1), 1-33. fang, v. w., tian, x., & tice, s. (2014). does stock liquidity enhance or impede firm innovation? the journal of finance, 69(5), 20852125. franc-dąbrowska, j., mądra-sawicka, m., & ulrichs, m. (2020). determinants of dividend payout decisions–the case of publicly quoted food industry enterprises operating in emerging markets. economic research-ekonomska istraživanja, 33(1), 1108-1129. gaver, j. j., & gaver, k. m. (1993). additional evidence on the association between the investment opportunity set and corporate financing, dividend, and compensation policies. journal of accounting and economics, 16(1-3), 125-160. 20 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 hashemi, a. s., & zadeh, f. z. f. (2012). the impact of financial leverage operating cash flow and size of company on dividend policy (case study of iran). interdisciplinary journal of contemporary research in business, 3(10) 264-270. huda, f., & farah, t. (2011). determinants of dividend decision: a focus on banking sector in bangdalesh. international research journal of finance and economics, 77, 33-46. hair, j., black, w., babin, b., & anderson, r. (2010). multivariate data analysis (7th ed.). pearson education. jensen, m. c. (1986). agency costs of free cash flow, corporate finance, and takeovers. the american economic review, 76(2), 323-329. jensen, m. c., & meckling, w. h. (1976). theory of the firm: managerial behavior, agency costs and ownership structure. journal of financial economics, 3(4), 305-360. jiang, f., ma, y., & shi, b. (2017). stock liquidity and dividend payouts. journal of corporate finance, 42, 295-314. kale, j. r., & noe, t. h. (1990). risky debt maturity choice in a sequential game equilibrium. journal of financial research, 13(2), 155-166. kuwari al-, d. (2010). to pay or not to pay: using emerging panel data to identify factors influencing corporate dividend payout decisions. international research journal of finance and economics, 42, 19-36. kuzucu, n. (2015). determinants of dividend policy: a panel data analysis for turkish listed firms. international journal of business and management, 10(11) 149-160. lintner, j. (1956). distribution of incomes of corporations among dividends, retained earnings, and taxes. the american economic review, 46(2), 97-113. malkawi al-, h. a. n. (2007). determinants of corporate dividend policy in jordan: an application of the tobit model. journal of economic & administrative sciences, 23(2), 44-70. mehrani, s., moradi, m., & esk, h. (2011). ownership structure and dividend policy: evidence from iran. african journal of business management, 5(17), 7516-7525. miller, m. h., & modigliani, f. (1961). dividend policy, growth, and the valuation of shares. the journal of business, 34(4), 411433. mitton, t. (2004). corporate governance and dividend policy in emerging markets. emerging markets review, 5(4), 409-426. 21 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 patra, t., poshakwale, s., & kean, o. (2012). determinants of corporate dividend policy in greece. applied financial economics, 22(13), 1079-1087. petrasek, l. (2012). do transparent firms pay out more cash to shareholders? evidence from international cross listings. financial management, 41(3), 615–636. pruitt, s. w., & gitman, l. j. (1991). the interactions between the investment, financing, and dividend decisions of major us firms. financial review, 26(3), 409-430. ramli, n. m. (2010). ownership structure and dividend policy: evidence from malaysian companies. international review of business research papers, 6(1), 170-180. rozeff, m. s. (1982). growth, beta and agency costs as determinants of dividend payout ratios. journal of financial research, 5(3), 249-259. security malaysia. (2016). malaysian capital market grows across all segments: expands to reach rm 2.82 trillion. https://www. sc.com.my/resources/media-releases-and-announcements/ malaysian-capital-market-grows-across-all-segmentsexpands-to-reach-rm282-trillion. security malaysia. (2018). malaysian capital market grew 12.6% to rm 3.2 trillion with strong growth across all segments. https:// www.sc.com.my/resources/media/media-release/malaysiancapital-market-grew-126-to-rm32-trillion-with-strong-growthacross-all-segments shabibi al-, b. k., & ramesh, g. (2011). an empirical study on the determinants of dividend policy in the uk. international research journal of finance and economics, 80,105-120. shubiri, al, f. n. (2011). determinants of changes dividend behavior policy: evidence from the amman stock exchange. far east journal of psychology and business, 4(1), 1-15. singla, h. k., & samanta, p. k. (2019). determinants of dividend payout of construction companies: a panel data analysis. journal of financial management of property and construction. smith jr, c. w., & watts, r. l. (1992). the investment opportunity set and corporate financing, dividend, and compensation policies. journal of financial economics, 32(3), 263-292. utami, s. r., & inanga, e. l. (2011). agency costs of free cash flow, dividend policy, and leverage of firms in indonesia. european 22 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 1–22 journal of economics, finance and administrative sciences, 33(6), 7-24. yarram, s. r., & dollery, y. b. (2015). corporate governance and financial policies. managerial finance, 41(3), 267-285. yusof, y., & ismail, s. (2016). determinants of dividend policy of public listed companies in malaysia. review of international business and strategy, 26(1), 88-99. 29 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 how to cite this article: aqilah, s., & fathoni, m. a. (2023). factors determining the use of mobile banking in indonesia. international journal of banking and finance, 18(2), 29-52. https://doi. org/10.32890/ ijbf2023.18.2.2 factors determining the use of mobile banking in indonesia 1shafa aqilah & 2muhammad anwar fathoni islamic economics program, faculty of economisc and business universitas pembangunan nasional veteran jakarta, indonesia 2corresponding author: mfathoni@upnvj.ac.id received: 1/3/2022 revised: 8/8/2022 accepted: 5/9/2022 published: 25/6/2023 abstract this study was aimed at determining the factors that could influence a person’s decision to use mobile banking. in this study, perceived usefulness, perceived ease of use, and perceived credibility were considered exogenous variables, and social factors as mediators. this research used quantitative methods. the population in this study were indonesian sharia bank customers who used bsi mobile banking services in the jakarta, bogor, depok, tangerang and bekasi areas. a sample of 200 people was obtained using the purposive sampling method. the analysis technique used was the inner model, outer model, and hypothesis testing using smartpls 3.2.9. the results obtained in this study showed that perceived usefulness had a significant effect on decisions to use mobile banking, perceived ease of use had no significant effect, and perceived credibility had a significant effect. in addition, this study has also shown that social factors did not succeed in mediating other variables in the decision to use mobile banking. https://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 30 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 keywords: credibility, ease of use, mobile banking, social factors, usefulness. jel classification: g21. introduction information technology is experiencing rapid development and has had a positive impact on the creation of business opportunities and transactions carried out electronically (setiawan, 2016). information technology is expected to provide benefits to its users in carrying out activities, for example enabling behaviors that require new techniques to make the task easier and not spending a lot of time executing it. the development of technology has increased the human need for information and requires business actors or even companies to continue to monitor the improvement of information technology and to optimize information technology services to consumers in order to sustain their businesses (romadhon & fitri, 2020). the role of islamic banking is very important in the economy and ensuring the continued existence of islamic financial institutions is paramount to help strengthen the indonesian economy. to date, islamic banking has experienced a considerable growth because indonesia has the advantage of being a country with the largest muslim population. currently, islamic banking in indonesia is keeping abreast of the development of information technology, and the sector has always tried to provide the best service by adopting advances in information technology (styarini & riptiono, 2020). clients can continue to use the traditional bank account services provided by the bank or switch to the newer platforms using digital bank accounts. currently, people are starting to experience changes in conducting banking transactions, from conventional transactions to digital transactions (sumadi, 2020). the emergence of digital banking is the inevitable response to the development of information technology, the increasing innovations related to the role of the digital base can help to reach a wider market potential (susilawaty & nicola, 2020). the financial services authority in indonesia stated that improving services to customers by prioritizing information technology generally leads banks into a new era, namely digital banking is attractive because it will enable customers to conduct transactions 31 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 independently. needless to say, having good service quality will encourage customers to continue to trust the digitalized banking companies (mawarni, 2021). one of the digital services currently offered by banks is mobile banking, which is an innovation resulting from optimizing information technology. mobile banking allows one to connect with the service provider, the bank with the mobile banking business management system from one’s mobile phone and one can access the most efficient banking transactions by independently managing one’s bank account (ningrum et al., 2021). in addition, the mobile banking users benefit the most as the new transactions are time saving and this has become a new paradigm and a new strategy for banking in facing new challenges in the digital era. this mobile banking service minimizes the costs incurred by both the bank and the customer, such as the cost of printing a prospective customer form to open an account or make transactions. mobile banking users will get other benefits, such as information related to their banking transactions that will be received quickly and effectively, customers do not waste time conducting numerous transactions. there are some banks in the world that offer technology with the help of banks such as bank syariah indonesia (bsi). bank syariah indonesia is the outcome from the merger of islamic state-owned banks in indonesia. as of july 2021, bsi mobile service users have reached 2.5 million users, with total transactions from january to june 2021 amounting to 41.99 trillion rupiah. if you are interested in using mobile banking, please contact bsi mobile for a digital transaction with 83.56 people using this feature (pasaribu, 2021). however, when compared to customers who have an account with bank syariah indonesia (bsi), this figure is still very large. the number of accounts at the bsi is calculated to range from 14.5 million to 15.5 million, after the completion of the migration carried out (walfajri & perwitasari, 2021). thus, it can be seen that there is a gap between the number of customers who have accounts and the number of users of the bsi mobile service, which is thought to be caused by other aspects. the credibility aspect is also one of the benefits that will be received by mobile banking users. in the aspect of credibility, there are two factors important for consideration, namely the security and privacy of customer data for mobile banking users (putri et al., 2020). this 32 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 aspect is very important because it is something that customers need to pay attention to before deciding to use mobile banking services. credibility, which includes security, is overshadowed by recent cases related to fraud and data breaches that have occurred to customers who use mobile banking. banks have an important role to play in maintaining the security of customer data of mobile banking users. in the context of mobile banking, a lack of credibility will cause customer concerns that their personal data and funds may be sent to other parties without their knowledge (gupta et al., 2019). in addition, social factors lead to the assumption that a person’s attitude in accepting new technology can be influenced by the social environment, which is one important aspect that can critically influence a consumer’s decision. social influence can be interpreted as the pressure that exists in one’s environment to dictate whether or not to do something in question (deameta, 2019). another factor that can influence customer decision to use mobile banking is the customer’s understanding of the use of the technology and its applications. the use of mobile banking is also an indication of one’s status in society (adzima & ariyanti, 2018). some studies have used similar variables as those in the present study. according to a research conducted by dewi et al. (2018) the variables that have been found to influence customers to use mobile banking were convenience, efficiency and security (dewi et al., 2018). however, the research conducted by latif pointed out that the convenience aspect is not part of the reasons for customer decisions to use mobile banking services (latif, 2017). another research conducted by adiwijaya (2018) showed that convenience and the success rate offered were able to increase customer interest in making transactions using mobile banking (adiwijaya, 2018). styarini and riptiono (2020) emphasied in their research that perceived usefulness was a very important aspect in one’s decision to use mobile banking. they also pointed out that one of the reasons that could hinder customers from using mobile banking was the perceived risk that would be generated (styarini & riptiono, 2020). in addition, the study by rahayu (2016) found that perceptions of usability and credibility were factors important in attracting interest in using mobile banking. meanwhile, the perception of convenience was felt to have no effect because the service often experienced system errors (rahayu, 2016). research conducted by hadi and novi (2015) revealed that variables such as accessibility, security and risk 33 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 variables would influence customer decisions. however, the ease of use of services had not been included in the aspects assessed by customers (hadi & novi, 2015). in light of the foregoing discussions of findings from previous research, it is clear that there are differences in the views regarding the influence of variables on the use of mobile banking services. for this reason, the researcher intends to further examine matters relating to the factors determining decision making to use mobile banking services. in this study, the focus of the analysis was on the following variables: perceived usefulness, the perceived ease of use, perceived credibility and other social factors. literature review mobile banking mobile banking is one of the digital services offered by financial institutions (tam & oliveira, 2017). mobile banking is a digital service that offers convenience in transactions, such as transfers between accounts, checking balances and other banking transactions. however, at this time the convenience provided in mobile banking services is not only about banking, but also extends to e-commerce such as e-wallet top up, paying for electricity and water, and even paying zakat can be done through these services. with the benefits obtained from this service, it makes it easier for customers to transact, and being less time consuming as compared to conducting conventional transactions (fadlan & dewantara, 2018). iriani (2019) stated that mobile banking has been defined as a service that provides convenience and speed of access, making it easier and faster to obtain information about banking and financial transactions (iriani, 2019). service options available in mobile banking include accessing balance information, interbank transfer services, electricity and water payments, insurance bill payments, purchase of credit and tickets, e-wallet top ups and even zakat payments, as well as several other service options (ningrum et al., 2021). with these various services, it can be said that the services offered are not confined only to the banking sector, but also in other fields. if so, these advantages not only make it easier for customers to carry out various types of transactions, but also benefits banks by helping to increase customer satisfaction, improve service operations and cost effectiveness (tam & oliveira, 2017). 34 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 purchase decision purchase decisions have been defined as a method of solving problems beginning with identifying problems, seeking information, evaluating a number of alternatives, making purchase decisions and ending with post-purchase attitudes (kotler & keller, 2016). decision making by consumers is defined as the process of solving a problem and composing a collaboration between knowledge and evaluation of two or more available alternatives (peter & olson, 2010). with the action of decision making, consumers have targets and behaviors that need to be achieved by themselves in order to receive satisfaction from the chosen action. in decision making, consumers play an important role in solving the problems they have. relationships that occur from decision-making actions are included in the category of reciprocal relationships that are intertwined around environmental factors and the resulting behavior (firmansyah, 2018). several indicators influence one in using a product or service. according to mongi (2014), the need to use a product or service is the main reason one makes a purchase or uses it. the need for a product or service makes a person know the actions that need to be taken to meet those needs. in today’s all-digital environment, the need for practical and efficient services often increases among the people. the service in question is a service that provides convenience in carrying out activities. one aspect that a person considers in deciding to buy is the quality of a product or service. if the product or service to be used has good quality, then one will not hesitate in deciding to buy. for that quality will be a priority when one decides to buy a product or use a service. a person’s decision to buy or use a product or service affects the behavior that will be generated afterward. if someone is satisfied with the quality of the product or service that will be used, it becomes a consideration in the future to reuse the product or service. on the other hand, if the product or service is not able to fulfill a person’s sense of satisfaction, there is the likelihood that there is no re-use of the product or service. technology acceptance model according to davis (1989), the theory in the technology acceptance model (tam) contains a basic theory that explains a person’s behavior in accepting the novelty of information. the technology acceptance 35 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 model or commonly abbreviated as tam, is a development of psychological theory and has several foundations, namely attitudes, beliefs, desires and user behavior (user behavior relationship). the tam mentions two factors that make a person accept the novelty of information systems, namely the theory of perceived ease of use and the theory of perceived usefulness. the model is a development of the theory of reasoned action which aims to determine the influence of external variables such as beliefs, attitudes and intentions of a person to use a new technological system. the difference between tam theory and the previous theories is that it shows more specific characteristics related to technology acceptance (wu et al., 2011). according to wu and song (2021), although the tam can well document and predict user attitudes in accepting technology, there are shortcomings in its theory. for example, most significantly, it cannot explain all the driving factors that influence user attitudes because the model used is very common and very simple (wu & song, 2021). perceived usefulness perceived usefulness has been defined as a view related to the use of a new technology that is believed to provide benefits to its users (davis, 1989). similarly, thompson et al. (1991) have also stated that the advantages of an increasingly developing information technology are clearly the expectations desired by its users in carrying out the desired activities. thompson and howell also explained the reasons individuals use information technology, namely being very much aware of the positive impacts from the use of the technology (thompson et al., 1991). basically, one will use something if one knows the benefits of the technology offered. it is hoped that the advent of the latest information technology can bring benefits to everyone who uses it (ernawati & noersanti, 2020). perceived ease of use perceived ease of use is defined as a feeling of trust in the technology, that it is not troublesome to understand and easy to use. perceived ease of use is included in the determinants of individuals accepting an information technology (jogiyanto, 2007). assessment of the 36 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 ease of use of technology is recognized if the technology is easy to understand for new users, therefore they are very likely to adopt the said technology (kristianti & pambudi, 2017). davis (1989) states that there are five indicators which are related to the perceived ease of use of technology. first, mobile banking is not difficult to understand for individuals who are using the service for the first time. offering this convenience to new users, it allows the user to immediately understand the features available in the service. second, mobile banking is easy for users to operate. the convenience offered is not only in terms of understanding the existing features, but also because it offers ease of operation (davis, 1989). furthermore, mobile banking can be operated according to the wishes of the user. this means that mobile banking always runs the system easily, and according to what the user wants. the advantages offered are clear and it is clear that users do not need much effort to carry out their banking activities. this can be seen because it does not take much time and there are no queues to make transactions because it is only done through smartphones. lastly, the use of mobile banking is flexible. in any mobile banking service on offer, flexibility is an inherent feature, namely the ease of using the service anywhere and anytime without wasting one’s precious time. perceived credibility perceived credibility is one of the benefits that will be received by mobile banking users. the credibility component put forward by wang et al. (2003), is that one believes that there is security of transactions and personal data. one can carry out all banking activities and the online banking system is well maintained. these factors will certainly influence the acceptance of the information technology used. the perception of credibility held by the public regarding systems that help complete financial transactions safely and with guaranteed confidentiality will greatly affect how customers voluntarily accept mobile banking services. thus the perception of credibility is predicted to have the ability to explain customers’ intentions to use mobile banking (hanudin et al., 2007). in conclusion, credibility is defined by users’ trust in companies that have been able to provide products/services and meet consumer needs easily by optimizing technological novelty. 37 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 social factors social factors are defined as the influence from a person or group of people on other people and the individual is then willing to imitate their habits (kotler & keller, 2016). social factors come from external parties which include the family, social groups and the role and social status of other people who encourage them to take action on purchasing decisions. kotler and keller (2016) have stated that there are three indicators of social factors as follows. in accordance with what was stated by kotler and keller (2016) that there were several aspects that needed to be considered, namely family, social group, and status. regarding family, it can be said that the closest environment a person begins with is the family environment. the family contains individuals who can influence other members in the group. so it can be said that family members have a strong influence in impacting consumer behavior. in addition, families are included as the main targets by marketers and this has been studied in depth. it can be exemplified that marketers are interested in attracting family members, both husband and wife, and their children to buy a product or service being offered. then there is one’s social group which has a direct influence on members who are part of the group. there are two categories of groups described by kotler and keller (2016), namely primary/regular groups and secondary/irregular groups. the primary group contains individuals who interact quite a lot, but can be seen as informal, for example, such as family, neighbors, friends and schoolmates or coworkers. it is different from the secondary group which contains individuals who can interact formally, such as recitation groups, community groups, certain organizations, and so on. in each secondary group there are individuals who serve as leaders in giving opinions and influences. this individual is not simply chosen without any criteria, but the selection can be made because of the personality, skills, knowledge and other characteristics of the individual. finally, roles and status also play an important role in social factors. in a collection of individuals, for example in families, organizations and communities, there will be someone who has an important role in the group. the role and status can indicate the position held by the person. so it takes someone who can reflect the group to the general public through the required role and status. therefore, group members 38 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 often decide to buy products or services that can help reflect their group in society. methodology data the type of data used in this research was the quantitative data generated from the online questionnaires that were distributed. quantitative data is defined as numerical data which can be in the form of ordinal, ratio, or interval data. the ordinal numerical data from this study was obtained from a likert scale used in the study. sources of data in this study came from primary and secondary data sources. the primary data was from the responses obtained from online questionnaires to prospective respondents who were targeted. the secondary data used in this research were sourced from scientific articles or companies or parties related to this research. the method of collecting data in this study was to use a questionnaire. gathering information is necessary for the research to obtain valid and actual items, based on facts. the questionnaire is a research instrument aimed at collecting data from respondents using a prepared set of questions. in this study, the sample used was the 200 respondents who were using the bsi mobile services. the present research used descriptive and quantitative methods in its analysis. the descriptive method was used to explain the data obtained from online questionnaires that have been filled out by respondents. the quantitative method used was the partial least square (pls) method, with the help of the smartpls 3.2.9 analytical tool. the partial least square is included in the structural equation model (sem) method. sem is often used in research in the humanities and social sciences. this is because the sem method can explain path analysis on the latent variables used. in addition, the sem method has high flexibility compared to other analytical methods. model development this study used the following exogenous variables, namely perceived usefulness (x1), perceived ease of use (x2), and perceived credibility (x3) and these variable will be tested for its influence on the endogenous variables affecting the decision to use mobile banking services (y). in addition, in this study there was a mediating variable, 39 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 namely the social factors (m) which would play a role in determining the effect of mediating variables on exogenous variables. the influence of perceived usefulness on decisions to use mobile banking services perceived usefulness is defined as the view that the use of a new technology is believed to provide benefits to its users (davis, 1989). perception of usability is the main factor for consideration in adopting the latest information technology. a study conducted by mohd thas thaker et al (2019) stated that perceived usefulness has a significant influence on the use of mobile banking. based on this, the following hypotheses have been proposed. h01 = perceived usefulness does not significantly affect the decision to use mobile banking services ha1 = perceived usefulness significantly influences the decision to use mobile banking services the influence of perceived ease of use on decisions to use mobile banking services perception of ease of use is included in the determinants of individuals adopting information technology (jogiyanto, 2007). if one has a sense of trust in information technology, then there are no qualms in carrying out a financial transaction through a payment system that has been modified so that it is easy to operate. a study conducted by alalwan et al. (2016) stated that the perception of ease of use has a significant influence on the use of mobile banking. based on this, the following hypotheses have been proposed. h02 = perceived ease of use does not significantly affect the decision to use mobile banking services ha2 = perceived ease of use significantly influences the decision to use mobile banking services the influence of perceived credibility on decisions to use mobile banking services the perception of credibility put forward by wang et al. (2003) is that one tends to believe that when the security of transactions and personal data are well maintained, it can influence one to accept the 40 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 information technology used. perceived credibility is an assessment of security and privacy issues when discussing the credibility of mobile banking services. studies related to the acceptance of technology offered by banks have been noted by many other researchers. a study conducted by wicaksono et al. (2021) stated that the perception of credibility has a significant influence on the use of mobile banking. based on this, the following hypotheses have been proposed. h03 = perceived credibility does not significantly influence the decision to use mobile banking services ha3 = perceived credibility significantly influences the decision to use mobile banking services the influence of social factors on decisions to use mobile banking services social factors are defined as the influence exerted by a person or group of people on others to want to imitate their habits. social factors come from external parties which include family, social groups and the role and social status of others that encourage them to take action on purchasing decisions (kotler & keller, 2016). a study conducted by gupta et al. (2019) stated that social factors have a significant influence on the use of mobile banking. based on this, the following hypothesis have been proposed. h04 = social factors do not significantly influence the decision to use mobile banking services ha4 = social factors significantly influence the decision to use mobile banking services the influence of social factors in the mediation of perceived usefulness, perceived ease of use and perceived credibility in decisions to use mobile banking services in light of the foregoing discussions, the present study has also proposed the following hypotheses: h05 = social factors mediate perceived usefulness, perceived ease of use and perceived credibility on decisions to use mobile banking services ha5 = social factors do not mediate perceived usefulness, perceived ease of use and perceived credibility on the decision to use mobile banking services 41 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 analysis the present research uses descriptive and quantitative methods in carrying out its analyses. the descriptive method was used to explain the data obtained from online questionnaires that have been filled out by respondents. the quantitative method used was the partial least square (pls) method, which has recourse to the smartpls 3.2.9 analytical tool. partial least square is included in the structural equaition model (sem) method. the partial least square (pls) analysis method is one of the methods incorporated in sem. when compared to the other sem methods, the pls is stronger because there are not many underlying assumptions. not only that, the data used in the pls test of the sample tested does not have to be on a large scale and the data does not have to be normally distributed. the plssem analysis consists of two sub-models, namely the measurement model (measurement model) or commonly called the outer model and the structural model (structural model) or the inner model. results data respondents based on table 1, it can be seen that the respondents based on gender are divided into two categories, namely men and women. of the 200 respondents, female respondents dominated, namely there were 129 respondents or a percentage of 64.5 percent, and the rest were male, there were 71 respondents with a percentage of 35.5 percent. it is therefore, clear from the results that women use mobile banking more in transactions. table 1 respondents by gender gender frequency percentage male 71 35.5% female 129 64.5% total 200 100.0% based on table 2, it can be seen that the respondents based on gender were divided into four categories, namely 12 to 23 years, 24 to 39 42 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 years, 40 to 55 years, and 56 to 74 years. of the 200 respondents, respondents aged 12-23 years (generation z) dominated as there were 122 respondents or a percentage of 61 percent, this was followed by 45 respondents aged 24-39 years, constituting a percentage of 22.5 percent, 29 respondents aged 40-55 years giving a percentage of 14.5 percent, and the remaining 4 respondents were aged 56 – 74, or 2 percent of the total respondents. it is clear that the data showed the dominance of respondents in the age range of 12 – 23 years. it is people in this age group who are included in the category generation z, and they are more sensitive to technological developments as that they are more willing adopt mobile banking services. table 2 respondents by age age frequency percentage 12 23 years (generation z) 122 61.0% 24 39 years (millenial generation) 45 22.5% 40 55 years (generation x) 29 14.5% 56 74 years (baby boomer generation) 4 2.0% total 200 100.0% convergent validity convergent validity is a test to determine the value of the construct of each latent variable indicator. generally, the construct value can be said to be ideal if it meets the criteria above 0.70, but actually the construct value range from 0.40 to 0.70 is acceptable. based on the test carried out, there was one indicator that produced a construct value of less than 0.40 so it had to be removed. in order to get good results it was necessary to do some gradual removal. as a result, the pu1, pu6, peou4 and ud2 indicators were abolished in order to get results that were in accordance with the provisions. discriminant validity discriminant validity aims to find out every question on the indicator does not confuse the answers given by the sample. discriminant validity can be seen from the cross loading and the average variance extracted (ave) value. 43 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 table 3 loading factors perceived usefulness perceived ease of use perceived credibility social factor use decision pu2 0.666 pu3 0.795 pu4 0.766 pu5 0.680 pu7 0.667 pu8 0.695 peou1 0.734 peou2 0.791 peou3 0.738 peou5 0.708 peou6 0.716 peou7 0.641 pc1 0.726 pc2 0.730 pc3 0.799 pc4 0.799 pc5 0.753 pc6 0.742 sf1 0.756 sf2 0.860 sf3 0.816 sf4 0.879 sf5 0.808 ud1 0.621 ud3 0.683 ud4 0.789 ud5 0.815 ud6 0.756 ud7 0.680 44 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 table 4 average variance extracted (ave) average variance extracted (ave) description perceived usefulness 0.509 valid perceived ease of use 0.522 valid perceived credibility 0.575 valid social factor 0.681 valid usage decision 0.529 valid based on the results of the processing of the data above, the results of the average variance extracted (ave) value were found to be above 0.50; it can then be concluded that the variables perceived usefulness, perceived ease of use, perceived credibility, social factors and usage decisions had values above 0.50, which means that all the variable data are valid. reliability table 5 composite reliability composite reliability description perceived usefulness 0.861 reliable perceived ease of use 0.867 reliable perceived credibility 0.890 reliable social factor 0.914 reliable usage decision 0.870 reliable in table 5 above, it can be seen that the composite reliability value for all variables was more than 0.70, which shows that all the variables in this test have met the requirements. from this value, it can be concluded that the composite reliability value can be said to be good for each construct. r-square (r2) the r-square test aims to see the ability of a model to explain a dependent variable. the r-square test in this study is as shown in table 6. 45 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 table 6 r-square on endogenous variable r-square usage decision 0.515 table 6 shows the value of r2 on the endogenous and mediating variables tested in this study. based on the r2 value shown in table 6 above, the decision to use variable shows good results with a value of 0.515, or it can be concluded that the exogenous variables, namely perceived usefulness, perceived ease of use, perceived credibility are able to explain the use decision variable by 51.5 percent and the rest is explained by other factors which are outside the scope of this research. f-square (f2) measurement of f-square or effect size aims to assess the relative impact of an exogenous variable on endogenous variables. the f-square test in this study is as shown in table 7. table 7 f-square on endogenous variable perceived usefulness perceived of ease of use perceived credibility usage decision pu 0.200 peou pc 0.041 sf ud the conclusion of the f-square test results in this study can be seen in table 7 above. it shows that the effect size of the perceived ease of use variable on social factors and the effect size of the perceived credibility variable on the decision to use mobile banking services were included in the low effect category with values of 0.026 and 0.041, respectively. hypothesis test hypothesis testing is used to test the truth of a statement. to make decisions from hypothesis testing, it can be seen from the t-statistic and p-value. 46 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 table 8 path coefficients original sample sample mean standard deviation t-table tstatistics p-values perceived usefulness usage decision 0.469 0.468 0.079 1.97 5.959 0.000 perceived ease of use usage decision 0.117 0.117 0.085 1.97 1.374 0.170 perceived credibility usage decision 0.194 0.204 0.081 1.97 2.389 0.017 social factors usage decision 0.060 0.052 0.061 1.97 0.972 0.331 the results of the hypothesis testing of the perceived usefulness variable showed a positive and significant effect. the coefficient value on the decision to use mobile banking services was 0.469. thus, it can be concluded that h1 in this study, that the perceived usefulness variable has a positive and significant effect on decisions to use mobile banking services can be accepted. meanwhile, the perceived ease of use variable on the decision to use mobile banking services has a path coefficient value of 0.117. the t-statistic value <t-table with a comparison of 1.374 <1.97, at the significance level the p-value (0.170)> 0.05 which did not meet the significance criteria. thus, it can be concluded that h2 in this study, that the perceived ease of use variable has a positive but not significant effect on decisions to use mobile banking services should be rejected. the test results indicate that the perceived credibility variable on the decision to use mobile banking services has a positive and significant relationship, based on the criteria that have been met. in the test conducted, the perceived credibility has a coefficient value of 0.194. the significance value obtained was 0.017 <0.05, which has met the significance criteria. thus, it can be concluded that h3 in this study, that the perceived credibility variable has a positive but not significant effect on decisions to use mobile banking services should be accepted. mediation test the test results show that social factors have no effect whatsoever on the relationship between perceived usefulness, perceived convenience and perceived credibility on the decision to use bsi mobile services. thus, it can be said that the last hypothesis in this study, that social 47 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 factors could not mediate perceived usefulness, perceived ease of use and perceived credibility in making decisions about using mobile services, was rejected. this is because the majority of the bsi mobile users in jabodetabek did not decide to use the bsi mobile services due to the recommendations from friends, family, relatives and others. people decide to use the bsi mobile services due to their awareness of the need for these services. table 9 indirect effect original sample sample mean standard deviation t-table t-statistics p-values perceived usefulness social factors usage decisions 0.011 0.010 0.014 1.97 0.775 0.438 perceived ease of use social factors usage decision 0.013 0.013 0.017 1.97 0.787 0.431 perceived credibility social factors usage decision -0.004 -0.004 0.009 1.97 0.426 0.670 discussions the results of this study are in line with the previous research conducted by styarini and riptiono (2020). they have pointed out that the better the perceived usefulness of the user, the greater the possibility of the decision to use mobile banking services. transactions using mobile banking services are faster so that urgent transaction needs can be resolved quickly without having to go to an atm or bank. the research conducted by haider et al. (2018) also found that perceived usefulness had a significant influence on the use of mobile banking services. the results of this study are related to the research conducted by makanyeza (2019), which revealed that the perception of ease of use has no influence on one in deciding to use mobile banking services. their findings are however, in contrast to the research conducted by putri (2020) and wardani (2021) which found that the perception of ease of use had an influence on decisions to use mobile banking services (wardani, 2021). the results of present study are in line with the previous research conducted by wicaksono (2021), styarini (2020) and koksal (2016) 48 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 which showed consensus in the finding that the perception of credibility would have an influence on decisions to use mobile banking services. one of the most important findings of the present study is that users agree that the bsi mobile is equipped with a professional security system so that the security and confidentiality of transactions are guaranteed. the responses given by bsi mobile users in this study were related to external influences in making decisions to use the bsi mobile. the majority of respondents gave responses that indicated that decisions to use bsi mobile were not influenced by external parties such as friends, relatives, family and others. the results of the present study are in line with the research conducted by rita and fitria (2021), which showed that there was no relationship between decisions to use mobile banking and social factors. this finding is different from the conclusions reached in the research conducted by makanyeza (2017), koksal (2016) and putri (2020), which all confirmed that social factors could influence the decision making in using mobile banking. the test results of the present study have shown that social factors had no effect whatsoever on the relationship between perceived usefulness, perceived convenience and perceived credibility on the decision to use bsi mobile services. thus, it can be said that the mediation hypothesis in this study, which stated that social factors could not mediate perceptions of usability, perceived ease of use and perceptions of credibility in making decisions about using mobile services was rejected. this is because the majority of the bsi mobile users in jabodetabek did not decide to use the bsi mobile services due to the recommendations from friends, family, relatives and others. people decide to use the bsi mobile services due to their awareness of the need for these services. conclusion based on the results of the analysis carried out and foregoing discussions, it is concluded that in this study the factors that influence a person’s decision to use mobile banking. based on the analysis, it can be concluded that perceived usefulness has a significant influence on decisions to use mobile banking. meanwhile, the perceived ease of use does not have a significant influence on the decision to use mobile banking. 49 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 social factors do not affect a person’s consideration of using mobile banking services because the decision is adjusted to the required needs. the analysis on social factors as mediation in the decision to use mobile banking shows that the mediating variable is not able to mediate the variables of perceived usefulness, perceived ease of use and perceived credibility. social factors do not affect a person’s consideration of using mobile banking services because the decision is adjusted to the required needs. analysis related to social factors as mediation in the decision to use mobile banking shows that the mediating variable is not able to mediate the variables of perceived usefulness, perceived ease of use and perceived credibility of the decision making to use mobile banking services. social factors do not affect a person’s consideration of using mobile banking services because the decision is adjusted to the required needs. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. references adiwijaya, i. g. b. p. (2018). kemudahan penggunaan, tingkat keberhasilan transaksi, kemampuan sistem teknologi, kepercayaan dan minat bertransaksi menggunakan mobile banking. jurnal manajemen bisnis, 15(3), 135–153. adzima, f., & ariyanti, m. (2018). analisis faktor-faktor yang mempengaruhi minat menggunakan aplikasi mobile banking pada nasabah bank bri purwakarta. eproceedings of management, 5(2). alalwan, a. a., dwivedi, y. k., rana, n. p. p., & williams, m. d. (2016). consumer adoption of mobile banking in jordan. journal of enterprise information management, 29(1), 118–139. davis, f. d. (1989). perceived usefulness, perceived ease of use and user acceptance of information technology. mis quarterly, 319–340. deameta, a. f. (2019). prediksi intensi perilaku terhadap penggunaan mobile banking menggunakan the unified theory of acceptance 50 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 and use of technology, trust dan network externalities. jasa (jurnal akuntansi, audit dan sistem informasi akuntansi), 3(2), 205–218. dewi, s., sabri, s., & winata, v. p. (2018). analisis kemudahan penggunaan, efisiensi dan keamanan terhadap minat nasabah bertransaksi menggunakan bsm mobile banking pada bank syariah mandiri kantor cabang payakumbuh. ekonomika syariah: journal of economic studies, 2(2), 203–210. ernawati, n., & noersanti, l. (2020). pengaruh persepsi manfaat, kemudahan penggunaan dan kepercayaan terhadap minat penggunaan pada aplikasi ovo. fadlan, a., & dewantara, r. y. (2018). pengaruh persepsi kemudahan dan persepsi kegunaan terhadap penggunaan mobile banking (studi pada mahasiswa pengguna mobile banking universitas brawijaya). firmansyah, m. a. (2018). perilaku konsumen (sikap dan pemasaran) (1st ed.). deepublish. gupta, k. p., manrai, r., & goel, u. (2019). factors influencing adoption of payments banks by indian customers: extending utaut with perceived credibility. journal of asia business studies, 13(2), 173–195. hadi, s., & novi, n. (2015). faktor-faktor yang mempengaruhi penggunaan layanan mobile banking. optimum: jurnal ekonomi dan pembangunan, 5(1), 55–65. haider, m. j., changchun, g., akram, t., & hussain, s. t. (2018). does gender differences play any role in intention to adopt islamic mobile banking in pakistan? journal of islamic marketing, 9(2), 439–460. hanudin, a., baba, r., & muhammad, m. z. (2007). an analysis of mobile banking acceptance by malaysian customers. sunway academic journal, 4, 1–12. iriani, a. f. (2019). minat nasabah dalam penggunaan mobile banking pada nasabah bank syariah mandiri kota palopo. dinamis: journal of islamic management and bussiness, 2(2). jogiyanto. (2007). sistem informasi keperilakuan (1st ed.). andi. koksal, m. h. (2016). the intentions of lebanese consumers to adopt mobile banking. international journal of bank marketing, 34(3), 327–346. kotler, p., & keller, k. l. (2016). marketing management (15th ed.). pearson education. kristianti, m. l., & pambudi, r. (2017). analisis pengaruh persepsi kemudahan, persepsi kemanfaatan, persepsi tingkat keamanan 51 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 dan fitur layanan terhadap penggunaan mobile banking pada mahasiswa di dki jakarta. jurnal akuntansi, 11(1), 50–67. latif, c. s. (2017). minat individu terhadap penggunaan mobile banking: pendekatan technology acceptance model (tam). jurnal ilmiah mahasiswa feb, 5(2). mawarni, r. (2021). penerapan digital banking bank syariah sebagai upaya customer retantion pada masa covid-19. al iqtishod: jurnal pemikiran dan penelitian ekonomi islam, 9(2), 39–54. mohd thas thaker, m. a. bin, allah pitchay, a. bin, mohd thas thaker, h. bin, & amin, m. f. bin. (2019). factors influencing consumers’ adoption of islamic mobile banking services in malaysia. journal of islamic marketing, 10(4), 1037–1056. mongi, l., mananeke, l., & repi, a. (2014). kualitas produk, strategi promosi dan harga pengaruhnya terhadap keputusan pembelian kartu simpati telkomsel di kota manado. jurnal emba: jurnal riset ekonomi, manajemen, bisnis dan akuntansi, 1(4). ningrum, r., bacmid, s., & jalil, a. (2021). pengaruh manfaat, kepercayaan dan kemudahan penggunaan terhadap minat nasabah menggunakan mobile banking di bank mega syariah cabang palu. jurnal ilmu perbankan dan keuangan syariah, 3(1), 30–45. pasaribu, a. (2021). pengguna bsi mobile banking tembus 2,5 juta. antara news.com. https://www.antaranews.com berita/2297578/ pengguna-bsi-mobile-banking-tembus-25-juta peter, j. p., & olson, j. c. (2010). consumer behavior and marketing strategy (9th ed.). mcgraw-hill. putri, a., anriza witi, s. e., & seri, e. (2020). pengaruh perceived usefulness, perceived ease of use dan perceived credibiity terhadap keputusan menggunakan m-banking di masa covid-19. 1(2), 53–62. rahayu, i. s. (2016). minat nasabah menggunakan mobile banking dengan menggunakan kerangka technology acceptance model (tam) (studi kasus pt bank syariah mandiri cabang yogyakarta). jesi (jurnal ekonomi syariah indonesia), 5(2), 137–150. rita, r., & fitria, m. h. (2021). analisis faktor-faktor utaut dan trust terhadap behavioral intention pengguna bni mobile banking pada pekerja migran indonesia. jurnal ekonomi dan ekonomi syariah, 4(2), 926–939. romadhon, f., & fitri, a. (2020). analisis peluang dan tantangan penggunaan financial technology sebagai upaya optimalisasi potensi umkm (studi kasus umkm di gresik). jurnal technobiz, 3(1), 30–44. 52 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 29–52 setiawan, h. (2016). pengaruh kualitas layanan, persepsi nilai dan kepercayaan terhadap kepuasan dan loyalitas pengguna layanan mobile banking. jurnal keuangan dan perbankan, 20(3), 518–528. styarini, f., & riptiono, s. (2020). analisis pengaruh customer trust terhadap keputusan menggunakan mobile banking melalui perceived risk dan perceived usefulness sebagai variabel intervening. jurnal ilmiah mahasiswa manajemen, bisnis dan akuntansi (jimmba), 2(4), 670–680. sumadi, s. (2020). menakar dampak fenomena pandemi covid-19 terhadap perbankan syariah. jurnal hukum ekonomi syariah, 3(2), 145–162. susilawaty, l., & nicola, n. (2020). pengaruh layanan perbankan digital pada kepuasan nasabah perbankan. jurnal manajemen maranatha, 19(2), 179–190. tam, c., & oliveira, t. (2017). literature review of mobile banking and individual performance. international journal of bank marketing, 35(7), 1044–1067. thompson, r. l., higgins, c. a., & howell, j. m. (1991). personal computing: toward a conceptual model of utilization. mis quarterly, 15(1), 125–143. walfajri, m., & perwitasari, a. s. (2021). merger rampung, begini strategi bank syariah indonesia (bsi) di tahun 2021. kontan. co.id.https://keuangan.kontan.co.id/news/merger-rampungbegini-strategi-bank-syariah-indonesia-di-tahun-2021 wang, y., wang, y., lin, h., & tang, t. (2003). determinants of user acceptance of internet banking: an empirical study. international journal of service industry management, 14(5), 501–519. wardani, d. (2021). faktor-faktor pengaruh penggunaan mobile banking. jurnal sistem informasi bisnis (junsibi), 2(1), 15–32. wicaksono, m. b., triwardhani, d., & nastiti, h. (2021). pengaruh kepercayaan dan kualitas layanan terhadap keputusan penggunaan m-banking mandiri di jakarta selatan. konferensi riset nasional ekonomi manajemen dan akuntansi, 2(1), 1090–1103. wu, j., & song, s. (2021). older adults’ online shopping continuance intentions: applying the technology acceptance model and the theory of planned behavior. international journal of human– computer interaction, 37(10), 938–948. wu, l., li, j.-y., & fu, c.-y. (2011). the adoption of mobile healthcare by hospital’s professionals: an integrative perspective. decision support systems, 51(3), 587–596. 31 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 how to cite this article: el msiyah, c., & madkour, j. (2022). dynamics of the moroccan industry indices network before and during the covid-19 pandemic. international journal of banking and finance, 18(1), 31-50. https://doi. org/10.32890/ ijbf2023.18.1.2 dynamics of the moroccan industry indices network before and during the covid-19 pandemic 1cherif el msiyah & 2jaouad madkour 1national school of commerce and management moulay ismail university, morocco 2 faculty of law and economics abdelmalek essaadi university, morocco 1corresponding author: c.elmsiyah@umi.ac.ma received: 28/7/2021 revised: 18/10/2021 accepted: 21/10/2021 published: 5/1/2023 abstract this paper studies the topological properties of the dynamics of the industry indices network at the moroccan stock exchange by using network theory. the minimum spanning tree (mst) was constructed from the metric distances which had been calculated for the different pairs of industrial indices. the dynamics of the mst were analysed over the period 2013 to 2020 using the sliding window technique. the period studied was divided into the pre-pandemic covid-19 period and the pandemic covid-19 period. connectivity and centrality indicators were calculated to track the connectivity structure over time and to identify the positioning and the importance of the industry indices studied. the result of this study indicates that the network of industry indices was relatively stable during the pre-pandemic covid-19 period https://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 32 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 before observing a sudden rapprochement between industries when the covid-19 pandemic was officially announced. the formation of star-shaped networks was also observed. these networks were centred on the banking industry, essentially during the pandemic covid-19 period. the banking industry was also positioned at the centre of the moroccan industry indices network. keywords: industry indices network, minimum spanning tree, covid-19, network connectivity, network centrality. jel classification: c45, g01, g11, g23. introduction the use of network theory to translate the financial taxonomy of stock markets has been adopted by many research studies. in addition to some seminal works (mantegna, 1999; bonanno et al., 2000; bonanno et al., 2001; onnela et al., 2003), a large body of the literature has confirmed that stock markets behave like complex network systems; and it is possible to study the topological properties of financial networks, which have an associated significant economic taxonomy (tabak et al., 2010). for arthur et al. (1997), financial markets can be characterized as complex evolving systems, therefore it is useful to use the tools of complex network systems to analyse the stock market dynamics. according to de carvalho and gupta (2018), the network representation of stock market asset returns not only retains the most essential and important features of a large set of asset return co-movements, but also helps to simplify the challenges for identifying the co-movement dynamics between assets, including those related to the number of assets and risk factors that must be assessed simultaneously and the non-stationarity of movement in the asset dynamics. as the main tools of complex network theory, the clustering and filtering algorithms of the minimum spanning tree (mst) technique have been used to study the different topological and structural aspects of the stock markets, and also to illustrate their ability to transmit and exploit significant economic information. for example, tumminello et al (2010) demonstrated the formation of clusters at different levels of trees and the possibility to define the taxonomy of stocks within 33 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 the network by applying the different procedures for grouping and for filtering the correlation matrix of daily returns. tola et al. (2008) used clustering algorithms to improve the reliability of portfolios in terms of the ratio of the expected risk to the realised risk. they showed that clustering portfolio optimization methods often outperforms the markowitz or random matrix theory methods. khashanah and miao (2011) applied the mst to the entire financial system, consisting of typical markets (stocks, bonds, derivatives, currencies, and commodities) to study the changes in the structure of the financial system during the economic downturn. with regard to the application of mst techniques on the stock market, a great deal of work has been done in both developed and emerging countries. for example, there has been work on certain emerging markets; galazka (2011) used the mst to identify the stocks that strongly influenced the price dynamics of other stocks on the same polish stock market, using a portfolio of 252 stocks in 2007. majapa and gossel (2016) applied the mst approach to study the topological evolution before, during and after the 2008-2009 crisis, by constructing a network map of the top 100 companies listed on the johannesburg stock exchange. sinha and pan (2007) used data from 201 stocks over the period 1996–2006 to assess the strength of dominance of the largest companies in the indian economy. tabak et al. (2010) constructed the mst by using the weekly prices of the 47 stocks on the brazilian stock exchange from january 7, 2000 to february 29, 2008, and by the correlation matrix for a variety of stocks of different industry indices. moreover, the study showed that stocks tended to cluster by industry. situngkir and surya (2005) found that the indonesian stock market became stable during the period from 2000 to 2004, just after the economic shock of the currency crisis. they also noted the dominance of several stocks in certain industries. the world has already faced health crises (sars, mers and ebola, among others) that impacted the stock markets and business activities, but less forcefully than the covid-19 pandemic (baker et al., 2020). ashraf (2020) used the daily covid-19 confirmed cases and death cases, as well as the stock markets returns data from 64 countries and found that stock markets quickly reacted to the covid-19 pandemic. liu et al. (2020) found that asia experienced the most negative abnormal returns among the 21 leading stock market indices in most affected countries. haroon and rizvi (2020) found that the overwhelming panic 34 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 generated by the news outlets has led to the increase in the volatility in the equity markets. zhang et al. (2020) found that covid-19 has led to an increase in the global financial market risk aslam et al. (2020) examined the effects of covid-19 on 56 global stock indices by using a complex network method and they revealed a structural change in the form of node changes, a reduced connectivity and significant differences in the topological characteristics of the network. as in yang et al. (2014), the present study aims to apply the mst technique directly to industry indices in order to analyse the structural change of the stock market and to identify the key sectors of the moroccan economy. the study of the dynamic evolution of the industry indices network will then allow one to see the impact of the crisis on the industrial structure of the stock market. in the second section, of this paper, the mst model is presented and its usefulness shown in terms of simplifying the dependency structure between nodes. the next section presents the details of the study data and their uses for the construction of msts of the moroccan industry indices. in the fourth section, the discussion will turn to the use of some indicators to analyse the dynamics of msts. the last section will present the results and draw the relevant conclusions. methodology minimum spanning tree model the use of the mst to translate the financial taxonomy consists in studying the financial assets connections through the evolution of their stock prices. in the context of the present study, the focus is on the industrial taxonomy of the moroccan stock market. in this regard, the daily returns of the industry indices were used to calculate the correlation and distance matrix and to trace the mst for the moroccan stock industries. let be the number of industry indices studied. for an industry the rate of return on day is: is the closing price of on day 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 35 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 the pearson correlation coefficient between two industries and : for an industry represents the average of the log-returns over the period studied. for trading days studied . the correlation coefficients calculated for each pair of industry indices and form a symmetric correlation matrix of dimension and in which its diagonal elements equal to unity. in total, the correlation matrix contains correlation coefficients fulfilling the condition . the correlation coefficient of a pair of industry indices cannot be used as the distance between the two industries and because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of the industry indices network, a variable is needed that can function as distance, satisfying the three axioms of euclidean distance: the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007), and it is permitted to relate the distance of two variables (i.e. industry indices) to their correlation coefficients, as in equation (1): (1) the correlation coefficients are therefore, used to calculate the distances and to construct a matrix of the distances of the industry indices of the same dimension as the correlation matrix .the small distances in the matrix imply the high correlations between industry indices. from the distance matrix, it can be seen that there is a directionless and fully connected graph in which the nodes are industry indices and the links between them are measured by the distances the mst is then constructed by using direct links between the nodes of ,such that the sum of the distances of these 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 36 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 links is minimal. in this way, the mst is constructed progressively by linking all the nodes together with the smallest distances. it can be started by linking the two closest (most correlated) nodes, the next node is added to the mst, with the condition that the added distance must be the smallest and also the addition of that node should not close the loop with the nodes that are already added to the mst. kruskal’s algorithm is used for this purpose. a detailed approach to constructing msts using kruskal’s algorithm is presented in yang et al. (2014). the number of links between nodes is considerably reduced by using the minimum spanning tree (mst). the present study therefore, will go from the d distances in the distance matrix to only ,the shortest distances in the mst. data the data for the study were the prices of industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1,945 trading days as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014), and for analysis and smoothing purposes, the data were divided into windows of width the step between two consecutive windows was set at trading days the choice of was motivated by the fact that the period of the covid-19 pandemic studied was 185 days (between march 01, 2020 to november 30, 2020); this period corresponded roughly to the time between the official announcement of the pandemic and the announcement of the vaccine availability. the step of corresponded roughly to the average number of trading days per month. thus, this distribution of data consisted of studying the evolution of 89 msts, each corresponding to a window of 185 stock market prices with a 20-day shift (overlap) from one window to another. the last window studied corresponded to the entire period of the covid-19 pandemic (march 01, 2020 to november 30, 2020). this choice was arbitrary with the fact that the choice of window width had been a compromise between too noisy and too smoothed data for small and large window widths, respectively (onnela et al., 2003). similarly, for the overlap step, the choice of very small might cause concerns related to its inability to identify the episodic changes in the networks because the network 1 dynamics of the moroccan industry indices network before and during the covid-19 pandemic minimum spanning tree (mst) model let 𝑁𝑁𝑁𝑁 be the number of industry indices studied. for an industry 𝑖𝑖𝑖𝑖 (𝑖𝑖𝑖𝑖 = 1, . . . , 𝑁𝑁𝑁𝑁), the rate of return on day 𝑡𝑡𝑡𝑡 is: 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡−1) � 𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) is the closing price of 𝑖𝑖𝑖𝑖 on day 𝑡𝑡𝑡𝑡 . the pearson correlation coefficient between two industries 𝑖𝑖𝑖𝑖 and 𝑗𝑗𝑗𝑗: 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥������−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� ��𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤 2����−𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤��� 2��𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥 2����−𝑅𝑅𝑅𝑅𝚥𝚥𝚥𝚥��� 2� for an industry 𝑖𝑖𝑖𝑖 , 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� represents the average of the log-returns 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡) over the period studied. for 𝑇𝑇𝑇𝑇 trading days studied, 𝑅𝑅𝑅𝑅𝚤𝚤𝚤𝚤� = 1 𝑇𝑇𝑇𝑇 ∑ 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)𝑇𝑇𝑇𝑇𝑡𝑡𝑡𝑡=1 . 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝑡𝑡𝑡𝑡 of 𝑁𝑁𝑁𝑁 × 𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁(𝑁𝑁𝑁𝑁 − 1)/2 1 ≤ 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 1 � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖𝑖𝑖 = 𝑗𝑗𝑗𝑗 (positive definiteness) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖+𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 (triangle inequality) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = �2�1 − 𝜌𝜌𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖� (1) 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖. ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖,𝑖𝑖𝑖𝑖)∈mst the 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 4 of the pandemic and the announcement of the vaccine. the step of 𝑑𝑑𝑑𝑑 = 20 roughly corresponds to the average number of trading days per month. thus, this distribution of data consists of studying the evolution of 89 msts each corresponding to a window of 185 stock market prices with a 20-day shift (overlap) from one window to another. the last window studied corresponds to the entire period of the covid-19 pandemic (march 01, 2020 to november 30, 2020). this choice remains arbitrary in view of the fact that the choice of window width is a compromise between too noisy and too smoothed data for small and large window widths respectively (onnela et al., 2003). similarly for the overlap step, the choice of very small dw may raise concerns about the inability to identify episodic changes in the networks as the network carries a large proportion of the observations between successive windows (de carvalho & gupta, 2018). the msts constructed for different time windows are not independent of each other (two successive msts share 165 trading days over an analyzed mst period of 185 trading days), but form a series over time. therefore, this multitude of trees is interpreted as a sequence of evolutionary steps of a single dynamic asset tree (onnela et al., 2003) the 21 stock market industry indices studied are classified as in table 1. the msts corresponding to the two periods, the pre-pandemic period (january 23, 2013 to february 29, 2020) and the pandemic period (march 1, 2020 to april 30, 2020) are given in figure 1. table 1 industry indices included in the study and their tickers. notes: index and ticker symbol are given by the authors and the industry indices correspond to that of the casablanca stock exchange. data analysis a simple comparison of these two msts allows us to see that the industries are much more centered on the banking industry during the period of the covid-19 pandemic than during the period before the covid-19 pandemic. it is also pointed that the industries with which the banking industry is directly linked before the pandemic are also linked during the pandemic period; this is the case for “real estate participation and development” industry (link 3-14), “building and construction materials” industry (link 3-4), “food/ production” industry (link 3-1) and “telecommunication” industry (link 320). this result is consistent with that of musmeci et al. (2015) who, using filtered correlations between us stocks over the period from january 1997 to december 2012, lead to a pattern of clustering that changes radically with the outbreak of the financial crisis of 2007. the change in structure is analysed in more detail below using some connectivity and centrality indicators. 4 of the pandemic and the announcement of the vaccine. the step of 𝑑𝑑𝑑𝑑 = 20 roughly corresponds to the average number of trading days per month. thus, this distribution of data consists of studying the evolution of 89 msts each corresponding to a window of 185 stock market prices with a 20-day shift (overlap) from one window to another. the last window studied corresponds to the entire period of the covid-19 pandemic (march 01, 2020 to november 30, 2020). this choice remains arbitrary in view of the fact that the choice of window width is a compromise between too noisy and too smoothed data for small and large window widths respectively (onnela et al., 2003). similarly for the overlap step, the choice of very small dw may raise concerns about the inability to identify episodic changes in the networks as the network carries a large proportion of the observations between successive windows (de carvalho & gupta, 2018). the msts constructed for different time windows are not independent of each other (two successive msts share 165 trading days over an analyzed mst period of 185 trading days), but form a series over time. therefore, this multitude of trees is interpreted as a sequence of evolutionary steps of a single dynamic asset tree (onnela et al., 2003) the 21 stock market industry indices studied are classified as in table 1. the msts corresponding to the two periods, the pre-pandemic period (january 23, 2013 to february 29, 2020) and the pandemic period (march 1, 2020 to april 30, 2020) are given in figure 1. table 1 industry indices included in the study and their tickers. notes: index and ticker symbol are given by the authors and the industry indices correspond to that of the casablanca stock exchange. data analysis a simple comparison of these two msts allows us to see that the industries are much more centered on the banking industry during the period of the covid-19 pandemic than during the period before the covid-19 pandemic. it is also pointed that the industries with which the banking industry is directly linked before the pandemic are also linked during the pandemic period; this is the case for “real estate participation and development” industry (link 3-14), “building and construction materials” industry (link 3-4), “food/ production” industry (link 3-1) and “telecommunication” industry (link 320). this result is consistent with that of musmeci et al. (2015) who, using filtered correlations between us stocks over the period from january 1997 to december 2012, lead to a pattern of clustering that changes radically with the outbreak of the financial crisis of 2007. the change in structure is analysed in more detail below using some connectivity and centrality indicators. 37 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 carried a huge portion of the observations between the successive windows (de carvalho & gupta, 2018). the constructed msts for different time windows were not independent of each other (two successive msts share 165 trading days over an analyzed mst period of 185 trading days), but form a series over time. therefore, this multitude of trees was interpreted as a sequence of evolutionary steps of a single dynamic asset tree (onnela et al., 2003) the 21 stock market industry indices studied are classified as in table 1. the msts corresponding to the two periods, the pre-pandemic period (january 23, 2013 to february 29, 2020) and the pandemic period (march 1, 2020 to april 30, 2020) are as given in figure 1. table 1 industry indices included in the study and their tickers index (1) ticker symbol (1) industry indice (1) index (2) ticker symbol (2) industry indice (1) 1 fp food / production 12 h&ss hardware, software and services 2 insur insurance 13 mines mines 3 bank banks 14 re real estate participation and development 4 b&cm building and construction materials 15 o&g oil and gas 5 drink drinks 16 util utilities 6 chem chemistry 17 fc&of funding company and other financial activities 7 distr distributors 18 pc&h portfolio companies and holding companies 8 eee electronic and electrical equipment 19 f&p forestry and paper 9 pharm pharmaceutical industry 20 tcom telecommunication 10 eie engineering and industrial equipment 21 trans transport 11 l&h leisure and hotels notes. index and ticker symbols are given by the authors and the industry indices correspond to that of the casablanca stock exchange. 38 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 data analysis a simple comparison of the two msts allows us to see that the industries were much more centred on the banking industry during the period of the covid-19 pandemic than during the period before the covid-19 pandemic. it should also bevpointed out that the industries linked directly to the banking industry before the pandemic were also linked during the pandemic period; this is the case for “real estate participation and development” industry (link 3-14), “building and construction materials” industry (link 3-4), “food/ production” industry (link 3-1) and “telecommunication” industry (link 3-20). this result is consistent with that of musmeci et al. (2015) who, using filtered correlations between us stocks over the period from january 1997 to december 2012, showed a pattern of clustering that changes radically with the outbreak of the financial crisis of 2007. the change in structure is analysed in more detail below using some of the connectivity and centrality indicators. figure 1 minimum spanning trees (mst) of the industry indices of the covid-19 pandemic period (b) and the covid-19 pre-pandemic period (a) results and analysis network connectivity dynamics of industry indices in addition to the calculation of the overall distance of msts, the present study used some key indicators to identify the most important 6 results and analysis network connectivity dynamics of industry indices in addition to the calculation of the overall distance of msts, the present study used some key indicators to identify the most important connections between the industry indices, and to analyse their evolution over the period under review. for the 20 distances of the links belonging to mst, the size of mst is calculated by equation (2): (2) for 89 msts generated between january 23, 2013 and november 30, 2020, the last 9 msts (mst 81 to mst 89) had part of the period covered (185 trading days) during the covid-19 pandemic. specifically, the 81st mst covered 160 days before the pandemic period (before march 1, 2020) and 25 trading days during the pandemic period (from march 1). the 82nd mst covered 140 days before the pandemic period and 45 trading days during the pandemic period and so on until the last mst (89th) which covered the pandemic period from march 1, 2020 to november 30, 2020. it can be seen that the mst size (mst-distance) of the last 9 msts decreased on average from 25.03 before the pandemic period to 19.5 during the pandemic period. it was also noted that from the 81st mst and for only 25 covered trading days, there was practically the same level of decline. indeed, the size of the 81st mst, which covered the period from september 21, 2019 to march 25, 2020, showed a comparable decline with the msts covering a larger fraction of the pandemic period. this can be explained by the brutal reaction of the stock market from the first days of the official announcement of the pandemic. for a given mst and a knowledge of n, i.e., the number of industry indices studied, the mean and the standard deviation of the link distances in mst is as expressed in equation (3) and equation (4): (3) (4) 39 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 connections between the industry indices, and to analyse their evolution over the period under review. for the 20 distances of the links belonging to mst, the size of mst is calculated by equation (2): (2) for 89 msts generated between january 23, 2013 and november 30, 2020, the last 9 msts (mst 81 to mst 89) had part of the period covered (185 trading days) during the covid-19 pandemic. specifically, the 81st mst covered 160 days before the pandemic period (before march 1, 2020) and 25 trading days during the pandemic period (from march 1). the 82nd mst covered 140 days before the pandemic period and 45 trading days during the pandemic period and so on until the last mst (89th) which covered the pandemic period from march 1, 2020 to november 30, 2020. it can be seen that the mst size (mst-distance) of the last 9 msts decreased on average from 25.03 before the pandemic period to 19.5 during the pandemic period. it was also noted that from the 81st mst and for only 25 covered trading days, there was practically the same level of decline. indeed, the size of the 81st mst, which covered the period from september 21, 2019 to march 25, 2020, showed a comparable decline with the msts covering a larger fraction of the pandemic period. this can be explained by the brutal reaction of the stock market from the first days of the official announcement of the pandemic. for a given mst and a knowledge of n, i.e., the number of industry indices studied, the mean and the standard deviation of the link distances in mst is as expressed in equation (3) and equation (4): (3) (4) where is the number of links in the mst. 2 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑇𝑇𝑇𝑇 − 𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (2) 𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (3) 𝜎𝜎𝜎𝜎𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ �𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 − 𝑑𝑑𝑑𝑑� 2 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (4) 𝜎𝜎𝜎𝜎𝑡𝑡𝑡𝑡 = 1 𝑁𝑁𝑁𝑁−1 × �𝐸𝐸𝐸𝐸𝑡𝑡𝑡𝑡 ∩ 𝐸𝐸𝐸𝐸(𝑡𝑡𝑡𝑡−1)� (5) 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖 = ∑ 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑁𝑁𝑁𝑁𝑖𝑖𝑖𝑖=1 (6) 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) = 1 𝑁𝑁𝑁𝑁 ∑ ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝑁𝑁𝑁𝑁 𝑖𝑖𝑖𝑖=1 (7) 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝐵𝐵𝐵𝐵𝐶𝐶𝐶𝐶(𝑖𝑖𝑖𝑖) = ∑ 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗(𝑖𝑖𝑖𝑖) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗 𝑖𝑖𝑖𝑖≠𝑖𝑖𝑖𝑖 (8) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖) 5 figure 1 minimum spanning trees (mst) of the industry indices of the covid-19 pandemic period (b) and the covid-19 pre-pandemic period (a). results and analysis network connectivity dynamics of industry indices in in addition to the calculation of the overall distance of msts, we use some key indicators to identify the most important connections between the industry indices and analyse their evolution over the period under review. for the 20 distances 𝑑𝑑𝑖𝑖𝑖𝑖 of the links belonging to mst, the size of mst is calculated by: 𝑀𝑀𝑀𝑀𝑀𝑀 − 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = ∑ 𝑑𝑑𝑖𝑖𝑖𝑖𝑑𝑑𝑖𝑖𝑖𝑖∈mst (2) for 89 msts generated between january 23, 2013 and november 30, 2020, the last 9 msts (mst 81 to mst 89) have part of the period covered (185 trading days) during the covid-19 pandemic. specifically, the 81st mst covers 160 days before the pandemic period (before march 1, 2020) and 25 trading days during the pandemic period (from march 1). the 82nd mst covers 140 days before the pandemic period and 45 trading days during the pandemic period and so on until the last mst (89th) which covers exactly the pandemic period (march 1, 2020 november 30, 2020). it can be seen that the mst size (mst-distance) of the last 9 msts increases on average from 25.03 before the pandemic period to 19.5 for the msts corresponding to the pandemic period. we also note that from the 81st mst and for only 25 covered trading days, we have practically the same level of decline. this can be explained by the brutal reaction of the stock market from the first days of official announcement of the pandemic. 2 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑇𝑇𝑇𝑇 − 𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (2) 𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (3) 𝜎𝜎𝜎𝜎𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ �𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 − 𝑑𝑑𝑑𝑑� 2 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (4) 𝜎𝜎𝜎𝜎𝑡𝑡𝑡𝑡 = 1 𝑁𝑁𝑁𝑁−1 × �𝐸𝐸𝐸𝐸𝑡𝑡𝑡𝑡 ∩ 𝐸𝐸𝐸𝐸(𝑡𝑡𝑡𝑡−1)� (5) 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖 = ∑ 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑁𝑁𝑁𝑁𝑖𝑖𝑖𝑖=1 (6) 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) = 1 𝑁𝑁𝑁𝑁 ∑ ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝑁𝑁𝑁𝑁 𝑖𝑖𝑖𝑖=1 (7) 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝐵𝐵𝐵𝐵𝐶𝐶𝐶𝐶(𝑖𝑖𝑖𝑖) = ∑ 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗(𝑖𝑖𝑖𝑖) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗 𝑖𝑖𝑖𝑖≠𝑖𝑖𝑖𝑖 (8) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖) 2 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑇𝑇𝑇𝑇 − 𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (2) 𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (3) 𝜎𝜎𝜎𝜎𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ �𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 − 𝑑𝑑𝑑𝑑� 2 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (4) 𝜎𝜎𝜎𝜎𝑡𝑡𝑡𝑡 = 1 𝑁𝑁𝑁𝑁−1 × �𝐸𝐸𝐸𝐸𝑡𝑡𝑡𝑡 ∩ 𝐸𝐸𝐸𝐸(𝑡𝑡𝑡𝑡−1)� (5) 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖 = ∑ 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑁𝑁𝑁𝑁𝑖𝑖𝑖𝑖=1 (6) 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) = 1 𝑁𝑁𝑁𝑁 ∑ ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝑁𝑁𝑁𝑁 𝑖𝑖𝑖𝑖=1 (7) 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝐵𝐵𝐵𝐵𝐶𝐶𝐶𝐶(𝑖𝑖𝑖𝑖) = ∑ 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗(𝑖𝑖𝑖𝑖) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗 𝑖𝑖𝑖𝑖≠𝑖𝑖𝑖𝑖 (8) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖) 3 for an industry 𝑖𝑖 , 𝑅𝑅�̅�𝑖 represents the average of the log-returns 𝑅𝑅𝑖𝑖 (𝑡𝑡) over the period studied. for 𝑇𝑇 trading days studied, 𝑅𝑅�̅�𝑖 = 1 𝑇𝑇 ∑ 𝑅𝑅𝑖𝑖 (𝑡𝑡)𝑇𝑇𝑡𝑡=1 . the correlation coefficients 𝜌𝜌𝑖𝑖𝑖𝑖 calculated for each pair of industry indices 𝑖𝑖 and 𝑗𝑗 form a symmetric correlation matrix 𝐶𝐶𝑡𝑡 of dimension 𝑁𝑁 × 𝑁𝑁 and whose diagonal elements equal to unity. in total, the correlation matrix contains 𝑁𝑁(𝑁𝑁 − 1)/2 correlation coefficients fulfilling the condition 1 ≤ 𝜌𝜌𝑖𝑖𝑖𝑖 ≤ 1. the correlation coefficient of a pair of industry indices 𝜌𝜌𝑖𝑖𝑖𝑖 cannot be used as the distance between the two industries 𝑖𝑖 and 𝑗𝑗 because it does not fulfill the three axioms that define a metric (mantegna, 1999). indeed, to analyse the topological taxonomy of our industry indices network, we need a variable that can function as distance, satisfying the three axioms of euclidean distance: { 𝑑𝑑𝑖𝑖𝑖𝑖 = 0 if and only if 𝑖𝑖 = 𝑗𝑗 (positive definiteness) 𝑑𝑑𝑖𝑖𝑖𝑖 = 𝑑𝑑𝑖𝑖𝑖𝑖 (symmetry) 𝑑𝑑𝑖𝑖𝑖𝑖 ≤ 𝑑𝑑𝑖𝑖𝑖𝑖 +𝑑𝑑𝑖𝑖𝑖𝑖 (triangle inequality) the metric distance, introduced by mantegna (1999), is determined from the euclidean distance (coelho et al., 2007) and makes it possible to relate the distance of two variables (i.e. industry indices) to their correlation coefficients: 𝑑𝑑𝑖𝑖𝑖𝑖 = √2(1 − 𝜌𝜌𝑖𝑖𝑖𝑖 ) (1) the correlation coefficients are therefore used to calculate the distances and to construct a matrix of the distances of the industry indices 𝐷𝐷𝑡𝑡 of the same dimension as the 𝑁𝑁 × 𝑁𝑁 correlation matrix. small distances in the matrix imply high correlations between industry indices. from the distance matrix 𝐷𝐷𝑡𝑡, we can draw a directionless and fully connected graph 𝐺𝐺 in which the nodes are industry indices and the links between them are measured by the distances 𝑑𝑑𝑖𝑖𝑖𝑖 . the mst is then constructed using (𝑁𝑁 − 1) direct links between the 𝑁𝑁 nodes of 𝐺𝐺 such that the sum of the (𝑁𝑁 − 1) distances of these links ∑ 𝑑𝑑𝑖𝑖𝑖𝑖(𝑖𝑖,𝑖𝑖)∈mst is minimal. the mst is constructed in this way by progressively linking all the nodes together with the smallest distances; starting by linking the two closest (most correlated) nodes, the next node is added to the mst provided that the added distance is the smallest and that the addition of this node does not close the loop with the nodes already added to the mst. kruskal's algorithm is used for this purpose. a detailed approach to constructing msts using kruskal's algorithm is presented in (yang et al., 2014). the number of links between nodes is considerably reduced by using minimum spanning tree (mst). we therefore go from 𝑁𝑁 × (𝑁𝑁 − 1)/2 distances in the distance matrix 𝐷𝐷𝑡𝑡 to only (𝑁𝑁 − 1) shortest distances in the mst. data the data we will study concern the prices of 𝑁𝑁 = 21 industry indices of the moroccan stock exchange from january 23, 2013 to november 30, 2020, i.e. a total of 1945 trading days 𝑡𝑡 = 1,2, . . . , 1945. as in onnela et al. (2003), tabak et al. (2010) and yang et al. (2014) and for analysis and smoothing purposes, the data are divided into 𝑤𝑤 = 1.2 … .89 windows of width 𝑊𝑊 = 185 . the step between two consecutive windows is set at 20 trading days 𝑑𝑑𝑊𝑊 = 20 . the choice of 𝑊𝑊 = 185 is motivated by the fact that the period of the covid-19 pandemic studied is 185 days (between march 01, 2020 to november 30, 2020); this period corresponds roughly to the time between the official announcement 40 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 figure 2 distance mean (a) and distance standard deviation (b) of mst links the mean and standard deviation of the link distances in the mst are represented as a function of time for the different msts generated in figure 2. the average distance of links in the msts decreased from 1.25 during the pre-pandemic period to 0.97 for msts in the pandemic period. this represents a decrease of about 22 percent from one period to the next, the industry indices were thus 22 percent closer on average during the pandemic period (indicating more correlation between indices). the decrease in the average distance of links is in line with the results of several studies, which have shown an increase in the mean correlation (decrease in average distance of links) during market crashes (see onnela et al. (2003), drozdz et al. (2000), and majapa and gossel (2016). the result is also partly consistent with that of ang and chen (2010) who, in studying the correlations between us equities and the overall us market, found that correlations increased following downside moves, particularly for extreme downside moves than for upside moves. this asymmetry of correlation movement was also demonstrated by longin and solnik (2001) who also found that correlations were higher during bear markets than in bull markets. the decrease in the average distance of links was accompanied by a very high variation in the distance standard deviation of links between the two periods (figure 2b). the standard deviation increased from 0.057 on average during the pre-pandemic period to 0.204 during the 41 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 pandemic period (for a minimum of 0.63 and a maximum of 1.32); which was an increase of 255 per cent. according to khashanah and miao (2011), this clearly visible jump in the distance standard deviation could be explained by the intensification of systemic risk; correlations thus, appeared more uncertain due to unexpected and urgent responses to market events. as for the distance standard deviation of links, the variability was also remarkable. for the prepandemic period, the minimum distance was 1.13 and the maximum distance was 1.34, whereas during the pandemic period, the minimum distance was 0.63 and the maximum distance was 1.32. the combined effect was reflected in the strengthening of the links between industries (nodes), as can be seen in figure 1b. this may occur either because of the strengthening of existing links in mst, as is the case for some industries whose linkage has become closer to the banking industry, or because weaker links (of high distance) are deactivated from the mst. for a better visualization of the successive topological changes in msts and in particular, the links with the banking industry, the survival rate of direct links in msts were calculated (onnela et al., 2003). in a single step, the simple survival rate calculates the percentage of links (between each 2 industry indices) that survive from one mst to another. for two consecutive msts at times and , the simple survival rate is given by equation (5): (5) where is the set of mst links at time, is the intersection operator and gives the number of elements in the set. in the short term, it can be said that the high level of simple survival rates reflected a certain topological stability in the configuration of the industry indices network. indeed, compared to a total number of direct links an average survival rate of 67.56 per cent (13.51 links from one mst to another over the prepandemic period) was noted. this was in comparison to the average survival rate of 80 per cent (16 links from one mst to another) during the pandemic period. thus, one could conclude that the strengthening of the links between industries (nodes) in successive msts occurred, to a large extent, as a result of the strengthening of already existing links. 2 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑇𝑇𝑇𝑇 − 𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (2) 𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (3) 𝜎𝜎𝜎𝜎𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ �𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 − 𝑑𝑑𝑑𝑑� 2 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (4) 𝜎𝜎𝜎𝜎𝑡𝑡𝑡𝑡 = 1 𝑁𝑁𝑁𝑁−1 × �𝐸𝐸𝐸𝐸𝑡𝑡𝑡𝑡 ∩ 𝐸𝐸𝐸𝐸(𝑡𝑡𝑡𝑡−1)� (5) 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖 = ∑ 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑁𝑁𝑁𝑁𝑖𝑖𝑖𝑖=1 (6) 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) = 1 𝑁𝑁𝑁𝑁 ∑ ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝑁𝑁𝑁𝑁 𝑖𝑖𝑖𝑖=1 (7) 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝐵𝐵𝐵𝐵𝐶𝐶𝐶𝐶(𝑖𝑖𝑖𝑖) = ∑ 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗(𝑖𝑖𝑖𝑖) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗 𝑖𝑖𝑖𝑖≠𝑖𝑖𝑖𝑖 (8) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖) 7 as for the distance standard deviation of links, the variability is also remarkable. for the pre pandemic period, the minimum distance is 1.13 and the maximum distance is 1.34, whereas during the pandemic period, the minimum distance is 0.63 and the maximum distance is 1.32. the combined effect is reflected in the strengthening of the links between industries (nodes) as seen in figure 1b. this may occur either because of the strengthening of existing links in mst as is the case for some industries whose linkage has become closer to the banking industry, or because weaker links (of high distance) are deactivated from mst. to better visualize successive topological changes in msts and in particular the links with the banking industry, we calculate the survival rate of direct links in msts (onnela et al., 2003). in a single step, the simple survival rate is calculated for two consecutive msts at times 𝑡𝑡 and 𝑡𝑡 − 1 (𝑡𝑡 = 2, … , 89) by: 𝜎𝜎𝑡𝑡 = 1 𝑁𝑁−1 × |𝐸𝐸𝑡𝑡 ∩ 𝐸𝐸(𝑡𝑡−1)| (5) where 𝐸𝐸𝑡𝑡 is the set of mst links at time 𝑡𝑡, ∩ is the intersection operator and | . . . | gives the number of elements in the set. figure 3 number of times the nodes have survived both during the covide-19 pandemic period and before the covide-19 pandemic period. in the short term, it can be said that the high level of simple survival rates reflects a certain topological stability in the configuration of the industry indices network. indeed, compared to a total number of direct links (𝑁𝑁 − 1 = 20), we note an average survival rate of 67.56 per cent (13.51 links from one mst to another over the prepandemic period). this compares with an average survival rate of 80 per cent (16 links from one mst to another) during the pandemic period. it can be concluded from this that the strengthening of links between industries (nodes) in successive msts occurs, to a large extent, as a result of the strengthening of already existing links. in both the prepandemic and pandemic periods, the banking industry is the most surviving node (figure 3). during the prepandemic period, the banking industry survived 197 times, of which 46 times with the “telecommunication” node (link 3-4), 31 times with the “building and construction materials” node (link 3-4), 29 times with the “real estate participation and development” node (link 7 as for the distance standard deviation of links, the variability is also remarkable. for the pre pandemic period, the minimum distance is 1.13 and the maximum distance is 1.34, whereas during the pandemic period, the minimum distance is 0.63 and the maximum distance is 1.32. the combined effect is reflected in the strengthening of the links between industries (nodes) as seen in figure 1b. this may occur either because of the strengthening of existing links in mst as is the case for some industries whose linkage has become closer to the banking industry, or because weaker links (of high distance) are deactivated from mst. to better visualize successive topological changes in msts and in particular the links with the banking industry, we calculate the survival rate of direct links in msts (onnela et al., 2003). in a single step, the simple survival rate is calculated for two consecutive msts at times 𝑡𝑡 and 𝑡𝑡 − 1 (𝑡𝑡 = 2, … , 89) by: 𝜎𝜎𝑡𝑡 = 1 𝑁𝑁−1 × |𝐸𝐸𝑡𝑡 ∩ 𝐸𝐸(𝑡𝑡−1)| (5) where 𝐸𝐸𝑡𝑡 is the set of mst links at time 𝑡𝑡, ∩ is the intersection operator and | . . . | gives the number of elements in the set. figure 3 number of times the nodes have survived both during the covide-19 pandemic period and before the covide-19 pandemic period. in the short term, it can be said that the high level of simple survival rates reflects a certain topological stability in the configuration of the industry indices network. indeed, compared to a total number of direct links (𝑁𝑁 − 1 = 20), we note an average survival rate of 67.56 per cent (13.51 links from one mst to another over the prepandemic period). this compares with an average survival rate of 80 per cent (16 links from one mst to another) during the pandemic period. it can be concluded from this that the strengthening of links between industries (nodes) in successive msts occurs, to a large extent, as a result of the strengthening of already existing links. in both the prepandemic and pandemic periods, the banking industry is the most surviving node (figure 3). during the prepandemic period, the banking industry survived 197 times, of which 46 times with the “telecommunication” node (link 3-4), 31 times with the “building and construction materials” node (link 3-4), 29 times with the “real estate participation and development” node (link 7 as for the distance standard deviation of links, the variability is also remarkable. for the pre pandemic period, the minimum distance is 1.13 and the maximum distance is 1.34, whereas during the pandemic period, the minimum distance is 0.63 and the maximum distance is 1.32. the combined effect is reflected in the strengthening of the links between industries (nodes) as seen in figure 1b. this may occur either because of the strengthening of existing links in mst as is the case for some industries whose linkage has become closer to the banking industry, or because weaker links (of high distance) are deactivated from mst. to better visualize successive topological changes in msts and in particular the links with the banking industry, we calculate the survival rate of direct links in msts (onnela et al., 2003). in a single step, the simple survival rate is calculated for two consecutive msts at times 𝑡𝑡 and 𝑡𝑡 − 1 (𝑡𝑡 = 2, … , 89) by: 𝜎𝜎𝑡𝑡 = 1 𝑁𝑁−1 × |𝐸𝐸𝑡𝑡 ∩ 𝐸𝐸(𝑡𝑡−1)| (5) where 𝐸𝐸𝑡𝑡 is the set of mst links at time 𝑡𝑡, ∩ is the intersection operator and | . . . | gives the number of elements in the set. figure 3 number of times the nodes have survived both during the covide-19 pandemic period and before the covide-19 pandemic period. in the short term, it can be said that the high level of simple survival rates reflects a certain topological stability in the configuration of the industry indices network. indeed, compared to a total number of direct links (𝑁𝑁 − 1 = 20), we note an average survival rate of 67.56 per cent (13.51 links from one mst to another over the prepandemic period). this compares with an average survival rate of 80 per cent (16 links from one mst to another) during the pandemic period. it can be concluded from this that the strengthening of links between industries (nodes) in successive msts occurs, to a large extent, as a result of the strengthening of already existing links. in both the prepandemic and pandemic periods, the banking industry is the most surviving node (figure 3). during the prepandemic period, the banking industry survived 197 times, of which 46 times with the “telecommunication” node (link 3-4), 31 times with the “building and construction materials” node (link 3-4), 29 times with the “real estate participation and development” node (link 7 as for the distance standard deviation of links, the variability is also remarkable. for the pre pandemic period, the minimum distance is 1.13 and the maximum distance is 1.34, whereas during the pandemic period, the minimum distance is 0.63 and the maximum distance is 1.32. the combined effect is reflected in the strengthening of the links between industries (nodes) as seen in figure 1b. this may occur either because of the strengthening of existing links in mst as is the case for some industries whose linkage has become closer to the banking industry, or because weaker links (of high distance) are deactivated from mst. to better visualize successive topological changes in msts and in particular the links with the banking industry, we calculate the survival rate of direct links in msts (onnela et al., 2003). in a single step, the simple survival rate is calculated for two consecutive msts at times 𝑡𝑡 and 𝑡𝑡 − 1 (𝑡𝑡 = 2, … , 89) by: 𝜎𝜎𝑡𝑡 = 1 𝑁𝑁−1 × |𝐸𝐸𝑡𝑡 ∩ 𝐸𝐸(𝑡𝑡−1)| (5) where 𝐸𝐸𝑡𝑡 is the set of mst links at time 𝑡𝑡, ∩ is the intersection operator and | . . . | gives the number of elements in the set. figure 3 number of times the nodes have survived both during the covide-19 pandemic period and before the covide-19 pandemic period. in the short term, it can be said that the high level of simple survival rates reflects a certain topological stability in the configuration of the industry indices network. indeed, compared to a total number of direct links (𝑁𝑁 − 1 = 20), we note an average survival rate of 67.56 per cent (13.51 links from one mst to another over the prepandemic period). this compares with an average survival rate of 80 per cent (16 links from one mst to another) during the pandemic period. it can be concluded from this that the strengthening of links between industries (nodes) in successive msts occurs, to a large extent, as a result of the strengthening of already existing links. in both the prepandemic and pandemic periods, the banking industry is the most surviving node (figure 3). during the prepandemic period, the banking industry survived 197 times, of which 46 times with the “telecommunication” node (link 3-4), 31 times with the “building and construction materials” node (link 3-4), 29 times with the “real estate participation and development” node (link 7 as for the distance standard deviation of links, the variability is also remarkable. for the pre pandemic period, the minimum distance is 1.13 and the maximum distance is 1.34, whereas during the pandemic period, the minimum distance is 0.63 and the maximum distance is 1.32. the combined effect is reflected in the strengthening of the links between industries (nodes) as seen in figure 1b. this may occur either because of the strengthening of existing links in mst as is the case for some industries whose linkage has become closer to the banking industry, or because weaker links (of high distance) are deactivated from mst. to better visualize successive topological changes in msts and in particular the links with the banking industry, we calculate the survival rate of direct links in msts (onnela et al., 2003). in a single step, the simple survival rate is calculated for two consecutive msts at times 𝑡𝑡 and 𝑡𝑡 − 1 (𝑡𝑡 = 2, … , 89) by: 𝜎𝜎𝑡𝑡 = 1 𝑁𝑁−1 × |𝐸𝐸𝑡𝑡 ∩ 𝐸𝐸(𝑡𝑡−1)| (5) where 𝐸𝐸𝑡𝑡 is the set of mst links at time 𝑡𝑡, ∩ is the intersection operator and | . . . | gives the number of elements in the set. figure 3 number of times the nodes have survived both during the covide-19 pandemic period and before the covide-19 pandemic period. in the short term, it can be said that the high level of simple survival rates reflects a certain topological stability in the configuration of the industry indices network. indeed, compared to a total number of direct links (𝑁𝑁 − 1 = 20), we note an average survival rate of 67.56 per cent (13.51 links from one mst to another over the prepandemic period). this compares with an average survival rate of 80 per cent (16 links from one mst to another) during the pandemic period. it can be concluded from this that the strengthening of links between industries (nodes) in successive msts occurs, to a large extent, as a result of the strengthening of already existing links. in both the prepandemic and pandemic periods, the banking industry is the most surviving node (figure 3). during the prepandemic period, the banking industry survived 197 times, of which 46 times with the “telecommunication” node (link 3-4), 31 times with the “building and construction materials” node (link 3-4), 29 times with the “real estate participation and development” node (link 7 as for the distance standard deviation of links, the variability is also remarkable. for the pre pandemic period, the minimum distance is 1.13 and the maximum distance is 1.34, whereas during the pandemic period, the minimum distance is 0.63 and the maximum distance is 1.32. the combined effect is reflected in the strengthening of the links between industries (nodes) as seen in figure 1b. this may occur either because of the strengthening of existing links in mst as is the case for some industries whose linkage has become closer to the banking industry, or because weaker links (of high distance) are deactivated from mst. to better visualize successive topological changes in msts and in particular the links with the banking industry, we calculate the survival rate of direct links in msts (onnela et al., 2003). in a single step, the simple survival rate is calculated for two consecutive msts at times 𝑡𝑡 and 𝑡𝑡 − 1 (𝑡𝑡 = 2, … , 89) by: 𝜎𝜎𝑡𝑡 = 1 𝑁𝑁−1 × |𝐸𝐸𝑡𝑡 ∩ 𝐸𝐸(𝑡𝑡−1)| (5) where 𝐸𝐸𝑡𝑡 is the set of mst links at time 𝑡𝑡, ∩ is the intersection operator and | . . . | gives the number of elements in the set. figure 3 number of times the nodes have survived both during the covide-19 pandemic period and before the covide-19 pandemic period. in the short term, it can be said that the high level of simple survival rates reflects a certain topological stability in the configuration of the industry indices network. indeed, compared to a total number of direct links (𝑁𝑁 − 1 = 20), we note an average survival rate of 67.56 per cent (13.51 links from one mst to another over the prepandemic period). this compares with an average survival rate of 80 per cent (16 links from one mst to another) during the pandemic period. it can be concluded from this that the strengthening of links between industries (nodes) in successive msts occurs, to a large extent, as a result of the strengthening of already existing links. in both the prepandemic and pandemic periods, the banking industry is the most surviving node (figure 3). during the prepandemic period, the banking industry survived 197 times, of which 46 times with the “telecommunication” node (link 3-4), 31 times with the “building and construction materials” node (link 3-4), 29 times with the “real estate participation and development” node (link 7 as for the distance standard deviation of links, the variability is also remarkable. for the pre pandemic period, the minimum distance is 1.13 and the maximum distance is 1.34, whereas during the pandemic period, the minimum distance is 0.63 and the maximum distance is 1.32. the combined effect is reflected in the strengthening of the links between industries (nodes) as seen in figure 1b. this may occur either because of the strengthening of existing links in mst as is the case for some industries whose linkage has become closer to the banking industry, or because weaker links (of high distance) are deactivated from mst. to better visualize successive topological changes in msts and in particular the links with the banking industry, we calculate the survival rate of direct links in msts (onnela et al., 2003). in a single step, the simple survival rate is calculated for two consecutive msts at times 𝑡𝑡 and 𝑡𝑡 − 1 (𝑡𝑡 = 2, … , 89) by: 𝜎𝜎𝑡𝑡 = 1 𝑁𝑁−1 × |𝐸𝐸𝑡𝑡 ∩ 𝐸𝐸(𝑡𝑡−1)| (5) where 𝐸𝐸𝑡𝑡 is the set of mst links at time 𝑡𝑡, ∩ is the intersection operator and | . . . | gives the number of elements in the set. figure 3 number of times the nodes have survived both during the covide-19 pandemic period and before the covide-19 pandemic period. in the short term, it can be said that the high level of simple survival rates reflects a certain topological stability in the configuration of the industry indices network. indeed, compared to a total number of direct links (𝑁𝑁 − 1 = 20), we note an average survival rate of 67.56 per cent (13.51 links from one mst to another over the prepandemic period). this compares with an average survival rate of 80 per cent (16 links from one mst to another) during the pandemic period. it can be concluded from this that the strengthening of links between industries (nodes) in successive msts occurs, to a large extent, as a result of the strengthening of already existing links. in both the prepandemic and pandemic periods, the banking industry is the most surviving node (figure 3). during the prepandemic period, the banking industry survived 197 times, of which 46 times with the “telecommunication” node (link 3-4), 31 times with the “building and construction materials” node (link 3-4), 29 times with the “real estate participation and development” node (link 42 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 in both the prepandemic and pandemic periods, the banking industry was the most surviving node (see figure 3). figure 3 number of times the nodes have survived both during the covide-19 pandemic period and before the covide-19 pandemic period figure 4 number of times the nodes have survived together with bank node, both during the covide-19 pandemic period and before the covide-19 pandemic period during the prepandemic period, the banking industry survived 197 times, of which 46 times with the “telecommunication” node (link 3-4), 31 times with the “building and construction materials” node (link 7 as for the distance standard deviation of links, the variability is also remarkable. for the pre pandemic period, the minimum distance is 1.13 and the maximum distance is 1.34, whereas during the pandemic period, the minimum distance is 0.63 and the maximum distance is 1.32. the combined effect is reflected in the strengthening of the links between industries (nodes) as seen in figure 1b. this may occur either because of the strengthening of existing links in mst as is the case for some industries whose linkage has become closer to the banking industry, or because weaker links (of high distance) are deactivated from mst. to better visualize successive topological changes in msts and in particular the links with the banking industry, we calculate the survival rate of direct links in msts (onnela et al., 2003). in a single step, the simple survival rate is calculated for two consecutive msts at times 𝑡𝑡 and 𝑡𝑡 − 1 (𝑡𝑡 = 2, … , 89) by: 𝜎𝜎𝑡𝑡 = 1 𝑁𝑁−1 × |𝐸𝐸𝑡𝑡 ∩ 𝐸𝐸(𝑡𝑡−1)| (5) where 𝐸𝐸𝑡𝑡 is the set of mst links at time 𝑡𝑡, ∩ is the intersection operator and | . . . | gives the number of elements in the set. figure 3 number of times the nodes have survived both during the covide-19 pandemic period and before the covide-19 pandemic period. in the short term, it can be said that the high level of simple survival rates reflects a certain topological stability in the configuration of the industry indices network. indeed, compared to a total number of direct links (𝑁𝑁 − 1 = 20), we note an average survival rate of 67.56 per cent (13.51 links from one mst to another over the prepandemic period). this compares with an average survival rate of 80 per cent (16 links from one mst to another) during the pandemic period. it can be concluded from this that the strengthening of links between industries (nodes) in successive msts occurs, to a large extent, as a result of the strengthening of already existing links. in both the prepandemic and pandemic periods, the banking industry is the most surviving node (figure 3). during the prepandemic period, the banking industry survived 197 times, of which 46 times with the “telecommunication” node (link 3-4), 31 times with the “building and construction materials” node (link 3-4), 29 times with the “real estate participation and development” node (link 8 3-14). during the pandemic period (for the last 9 msts), the banking industry survived 77 times, including 8 times (out of the 8 msts) with 6 nodes including the nodes “telecommunication”, “building and construction materials” and “real estate participation and development” (figure 4). figure 4 number of times the nodes have survived with bank both during the covide-19 pandemic period and before the covide-19 pandemic period. dynamic analysis of network centrality in addition to the connectivity indicators discussed above, centrality indicators contain additional very useful information about the structure of the network topology. the node degree is the basic indicator for measuring clustering and centrality in a network, it is equal to the number of direct links connecting a given node to other nodes in the network. for 𝑁𝑁 nodes of mst, the degree of node 𝑗𝑗 is given by: 𝐷𝐷𝐷𝐷𝐷𝐷𝑗𝑗 = ∑ 𝐼𝐼𝑗𝑗𝑗𝑗𝑁𝑁𝑗𝑗=1 (6) where 𝐼𝐼𝑗𝑗𝑗𝑗 denotes the indicator function that gives 1 if node 𝑗𝑗 is directly linked to 𝑘𝑘, and 0 otherwise. table 2 summarizes the degrees of the main nodes over the entire prepandemic and the entire pandemic periods (figure 1). it can be seen that the banking industry dominates the dynamics of the industry indices mainly during the pandemic period. the other industries have a lower weight in the network. the case of the “oil and gas” industry shows a disconnected evolution possibly due to the deep price drop. 43 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 3-4), 29 times with the “real estate participation and development” node (link 3-14). during the pandemic period (for the last 9 msts), the banking industry survived 77 times, including 8 times (out of the 8 msts) with 6 nodes including the nodes “telecommunication”, “building and construction materials” and “real estate participation and development” (see figure 4). dynamic analysis of network centrality in addition to the connectivity indicators discussed above, centrality indicators contained very useful additional information about the structure of the network topology. the node degree is the basic indicator for measuring the clustering and the centrality in a network, it is equal to the number of direct links connecting a given node to other nodes in the network. for nodes of the mst, the degree of node is given by the equation (6): (6) where denotes the indicator function that gives in a condition that the node is directly linked to and otherwise. table 2 summarizes the degrees of the main nodes over the entire pre pandemic and the entire pandemic periods (see figure 1). it can be seen that the banking industry dominates the dynamics of the industry indices mainly during the pandemic period. the other industries have a lower weight in the network. the case of the “oil and gas” industry shows a disconnected evolution possibly due to the deep price drop. table 2 degrees of the main nodes bank fp o&g b&cm re january 23, 2013 to february 29, 2020 4 3 4 3 2 march 1, 2020 to november 30, 2020 12 3 1 1 2 notes. table 2 uses the tickers defined in table 1 dynamic monitoring of the time sequence of msts allows us to identify the central node for the 89 msts and the degree of centrality 8 3-14). during the pandemic period (for the last 9 msts), the banking industry survived 77 times, including 8 times (out of the 8 msts) with 6 nodes including the nodes “telecommunication”, “building and construction materials” and “real estate participation and development” (figure 4). figure 4 number of times the nodes have survived with bank both during the covide-19 pandemic period and before the covide-19 pandemic period. dynamic analysis of network centrality in addition to the connectivity indicators discussed above, centrality indicators contain additional very useful information about the structure of the network topology. the node degree is the basic indicator for measuring clustering and centrality in a network, it is equal to the number of direct links connecting a given node to other nodes in the network. for 𝑁𝑁 nodes of mst, the degree of node 𝑗𝑗 is given by: 𝐷𝐷𝐷𝐷𝐷𝐷𝑗𝑗 = ∑ 𝐼𝐼𝑗𝑗𝑗𝑗𝑁𝑁𝑗𝑗=1 (6) where 𝐼𝐼𝑗𝑗𝑗𝑗 denotes the indicator function that gives 1 if node 𝑗𝑗 is directly linked to 𝑘𝑘, and 0 otherwise. table 2 summarizes the degrees of the main nodes over the entire prepandemic and the entire pandemic periods (figure 1). it can be seen that the banking industry dominates the dynamics of the industry indices mainly during the pandemic period. the other industries have a lower weight in the network. the case of the “oil and gas” industry shows a disconnected evolution possibly due to the deep price drop. 8 3-14). during the pandemic period (for the last 9 msts), the banking industry survived 77 times, including 8 times (out of the 8 msts) with 6 nodes including the nodes “telecommunication”, “building and construction materials” and “real estate participation and development” (figure 4). figure 4 number of times the nodes have survived with bank both during the covide-19 pandemic period and before the covide-19 pandemic period. dynamic analysis of network centrality in addition to the connectivity indicators discussed above, centrality indicators contain additional very useful information about the structure of the network topology. the node degree is the basic indicator for measuring clustering and centrality in a network, it is equal to the number of direct links connecting a given node to other nodes in the network. for 𝑁𝑁 nodes of mst, the degree of node 𝑗𝑗 is given by: 𝐷𝐷𝐷𝐷𝐷𝐷𝑗𝑗 = ∑ 𝐼𝐼𝑗𝑗𝑗𝑗𝑁𝑁𝑗𝑗=1 (6) where 𝐼𝐼𝑗𝑗𝑗𝑗 denotes the indicator function that gives 1 if node 𝑗𝑗 is directly linked to 𝑘𝑘, and 0 otherwise. table 2 summarizes the degrees of the main nodes over the entire prepandemic and the entire pandemic periods (figure 1). it can be seen that the banking industry dominates the dynamics of the industry indices mainly during the pandemic period. the other industries have a lower weight in the network. the case of the “oil and gas” industry shows a disconnected evolution possibly due to the deep price drop. 2 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑇𝑇𝑇𝑇 − 𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (2) 𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (3) 𝜎𝜎𝜎𝜎𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ �𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 − 𝑑𝑑𝑑𝑑� 2 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (4) 𝜎𝜎𝜎𝜎𝑡𝑡𝑡𝑡 = 1 𝑁𝑁𝑁𝑁−1 × �𝐸𝐸𝐸𝐸𝑡𝑡𝑡𝑡 ∩ 𝐸𝐸𝐸𝐸(𝑡𝑡𝑡𝑡−1)� (5) 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖 = ∑ 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑁𝑁𝑁𝑁𝑖𝑖𝑖𝑖=1 (6) 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) = 1 𝑁𝑁𝑁𝑁 ∑ ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝑁𝑁𝑁𝑁 𝑖𝑖𝑖𝑖=1 (7) 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝐵𝐵𝐵𝐵𝐶𝐶𝐶𝐶(𝑖𝑖𝑖𝑖) = ∑ 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗(𝑖𝑖𝑖𝑖) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗 𝑖𝑖𝑖𝑖≠𝑖𝑖𝑖𝑖 (8) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖) 8 3-14). during the pandemic period (for the last 9 msts), the banking industry survived 77 times, including 8 times (out of the 8 msts) with 6 nodes including the nodes “telecommunication”, “building and construction materials” and “real estate participation and development” (figure 4). figure 4 number of times the nodes have survived with bank both during the covide-19 pandemic period and before the covide-19 pandemic period. dynamic analysis of network centrality in addition to the connectivity indicators discussed above, centrality indicators contain additional very useful information about the structure of the network topology. the node degree is the basic indicator for measuring clustering and centrality in a network, it is equal to the number of direct links connecting a given node to other nodes in the network. for 𝑁𝑁 nodes of mst, the degree of node 𝑗𝑗 is given by: 𝐷𝐷𝐷𝐷𝐷𝐷𝑗𝑗 = ∑ 𝐼𝐼𝑗𝑗𝑗𝑗𝑁𝑁𝑗𝑗=1 (6) where 𝐼𝐼𝑗𝑗𝑗𝑗 denotes the indicator function that gives 1 if node 𝑗𝑗 is directly linked to 𝑘𝑘, and 0 otherwise. table 2 summarizes the degrees of the main nodes over the entire prepandemic and the entire pandemic periods (figure 1). it can be seen that the banking industry dominates the dynamics of the industry indices mainly during the pandemic period. the other industries have a lower weight in the network. the case of the “oil and gas” industry shows a disconnected evolution possibly due to the deep price drop. 8 3-14). during the pandemic period (for the last 9 msts), the banking industry survived 77 times, including 8 times (out of the 8 msts) with 6 nodes including the nodes “telecommunication”, “building and construction materials” and “real estate participation and development” (figure 4). figure 4 number of times the nodes have survived with bank both during the covide-19 pandemic period and before the covide-19 pandemic period. dynamic analysis of network centrality in addition to the connectivity indicators discussed above, centrality indicators contain additional very useful information about the structure of the network topology. the node degree is the basic indicator for measuring clustering and centrality in a network, it is equal to the number of direct links connecting a given node to other nodes in the network. for 𝑁𝑁 nodes of mst, the degree of node 𝑗𝑗 is given by: 𝐷𝐷𝐷𝐷𝐷𝐷𝑗𝑗 = ∑ 𝐼𝐼𝑗𝑗𝑗𝑗𝑁𝑁𝑗𝑗=1 (6) where 𝐼𝐼𝑗𝑗𝑗𝑗 denotes the indicator function that gives 1 if node 𝑗𝑗 is directly linked to 𝑘𝑘, and 0 otherwise. table 2 summarizes the degrees of the main nodes over the entire prepandemic and the entire pandemic periods (figure 1). it can be seen that the banking industry dominates the dynamics of the industry indices mainly during the pandemic period. the other industries have a lower weight in the network. the case of the “oil and gas” industry shows a disconnected evolution possibly due to the deep price drop. 8 3-14). during the pandemic period (for the last 9 msts), the banking industry survived 77 times, including 8 times (out of the 8 msts) with 6 nodes including the nodes “telecommunication”, “building and construction materials” and “real estate participation and development” (figure 4). figure 4 number of times the nodes have survived with bank both during the covide-19 pandemic period and before the covide-19 pandemic period. dynamic analysis of network centrality in addition to the connectivity indicators discussed above, centrality indicators contain additional very useful information about the structure of the network topology. the node degree is the basic indicator for measuring clustering and centrality in a network, it is equal to the number of direct links connecting a given node to other nodes in the network. for 𝑁𝑁 nodes of mst, the degree of node 𝑗𝑗 is given by: 𝐷𝐷𝐷𝐷𝐷𝐷𝑗𝑗 = ∑ 𝐼𝐼𝑗𝑗𝑗𝑗𝑁𝑁𝑗𝑗=1 (6) where 𝐼𝐼𝑗𝑗𝑗𝑗 denotes the indicator function that gives 1 if node 𝑗𝑗 is directly linked to 𝑘𝑘, and 0 otherwise. table 2 summarizes the degrees of the main nodes over the entire prepandemic and the entire pandemic periods (figure 1). it can be seen that the banking industry dominates the dynamics of the industry indices mainly during the pandemic period. the other industries have a lower weight in the network. the case of the “oil and gas” industry shows a disconnected evolution possibly due to the deep price drop. 8 3-14). during the pandemic period (for the last 9 msts), the banking industry survived 77 times, including 8 times (out of the 8 msts) with 6 nodes including the nodes “telecommunication”, “building and construction materials” and “real estate participation and development” (figure 4). figure 4 number of times the nodes have survived with bank both during the covide-19 pandemic period and before the covide-19 pandemic period. dynamic analysis of network centrality in addition to the connectivity indicators discussed above, centrality indicators contain additional very useful information about the structure of the network topology. the node degree is the basic indicator for measuring clustering and centrality in a network, it is equal to the number of direct links connecting a given node to other nodes in the network. for 𝑁𝑁 nodes of mst, the degree of node 𝑗𝑗 is given by: 𝐷𝐷𝐷𝐷𝐷𝐷𝑗𝑗 = ∑ 𝐼𝐼𝑗𝑗𝑗𝑗𝑁𝑁𝑗𝑗=1 (6) where 𝐼𝐼𝑗𝑗𝑗𝑗 denotes the indicator function that gives 1 if node 𝑗𝑗 is directly linked to 𝑘𝑘, and 0 otherwise. table 2 summarizes the degrees of the main nodes over the entire prepandemic and the entire pandemic periods (figure 1). it can be seen that the banking industry dominates the dynamics of the industry indices mainly during the pandemic period. the other industries have a lower weight in the network. the case of the “oil and gas” industry shows a disconnected evolution possibly due to the deep price drop. 8 3-14). during the pandemic period (for the last 9 msts), the banking industry survived 77 times, including 8 times (out of the 8 msts) with 6 nodes including the nodes “telecommunication”, “building and construction materials” and “real estate participation and development” (figure 4). figure 4 number of times the nodes have survived with bank both during the covide-19 pandemic period and before the covide-19 pandemic period. dynamic analysis of network centrality in addition to the connectivity indicators discussed above, centrality indicators contain additional very useful information about the structure of the network topology. the node degree is the basic indicator for measuring clustering and centrality in a network, it is equal to the number of direct links connecting a given node to other nodes in the network. for 𝑁𝑁 nodes of mst, the degree of node 𝑗𝑗 is given by: 𝐷𝐷𝐷𝐷𝐷𝐷𝑗𝑗 = ∑ 𝐼𝐼𝑗𝑗𝑗𝑗𝑁𝑁𝑗𝑗=1 (6) where 𝐼𝐼𝑗𝑗𝑗𝑗 denotes the indicator function that gives 1 if node 𝑗𝑗 is directly linked to 𝑘𝑘, and 0 otherwise. table 2 summarizes the degrees of the main nodes over the entire prepandemic and the entire pandemic periods (figure 1). it can be seen that the banking industry dominates the dynamics of the industry indices mainly during the pandemic period. the other industries have a lower weight in the network. the case of the “oil and gas” industry shows a disconnected evolution possibly due to the deep price drop. 44 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 of the different industries. from gilmorea et al. (2008), the central node was defined as the node (industry) with the greatest number of direct links (the maximum degree of node). in the cases where the two nodes have the same number of direct links, the node with the lowest sum of distances of its links is selected. this is the same principle used in the construction of msts because the short connections link the node more closely to its neighbourhood than long connections (onnela et al., 2003). figure 5 msts central nodes during the whole period studied in figure 5, the banking industry dominated the share of central nodes of msts (28.09 percent), followed by “building and construction materials” industry and “real estate participation and development” industry with shares equal to 11.236 per cent. these three industries together constituted 50 percent of the central nodes. during the pandemic period, the banking industry monopolized almost half of the direct links (9 to 12, out of 21) while the average for the prepandemic period was 4,737 direct links. thus, the banking industry positioned itself as the central node for all msts during the pandemic period. the network connecting the bank node during the pandemic period demonstrates the formation of a very starlike network structure (see figure 1b) compared to the shape of the network before the pandemic period (see figure 1a). the formation of star-shaped networks was directly related to a higher maximum node degree (de carvalho & gupta, 2018). 9 table 2 degrees of the main nodes. notes: this table uses the tickers defined in table 1. dynamic monitoring of the time sequence of msts allows us to identify the central node for the 89 msts and the degree of centrality of the different industries. from gilmorea et al. (2008), we define the central node as the node (industry) with the greatest number of direct links (the maximum degree of node). in cases where two nodes have the same number of direct links, the node with the lowest sum of distances of its links is selected. this is the same principle used in the construction of msts because short connections link the node more closely to its neighbourhood than long connections (onnela et al., 2003). figure 5 msts central nodes. during the whole period studied in figure 5, the banking industry dominates the share of central nodes of msts with (28.09 per cent), followed by “building and construction materials” industry and “real estate participation and development” industry with shares equal to 11.236 per cent. these three industries together constitute 50 per cent of the central nodes. during the pandemic period, the banking industry monopolizes almost half of the direct links (9 to 12 out of 21) while the average for the prepandemic period is 4,737 direct links. the banking industry thus positions itself as the central node for all msts during the pandemic period. the degree tracking of the “bank” node demonstrates the formation of a very star-like network structure (figure 1b) compared to the shape of the network before the pandemic period (figure 1a). the formation of star-shaped networks is directly related to a higher maximum node degree (de carvalho & gupta, 2018). 45 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 for a more global measure of centrality, onnela et al. (2003) has proposed to calculate mean occupation layer (mol), which is the average number of links crossed by the centre of the tree (central node) to reach the various other nodes. this is expressed as equation (7). (7) where is the mol for the central node of mst during period and denotes the level of node (other than the central node) in relation to the central node. this level is 1 if is linked to by a direct link, 2 if passes through another node before reaching and so on. the average mol during the pandemic period dropped significantly to 1.76 compared to a pre-pandemic average of 2.88. moreover, the average distance of mol links also decreased by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry became very close to other industries during the pandemic period, both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and confirmed in several studies such as in onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node, the betweenness centrality is given by equation (8): (8) where is the total number of shortest paths from node to node and is the number of shortest paths from node to node through the usefulness of the betweenness centrality is that some less connected nodes can be very central, i.e.,the fact that they link different parts of the network together. again, the banking sector was the most central (see figure 6); it had 36.85 per cent of the shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of the centrality and the systemic risk that may result. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 2 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑇𝑇𝑇𝑇 − 𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (2) 𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (3) 𝜎𝜎𝜎𝜎𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ �𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 − 𝑑𝑑𝑑𝑑� 2 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (4) 𝜎𝜎𝜎𝜎𝑡𝑡𝑡𝑡 = 1 𝑁𝑁𝑁𝑁−1 × �𝐸𝐸𝐸𝐸𝑡𝑡𝑡𝑡 ∩ 𝐸𝐸𝐸𝐸(𝑡𝑡𝑡𝑡−1)� (5) 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖 = ∑ 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑁𝑁𝑁𝑁𝑖𝑖𝑖𝑖=1 (6) 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) = 1 𝑁𝑁𝑁𝑁 ∑ ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝑁𝑁𝑁𝑁 𝑖𝑖𝑖𝑖=1 (7) 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝐵𝐵𝐵𝐵𝐶𝐶𝐶𝐶(𝑖𝑖𝑖𝑖) = ∑ 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗(𝑖𝑖𝑖𝑖) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗 𝑖𝑖𝑖𝑖≠𝑖𝑖𝑖𝑖 (8) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖) 2 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑇𝑇𝑇𝑇 − 𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (2) 𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (3) 𝜎𝜎𝜎𝜎𝑑𝑑𝑑𝑑 = 1 𝑁𝑁𝑁𝑁−1 ∑ �𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 − 𝑑𝑑𝑑𝑑� 2 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖∈mst (4) 𝜎𝜎𝜎𝜎𝑡𝑡𝑡𝑡 = 1 𝑁𝑁𝑁𝑁−1 × �𝐸𝐸𝐸𝐸𝑡𝑡𝑡𝑡 ∩ 𝐸𝐸𝐸𝐸(𝑡𝑡𝑡𝑡−1)� (5) 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖𝑖𝑖 = ∑ 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑁𝑁𝑁𝑁𝑖𝑖𝑖𝑖=1 (6) 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) = 1 𝑁𝑁𝑁𝑁 ∑ ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝑁𝑁𝑁𝑁 𝑖𝑖𝑖𝑖=1 (7) 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀(𝐷𝐷𝐷𝐷𝑙𝑙𝑙𝑙𝑡𝑡𝑡𝑡) ℒ[𝑙𝑙𝑙𝑙𝑖𝑖𝑖𝑖(𝑡𝑡𝑡𝑡)] 𝐵𝐵𝐵𝐵𝐶𝐶𝐶𝐶(𝑖𝑖𝑖𝑖) = ∑ 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗(𝑖𝑖𝑖𝑖) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑗𝑗𝑗𝑗 𝑖𝑖𝑖𝑖≠𝑖𝑖𝑖𝑖 (8) 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝜎𝜎𝜎𝜎𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖(𝑖𝑖𝑖𝑖) 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 46 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 figure 6 betweenness centrality of mst nodes this central weight of the banking industry underscored its importance in the moroccan economy, and even at the african level where moroccan banks are among the most solid and most visible. things may be different in other countries. yang et al. (2014) found that durable consumer goods were at the heart of the chinese stock market. coelho et al. (2007) found that the financial industry was located as a central node in the london stock exchange. on the brazilian stock market, tabak et al. (2010) showed that telecommunications was positioned at a central location, followed by the materials industry and the finance industry. conclusion this study has provided useful information on how different industry indices have reacted together in light of the pandemic events like that of covid-19. this reaction was reflected in a sudden rapprochement between industries from the beginning of the covid-19 pandemic. the construction of the minimum spanning trees (mst) for the pandemic and prepandemic periods effectively shows a considerable shrinkage of the tree from the first days of the announcement of pandemic measures. the study of the dynamic evolution of the moroccan industry indices network shows that the network was relatively stable during the pre 10 for a more global measure of centrality, onnela et al. (2003) propose to calculate mean occupation layer (mol) which is the average number of links crossed by the center of the tree (central node) to reach the various other nodes. 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) = 1 𝑁𝑁 ∑ ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)]𝑁𝑁𝑖𝑖=1 (7) where 𝑀𝑀𝑀𝑀𝑀𝑀(𝑐𝑐𝑐𝑐𝑡𝑡 ) is the mol for the central node of mst during period 𝑡𝑡 and ℒ[𝑐𝑐𝑖𝑖 (𝑡𝑡)] denotes the level of node 𝑐𝑐𝑖𝑖 (other than the central node) in relation to the central node. this level is 1 if 𝑐𝑐𝑖𝑖 is linked to 𝑐𝑐𝑐𝑐 by a direct link, 2 if 𝑐𝑐𝑖𝑖 passes through another node before reaching 𝑐𝑐𝑐𝑐 and so on. the average mol during the pandemic period drops significantly to 1.76 compared to a prepandemic average of 2.88. moreover, the average distance of mol links also decreases by almost 28 per cent (from 1.23 to 0.89). this shows that the banking industry becomes very close to other industries during the pandemic period both in terms of the decrease in the number of links crossed on average (mol) and the average distance per link crossed. this double effect explains the topological shrinkage of mst during the pandemic period, initially explained by the mst-distance indicator and proven by several studies such as onnela et al. (2003) and majapa and gossel (2016). another indicator given by barthélemy (2004) to measure the centrality of nodes is the betweenness centrality which is the fraction of shortest paths passing through a given node. for a node 𝑖𝑖, the betweenness centrality is given by: 𝐵𝐵𝐵𝐵(𝑖𝑖) = ∑ 𝜎𝜎𝑗𝑗𝑗𝑗(𝑖𝑖)𝜎𝜎𝑗𝑗𝑗𝑗𝑗𝑗≠𝑘𝑘 (8) where 𝜎𝜎𝑗𝑗𝑘𝑘 is the total number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 and 𝜎𝜎𝑗𝑗𝑘𝑘 (𝑖𝑖) is the number of shortest paths from node 𝑗𝑗 to node 𝑘𝑘 through 𝑖𝑖. the usefulness of betweenness centrality is that some less connected nodes can be very central by the fact that they link different parts of the network together. again, the banking sector is the most central (figure 6); it has 36.85 per cent of shortest paths in the pre-pandemic period and 78.47 per cent in the pandemic period. this shows a particular importance both in terms of centrality and the systemic risk that may result. figure 6 betweenness centrality of mst nodes. 47 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 pandemic period, before observing great volatility from the start of the pandemic. this is valid for all the indicators used to study this dynamic. the results show that from one period to another (before pandemic to pandemic), the mean distance dropped by 21.9 per cent, the standard deviation increased by 255 per cent and the mean proximity to the centre of the mst (mol) decreased to 1.76 links, instead of 2.88 before the pandemic. the other important result is the noted importance of certain industries, and the central role played by the banking industry. in addition to the observed importance of certain industries (“real estate participation and development”, “building and construction materials”, “telecommunication”, “ food / production”) due to their central weight and their survival in msts, the banking industry was positioned as the most central node throughout the period studied. this centrality is perfectly reinforced from the point that it became the only central node during the period of the covide-19 pandemic. this was also the industry that has survived in all of the msts studied. considering that financial systems in low-income countries tend to be more bank-based than stock market-based, the moroccan case study is interesting for two reasons. firstly, for a modest income country like morocco and with the observed importance of banks as a very central industry in the industry indices network, it can be concluded that it is rather the banking system that is still at the centre of the financial system. secondly, the importance given by the economic policy in morocco to the banking industry as a lever for financing and development is compatible with the role that this industry plays in the evolution of the industry indices network. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. references ang, a., & chen, j. (2002). asymmetric correlations of equity portfolios. journal of financial economics, 63(3), 443–494. https://doi.org/10.1016/s0304-405x(02)00068-5 48 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 arthur, w. b., durlauf, s. n., & lane, d. a. (1997). the economy as an evolving complex system ii (1st ed.). santa fe institute series. ashraf, b. n. (2020). stock markets’ reaction to covid-19: cases or fatalities? research in international business and finance, 54, 101249. https://doi.org/10.1016/j.ribaf.2020.101249 aslam, f., mohmand, y. t., ferreira, p., memon, b. a., khan, m., & khan, m. (2020). networkanalysis of global stock markets at the beginning of the coronavirus disease (covid-19) outbreak, borsa istanbul review, 20, 49-61. https://doi.org/10.1016/j. bir.2020.09.003 baker, s. r., bloom, n., davis, s. j., kost, k. j., sammon, m. c., & viratyosin, t. (2020). the unprecedented stock market impact of covid19, national bureau of economic research, working paper series, number 26945. barthélemy, m. (2004). betweenness centrality in large complex networks. eur. phys. j. b, 38, 163-168. https://doi.org/10.1140/ epjb/e2004-00111-4 bonanno, g., vandewalle, n., & mantegna, r. n. (2000). taxonomy of stock market indices. physical review e62, 7615-7618. https://doi.org/10.1103/physreve.62.r7615 bonanno, g., lillo, f., & mantegna, r. n. (2001). high-frequency cross-correlation in a set of stocks. quantitative finance, 1(1), 96–104. https://doi.org/10.1080/713665554 coelho, r., hutzler, s., repetowicz, p., & richmond, p. (2007). sector analysis for a ftse portfolio of stocks. physica a. 373, 615-626. https://doi.org/10.1016/j.physa.2006.02.050 de carvalho, p. j. c., & gupta, a. (2018). a network approach to unravel asset price co-movement using minimal dependence structure. journal of banking and finance, 91, 119-132. https:// doi.org/10.1016/j.jbankfin.2018.04.012 drozdz, s., gruemmer, f., ruf, f., & speth, j. (2000). dynamics of competition between collectively and noise in the stock market. physica a, 287, 440-449. https://doi.org/10.1016/ s0378 4371(00)00383-6 galazka, m. (2011). characteristics of the polish stock market correlations. international review of financial analysis, 20(1), 1–5. https://doi.org/10.1016/j.irfa.2010.11.002 gilmore, c. g., lucey, b. m., & boscia, m. (2008). an ever-closer union? examining the evolution of linkages of european equity markets via minimum spanning trees. physica a, 387. 6319–6329. https://doi.org/10.1016/j.physa.2008.07.012 49 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 haroon, o., & rizvi, s. a. r. (2020). covid-19: media coverage and financial markets behaviour: a sectoral inquiry. journal of behavioral and experimental finance, 27, 100343. https://doi. org/10.1016/j.jbef.2020.100343 khashanah, k., & miao, l. (2011). dynamic structure of the us financial systems. studies in economics and finance, 28(4), 321–339. https://doi.org/10.1108/10867371111171564 liu, h., manzoor, a., wang, c., zhang, l., & manzoor, z. (2020). the covid-19 outbreak and affected countries stock markets response. international journal of environmental research and public health, 17(8), 2800. https://doi.org/10.3390/ ijerph17082800 longin, f., & solnik, b. (2001). extreme correlation of international equity markets. the journal of finance, 56(2), 649–676. https://doi.org/10.1111/0022-1082 majapa, m., & gossel, s. j. (2016).topology of the south african stock market network across the 2008 financial crisis. physica a, 445, 35–47. https://doi.org/10.1016/j.physa.2015.10.108 mantegna, r. n. (1999). hierarchical structure in financial markets. european physical journal b11, 193-197. https://doi. org/10.1007/s100510050929 musmeci, n., aste, t., & di matteo, t. (2015). risk diversification: a study of persistence with a filtered correlation-network approach. journal of network theory in finance, 1(1), 77–98. https://doi.org/10.21314/jntf.2015.005 onnela, j. p., chakraborti, a., kaski, k., kertesz, j., & kanto, a. (2003). dynamics of market correlations: taxonomy and portfolio analysis. physical review e68, 056110. https://doi. org/10.1103/physreve.68.056110 sinha, s., & pan, r. k. (2007). uncovering the internal structure of the indian financial market: cross correlation behaviour in the nse. econophysics of markets and business networks. new economic windows. springer, milano. https://doi. org/10.1007/978-88-470-0665-2_1 situngkir, h., & surya, y. (2005). indonesian stock market dynamics through ultrametricity of minimum spanning tree. https://ssrn. com/abstract=768204 or http://dx.doi.org/10.2139/ssrn.768204 tabak, b. m., serrac, t. r., & cajueiro, d. o. (2010).topological properties of stock market networks: the case of brazil. physicaa, 389, 3240–3249. https://doi.org/10.1016 jphysa.2010.04.002 50 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 31–50 tola, v., lillo, f., gallegati, m., & mantegna, r. n. (2008). cluster analysis for portfolio optimization. journal of economic dynamics and control, 32, 235-258. https://doi.org/10.1016/j. jedc.2007.01.034 tumminello, m., lillo, f., & mantegna, r. n. (2010). correlation, hierarchies, and networks in financial markets. journal of economic behavior and organization, 75, 40–58. https://doi. org/10.48550/arxiv.0809.4615 yang, r., lia, x., & zhang, t. (2014). analysis of linkage effects among industry sectors in china’s stock market before and after the financial crisis. physica a, 411, 12–20. https://doi. org/10.1016/j.physa.2014.05.072 zhang, d., hu, m., & ji, q. (2020). financial markets under the global pandemic of covid-19, finance research letters, 36,101528. https://doi.org/10.1016/j.frl.2020.101528 cetak a4 ijbf6(2)09.indd the international journal of banking and finance, 2008/09 vol. 6. number 2: 2009: 1-36 1 modelling the behaviour of technicians and fundamentalists in the shanghai stock market imad moosa and larry li monash university and latrobe university abstract this paper provides empirical evidence on the role of fundamentalists and technicians in the chinese stock market. three econometric models are used to differentiate the stock price effect between the actions of traders who act on the basis of fundamental analysis and those acting on the basis of technical analysis. the models are estimated using randomly selected monthly and daily data on the stock prices of one hundred companies listed on the shanghai stock exchange. the results reveal that both fundamentalists and technicians have roles to play in stock price formation, although technicians appear to play a more important role. this result holds even if the government intervention is allowed for. some explanations are presented for the dominance of technicians. keywords: technical analysis, fundamentalists, share prices, shanghai exchange, government intervention jel classifi cation: d53, g24 1. introduction the objective of this paper is to present empirical evidence on the role played by fundamentalists and technicians, as market analysts, in the shanghai stock market. the evidence is based on three versions of the model proposed by moosa and korczak (2000) and moosa and al-loughani (2003), which follows, in part, the model proposed by frankel and froot (1990) to differentiate the roles played by fundamentalists and technicians in the foreign exchange market. the modifi cation of the basic version of the model is takes into account two issues: (i) the possibility that the price effect measures may be sensitive to the specifi cation of the reaction function of the technicians, and (ii) the role of government intervention in stock price determination, the latter being a potential factor in this ijbf ht tp :// ijb f.u um .e du .m y 2 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 market. the proposed models are estimated and tested using randomly selected time series data on the stock prices of one hundred chinese companies covering two frequencies (daily and monthly) and various time periods. the behaviour of traders who act on the basis of fundamental analysis and those who act on the basis of technical charting analysis may differ drastically. fundamental analysts watch deviations from an equilibrium price, as implied by an assumed fundamental model. if the current price is above its equilibrium level, the asset is sold, which should lead to a decline in the price. this is not necessarily the case, however. if, by watching technician’s charts or following quantitative technical indicators, technical analysts believe that there is no indication of a trend reversal, they will keep on buying the asset, lending support to the price. how the price moves depends on the net effect of the forces of supply and demand resulting from the actions of both fundamental analysts and technical analysts. the same argument is valid if the price falls below its equilibrium level. economists, who are essentially fundamentalists, have been studying technical analysis and the role of technicians with increasing enthusiasm, motivated by the desire to come up with an explanation for the deviation of fi nancial prices from the values implied by fundamental models of fi nancial price determination. in studying technical analysis, economists seek to address three issues: (i) the extent to which technical analysis is used in practice to predict price movements; (ii) the forecasting power of technical models as opposed to fundamental models; and (iii) the role played by technicians in fi nancial price determination. while this study is concerned primarily with the third question, the empirical fi ndings should shed some light on the other two issues. if, for example, the results indicate that the actions of technicians are more powerful in determining fi nancial prices, we may conclude that (i) technical analysis must be used extensively, and (ii) the forecasting power of technical models is superior to that of fundamental models. the rest of the paper is divided as follows. in the next section, the reader will fi nd a brief review of relevant prior studies on the subject. in section 3 can be found the models to be used in this study. the fi ndings are presented and commented upon in section 4. the paper ends with a concluding fi nal section. 2. the empirical evidence empirical evidence on the role played by fundamentalists and technicians in fi nancial markets is found in studies following three approaches: (i) those based on econometric models; (ii) survey studies; and (iii) those analysing the comparative profi tability of fundamental and technical trading rules. one of the pioneering studies based on econometric modelling is that of frankel and froot (1990), who designed a model that is founded on the same idea to explain the sharp rise in the demand for the dollar in the fi rst half of the 1980s. on the basis of the results produced by their model, they attributed the increase in the demand for the u.s. currency to the overwhelming role played ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 3 by technical analysts during that period. the rise to glory of technical analysis is frequently attributed to the remarkable rise of the dollar between 1981 and 1985 to its highest level ever. the relentless appreciation of the dollar during that time period was unexplainable in terms of market fundamentals, such as interest rates and growth rates. frustration with fundamental explanations of the dollar behaviour (and the consequent trading losses) led many analysts and investment managers to resort to technical analysis, a tendency that has created the status quo where technical analysis is taken seriously, even by anti-technical hardliners. the frankel-froot model was used by vigfusson (1997), whereas kirman (1991) presented an extension of the model. levin (1997) developed a model involving the interaction between the expectations of chartists and fundamentalists. by using a modifi ed version of the frankel-froot model, moosa and korczak (2000) presented evidence indicating that (i) the exchange rate is determined by both technicians and fundamentalists, and (ii) fundamentalists play a bigger role in this respect. the second fi nding is justifi ed on the grounds of using low-frequency data, which implies a long investment horizon. moosa and al-loughani (2003) found some evidence indicating that the role of technicians is slightly more important than that of the fundamentalists. variable addition tests revealed that the addition of the technicians’ activity is more important than the addition of the fundamentalists’ activity even though the coeffi cients on the variables representing the activities of fundamentalists and technicians turned out to be equal. guest (2004) tested the same model using australian spot and futures stock prices, producing evidence for the hypothesis that both fundamentalists and technicians have roles to play in price determination. al-muraikhi (2005) found similar results for the emerging stock and foreign exchange markets of kuwait. survey evidence on the role played by fundamentalists and technicians is provided by a number of studies conducted on various markets. harvey (1993, p.680) presents an interesting argument for survey evidence as an alternative to econometric modelling, stating that “more can be learned about the mysteries of international fi nance through understanding the propensities and proclivities of the traders than can be gleaned from a hundred multivariate regressions”. allen and taylor (1989, 1990) and taylor and allen (1992) surveyed more than 400 foreign exchange dealers in london, with 60 per cent saying that charts are at least as important as fundamental analysis. cheung et al. (2004) conducted a u.k.-based survey to fi nd that technical-based trading was the preferred trading model. a subsequent survey by oberlechner (2001) incorporated traders on all hierarchy levels in european foreign exchange markets, fi nding strong support for the use of technical analysis. menkhoff (1997) and lui and mole (1998) obtained similar results from surveys conducted in the german and hong kong foreign exchange markets, respectively. evidence on the role played by fundamentalists and technicians is also provided by studies comparing the profi tability of fundamental and technical trading rules, as well as studies stipulating heterogeneity in fi nancial markets ht tp :// ijb f.u um .e du .m y 4 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 (see moosa, 2003, pp. 236-240). the general conclusion of these studies is that both fundamentalists and technicians play a role in fi nancial markets and that fi nancial market participants are heterogenous with respect to the trading strategies they use. pilbeam (1995a, 1995b) based his study of the profi tability of foreign exchange trading on the notion of trader heterogeneity. in pilbeam (1995b) traders are supposed to follow three different exchange rate determination models (the fl exible-price monetary model, the sticky-price monetary model and the sticky-price portfolio balance model) in conjunction with six expectation formation mechanisms (static, extrapolative, adaptive, regressive, rational and heterogenous). in pilbeam (1995a), traders are classifi ed into chartists, fundamentalists and simpletons. the same idea forms the basis of the post-keynesian theory of exchange rate determination (see for example, harvey, 1993). moosa (2002) presents a simple theoretical model that is based on the micro foundations of exchange rate determination to illustrate the relation between the heterogeneity of traders and volatility. the model is founded on the idea that observed exchange rate volatility can only result from erratic shifts in the market’s excess demand function that is made up of the excess demand functions of heterogenous traders. the heterogeneity of traders means that they have different sentiments and different expectations at any point in time. hence, they are likely to react differently to new developments: some want to buy (thus raising excess demand) and some want to sell (thus reducing excess demand). the net effect of their actions is to shift the aggregate excess demand function by a certain amount in a certain direction. in describing the model, moosa assumes the presence of four kinds of traders: technicians using fi lter rules, technicians using moving average rules, fundamentalists using rules, and fundamentalists using discretion. this model was tested by moosa and shamsuddin (2003) who found that fi nancial price formation results from the interaction of traders using a wide variety of technical and fundamental models. the conclusion that can be safely derived from the preceding examination of the literature is that ample survey and econometric evidence show that both fundamentalists and technicians have roles to play in fi nancial price determination. the evidence also indicates that the “balance of power” may tip in favour of one group of traders under certain circumstances. for example, the role of technicians is more important in situations involving short investment horizons. the empirical work conducted in this paper is designed to fi nd out if this general conclusion holds for the chinese stock market, which is attracting a lot of attention (see, for example, gao, 2002; fernald & rogers, 2002). this constitutes the motivation for writing this paper, as the fi ndings will be useful for trading and policy purposes. 3. model specifi cation, hypotheses and estimation the specifi cation of the models is derived from the following propositions. first, fundamentalists base their decisions (with respect to buying and selling stocks or ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 5 fi nancial assets in general) on the difference between the equilibrium price and the actual price. hence, the current period’s change in the price that is due to the operations of fundamentalists is given by (1) where p is the logarithm of the price, a bar denotes the equilibrium price, f denotes fundamentalists and α is a positive parameter that refl ects the speed of adjustment of the actual price to the deviation from the equilibrium price. technicians, on the other hand, base their decisions on previous changes in the price whether they use trading rules, follow quantitative technical indicators or simply observe charts. while this kind of behaviour can be represented by a number of specifi cations, our initial choice falls on a geometrically-declining distributed lag representation as proposed by moosa and korczak (2000). this specifi cation is related to the use of an exponential moving average rule, which is a technical trading rule (buying and selling on signals indicated by the points of intersection of the price and its exponential moving average). hence, the change in the price due to the activities of the technicians can be represented by the equation: (2) where 0 < β < 1 and the superscript t denotes technicians. if the activities of both fundamentalists and technicians contribute to price changes, the model can be written in a testable stochastic form as (3) such that γ 1 > 0 and γ 2 > 0. if 21 γγ > , we may safely conclude that fundamentalists play a more signifi cant role in price determination, and vice versa. the model represented by equation (3) will be called model i. the use of a geometrically-declining distributed lag to represent the reaction function of the technicians implies that technicians, who base their decisions on the past behaviour of prices, assign more signifi cant weight to recent observations. this, however, may not be the case, as it could be that technicians look at a fi nite history of prices and assign similar weights to recent and less recent observations (hence the difference between an exponential and simple moving average rules). allowing for the possibility of a different reaction function of technicians may be useful as a robustness check. hence, the price change due to the activities of technicians may be represented by the equation (4) )()( 11 tt f t ppp 1 )( i it it t pp 1 21110 )( i tit i ttt pppp n i it m i it t t pn p m p 11 11 )( ht tp :// ijb f.u um .e du .m y 6 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 such that m < n . equation (4) resembles the process used to generate buy and sell signals on the basis of moving average cross-over, as the fi rst term on the right hand side of the equation represents the short moving average whereas the second term represents the long moving average. simply stated, equation (4) represents another way of looking at the past behaviour of prices, which is what technicians do. in this case, the model becomes (5) such that γ1 > 0 and γ2 > 0 . equation (5) will be called model ii. another modifi cation to the model can be introduced by allowing an explicit role for the government in the stock price determination process, which may be necessary because the chinese government has been keeping an open eye on trading to ensure a well-functioning market, and has power to intervene. however, any attempt to allow for the role played by the government in the stock market will be hampered by at least two problems. the fi rst is that the chinese government does not have an explicit objective with respect to the behaviour of stock prices. the second is that, even if such an objective exists, it is not easy to incorporate in the model. it is, however, plausible to assume that an objective of the chinese government is to reduce market volatility, to maintain a smooth time path for stock prices. if this is the case, the role of government intervention can be modelled by borrowing an idea suggested by hodrick and prescott (1997) to construct the hp fi lter. they proposed, as a measure of the smoothness of the time path (trend) of a variable, the sum of squares of its second difference. as a modifi cation of this idea, we suggest the use of a geometrically-declining distributed lag of the squares of the second difference of stock prices, which gives (6) where (∆p t )gis the change in price due to government intervention. in this case the model is written as (7) where γ 3 < 0 . equation (7) will be called model iii. the models represented by equations (3), (5) and (7) are estimated by ols, which is a valid procedure since the underlying variables are stationary. the unobserved equilibrium price, , is estimated by applying the hodrick-prescott (1997) fi lter to p. this is a detrending technique that is used to decompose an observed time series into trend and cycle. formally, the hp fi lter is used to t n i it m i itttt pn p m ppp 11 21110 11 )( 1 22 )()( i it ig t pp 1 1 22 321110 )()( i i tit i it i ttt ppppp p ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 7 estimate the trend path of the time series, subject to the constraint that the sum of the squared second differences of the time series is not too large.1 the trend is calculated from the observed time series by solving the optimisation problem (8) where the smoothing parameter, λ , is typically determined by the frequency of the observations. in essence, the equilibrium price is taken to be the fi tted hp trend. although the hp fi lter is not the only means whereby a proxy for the equilibrium price can be obtained, it is assumed here that the long-run trend of the price refl ects the behaviour of the fundamental variables determining its equilibrium price. by using a specifi c fundamental model to estimate the equilibrium value, we run the risk of not capturing all of the fundamental variables, and hence the risk of misrepresenting the equilibrium price. for estimating the geometrically-declining distributed lags, the choice of the values of the parameters β and φ is arbitrary, but a value of 0.8 seems reasonable because it means that the technicians assign more weight to more recent observations on ∆pt-i and (∆ 2 pt-i) 2 . as a robustness check, other values of the parameter were tried but made no qualitative difference to the results. similar procedures are used for the determination of the values of m and n in equation (5).2 following the estimation of the models, the importance of fundamentalists relative to technicians is determined by testing the null hypothesis (9) which amounts to a wald test where the test statistic has a χ2(1) distribution, since there is only one restriction on the values of the estimated coeffi cients. if h 0 is rejected such that γ 1 > γ 2 , then the role of fundamentalists is more important than the role of technicians, and vice versa. also used are variable deletion tests. to fi nd out if fundamentalists play a role in stock price determination, the coeffi cient restriction γ 1 = 0 is imposed. a signifi cant test statistic in this case implies that the restriction is invalid, meaning that fundamentalists play a role in price determination (and vice versa). on the basis of the residual sum of squares of the unrestricted models (3, 5 and 7) and ntpt ,,2,1, * ntpt ,,2,1, n t n t ttt ppp ppp n 1 1 2 2* 1 22* ,, )()(min ** 2 * 1 210 :h 1 this is the idea of representing smoothness by the sum of the squared second differences of the underlying variable. this representation was used earlier to specify a proxy for the role of government intervention in the stock market. 2 theory does not provide any guidance on the values of β, φ , m and n. intuition and robustness checks provide the key to the selection of appropriate values of these parameters. ht tp :// ijb f.u um .e du .m y 8 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 the restricted models (resulting from the imposition of the restriction γ 1 = 0 ), two test statistics are calculated: a lagrange multiplier (lm) and a likelihood ratio (lr) test statistics. both of these statistics follow a χ2 (1) distribution, since only one coeffi cient restriction is imposed. likewise, the restriction γ 2 = 0 is imposed to fi nd out if technicians play a role in price determination. non-nested model selection tests are used to fi nd out if either technicians or fundamentalists can on their own determine stock prices. consider the following two models, m1 and m2: (10) (11) where x is an observation matrix on the variable representing the role of fundamentalists (plus the role of the government if necessary), z is an observation matrix on the variable representing the role of technicians (plus the role of the government if necessary), δ 1 and δ 2 are unknown regression coeffi cient vectors, and ξ 1 and ξ 2 are disturbance vectors. in the particular case under investigation here, m1 and m2 are related to the models represented by equations (3), (5) and (7), as the former can be obtained by imposing some restrictions on the latter. the m1 and m2 corresponding to equation (3) are given respectively by (12) (13) which tell us that stock prices are determined exclusively by the activities of fundamentalists (m1 as represented by equation 12) or exclusively by the activities of the technicians (m2 as represented by equation 13). the m1 model corresponding to equation (5) is the same as equation (12) but m2 becomes (14) likewise, the m1 and m2 corresponding to equation (7) are given respectively by (15) m1: 11xpt m2: 22zpt m1: tttt ppp )( 1110 m2: 1 20 i tit i t pp m2: t n i it m i itt pn p m p 11 20 11 m1: 1 22 31110 )()( i tit i ttt pppp ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 9 (16) the models m1 and m2 are said to be non-nested if the regressors of either of them cannot be expressed as an exact linear combination of the regressors of the other. obviously, m1 and m2 as represented by equations (12)-(16) are nonnested because either the explanatory variables are different (as in 12 and 13) or because there is a common explanatory variable but the other one appears in one equation only (as in 15 and 16). six model selection tests are used: n is the cox test derived in pesaran (1974); nt is the adjusted cox test derived in godfrey and pesaran (1983); w is the wald-type test proposed by godfrey and pesaran (1983); j is the davidson and mackinnon (1981) test; ja is the fisher-mcaleer (1981) test; and en is the encompassing test proposed, inter alia, by mizon and richard (1986). all of the test statistics have t distribution, except for the encompassing test that has f distribution. the tests are run both ways by testing m1 versus m2 and m2 versus m1. when m1 is tested versus m2, the null hypothesis is that m1 is a better model (in terms of specifi cation) than m2. a signifi cant test statistic indicates that m1 is not a better model than m2. when m2 is tested against m1, the null is that m2 is a better model than m1. a signifi cant test statistic indicates that m2 is not a better model than m1. if we obtain signifi cant test statistics both ways, this means that the two models are misspecifi ed (that is, neither fundamentalists nor technicians can determine prices on their own). if we get insignifi cant test statistics by testing m1 versus m2 and signifi cant statistics by testing m2 versus m1, this means that m1 is preferred to m2 (that is, fundamentalists alone determine stock prices). the econometrics of non-nested model selection tests can be found in pesaran and pesaran (1997). 4. empirical results the empirical work conducted in this paper is based on monthly and daily data on the stock prices of one hundred listed companies, obtained from the taiwan economic journal database. the appendix displays the names of the companies, the sectors they belong to, their markets values (in rmb billion) and the sample periods. the listing of market values is useful to fi nd out if size matters, in the sense that the results are (or are not) affected by the market value of the company used to carry out the test.3 the stocks were selected randomly from the full list of m2: 1 1 22 320 )( i i tit i it i t ppp 3 there is no obvious theoretical reason to believe that fundamentalists play a more important role than technicians in the determination of the stock prices of large companies than those of small companies or vice versa. but it could be that large companies are better audited and that they publish more detailed fi nancial reports than small companies, which makes them better suited to fundamental analysis than small companies. in any case, this is an issue that is worth exploring. ht tp :// ijb f.u um .e du .m y 10 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 the stocks traded on the shanghai stock exchange. to ensure random selection, a random number generator was used to pick the companies according to their shanghai stock exchange codes. with one hundred companies, two frequencies and three models, the amount of results is too big to report in a tabular form. what we have decided to do is to report the results primarily in a graphical form, while reporting a sample of the results of estimating model i in a tabular form for eight selected companies, chosen as representatives of eight different percentiles of market values, including the smallest and largest companies.4 the selected companies for which the results are reported are 1, 17, 22, 31, 40, 55 72 and 85. the details of these companies can be found in the appendix. the results of estimating model i, as represented equation (3), for the eight selected companies are reported in table 1, including the estimated coeffi cients and their t statistics (in parentheses), as well as the coeffi cient of determination (r2) and the diagnostics for serial correlation (sc), functional form (ff) and heteroscedasticity (hs), all of which have χ2 distribution (with one degree of freedom for daily data and the same for monthly data, except for the serial correlation test statistic that has 12 degrees of freedom for monthly data). the estimated equations pass the diagnostic tests, producing signifi cantly positive coeffi cients in all cases. the null hypothesis h 0 :γ 1 = γ 2 is rejected across the board, which means that although both fundamentalists and technicians have roles to play, the latter play a more important role in terms of exerting more infl uence on the market price (γ 2 > γ 1 in all cases). table 2 reports the results of the variable deletion tests for the eight selected companies. in all cases, the lm and lr statistics are signifi cant, which means that the null hypothesis is rejected, implying that both fundamentalists and technicians have roles to play. these results are supported by the results of the non-nested model selection tests, which are presented in table 3. with few exceptions, the test statistics are signifi cant (both ways), implying that both m1 and m2 are misspecifi ed and that the correctly-specifi ed model should include terms that represent the activities of fundamentalists and technicians.5 4 the full set of results, in a tabular form, encompassing 100 companies, two frequencies and three models is available from the authors upon request. 5 the j, ja and en tests are consistent in rejecting m1 versus m2 and m2 versus m1. however, the n, nt and w tests show cases where m2 is not rejected versus m1 (31, 40 and 85), implying that the role of fundamentalists can be ignored. however, the bulk of the evidence indicates that the activities of both fundamentalists and technicians do matter. ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 11 t ab le 1 : m od el i e st im at io n r es ul ts f or s el ec te d c om pa ni es 1 17 22 31 40 55 72 85 m on th ly γ 0 -0 .0 16 0. 04 1 0. 01 2 0. 00 6 0. 05 1 0. 10 7 -0 .0 02 -0 .0 06 (1. 88 ) (1 .9 7) (0 .9 7) (0 .8 9) (8 .3 4) (7 .7 7) (0. 31 ) (0. 99 ) γ 1 0. 68 8 0. 41 6 0. 51 1 0. 39 2 0. 57 9 0. 63 7 0. 45 7 0. 37 8 (1 3. 98 ) (7 .9 3) (8 .0 4) (9 .0 5) (1 2. 58 ) (9 .6 5) (1 0. 13 ) (7 .4 6) γ 2 0. 79 6 0. 63 2 0. 68 0 0. 47 0 0. 74 3 0. 79 3 0. 95 1 0. 56 0 (1 7. 35 ) (1 4. 65 ) (8 .1 0) (1 3. 57 ) (1 6. 79 ) (1 3. 36 ) (1 3. 69 ) (1 0. 84 ) r 2 0. 89 0. 79 0. 82 0. 66 0. 86 0. 85 0. 72 0. 71 s c 3. 55 1. 78 2. 68 3. 79 3. 67 3. 14 3. 25 2. 44 f f 2. 20 1. 15 1. 67 3. 50 2. 60 0. 88 1. 91 2. 35 h f 2. 94 0. 36 2. 19 1. 18 0. 17 0. 03 1. 67 1. 94 h 0 : γ 1 = γ 2 4. 39 20 .3 8 5. 73 4. 05 10 .6 7 4. 47 6. 30 9. 45 d ai ly γ 0 -0 .0 00 5 0. 00 07 0. 00 04 -0 .0 00 3 0. 00 1 0. 00 1 -0 .0 00 2 -0 .0 00 4 (1. 36 ) (1 .7 7) (0 .4 9) (0. 34 ) (1 .4 3) (1 .5 9) (0. 65 ) (1. 19 ) γ 1 0. 31 9 0. 31 6 0. 29 5 0. 33 5 0. 33 5 0. 31 0 0. 34 3 0. 30 7 (2 7. 23 ) (2 39 .7 2) (1 9. 92 ) (1 9. 35 ) (3 1. 22 ) (3 8. 09 ) (3 8. 08 ) (2 8. 43 ) γ 2 0. 47 8 0. 46 9 0. 45 2 0. 47 7 0. 49 6 0. 47 4 0. 49 0 0. 47 0 (3 8. 80 ) (4 2. 56 ) (2 9. 15 ) (2 7. 47 ) (4 3. 19 ) (5 4. 52 ) (5 3. 46 ) (4 1. 13 ) r 2 0. 64 0. 63 0. 61 0. 64 0. 65 0. 64 0. 64 0. 62 s c 2. 61 2. 89 2. 85 2. 24 1. 39 2. 78 2. 21 2. 87 f f 2. 69 2. 36 1. 36 0. 36 2. 44 1. 60 3. 21 2. 67 h f 1. 96 2. 75 2. 58 0. 07 3. 13 3. 50 2. 79 3. 85 h 0 : γ 1 = γ 2 13 7. 45 15 3. 99 81 .1 7 51 .6 0 16 4. 31 29 9. 09 19 5. 22 16 5. 04 n ot e: t he n um be rs 1 , … , 8 5 re pr es en t th e co m pa ni es l is te d in t he a pp en di x un de r th e sa m e nu m be rs . ht tp :// ijb f.u um .e du .m y 12 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 t ab le 2 : v ar ia bl e d el et io n t es t r es ul ts r es tr ic ti on γ 1 = 0 γ 1 = 0 γ 2 = 0 γ 2 = 0 l m : χ 2 ( 1) l r : χ2 ( 1) l m : χ2 (1 ) l r : χ2 (1 ) m on th ly 1 40 .5 1 83 .1 1 43 .3 0 10 0. 51 17 31 .7 3 44 .7 9 48 .0 3 94 .4 4 22 21 .1 7 36 .6 8 21 .2 5 36 .9 8 31 47 .3 1 62 .2 4 68 .7 7 10 9. 36 40 41 .9 5 77 .4 5 47 .1 4 10 3. 25 55 28 .6 2 93 .0 8 33 .1 3 70 .4 9 72 49 .8 4 71 .0 1 63 .2 8 10 5. 14 85 28 .9 2 40 .3 6 39 .0 6 65 .8 8 d ai ly 1 42 2. 92 55 3. 55 54 9. 32 91 3. 90 17 50 8. 24 66 2. 30 71 9. 98 11 03 .6 0 22 24 0. 54 30 6. 47 35 4. 47 53 2. 86 31 21 4. 66 28 0. 66 30 1. 06 46 1. 40 40 54 0. 80 71 6. 46 74 4. 49 11 55 .2 0 55 82 5. 04 10 80 .4 0 11 63 .1 0 17 94 .0 72 81 5. 09 10 73 .0 11 26 .6 0 17 32 .3 0 85 47 8. 22 61 5. 33 69 1. 31 10 47 .3 0 ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 13 t ab le 3 : n on -n es te d m od el s el ec ti on t es t r es ul ts ( co n ti n u ed ) 1 17 22 31 40 55 72 85 m on th ly m 1 vs m 2 n -2 0. 50 -2 16 .5 2 -3 7. 91 -3 0. 93 -1 1. 22 -1 9. 99 -2 7. 13 -2 1. 80 n t -1 5. 12 -3 3. 94 -1 3. 32 -2 3. 71 -1 4. 60 -1 3. 20 -2 1. 35 -1 4. 79 w -1 1. 54 -2 3. 34 -1 0. 69 -1 8. 77 -1 1. 13 -9 .8 5 -1 6. 84 -1 1. 59 j 17 .3 5 14 .6 )5 8. 10 13 .5 7 16 .7 9 13 .6 13 .6 9 10 .8 4 ja -1 7. 35 -1 4. 65 -8 .1 0 -1 3. 57 -1 6. 79 -1 3. 36 -1 3. 69 -1 0. 84 e n 30 0. 94 21 4. 76 65 .6 1 18 4. 04 28 1. 97 17 8. 52 18 7. 49 11 7. 53 m 2 vs m 1 n -6 .1 8 2. 37 -3 7. 02 -0 .4 5 -1 .7 2 -2 .1 2 -4 .5 5 -0 .7 8 n t -5 .1 9 2. 38 -1 3. 15 -0 .2 8 -1 .4 1 -1 .6 5 -4 .0 1 -0 .5 2 w -4 .6 8 2. 58 -1 0. 59 -0 .2 7 -1 .3 6 -1 .5 8 -3 .7 8 -0 .5 1 j 13 .9 8 7. 93 8. 04 9. 05 12 .5 8 9. 65 10 .1 3 7. 46 ja -1 3. 98 -7 .9 3 -8 .0 4 -9 .0 5 -1 2. 58 -9 .6 5 -1 0. 13 -7 .4 6 e n 19 4. 43 62 .8 8 64 .7 0 81 .8 4 15 8. 32 93 .0 8 10 2. 70 55 .6 2 d ai ly m 1 vs m 2 1 17 22 31 40 55 72 85 n -7 6. 39 -8 5. 61 -6 4. 62 -5 4. 51 -8 2. 74 -1 07 .3 7 -1 04 .8 1 -8 7. 17 n t -7 4. 10 -8 3. 45 -6 1. 17 -5 1. 43 -8 0. 83 -1 05 .6 7 -1 03 .1 9 -8 4. 79 w -6 0. 06 -6 7. 58 -4 9. 54 -4 1. 71 -6 5. 50 -8 5. 60 -8 3. 53 -6 8. 68 t a b le c o n ti n u es o n t h e n ex t p a g e ht tp :// ijb f.u um .e du .m y 14 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 j 38 .8 0 42 .5 6 29 .1 5 27 .4 7 43 .9 1 54 .5 2 53 .4 6 41 .1 3 ja -3 8. 80 -4 2. 56 -2 9. 15 -2 7. 47 -4 3. 91 -5 4. 52 -5 3. 46 -4 1. 13 e n 15 05 .5 0 18 11 .3 0 84 9. 75 75 4. 40 19 27 .8 0 29 72 .2 0 28 57 .5 0 16 91 .6 0 m 2 vs m 1 n -2 .9 1 -3 .2 1 -2 .3 2 -2 .7 3 -4 .0 4 -2 .9 1 -7 .4 2 -3 .4 4 n t -2 .8 3 -3 .1 4 -2 .2 3 -2 .6 2 -3 .7 9 -2 .8 7 -7 .3 5 -3 .3 7 w -2 .7 9 -3 .1 0 -2 .1 9 -2 .5 7 -3 .8 9 -2 .8 3 -7 .1 6 -3 .3 1 j 27 .2 3 29 .7 2 19 .9 2 19 .3 5 31 .2 2 38 .0 9 38 .0 8 28 .4 3 ja -2 7. 23 -2 9. 72 -1 9. 92 -1 9. 35 -3 1. 22 -3 8. 09 -3 8. 08 -2 8. 43 e n 74 1. 72 88 3. 24 39 6. 88 37 4. 45 97 4. 77 14 50 .7 14 49 .8 0 80 8. 17 t ab le 4 . p ea rs on c or re la ti on b et w ee n c om pa ny s iz e an d th e e st im at ed c oe ffi c ie nt s/ s ta ti st ic s c oe ff /s ta t m od el i (m ) m od el i (d ) m od el i i (m ) m od el i i (d ) m od el i ii ( m ) m od el i ii (d ) γ 1 -0 .0 26 -0 .0 54 -0 .0 49 -0 .0 58 0. 00 2 -0 .0 72 γ 2 -0 .0 27 -0 .0 47 -0 .0 25 -0 .0 63 0. 04 5 -0 .0 25 γ 3 -0 .0 85 -0 .0 62 t( γ 1 ) -0 .0 35 -0 .1 00 -0 .0 68 -0 .0 98 -0 .0 08 -0 .1 09 t( γ 2 ) -0 .0 32 -0 .0 99 -0 .0 85 -0 .0 86 -0 .0 03 -0 .1 12 t( γ 3 ) -0 .0 37 0. 01 8 χ2 (γ 1 = γ 2) -0 .0 35 -0 .0 89 0. 04 8 0. 04 5 0. 06 3 0. 08 7 t ab le 3 : n on -n es te d m od el s el ec ti on t es t r es ul ts ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 15 fundamentalists technicians the full results are presented in a graphical form in figures 1 to 6, two tables for models i, ii and iii, covering the monthly and daily frequencies. figure 1 reports the results for model i using monthly data. it consists of three parts (a, b and c), respectively showing the estimated values of γ 1 and γ 2 (figure 1(a)), their t statistics (figure 1(b)) and the χ2 statistics for the null hypothesis γ 1 = γ 2 (figure 1(c)). to make the graphs easier to follow visually, the estimated values of the coeffi cients and the test statistics are sorted in an ascending order whenever necessary. figure 1: the empirical results (model i, monthly data) (a) a radar chart of the estimated coeffi cients on the activities of fundamentalists (sorted) and technicians (b) the sorted t statistics of the coeffi cients on the activities of fundamentalists and technicians with the 5% critical value 0 5 10 15 20 25 fundamentalists technicians cvfundamentalists ht tp :// ijb f.u um .e du .m y 16 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 figure 1: (c) the sorted chi-square test statistics for the equality of the coeffi cients with the 5% critical value figure 1(a) is a radar chart showing the sorted estimated values of γ 1 (fundamentalists) and the corresponding estimated values of γ 2 (technicians). the chart shows that γ 2 > γ 1 in all of the 100 cases, as the two loops representing γ 1 and γ 2 do not intersect (the loop representing γ 1 is nested within the loop representing γ 2 ). figure 1(b) shows the sorted t statistics of the estimated values of γ 1 (fundamentalists) and γ 2 (technicians), together with the approximate 5% critical value of the t statistic, represented by the horizontal line cutting the vertical axis at 1.96. across the board, the t statistics are above the critical value, indicating the signifi cance of γ 1 and γ 2 . figures 1(a) and 1(b) show unequivocally that while both fundamentalists and technicians play a role in stock price determination, technicians play a more important role, in the sense that their activities have a quantitatively more signifi cant impact on stock prices. figure 1(c) displays the sorted χ2 statistics for the null hypothesis γ 1 = γ 2 as well as its 5% critical value, represented by the horizontal line cutting the vertical axis at 3.84. the χ2 statistics are insignifi cant in 12 cases out of 100, implying that the bulk of the evidence indicates that both fundamentalists and technicians have roles to play in stock price determination. 0 5 10 15 20 25 30 ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 17 figure 2: the empirical results (model i, daily data) (a) a radar chart of the estimated coeffi cients on the activities of fundamentalists (sorted) and technicians figure 2: (continued) (b) the sorted t statistics of the coeffi cients on the activities of fundamentalists and technicians with the 5% critical value fundam entalis ts technicians 0 10 20 30 40 50 60 70 fundamentalists technicians cv ht tp :// ijb f.u um .e du .m y 18 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 figure 2 is exactly like figure 1, except that it shows the results of estimating model i using daily data. the results are unanimous: both fundamentalists and technicians play a role in stock price determination, but technicians play a quantitatively more important role without any exception. of a particular interest is figure 3(c), which shows that the χ2 statistic for the null hypothesis γ 1 = γ 2 is not insignifi cant in any of the 100 cases (compared with insignifi cant statistics in 12 cases when monthly data are used, as shown in figure 1(c)). there is a simple explanation for this observation. the results are more strongly in favour of a more signifi cant role for the technicians when daily data are used, because it is not possible to observe fundamentals on a daily basis. indeed, the available survey evidence on the role of technicians and fundamentalists in fi nancial price determination indicates that the tendency to use technical analysis increases as the investment horizon becomes shorter. (c) the sorted chi-square test statistics for the equality of the coeffi cients with the 5% critical value figure 3 and figure 4 show exactly the same fi ndings for model ii when monthly and daily data are used, respectively. 0 100 200 300 400 500 ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 19 figure 3: the empirical results (model ii, monthly data) (a) a radar chart of the estimated coeffi cients on the activities of fundamentalists (sorted) and technicians figure 3: (continued) (b) the sorted t statistics of the coeffi cients on the activities of fundamentalists and technicians with the 5% critical value fundamentalists technicians 0 2 4 6 8 10 12 fundamentalists technicians cv ht tp :// ijb f.u um .e du .m y 20 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 (c) the sorted chi-square test statistics for the equality of the coeffi cients with the 5% critical value figure 4: the empirical results (model ii, daily data) (a) a radar chart of the estimated coeffi cients on the activities of fundamentalists (sorted) and technicians the fi rst point to observe is that changing the representation of the activities of technicians does not make any qualitative difference for the results, which consistently show γ 1 and γ 2 to be signifi cantly positive and that γ 2 > γ 1 in all cases. like figure 1(c), and unlike figures 2(c) and 4(c), figure 3(c) shows some insignifi cant χ2 statistics for the null hypothesis γ 1 = γ 2 , implying that 0 5 10 15 20 25 30 35 fundamentalists technicians ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 21 fundamentalists when monthly data are used there is some indication (though not overwhelming) that technicians and fundamentalists play similarly signifi cant roles in stock price determination. this fi nding is not apparent in figure 4(c), which is based on daily data. (b) the sorted t statistics of the coeffi cients on the activities of fundamentalists and technicians with the 5% critical value figure 4: (continued) (c) the sorted chi-square test statistics for the equality of the coeffi cients with the 5% critical value 0 10 20 30 40 50 60 70 fumdam entalis ts technicians cv 0 50 100 150 200 250 fundamentalists ht tp :// ijb f.u um .e du .m y 22 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 figure 5: the empirical results (model iii, monthly data) (a) a radar chart of the estimated coeffi cients on the activities of fundamentalists (sorted) and technicians (b) the sorted t statistics of the coeffi cients on the activities of fundamentalists and technicians with the 5% critical value fundam entalists technicians 0 5 10 15 20 25 30 fum dam entalists technicians cvfundamentalistsht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 23 figures 5 and 6 report the empirical results of estimating model iii using monthly and daily data, respectively. exactly the same results hold, confi rming the overall importance of technicians relative to fundamentalist. one additional advantage of model iii is that it allows for an explicit role for government intervention in the stock market. referring to equation (7), if the objective of government intervention is to reduce the volatility of (and maintain a smooth time path for) stock prices, the government intervention variable should be negatively related to the dependent variable. this means that if government intervention is effective then γ 3 < 0. figures 5(c) and 6(c) show the sorted t statistics of γ 3 , together with the positive and negative 5% critical values of the t statistic (cutting the vertical axis at 1.96 and -1.96, respectively). the results are supportive of the role of government intervention as γ 3 is signifi cantly negative in just under half of the 100 cases. not in one case does γ 3 turn out to be is signifi cantly positive. figure 5: (continued) (c) the sorted t statistics of the coeffi cient on the government intervention variables with the positive and negative 5% critical values -7.5 -5.0 -2.5 0.0 2.5 5.0 ht tp :// ijb f.u um .e du .m y 24 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 (d) the sorted chi-square test statistics for the equality of the coeffi cients with the 5% critical value figure 6: the empirical results (model iii, daily data) (a) a radar chart of the estimated coeffi cients on the activities of fundamentalists (sorted) and technicians fundam entalists technicians 0 5 10 15 20 25 ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 25 (b) the sorted t statistics of the coeffi cients on the activities of fundamentalists and technicians with the 5% critical value figure 6. (continued) (c) the sorted t statistics of the coeffi cient on the government intervention variables with the positive and negative 5% critical values 0 10 20 30 40 50 60 70 fumdam entalis ts technicians cv -7.5 -5.0 -2.5 0.0 2.5 5.0 fundamentalists ht tp :// ijb f.u um .e du .m y 26 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 (d) the sorted chi-square test statistics for the equality of the coeffi cients with the 5% critical value the issue that remains to be considered is whether or not the results are sensitive to company size as measured by market capitalisation. as pointed out earlier, it could be that big companies are more transparent than small companies with respect to fi nancial reporting. if this is the case, then one would expect traders to pay more attention to fundamental factors in the case of big companies, which means that the results should favour the role of technicians less in the case of big companies. to investigate this issue we calculated the pearson correlation coeffi cient between market capitalisation and the estimated coeffi cients (γ 1 , γ 2 and γ 3 ), their t statistics (t(γ 1 ), t(γ 2 ) and t(γ 3 )), and the χ2 statistics for the null hypothesis γ 1 – γ 2 (χ 2 (γ 1 = γ 2 ). the results are reported in table 4, showing very little correlation between the results and market capitalisation. company size, it seems, does not matter. 5. discussion of the results and concluding remarks in seeking an explanation for the dominance of technicians in the chinese stock market, one may resort to the key role played by the government, which determines how many companies can be listed each year and how many shares can be issued (su and fleisher, 1999). moreover, signifi cant proportions of the listed companies’ holdings belong to the government, directly and indirectly.6 0 50 100 150 200 250 300 350 400 6 gao (2002) argues that the government “seems to have too much infl uence on the market” by controlling ipos and holding signifi cant stakes in publicly-listed companies. ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 27 therefore, ordinary investors expect the government to be responsible for market performance. on the other hand, the government strives to achieve multiple objectives via the development of the stock market, such as improving government credibility, maintaining social stability and creating new jobs. this means that the behaviour of the market is independent from economic fundamentals and the performance of the real sector of the economy. figure 7 confi rms this observation, showing a fl at market over the period 2000-2005, followed by the surge of 2006-07, which accelerated in the fi rst half of 2007 (before the big decline of early june). this is by no means a refl ection of the performance of the chinese economy, as it has been growing rapidly throughout the sample period. trading activity in the chinese market is dominated by individual investors, because the shares controlled by the government are not traded. however, chinese individual investors lack investment knowledge and skill in general (because they have not been in this business for long). this characteristic results in the dominance of “noise traders”, who are closer to be technicians, if anything at all. as de long et al. (1990) have shown, the behaviour of noise traders is typically unpredictable, and it can lead to a large divergence between market prices and fundamental values. moreover, noise traders can survive for a long time and earn higher returns than sophisticated investors for bearing the extra risks created by them.7 actually, some chinese market anomalies can be explained by the behaviour of noise traders, including extensive speculation and excess volatility.8 7 a signifi cant amount of work has been done on noise trading, which is defi ned by black (1986) as “trading on noise as if it were information”. palomino (1996) defi nes noise traders as “agents with unpredictable beliefs”, whereas de long et al. (1990) them as those who “falsely believe that they have special information about the future price of risky assets”. while black (1986) does not present a reason why investors would rationally want to engage in noise trading, trueman (1988) provides an explanation for why managers of investment funds tend to indulge in noise trading. 8 the characteristics of extensive speculation and excess volatility are examined by gao (2002). he shows that the chinese market exhibits exceptionally high volatility as measured by the annualised standard deviation of monthly return and the frequency, magnitude and duration of market crashes. he argues that this kind of volatility “encompasses a speculative mentality in investors who are led to make short-term bets rather than longterm investments”. a measure of what gao calls “incredible speculation” is the average annual stock turnover divided by the average market capitalisation, which he shows to be more than 500 percent (roughly ten times what is found in most developed markets). fernald and rogers (2002) show that the higher level of volatility of chinese stock prices is consistent with the “simplest asset pricing model”. ht tp :// ijb f.u um .e du .m y 28 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 figure 7: the shanghai composite index some aspects of behavioural fi nance may also explain the dominance of technicians. the characteristics of the chinese market provide fertile grounds for making investors overconfi dent in their judgements, the consequence of which is to overestimate the precision of information (odean, 1998; graham & harvey, 2002). moreover, these investors appear to judge uncertain events based on their experience, which may generate representiveness bias and lead to signifi cant divergence of market prices from the fundamental values of stocks. confi rmation bias and conservatism are also important. confi rmation bias exists when investors only believe information that supports their opinion and ignore information that does not. on the other hand, investors with conservatism bias have a relatively slow reaction to market changes because they may think that new information is irrelevant and stick to what they believed previously. all of these phenomena seem to be thriving in the chinese stock market (see, for example, lin, 2005). the chinese corporate governance system may also explain the dominance of technicians. the major shareholders in china are likely to be institutions and the state rather than individuals. individual investors account for around 30 percent of total shares, and these shares are scattered over potentially many thousands of shareholders for each company (hovey et al. 2003). in addition, it is very hard to fi nd an individual shareholder on the board of directors, and it is believed that the interests of individual investors are not well protected. therefore, mok and hui (1998) conclude that chinese individual investors mainly look for the short-term capital speculation rather than long-term investment. this makes these investors more of technicians than fundamentalists. it can be concluded, therefore, that the results presented in this paper support the established view and the available survey and econometric evidence on the roles played by fundamentalists and technicians in fi nancial price determination. 0 1000 2000 3000 4000 5000 jan-00 jan-01 jan-02 jan-03 jan-04 jan-05 jan-06 jan-07 ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 29 however, some special features of the chinese market give the technicians the upper hand, in the sense that their activities have a more signifi cant impact on stock prices than those of the fundamentalists. author statement: imad moosa is a professor in the department of accounting and finance, monash university, melbourne. e-mail: imad.moosa@buseco. monash.edu.au. larry li is a teaching staff at the latrobe university, melbourne. e-mail: l.li@latrobe.edu.au. ht tp :// ijb f.u um .e du .m y 30 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 a p p en d ix : t he d at a s am pl es n o c om pa ny m on th ly ( y y /m m ) d ai ly (y y /m m /d d ) s ec to r m ar ke t v al ue (r m b b il li on ) 1 m er ch an ts b an k 02 /0 406 /0 6 02 /0 4/ 09 -0 6/ 07 /0 5 b an ki ng 94 .6 77 2 m in sh en g b an ki ng 00 /1 206 /0 5 00 /1 2/ 19 -0 6/ 07 /0 5 b an ki ng 44 .4 25 3 c it ic s s ec ur it ie s 03 /0 -0 6/ 05 03 /0 1/ 06 -0 6/ 07 /0 5 b an ki ng 37 .8 18 4 h ua x ia b an k 03 /0 906 /0 4 03 /0 9/ 12 -0 6/ 07 /0 5 b an ki ng 23 .8 56 5 a j c o. 93 /0 406 /0 5 95 /0 1/ 03 -0 6/ 07 /0 7 b an ki ng 3. 12 8 6 l uj ia zu i 93 /0 606 /0 5 95 /0 1/ 03 -0 6/ 07 /0 7 r ea l e st at e 13 .5 22 7 g ua ng hu i c o. l td . 00 /0 506 /0 6 00 /0 5/ 26 -0 6/ 07 /0 6 r ea l e st at e 5. 58 6 8 n ew h ua ng pu 93 /0 306 /0 5 95 /0 1/ 03 -0 6/ 07 /0 7 r ea l e st at e 2. 96 3 9 t ia nj in r ea li ty d ev el op m en t 01 /0 906 /0 6 01 /0 9/ 10 -0 6/ 07 /0 6 r ea l e st at e 1. 70 3 10 t ia nh on g b ao ye 01 /0 3 6 /0 6 01 /0 3/ 12 -0 6/ 07 /0 6 r ea l e st at e 0. 97 9 11 c hi xi a d ev el op m en t 02 /0 3 6 /0 6 02 /0 3/ 28 -0 6/ 07 /0 7 r ea l e st at e 2. 34 4 12 s uz ho u h i n ew & t ec h. 96 /0 806 /0 5 96 /0 8/ 15 -0 6/ 07 /0 7 r ea l e st at e 2. 23 7 13 h ua fa l td . 04 /0 206 /0 5 04 /0 2/ 25 6/ 07 /0 6 r ea l e st at e 1. 92 4 14 m ei du h ol di ng 99 /0 405 /1 0 99 /0 4/ 28 -0 6/ 07 /0 6 r ea l e st at e 0. 71 2 15 d on gh ua e nt er pr is e 0 1/ 03 -6 /0 6 01 /0 3/ 19 -0 6/ 07 /0 6 r ea l e st at e 0. 89 7 16 c ha ng ch un ji ng ka i 99 /0 906 /0 5 99 /0 9/ 19 -0 6/ 07 /0 6 r ea l e st at e 1. 99 2 17 c hi na r ai lw ay e rj u 01 /0 506 /0 6 01 /0 5/ 28 -0 6/ 07 /0 7 c on st ru ct io n 3. 66 5 18 b ei ji ng u rb an c on st ru ct io n 99 /0 206 /0 5 99 /0 2/ 03 -0 6/ 07 /0 6 c on st ru ct io n 3. 59 4 19 s ha ng ha i c on st ru ct io n 98 /0 606 /0 5 98 /0 6/ 23 -0 6/ 07 /0 6 c on st ru ct io n 3. 05 ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 31 n o. c om pa ny m on th ly (y y /m m ) d ai ly (y y /m m /d d ) s ec to r m ar ke t v al ue (r m b b il li on ) 20 c r b c 0 0/ 006 /0 5 00 /0 7/ 25 -6 /0 7/ /0 6 c on st ru ct io n 1. 76 7 21 a nh ui w at er r es ou rc es 03 /0 306 /0 6 03 /0 4/ 15 -0 6/ 07 /0 6 c on st ru ct io n 0. 79 7 22 h an gx ia o s te el s tr uc tu re 03 /1 -0 6/ 05 03 /1 1/ 10 -0 6/ 07 /0 6 c on st ru ct io n 0. 86 7 23 s ic hu an r oa d & b ri dg e 03 /0 306 /0 5 03 /0 3/ 25 -6 /0 7/ 05 c on st ru ct io n 1. 20 3 24 p ud on g c on st ru ct io n 04 /0 306 /0 5 04 /0 3/ 16 -0 6/ 07 /0 6 c on st ru ct io n 1. 08 5 25 c hi na u ni te d t el ec om 02 /1 006 /0 5 02 /1 0/ 19 -0 6/ 07 /0 5 in fo t ec h 50 .4 48 26 u f id a s of tw ar e 01 /0 506 /0 6 01 /0 5/ 18 -0 6/ 07 /0 7 in fo t ec h 4. 63 7 27 f ou nd er t ec h. 93 /0 106 /0 5 95 /0 1/ 03 -0 6/ 07 /0 7 in fo t ec h 3. 58 1 28 f ei lo c o. 93 /0 106 /0 5 95 /0 1/ 03 -0 6/ 07 /0 7 in fo t ec h 1. 82 7 29 c ha ng an i nf or m at io n 96 /0 506 /0 3 96 /0 5/ 16 -0 6/ 07 /0 7 in fo t ec h 0. 35 6 30 k in gd om c o. 03 /1 206 /0 5 03 /1 2/ 24 -0 6/ 07 /0 6 in fo t ec h 0. 69 8 31 g uo ne ng g ro up 97 /0 506 /0 5 04 /0 5/ 17 -0 6/ 07 /0 5 in fo t ec h 0. 86 5 32 c hi na n at io na l s of tw ar e 02 /0 506 /0 6 02 /0 5/ 17 -0 6/ 07 /0 7 in fo t ec h 2. 00 5 33 g eh ua c a t v 01 /0 206 /0 6 01 /0 2/ 08 -0 6/ 07 /0 5 o th er 9. 75 4 34 ir ic o d is pl ay c o. 96 /0 506 /0 5 96 /0 5/ 20 -0 6/ 07 /0 6 o th er 1. 39 35 s il an 03 /0 306 /0 5 03 /0 3/ 11 -0 6/ 07 /0 6 o th er 3. 65 7 36 s in om ic ro el ec tr on ic s 01 /0 306 /0 6 01 /0 3/ 16 -0 6/ 07 /0 6 o th er 2. 78 37 u ni da 98 /0 906 /0 4 98 /0 9/ 28 -0 6/ 04 /2 8 o th er 0. 35 5 38 ji an gs ha n c o. 01 /0 106 /0 6 01 /0 1/ 10 -0 6/ 07 /0 6 o th er 0. 90 7 39 d in gl i d ev el op m en t 93 /0 106 /0 6 95 /0 1/ 03 -0 6/ 07 /0 7 o th er 0. 48 1 a p p en d ix : t he d at a s am pl es ( co n ti n u ed ) a p p en d ix c o n ti n u es o n t h e n ex t p a g e ht tp :// ijb f.u um .e du .m y 32 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 n o. c om pa ny m on th ly (y y /m m ) d ai ly (y y /m m /d d ) s ec to r m ar ke t v al ue (r m b b il li on ) 40 e er du os i 01 /0 406 /0 5 01 /0 4/ 26 -0 6/ 07 /0 6 o th er 4. 28 6 41 m ai ly ar d 97 /1 106 /0 5 97 /1 1/ 06 -0 6/ 07 /0 6 o th er 1. 24 6 42 l ia on in g t im es 00 /1 106 /0 5 00 /1 1/ 28 -0 6/ 07 /0 6 o th er 0. 75 5 43 v ek en e li te 98 /0 606 /0 5 98 /0 6/ 09 -0 6/ 07 /0 6 o th er 1. 14 8 44 x in gy e p ol ye st er 00 /0 506 /0 5 00 /0 5/ 25 -0 6/ 07 /0 6 o th er 0. 45 45 b la ck p eo ny 02 /0 606 /0 6 02 /0 6/ 18 -0 6/ 07 /0 7 o th er 1. 85 9 46 y ib in p ap er i nd us tr y 97 /0 206 /0 5 97 /0 2/ 20 -0 6/ 07 /0 7 o th er 0. 67 8 47 ji ao y un 93 /0 906 /0 5 95 /0 1/ 03 -0 6/ 07 /0 7 o th er 1. 11 5 48 d on gf an g b oi le r 96 /1 206 /0 5 96 /1 2/ 17 -0 6/ 07 /0 7 o th er 10 .6 01 49 w an do ng m ed ic al 97 /0 506 /0 5 97 /0 5/ 19 -0 6/ 07 /0 5 o th er 0. 67 50 d on gf en g t ec hn ol og y 97 /0 706 /0 5 97 /0 7/ 03 -0 6/ 07 /0 6 o th er 1. 30 1 51 p in gg ao e le ct ri c 01 /0 206 /0 5 01 /0 2/ 21 -0 6/ 07 /0 6 o th er 4. 06 8 52 w an ji e h ig ht ec h 00 /0 106 /0 5 00 /0 1/ 03 -0 6/ 07 /0 6 d iv er si fi ed 1. 30 8 53 c he ng du c on st ru ct io n 97 /0 806 /0 5 97 /0 8/ 07 -0 6/ 07 /0 6 d iv er si fi ed 0. 46 54 d az ho ng p ub li c u ti li ti es 93 /0 306 /0 6 95 /0 1/ 03 -0 6/ 07 /0 7 d iv er si fi ed 3. 24 9 55 h ai na n o ve rs ea s c hi ne se 03 /0 106 /0 5 96 /1 0/ 08 -0 6/ 07 /0 7 d iv er si fi ed 0. 33 8 56 h um an w el l h it ec h 97 /0 606 /0 5 97 /0 6/ 06 -0 6/ 07 /0 6 d iv er si fi ed 0. 90 3 57 c om m od it ie s c it y 02 /0 506 /0 6 02 /0 5/ 09 -0 6/ 07 /0 6 d iv er si fi ed 7. 90 7 58 p ro so la r t ec hn ol og y 99 /0 506 /0 5 99 /0 5/ 27 -0 6/ 07 /0 6 d iv er si fi ed 0. 59 6 59 b ri gh t d ai ry & f oo d 02 /0 806 /0 6 02 /0 8/ 28 -0 6/ 07 /0 7 f oo d 8. 28 3 a p p en d ix : t he d at a s am pl es ( co n ti n u ed ) a p p en d ix c o n ti n u es o n t h e n ex t p a g e ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 33 n o. c om pa ny m ot hl y (y y /m m ) d ai ly (y y /m m /d d ) s ec to r m ar ke t v al ue (r m b b il li on ) 60 s ha ng ha i m al in g 97 /0 706 /0 5 97 /0 7/ 04 -0 6/ 07 /0 5 f oo d 2. 14 8 61 t si ng ta o b re w er y 93 /0 806 /0 6 95 /0 1/ 03 -0 6/ 07 /0 7 f oo d 14 .0 29 62 h on gh e g ua ng m in g 99 /1 206 /0 5 99 /1 2/ 02 -0 6/ 07 /0 6 f oo d 0. 60 3 63 s an yu an c o. 03 /0 906 /0 6 03 /0 9/ 15 -0 6/ 07 /0 6 f oo d 1. 73 4 64 y il it e 99 /0 906 /0 5 99 /0 9/ 16 -0 6/ 07 /0 6 f oo d 2. 75 2 65 t on gh ua g ra pe c o 01 /0 -0 6/ 05 01 /0 1/ 15 -0 6/ 07 /0 6 f oo d 0. 84 1 66 x in ji an g t un he 96 /0 706 /0 5 96 /0 7/ 31 -0 6/ 07 /0 7 f oo d 4. 67 3 67 h ar bi n p ha rm g ro up 93 /0 606 /0 5 95 /0 1/ 03 -0 6/ 07 /0 7 p ha rm ac eu ti ca l 10 .1 72 68 t on gr en ta ng 97 /0 606 /0 5 97 /0 6/ 25 -0 6/ 07 /0 6 p ha rm ac eu ti ca l 6. 79 2 69 p ie nt ze hu an g 03 /0 606 /0 5 03 /0 6/ 16 -0 6/ 07 /0 6 p ha rm ac eu ti ca l 2. 74 3 70 y ib ai p ha rm 04 /0 306 /0 6 04 /0 3/ 23 -0 6/ 07 /0 7 p ha rm ac eu ti ca l 1. 48 7 71 l in gr ui c o. 00 /1 006 /0 5 00 /0 1/ 18 -0 6/ 07 /0 6 p ha rm ac eu ti ca l 0. 49 1 72 f os un p ha rm 98 /0 806 /0 6 98 /0 8/ 07 -0 6/ 07 /0 6 p ha rm ac eu ti ca l 6. 21 2 73 z he ji an g m ed ic in e 99 /1 006 /0 5 99 /1 0/ 21 -0 6/ 07 /0 6 p ha rm ac eu ti ca l 2. 33 1 74 y ab ao p ha rm 02 /0 906 /0 6 02 /0 9/ 26 -0 6/ 07 /0 6 p ha rm ac eu ti ca l 0. 67 5 75 m in m et al s d ev el op t. 97 /0 506 /0 5 97 /0 5/ 08 -0 6/ 07 /0 5 r et ai li ng 6. 92 2 76 n ew w or ld 93 /0 106 /0 6 96 /0 1/ 02 -0 6/ 07 /0 7 r et ai li ng 4. 06 5 77 b ei ji ng x id an m ar ke t 96 /0 706 /0 5 96 /0 7/ 16 -0 6/ 06 /2 2 r et ai li ng 2. 19 6 78 n an ji ng x in ji ek ou d ep t 93 /1 006 /0 5 95 /0 1/ 03 -0 6/ 07 /0 7 r et ai li ng 2. 09 5 79 f ir st i nv es tm en t 02 /0 806 /0 6 02 /0 8/ 06 -0 6/ 07 /0 7 r et ai li ng 0. 54 1 a pp en di x: t he d at a s am pl es (c on tin ue d) a p p en d ix c o n ti n u es o n t h e n ex t p a g e ht tp :// ijb f.u um .e du .m y 34 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 n o. c om pa ny m on tl y (y y /m m ) d ai ly (y y /m m /d d ) s ec to r m ar ke t v al ue (r m b b il li on ) 80 z he ji an g o ri en t 97 /1 206 /0 5 97 /1 2/ 01 -0 6/ 07 /0 6 r et ai li ng 1. 51 1 81 o ri en t e nt er pr is e 00 /0 706 /0 5 00 /0 7/ 12 -0 6/ 07 /0 6 r et ai li ng 1. 15 5 82 h ua li an h yp er m ar ke t 01 /1 106 /0 6 01 /1 1/ 29 -0 6/ 07 /0 6 r et ai li ng 6. 73 5 83 f ir st p ro vi si on 93 /0 106 /0 6 95 /0 1/ 03 -0 6/ 07 /0 7 r et ai li ng 5. 74 7 84 z ho ng da g ro up 96 /0 606 /0 5 96 /0 6/ 06 -0 6/ 07 /0 7 r et ai li ng 1. 35 7 85 c hi na p et ro le um & c he m 01 /0 806 /0 5 01 /0 8/ 08 -0 6/ 07 /0 5 m in in g 53 2. 19 86 y an zh ou c oa l 98 /0 706 /0 5 98 /0 7/ 01 -0 6/ 07 /0 6 m in in g 33 .7 25 87 l an hu a s ci -t ec h 98 /1 206 /0 5 98 /1 2/ 17 -0 6/ 07 /0 6 m in in g 5. 04 5 88 s ha ng ha i e ne rg y r es ou rc es 01 /0 806 /0 6 01 /0 8/ 2 -0 6/ 07 /0 7 m in in g 5. 13 9 89 p an ji an g c o. 01 /0 506 /0 6 01 /0 5/ 31 -0 6/ 07 /0 6 m in in g 2. 46 5 90 z ho ng ji n g ol d 03 /0 806 /0 6 03 /0 8/ 14 -0 6/ 07 /0 6 m in in g 5. 27 8 91 g uo ya ng n ew e ne rg y 03 /0 806 /0 6 03 /0 8/ 21 -0 6/ 07 /0 6 m in in g 6. 41 2 92 s ha nd on g e xp re ss w ay 02 /0 306 /0 6 02 /0 3/ 18 -0 6/ 07 /0 6 p ub li c u ti li ty 12 .2 78 93 g ui gu an e le ct ri c p ow er 00 /0 306 /0 6 00 /0 3/ 23 -0 6/ 07 /0 6 p ub li c u ti li ty 8. 24 5 94 ji nj ia ng h ot el s 96 /1 006 /0 5 96 /1 0/ 11 -0 6/ 07 /0 7 p ub li c u ti li ty 5. 08 9 95 b as hi c o. 96 /0 806 /0 5 96 /0 8/ 26 -0 6/ 07 /0 7 p ub li c u ti li ty 2. 67 2 96 h ua ng sh an t ou ri sm 97 /0 506 /0 5 97 /0 5/ 07 -0 6/ 07 /0 5 p ub li c u ti li ty 2. 44 9 97 c hi na w or ld t ra de 99 /0 306 /0 5 99 /0 3/ 12 6/ 07 /0 5 p ub li c u ti li ty 5. 01 6 98 u ni te d t ra ve l 00 /0 906 /0 6 00 /0 9/ 12 -0 6/ 06 /0 9 p ub li c u ti li ty 1. 79 7 99 d al ia n s he ng ya 02 /0 706 /0 6 02 /0 7/ 11 -0 6/ 07 /0 3 p ub li c u ti li ty 0. 60 8 10 0 c yt s t ou rs h ol di ng 97 /1 206 /0 5 97 /1 2/ 03 -0 6/ 07 /0 6 p ub li c u ti li ty 1. 65 3 a pp en di x: t he d at a s am pl es ht tp :// ijb f.u um .e du .m y modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 35 references al-muraikhi, h. (2005). speculation in emerging financial markets: the case of the kuwait stock and foreign exchange markets (doctoral dissertation, la trobe university). black, f. (1986). noise. journal of finance, 41, 529-543. cheung, y-w, chinn, m.d., and marsh, i.w. (2004). how do uk-based exchange dealers think their market operates? international journal of finance and economics, 9, 289-306. dastoor, n. (1983). some aspects of testing non-nested hypotheses. journal of econometrics, 21, 213-228. davidson, r., and mackinnon, j.g. (1981). several tests for models specifi cation in the presence of alternative hypotheses. econometrica, 49, 781-793. de long, j. b., shleifer, a., summers, l.h., and waldmann, r. (1990). noise trader risk in fi nancial markets. journal of political economy, 98, 703738. deaton, a. (1982). model selection procedures, or does the consumption function exist? in chow, g.c. and corsi, p. (eds), evaluating the reliability of macroeconometric models. new york: wiley. fernald, j., and rogers, j. (2002). puzzles in the chinese stock market. review of economics and statistics, 84, 416-432. fisher, g.r., and mcaleer, m. (1981). alternative procedures and associated tests of signifi cance for non-nested hypotheses. journal of econometrics, 16, 103-119. frankel, j., and froot, k. (1990). the rationality of the foreign exchange rate: chartists, fundamentalists and trading in the foreign exchange market. american economic review, 80, 181-185. gao, s. (2002, sept.) china stock market in a global perspective. dow jones indexes, 1-48. godfrey, l.g., and pesaran, m.h. (1983). tests of non-nested regression models: small sample adjustments and monte carlo evidence, journal of econometrics, 21, 133-154. graham, j.r., and harvey, c. (2002). how do cfos make capital budgeting and capital structure decisions? journal of applied corporate finance, 15, 8-23. guest, o. (2004). the time series properties of the spi futures market and implications for fi nancial decisions (doctoral dissertation, la trobe university). harvey, j.t. (1993). the institution of foreign exchange trading. journal of economic issues, 27, 679-698. hodrick, r., and prescott e. (1997). post-war u.s. business cycles: an empirical investigation. journal of money, credit and banking, 29, 1-16. hovey, m., li., l., and naughton, t. (2003). the relationship between valuation and ownership of listed fi rms in china. corporate governance: an international review, 11, 112-122. ht tp :// ijb f.u um .e du .m y 36 modelling the behaviour of technicians and fundamentalists in the shanghai stock market: 1-36 kirman, a. (1991). epidemics of opinion and speculative bubbles in fi nancial markets. in m.p. taylor (ed.), money and financial markets, oxford: blackwell. levin, j.h. (1997). chartists, fundamentalists and exchange rate dynamics. international journal of finance and economics, 2, 281-290. lin, w. (2005). manipulation, price limits and the weekend effect: a study of the chinese stock market (doctoral dissertation, la trobe university). mizon, g.e., and richard, j.f. (1986). the encompassing principle and its application to non-nested hypotheses. econometrica, 54, 657-678. mok, h.m.k., and hui, y.v. (1998). underpricing and after market performance of ipos in shanghai, china. pacifi c-basin financial journal, 6, 453-474. moosa, i.a. (2002). exchange rates and fundamentals: a microeconomic approach. economia internazionale, 5, 551-571. moosa, i.a. (2003). international fi nancial operations: arbitrage, hedging, speculation, investment and fi nancing. london: palgrave. moosa, i., and al-loughani, n. (2003). the role of fundamentalists and technicians in the foreign exchange market when the domestic currency is pegged to a basket. applied financial economics, 13, 79-84. moosa, i., and korczak, m. (2000). the role of fundamentalists and technicians in exchange rate determination. economia internazionale, 53, 97-106. moosa, i., and shamsuddin, a. (2003). heterogeneity of traders as a source of exchange rate volatility: some simulation results based on a descriptive model. journal of financial studies, 11, 43-69. odean, t. (1998). are investors reluctant to realize their losses? journal of finance, 53, 1775-1798. palomino, f. (1996). noise trading in small markets. journal of finance, 51, 1537-1550. pesaran, m.h. (1974). on the general problem of model selection. review of economic studies, 41, 153-171. pesaran, m.h., and pesaran, b. (1997). working with microfi t 4.0: interactive econometric analysis. oxford: oxford university press. pilbeam, k. (1995a). exchange rate models and exchange rate expectations: an empirical investigation. applied economics, 27, 1009-1115. pilbeam, k. (1995b). the profi tability of trading in the foreign exchange market: chartists, fundamentalists, and simpletons. oxford economics papers, 47, 437-452. trueman, b. (1988). a theory of noise trading in securities markets. journal of finance, 43, 83-95. vigfusson, r. (1997). switching between chartists and fundamentalists: a markov regime-switching approach. international journal of finance and economics, 2, 291-305. ht tp :// ijb f.u um .e du .m y 1 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: ozili, p. k. (2022). financial inclusion in nigeria: an overview. international journal of banking and finance, 17(2), 1-24. https://doi. org/10.32890/ ijbf2022.17.2.1 financial inclusion in nigeria: an overview peterson k. ozili central bank of nigeria, abuja, nigeria corresponding author: petersonkitakogelu@yahoo.com received: 1/12/2020 revised: 25/2/2021 accepted: 22/4/2021 published: 27/6/2022 abstract this paper analyses the level of financial inclusion in nigeria using data from the global findex indicators. the findings reveal that nigeria witnessed growth in several financial inclusion indicators in the early years of financial inclusion in 2014, but the benefits were not sustained in the later years, especially in 2017. nigeria’s level of financial inclusion is very low compared to the world average. in the population group analysis, it was observed that the female, poorest, male, older and uneducated members of the population were worseoff in all indicators of financial inclusion in 2017. the implication of the observed decline in the level of financial inclusion in 2017 seems to suggest that there are barriers to financial inclusion in the post-2014 years. keywords: financial inclusion, nigeria, gender, savings. jel classification: o16, g00, g21. introduction financial inclusion is defined as the use of formal financial services (allen et al, 2016). another definition states that financial inclusion is 2 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 the ease of access, availability and usage of the formal financial system by all members of the economy (sarma, 2008; sahay et al, 2015). financial inclusion may also be defined as the provision of affordable financial services to people (ozili, 2018). the first step to achieve financial inclusion is ensuring that each household and individuals own a bank account, which would give them access to basic formal financial services (reddy, 2007), and such access to formal finance can build self-confidence, empower poor individuals and households, and promote social inclusion (paramasivan & ganeshkumar, 2013). financial inclusion is a policy agenda for development in many countries (marron, 2013; ozili, 2020a), and financial inclusion objectives are often achieved through a combination of activities, policies and programs designed to help reach the unbanked population. financial inclusion, when done correctly, can uplift poor people from poverty. however, when done incorrectly it can expose poor people to risk in the formal financial system, particularly risks associated with the use of financial products and services. such risks, when they materialize, can worsen the welfare of the poorest, increase income inequality and may lead to social exclusion, which is undesirable (ozili, 2020b). in nigeria, some commentators believe that financial inclusion can accelerate economic diversification in the country, yield economic gains and bring about shared prosperity in nigeria (adeola & evans, 2017). other commentators believe that financial inclusion can promote economic development through poverty reduction, wealth creation and improved standard of living. however, there are obstacles to financial inclusion in nigeria. on the demand side, challenges include the lack of awareness about existing financial services, rising financial illiteracy, superstitious and religious beliefs about banking services, high transaction costs and the general lack of interest in banking products and services by some segment of the rural population. on the supply side, there are issues such as the lack of bank branch penetration, unwillingness of banks to sustain financial inclusion programs due to high costs, and the unwillingness of banks to bear the social cost of a bank-led financial inclusion program. to analyze the level of financial inclusion in nigeria, data on financial inclusion was extracted from the global findex indicators of the world bank database. data from the early years of financial inclusion are compared with data of the later years of financial inclusion to identify 3 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 any observed changes in financial inclusion. using percentage change analysis (pca), the findings seem to suggest that the majority of the population had greater access to basic financial services in the early years of financial inclusion, but the benefits were not sustained in the later years. the present paper makes several contributions to the literature in the field. firstly, this study contributes to the financial inclusion literature. the study shows that education levels, being rich and being employed are associated with greater financial inclusion in nigeria. these findings are consistent with the results in various other studies found in the literature. secondly, it also contributes to the literature on the measure of financial inclusion, by looking at access to finance indicators (see grimes et al., 2010; klapper et al., 2015; ozili, 2021a; etc.). it does this by analyzing a number of access routes to finance indicators, and show that higher access to finance benefits some group of the population at the expense of other groups in nigeria. finally, the study contributes to the financial inclusion policy literature (see. langley, 2008; montgomerie, 2008; ozili, 2018; etc.), by showing that the gains of financial inclusion diminish after some years. there are policy lessons to be gained from this. the rest of the paper is structured in the following way. section 2 discusses the theory and existing literature review. section 3 presents some discussions on the sample data and research methodology. section 4 reports the empirical results. section 5 presents some recommendations. section 6 is the conclusion. theory and literature review theoretical underpinning buera et al. (2011) have argued that income differences across countries and cross-country differences in financial development would affect the level of development in countries, while demirguckunt and klapper (2012) have suggested that better access to finance for everyone could spur development in a country. demirguckunt and klapper (2012) have also argued that a well-functioning financial system would offer valuable savings, credit, payment and risk management products to individuals and households with diverse needs. such financial systems have allowed broad access to financial 4 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 services, without any significant price or non-price barriers in the use of financial services (demirguc-kunt & klapper, 2012). in the absence of an inclusive financial system, poor individuals and households will rely on their own limited savings to invest in their education or become entrepreneurs, while small businesses will rely on their small profit to grow their business (demirguc-kunt & klapper, 2012), and this can contribute to the widening of the income inequality gap and slow economic growth (demirguc-kunt & klapper, 2012). literature review recent evidence shows some consensus that financial inclusion can improve the welfare of people. demirgüç-kunt and klapper (2012) observed that only few adults in africa owned an account with a formal financial institution, and many adults used informal methods to save and borrow. they also observed that many small and medium enterprises in africa were financially excluded and faced major obstacles in obtaining funds to do business. grimes et al. (2010) examined whether economic education influenced the decision to own a bank account. they used a nationwide telephone survey, and found that adults who took a course in economics and business were less likely to be unbanked and vice versa; in addition, adults who had some good understanding of economic concepts were more likely to be financially included. klapper et al. (2015) observed that financial literacy was higher among the rich and educated population. ozili (2020a) showed that the level of financial inclusion was affected by the level of financial innovation, poverty, financial literacy and regulation. niankara and muqattash (2018) highlighted the impact of financial inclusion on individuals’ borrowing and saving decisions in the united states and the united arab emirates, and found that u.s. residents were more likely to save than their united arab emirates residents. they also found the prevalence of a gender based saving and borrowing inequality in favor of the male gender, while access to a bank account and a debit card increased the likelihood to save and borrow. dev (2006) argued that financial inclusion should be a business opportunity and a social responsibility through better regulation and the de-politicization of the financial system. fungáčová and weill (2015) analyzed financial inclusion in china, including comparisons 5 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 with the other brics countries, and found that the level of financial inclusion in china was high and it was manifested as greater formal account usage and formal saving than in the other brics countries. they also found that higher income, better education, being a man, and being older were associated with greater use of formal accounts and formal credit in china. bachas et al. (2018) assessed whether debit cards helped to improve the level of financial inclusion. they found that debit cards helped to lower transaction costs by reducing the distance to access bank accounts in mexico. however, few research has been done on financial inclusion in nigeria. for instance, efobi et al. (2018) examine the impact of financial inclusion on the export capacity of firms in the manufacturing sector in nigeria, and find that greater access to financial services increase the export capacity of firms, but the impact on each manufacturing firm depends on the location of the firm. adegbite and machethe (2020) analyse the gender gap in financial inclusion in nigeria, and find that there is increasing gender gap in financial inclusion in nigeria as more men had greater access to finance than women. adetunji and davidwest (2019) show that financial literacy affects savings patterns in financial institutions in nigeria. other studies have focused on the effect of financial inclusion on the macro economy. ibrahim and aliero (2020) found a strong positive relationship between the level of financial inclusion and the size of per capita income. ozili (2020c) found that the unemployed and people with at least a secondary education had greater debit card ownership and greater account ownership in a financial institution. david et al. (2018) examined the determinants of financial inclusion in nigeria from 1990 to 2016, and found that financial inclusion had increased gdp growth and broad money. fowowe (2020) showed that financial inclusion had a positive impact on agricultural productivity in nigeria. ozili (2021b) showed that fintechs were driving the growth in financial inclusion in african countries despite the political, infrastructure and structural problems hindering full financial inclusion in african countries. however, these nigerian studies have not examined how the financial inclusion indicators differ within several groups in the population. therefore, this paper will provide an analysis of the level of financial inclusion in nigeria, taking into account the group characteristics of the population categories. 6 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 research design data financial inclusion data for nigeria was collected from the world bank’s global financial inclusion (findex) database. the population is nigeria. the data was collected from 2011 to 2017. the financial inclusion data in the database had been reported tri-annually in 2011, 2014 and 2017. six (6) financial inclusion indicators were used in this study. the indicators were as follows: (i) account ownership, (ii) extent of borrowings from a formal financial institution, (iii) extent of borrowings from family and friends, (iv) credit card ownership, (v) debit card ownership, and (vi) savings in a formal financial institution. some financial inclusion indicators were excluded in the analyses because the indicators did not have full data reported for all the three years. the total number of observations was 213, that is, 71 indicators over three years. the description of variables is as presented in table 1. table 1 description of variables indicator meaning source account ownership households/adults who report having an account (by themselves or together with someone else) at a bank or another type of financial institution, or report personally using a mobile money service in the .past 12 months. global findex database formal borrowing (or borrowing from a financial institution) households/adults who report borrowing any money from a bank or another type of financial institution in the past 12 months. global findex database borrowing from family or friends households/adults who report borrowing any money from family, relatives, or friends in the past 12 months. global findex database credit card ownership households/adults who report owning a credit card in the past 12 months global findex database debit card ownership households/adults who report owning a debit card in the past 12 months global findex database (continued) 7 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 indicator meaning source formal savings (or savings at a formal financial institution) households/adults who report saving any money at a bank or another type of financial institution in the past 12 months. global findex database population category: all adults (% age 15+) all adult population global findex database out of labor force (% age 15+) unemployed adult population global findex database female (% age 15+) female adult population global findex database in labor force (% age 15+) employed adult population global findex database income, poorest 40% (% age 15+) poor adult population global findex database income, richest 60% (% age 15+) rich adult population global findex database male (% age 15+) male adult population global findex database older adults (% age 25+) older adult population global findex database primary education or less(% age 15+) uneducated adult population global findex database rural (% age 15+) rural adult population global findex database secondary education or more (% age 15+) educated adult population global findex database young adults (% age 15-24) young adult population global findex database real gdp growth macroeconomic indicator world bank method of analysis percentage change analysis (pca) was the method used to analyze the financial inclusion data. a pca will show how one or more indicators change as a percentage from one period to another period. regression analysis was not used because of the short data period and small number of observations, which together could bias the coefficient of the regression estimations. 8 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 discussion of results account ownership table 2 shows that account ownership in nigeria increased between 2011 and 2014 in all population categories. the largest increase in account ownership was recorded among the poorest, male and adult population categories. the observed increase in all the population categories of account ownership in 2014 was caused by the effective implementation of the national financial inclusion strategy led by the central bank of nigeria. the implementation of the strategy led to the expansion of account opening services across the country. a major part of the national financial inclusion strategy required the opening of more bank branch networks, introduction of mobile money service, mobile banking, banking agents and the issuance of operating license to payment service providers. the implication of these efforts has been that government intervention through increase bank branch networks and the issuance of license to agents has made it possible for banks and agents to reach the unbanked adults that need to open an account in nigeria. the level of account ownership decreased in 2017 in all population categories. the largest decrease in account ownership was recorded among the poorest, uneducated and female population. the decrease in all the categories of account ownership was caused by the effect of the 2016 recession, as well as the slow recovery from the recession. the economic difficulty brought about by the recession led to the merger of weak banks and the closure of some bank branches, which had in turn led to a reduction in account opening activities across the country in the years after the 2016 recession, such as in 2017. the implication is that a recession usually has a negative effect on the level of financial inclusion in nigeria, and the effects tend to extend beyond the crisis year. comparing nigeria with the global average, table 2 shows that the level of account ownership in nigeria is low compared to the global average in all population categories in 2011, 2014 and 2017. the implication is that, despite the government’s effort to increase the number of citizens that own an account, the government’s effort to increase the level of account ownership is still very low. this means the government needs to do a lot more. 9 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 finally, the values for real gdp growth and the growth in account ownership was positive in 2014. this suggests that the level of account ownership in nigeria tend to be higher during periods of economic prosperity. in 2017, real gdp growth remained positive, although lower in magnitude due to the after-effects of the 2016 recession, while the growth in account ownership was negative across all categories of account ownership in 2017. interestingly, nigeria’s real gdp growth was higher than the global average in 2011 and 2014. table 2 account ownership indicator / country nigeria world (average) % change / growth (nigeria) 2011 2014 2017 2011 2014 2017 (2014) (2017) all adults (% age 15+) 29.6 44.4 39.7 50.6 62.0 68.5 49.8 -10.7 female (% age 15+) 25.9 34.0 27.2 46.6 58.4 64.8 30.9 -19.8 in labor force (% age 15+) 34.1 52.6 42.9 55.9 68.6 73.8 54.3 -18.4 income, poorest 40% (% ages 15+) 13.9 30.5 24.5 41.3 54.6 60.5 118.5 -19.5 income, richest 60% (% ages 15+) 40.1 53.7 49.7 57.1 66.9 73.8 33.8 -7.4 male (% age 15+) 33.2 54.3 51.4 54.7 65.5 72.3 63.3 -5.4 older adults (% ages 25+) 33.8 48.8 44.0 54.4 66.2 71.9 44.6 -9.9 out of labor force (% age 15+) 17.6 29.01 30.1 38.4 49.3 59.3 64.7 4.1 primary education or less (% ages 15+) 12.2 28.1 16.1 37.03 48.4 55.9 130 -42.8 rural (% age 15+) 23.5 38.7 33.3 43.9 57.9 65.9 64.2 -13.9 secondary education or more (% ages 15+) 44.1 56.02 59.2 66.1 73.2 78.6 26.9 5.8 young adults (% ages 15-24) 21.4 35.7 32.5 37.1 46.9 56.2 66.4 -8.8 real gdp growth rate 5.31 6.31 0.81 3.13 2.83 3.11 formal borrowing (or borrowing from a financial institution) table 3 shows that the level of formal borrowings in nigeria increased between 2011 and 2014 in all population category. the largest increase in formal borrowings was recorded among the uneducated, unemployed, male and older population categories. 10 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 the observed increase in formal borrowings in all the population categories in 2014 was caused by the proliferation of banks with many branch networks across the country. the presence of many bank branches made it easier for citizens to access formal credit in 2014. this achievement was partly attributed to the implementation of the national financial inclusion strategy led by the cbn. the implication is that the expansion of bank branch networks is crucial to expand access to formal credit for nigerian citizens. the level of formal borrowings decreased in 2017 in some population categories, while the largest decrease in formal borrowings was recorded among the poorest, uneducated and unemployed population. the decrease in formal borrowings in some population categories was caused by the effect of the 2016 recession, as well as the slow recovery from the recession. the economic difficulty brought about by the recession led to a reduction in bank credit supply. banks reduced the amount of credit given to retail borrowers due to concerns about rising nonperforming loans. the implication is that the level of formal borrowing falls in a crisis year in nigeria, and the effect tends to extend beyond the crisis year, such as in 2017. the level of formal borrowings increased in 2017 in only in few population categories, such as in the rich and educated population categories. the increase in formal borrowings by these two groups was due to their superior ability to manage risks, and their good financial management skills during the crisis. comparing nigeria with the global average, table 3 shows that the level of formal borrowings in nigeria was very low compared to the global average in all population categories in 2011, 2014 and 2017. the implication is that, despite the government’s effort to increase the size of formal credit through the expansion of bank branch networks, the country’s effort to increase the level of formal borrowing is still very low. finally, the values for real gdp growth and the growth in formal borrowings were positive in 2014. this suggest that the level of formal borrowings in nigeria tend to be higher during periods of economic prosperity. in 2017, real gdp growth remained positive, although lower in magnitude due to the after-effects of the 2016 recession, while the growth in formal borrowings was negative in most population categories and positive in other population categories in 2017. interestingly, nigeria’s real gdp growth was higher than the global average in 2011 and 2014. 11 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 table 3 formal borrowings indicator / country nigeria world (average) % change / growth (nigeria) 2011 2014 2017 2011 2014 2017 (2011-2014) (2014-2017) female (% age 15+) 1.9 4.1 3.7 8.3 9.7 9.4 111.9 -8.8 in labor force (% age 15+) 2.3 5.8 4.6 10.9 13.1 13.4 154.3 -20.5 income, poorest 40% (% age 15+) 1.7 6.1 2.8 8.8 9.1 9.1 246.4 -54.0 income, richest 60% (% age 15+) 2.2 4.7 4.7 9.6 12.1 11.9 109.4 0.2 male (% age 15+) 2.1 6.3 4.1 10.2 12.1 12.2 193.4 -34.6 older adults (% age 25+) 2.03 5.7 3.8 10.6 12.4 12.1 183.2 -33.4 out of labor force (% age 15+) 1.3 4.1 1.9 6.2 6.8 6.2 200.5 -53.3 primary education or less (% age 15+) 1.1 5.9 1.7 7.4 7.4 7.4 412.8 -71.4 rural (% age 15+) 1.8 5.4 2.8 9.0 9.6 10.3 193.3 -48.7 secondary education or more (% age 15+) 2.8 4.8 5.9 11.4 13.7 13.5 71.1 22.6 young adults (% age 15-24) 2.1 4.4 4.3 4.7 5.4 6.3 109.7 -1.2 real gdp growth 5.31 6.31 0.81 3.13 2.83 3.11 borrowing from family or friends table 4 shows that the level of borrowing from family or friends in nigeria decreased between 2011 and 2014 in all population categories. the largest decrease in formal borrowings was recorded among the rural, educated and young adults population categories. the observed decrease in borrowing from family or friends was caused by the presence of banks with many branch networks across the country. the presence of many bank branches provided an alternative source of borrowing for citizens. it reduced the reliance on family and friends for borrowing. the implication is that the expansion of 12 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 bank branch networks is crucial as it provides an alternative source of borrowing for citizens, thereby reducing their reliance on family and friends for borrowings in nigeria. the level of borrowing from family or friends decreased in 2017 in some population categories. the largest decrease in borrowing from family and friends was recorded among the poorest, uneducated and older population categories. the observed decrease was also caused by the expansion of bank branch networks, which provided an alternative source of borrowing for citizens. the implication is that it reduced the dependence on family and friends for borrowings. table 4 borrowings from family and friends indicator / country nigeria world (average) % change / growth (nigeria) 2011 2014 2017 2011 2014 2017 (2011-2014) (2014-2017) adults (% age 15+) 44.1 37.5 28.2 23.0 26.4 25.7 -14.8 -24.6 female (% age 15+) 42.1 37.2 27.8 21.8 25.3 24.4 -11.7 -25.2 in labor force (% age 15+) 48.6 42.3 30.7 26.1 29.2 28.5 -12.8 -27.3 income, poorest 40% (% age 15+) 43.6 37.7 26.2 25.1 28.8 28.9 -13.5 -30.3 income, richest 60% (% age 15+) 44.3 37.4 29.5 21.5 24.8 23.6 -15.6 -20.8 male (% age 15+) 45.9 37.8 28.7 24.2 27.5 27.1 -17.6 -24.2 older adults (% age 25+) 47.8 42.2 26.6 22.4 25.7 25.6 -11.6 -37.1 out of labor force (% age 15+) 31.8 28.5 21.1 18.1 21.5 20.9 -10.4 -26.1 primary education or less (% age 15+) 46.1 44.6 26.8 24.5 29.5 29.1 -3.2 -39.9 rural (% age 15+) 45.4 36.0 28.7 24.1 28.1 27.6 -20.7 -20.1 secondary education or more (% age 15+) 42.3 32.5 30.4 21.2 23.9 23.2 -23.1 -6.8 young adults (% age 15-24) 36.5 28.1 31.8 25.1 28.9 26.5 -23.2 13.2 real gdp growth 5.31 6.31 0.81 3.13 2.83 3.11 comparing nigeria with the global average, table 4 shows that the level of borrowings from family and friends in nigeria was higher than 13 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 the global average in all population categories in 2011. in 2014, the level of borrowings from family and friends in nigeria was also higher than the global average in all population categories except in the young adult population category. in 2017, the level of borrowings from family and friends in nigeria was higher than the global average in the young adult, educated, rural, unemployed, older, male, rich, employed and female population categories. the implication is that a large number of nigerians still rely on family and friends for borrowings. despite the presence of many bank branches in nigeria, the reliance on family and friends for borrowings in nigeria is still very high relative to the world. finally, the value for real gdp growth was positive while the growth in borrowings from family and friends was negative in 2014. this suggest that the level of borrowings from family and friends in nigeria tends to decline during periods of economic prosperity. in 2017, real gdp growth remained positive, although lower in magnitude due to the after-effects of the 2016 recession, while the growth in borrowing from family and friends was negative for most population categories. interestingly, nigeria’s real gdp growth was higher than the global average in 2011 and 2014, but not in 2017. credit card ownership table 5 shows that credit card ownership in nigeria increased between 2011 and 2014 in all population categories. the largest increase in credit card ownership was recorded among the unemployed, male and rich population categories. the observed increase in credit card ownership was due to citizens’ awareness of available overdraft facilities, and easy access to overdraft facilities linked to a credit card which were offered by financial institutions in 2014. therefore, in 2014, the implication was that nigerian citizens were able to use credit cards to meet their basic subsistence and emergency expenditure with the expectation of repayment from future income, thereby improving their access to short term formal credit. in 2017, credit card ownership decreased in many population categories, such as in the female, poor, male, older, unemployed, uneducated, and young adult population categories. the largest decrease in credit card ownership was recorded among the poorest, 14 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 uneducated and unemployed population categories. the observed decrease in credit card ownership in 2017 was caused by the rising interest rate on credit card repayment, as well as banks’ refusal to increase the supply of credit cards due to low demand during the 2016 recession. meanwhile, the richest, rural and educated population categories witnessed an increase in credit card ownership in 2017. comparing nigeria with the global average, table 5 shows that credit card ownership in nigeria was significantly lower than the global average in all population categories in 2011, 2014 and 2017. the implication is that credit cards are not used extensively in nigeria compared to other countries of the world. one reason for this might be because credit cards are not very popular in nigeria. table 5 credit card ownership indicator / country nigeria world (average) % change / growth (nigeria) 2011 2014 2017 2011 2014 2017 (2011-2014) (2014-2017) all adults (% age 15+) 0.8 2.7 2.5 14.9 17.5 18.3 248.7 -6.2 female (% age 15+) 0.8 1.7 1.7 13.4 16.1 16.9 111.3 -2.2 in labor force (% age 15+) 0.9 2.4 3.1 16.4 20.5 22.2 165.6 28.2 income, poorest 40% (% age 15+) 0.7 1.4 0.7 10.7 11.5 12.1 91.2 -48.1 income, richest 60% (% age 15+) 0.8 3.6 3.8 17.7 21.5 22.5 346.2 4.9 male (% age 15+) 0.7 3.7 3.4 16.4 19.0 19.8 392.6 -8.0 older adults (% age 25+) 0.91 3.2 3.04 17.0 19.8 21.2 244.2 -3.6 out of labor force (% age 15+) 0.4 3.3 0.9 10.6 11.5 11.5 684.3 -71.4 primary education or less (% age 15+) 0.6 2.3 0.1 4.1 5.1 6.5 295.9 -94.3 rural (% age 15+) 0.8 1.7 1.7 9.2 12.0 13.9 122.4 4.8 secondary education or more (% age 15+) 0.9 3.1 4.5 27.2 27.8 27.7 221.3 47.1 young adults (% age 15-24) 0.5 2.0 1.8 7.4 9.2 8.2 275.7 -9.3 real gdp growth 5.31 6.31 0.81 3.13 2.83 3.11 finally, the value for real gdp growth was positive, and the growth in credit card ownership was also positive in 2014. this suggest that 15 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 credit card ownership in nigeria tends to increase during periods of economic prosperity. in 2017, real gdp growth remained positive, although lower in magnitude due to the after-effects of the 2016 recession, while the growth in credit card ownership was negative for most population categories. interestingly, nigeria’s real gdp growth was higher than the global average in 2011 and 2014, but not in 2017. debit card ownership table 6 shows that debit card ownership in nigeria increased between 2011 and 2014 in all population categories. the largest increase in debit card ownership was recorded among the uneducated, male and poor population categories. table 6 debit card ownership indicator / country nigeria world (average) % change / growth (nigeria) debit card ownership 2011 2014 2017 2011 2014 2017 (2011-2014) (2014-2017) all adults (% age 15+) 18.5 35.6 31.5 30.7 40.6 47.6 91.9 -11.3 female (% age 15+) 17.7 25.2 22.5 28.1 37.2 43.3 41.9 -10.7 in labor force (% age 15+) 21.3 43.6 34.1 35.1 46.8 54.6 104.9 -21.7 income, poorest 40% (% age 15+) 6.5 22.5 19.6 23.1 31.9 36.3 245.5 -12.6 income, richest 60% (% age 15+) 26.5 44.3 39.4 36.1 46.3 55.2 66.7 -10.9 male (% age 15+) 19.3 45.4 40.1 33.4 44.1 52.1 135.3 -11.8 older adults (% age 25+) 20.6 39.3 35.6 32.6 43.1 51.3 91.1 -9.5 out of labor force (% age 15+) 11.1 20.4 23.9 19.4 28.6 35.6 84.1 17.4 primary education or less (% age 15+) 4.1 18.1 9.6 16.5 22.7 29.3 341.5 -46.6 rural (% age 15+) 13.9 30.6 25.8 23.4 34.5 41.8 119.2 -15.6 secondary education or more (% age 15+) 30.5 48.0 49.5 46.8 55.5 62.4 57.1 3.2 young adults (% age 15-24) 14.5 28.1 25.1 23.7 31.4 34.9 93.5 -10.5 real gdp growth 5.31 6.31 0.81 3.13 2.83 3.11 16 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 the observed increase in debit card ownership was due to citizens’ awareness of available debit cards, which offered convenience when making cash withdrawals. the implication is that nigerian citizens increased their use of debit cards to access their cash deposits at any time without having to visit their physical bank branch. this reduced the number of customers visiting a bank, reduced the waiting hours on long queues, and improved customer satisfaction with banks in 2014. in 2017, debit card ownership decreased in many population categories, such as in the female, poor, male, older, employed, uneducated, and young adult population categories. the largest decrease in debit card ownership was recorded among the uneducated, employed and rural population categories. the observed decrease in debit card ownership in 2017 was caused by a fall in household income. the 2016 recession led to a fall in household disposable income, which reduced the bank account balances of households, thereby leading to a less frequent use of debit cards during the 2016 and 2017 period. meanwhile, the unemployed and educated population categories witnessed an increase in debit card ownership in 2017. comparing nigeria with the global average, table 6 shows that the debit card ownership in nigeria was lower than the global average in all population categories in 2011. in 2014, debit card ownership was also lower than the global average in all population categories except in the male population category. in 2017, debit card ownership was lower than the global average in all population categories. the implication is that, despite the popularity of debit cards in nigeria, a large number of nigerian citizens do not use debit cards extensively compared to citizens in other countries. finally, the value for real gdp growth was positive, and the growth in debit card ownership was also positive in 2014. this suggests that debit card ownership in nigeria tends to increase during periods of economic prosperity. in 2017, real gdp growth remained positive, although lower in magnitude due to the after-effects of the 2016 recession, while the growth in debit card ownership was negative for most population category. interestingly, nigeria’s real gdp growth was higher than the global average in 2011 and 2014, but not in 2017. formal savings (or savings at a formal financial institution) table 7 shows that the level of formal savings in nigeria increased between 2011 and 2014 in all population categories, except in the 17 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 educated population category that witnessed a decrease in formal savings. the decrease in formal savings by the educated population category may be due to their preference for investing rather than saving their money in a financial institution. table 7 formal savings indicator / country nigeria world (average) % change / growth (nigeria) 2011 2014 2017 2011 2014 2017 (2011-2014) (2014-2017) all adults (% age 15+) 23.5 27.1 20.6 22.3 27.3 26.6 14.7 -24.1 out of labor force (% age 15+) 11.9 13.7 12.3 13.9 17.7 17.8 15.02 -10.7 female (% age 15+) 20.9 21.4 13.4 20.9 25.7 24.1 2.4 -37.3 in labor force (% age 15+) 27.8 34.1 23.4 26.1 32.4 31.7 22.4 -31.4 income, poorest 40% (% age 15+) 9.5 17.8 13.4 15.4 19.1 16.6 87.7 -25.4 income, richest 60% (% age 15+) 32.9 33.1 25.3 27.1 32.8 33.4 0.6 -23.6 male (% age 15+) 26.2 32.5 27.3 23.8 29.0 29.3 23.7 -15.8 older adults (% age 25+) 26.7 30.5 22.03 24.5 29.9 29.01 13.9 -27.7 primary education or less (% age 15+) 8.2 17.5 7.5 14.5 19.1 15.7 113.7 -57.5 rural (% age 15+) 18.7 24.5 17.2 18.8 24.6 23.8 31.2 -29.9 secondary education or more (% age 15+) 36.3 33.6 31.3 31.3 34.3 35.4 -7.4 -6.9 young adults (% age 15-24) 17.2 20.02 18.5 15.1 18.1 18.3 15.7 -7.2 the largest increase in formal savings in 2014 was recorded among the uneducated, poor and rural population categories. the observed increase in formal savings borrowings in most of the population categories in 2014 was caused by the general preference for savings than investing. the implication is that such formal savings provided sufficient liquidity to banks and other financial institutions, as they used the saved money as loan for business purposes. in 2017, the level of formal savings decreased in all population categories, while the largest decrease in formal savings was recorded among the female, uneducated and employed population categories. 18 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 the decrease in formal savings in all population categories was caused by the effect of the 2016 recession, as well as the slow recovery from the recession. the economic difficulty brought about by the recession led to the withdrawal of saved monies, for the purpose of consumption in 2016 and in 2017. the implication is that this action reduced the liquidity of banks, and subsequently affected loan supply. this suggests that the level of formal savings falls in a crisis year in nigeria, and the effect tend to extend beyond the crisis year, such as in 2017. comparing nigeria with the global average, table 7 shows that the level of formal savings in nigeria was lower than the global average in most population categories, except in the employed, rich, male, older adult, educated and young adult population categories in 2011. in 2014, the level of formal savings in nigeria was lower than the global average in most population categories, except in the employed, rich, male, older adult, and young adult population categories. in 2017, the level of formal savings was lower than the global average in all population categories. the implication is that a large number of nigerians still rely on savings to escape poverty, rather than using investment as a tool to increase wealth. finally, the values for real gdp growth and the growth in formal savings were positive in 2014. this suggests that the level of formal savings in nigeria tends to be higher during periods of economic prosperity. in 2017, real gdp growth remained positive, although lower in magnitude due to the after-effects of the 2016 recession, while the growth in formal borrowings was negative in all population categories. interestingly, nigeria’s real gdp growth was higher than the global average in 2011 and 2014. population-specific effect: cluster analyses table 8 shows that all adults/households (age 15+) enjoyed greater levels of financial inclusion. it also shows that most of the population categories enjoyed greater account ownership, greater formal borrowing, higher levels of credit card ownership, higher levels of debit card ownership, reduced borrowing from family and friends, and greater formal savings in 2014. the only exception was the young adult population category, which witnessed negative benefit in formal 19 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 savings in 2017. overall, this indicates that nigeria had high levels of financial inclusion in 2014. table 8 population summary matrix for growth in 2014 account ownership formal borrowing borrowing family/friends credit card debit card formal saving population category: all adults (% age 15+) + + + + + out of labor force (% age 15+) + + + + + female (% age 15+) + + + + + in labor force (% age 15+) + + + + + income, poorest 40% (% age 15+) + + + + + income, richest 60% (% age 15+) + + + + + male (% age 15+) + + + + + older adults (% age 25+) + + + + + primary education or less (% age 15+) + + + + + rural (% age 15+) + + + + + secondary education or more (% age 15+) + + + + young adults (% age 15-24) + + + + + in sharp contrast, table 9 shows that all adults/households (age 15+) witnessed negative growth in all indicators of financial inclusion. table 9 shows that the female, poorest, male, older uneducated population categories recorded negative growth in all indicators of financial inclusion in 2017. also, in the unemployed, employed, rich, and educated population categories recorded some positive growth in some financial inclusion indicators, and negative growth in other indicators of financial inclusion indicators. overall, this indicates that nigeria witnessed a decrease in the level of financial inclusion in 2017. 20 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 table 9 population summary matrix for growth in 2017 account ownership formal borrowing borrowing family/friends credit card debit card formal saving population category: all adults (% age 15+) out of labor force (% age 15+) + + female (% age 15+) in labor force (% age 15+) + income, poorest 40% (% age 15+) income, richest 60% (% age 15+) + + male (% age 15+) older adults (% age 25+) primary education or less (% age 15+) rural (% age 15+) + secondary education or more (% age 15+) + + + + young adults (% age 15-24) + conclusion this paper analyzed the level of financial inclusion in nigeria using data from the global findex indicators. the findings reveal that nigeria witnessed growth in several financial inclusion indicators in the early years of financial inclusion in 2014, but the benefits were not sustained in the later years, especially in 2017. in the population categories, the female, poorest, male, older and uneducated population categories were worse-off in all indicators of financial inclusion in 2017. the implication of the observed decline in the level of financial inclusion in 2017 seems to suggest that there are barriers to financial inclusion in the post-2014 years. 21 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 one implication of the findings for policy is that the nigerian government should use appropriate policies that promote sustained financial inclusion. there is no doubt that the nigerian government has made efforts to increase the level of financial inclusion to ensure that all households have access to appropriate financial services they need to improve their lives. however, the efforts made so far are not enough, especially when compared to the global average. to increase financial inclusion, the government should first remove all existing barriers to financial inclusion. then, the government should use appropriate policies to solve existing supply and demand problems in the market for formal financial services in nigeria. the government should also implement financial policies and adopt financial literacy programs that will help to make people know that they can access formal credit and other basic financial services. also, banks should be encouraged to simplify the process of obtaining formal credit, and they should look at financial inclusion as a social responsibility, not just as a profit making opportunity. apart from banks, other players should also be allowed to play an important role in improving the level of financial inclusion. finally, given the impact of financial inclusion on the economy, the government may need more funds or use foreign borrowings to create supporting infrastructure for sustained financial inclusion. this will help to extend financial inclusion programs to a larger segment of the population. one limitation of the study is its focus on financial inclusion at the household levels. it did not examine financial inclusion for small and medium scale enterprises (smes). future studies can investigate other dimensions of financial inclusion in nigeria, particularly how small businesses are affected by the lack of access to basic financial services. also, future studies can identify how the level of financial inclusion can be enhanced through digital finance channels and fintech innovations. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. 22 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 references adegbite, o. o., & machethe, c. l. (2020). bridging the financial inclusion gender gap in smallholder agriculture in nigeria: an untapped potential for sustainable development. world development, 127, 104755. adeola, o., & evans, o. (2017). financial inclusion, financial development, and economic diversification in nigeria. the journal of developing areas, 51(3), 1-15. adetunji, o. m., & david-west, o. (2019). the relative impact of income and financial literacy on financial inclusion in nigeria. journal of international development, 31(4), 312-335. allen, f., demirguc-kunt, a., klapper, l., & peria, m. s. m. (2016). the foundations of financial inclusion: understanding ownership and use of formal accounts. journal of financial intermediation, 100(27), 1-30. bachas, p., gertler, p., higgins, s., & seira, e. (2018, may). digital financial services go a long way: transaction costs and financial inclusion. in aea papers and proceedings (vol. 108, pp. 444-48). buera, f. j., kaboski, j. p., & shin, y. (2011). finance and development: a tale of two sectors. american economic review, 101(5), 1964-2002. david, o. o., oluseyi, a. s., & emmanuel, a. (2018). empirical analysis of the determinants of financial inclusion in nigeria: 1990-2016. journal of finance and economics, 6(1), 19-25. demirguc-kunt, a., & r. levine. (2009). finance and inequality: theory and evidence. annual review of financial economics, 1, 287–318. dev, s. m. (2006). financial inclusion: issues and challenges. economic and political weekly, 4310-4313. demirgüç-kunt, a., & klapper, l. (2012). financial inclusion in africa: an overview. the world bank. demirguc-kunt, a., & klapper, l. (2012). measuring financial inclusion: the global findex database. the world bank. efobi, u. r., orkoh, e., & atata, s. (2018). financial inclusion and foreign market participation of firms: a quasi-experiment from nigeria. in financing sustainable development in africa (pp. 39-61). palgrave macmillan, cham. fowowe, b. (2020). the effects of financial inclusion on agricultural productivity in nigeria. journal of economics and development. 22(1), 2020. doi 10.1108/jed-11-2019-0059 23 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 fungáčová, z., & weill, l. (2015). understanding financial inclusion in china. china economic review, 34, 196-206. grimes, p. w., rogers, k. e., & smith, r. c. (2010). high school economic education and access to financial services. journal of consumer affairs, 44(2), 317-335. ibrahim, s. s., & aliero, h. m. (2020). testing the impact of financial inclusion on income convergence: empirical evidence from nigeria. african development review, 32(1), 42-54. klapper, l., laeven, l., & rajan, r. (2006). entry regulation as a barrier to entrepreneurship. journal of financial economics, 82, 591-629. klapper, l., lusardi, a., & van oudheusden, p. (2015). financial literacy around the world. standard & poor’s ratings services global financial literacy survey. langley, p. (2008). the everyday life of global finance: saving and borrowing in angloamerica. oxford university press marron, d. (2013). governing poverty in a neoliberal age: new labour and the case of financial exclusion. new political economy, 18, 6, 785-810 montgomerie, j. (2008). bridging the critical divide: global finance, financialisation and contemporary capitalism. contemporary politics, 14, 3, 233-52. niankara, i., & muqattash, r. (2018). financial inclusion, saving and borrowing behaviors in the united arab emirates and the united states: a comparative analysis. doi:10.20944/ preprints201807.0445.v1 ozili, p. k. (2018). impact of digital finance on financial inclusion and stability. borsa istanbul review, 18(4), 329-340. ozili, p. k. (2020a). financial inclusion research around the world: a review. in forum for social economics (pp. 1-23). routledge. ozili, p. k. (2020b). contesting digital finance for the poor. digital policy, regulation and governance. ozili, p. k. (2020c). financial inclusion in nigeria: determinants, challenges and achievements. challenges and achievements (march 2020). ozili, p. k. (2021a). has financial inclusion made the financial sector riskier? journal of financial regulation and compliance. 29(3), 237-255. doi 10.1108/jfrc-08-2020-0074 ozili, p. k. (2021b). financial inclusion in africa: what we know. academia.eduhttps://www.academia.edu paramasivan, c., & ganeshkumar, v. (2013). overview of financial inclusion in india. international journal of management and development studies, 2(3), 45-49. 24 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 1–24 reddy, y. v. (2007, february). what rbi means to the common person. speech presented at karamchedu village, andhra pradesh. sarma, m. (2008). index of financial inclusion (no. 215). working paper. sharma, a., & kukreja, s. (2013). an analytical study: relevance of financial inclusion for developing nations. international journal of engineering and science, 2(6), 15-20. sharma, a., & kukreja, s. (2013). an analytical study: relevance of financial inclusion for developing nations. international journal of engineering and science, 2(6), 15-20. sahay, r., čihák, m., n’diaye, p. m. b. p., barajas, a., mitra, s., kyobe, a., mooi, y. n., & yousefi, s. r. (2015). financial inclusion: can it meet multiple macroeconomic goals? international monetary fund. world bank. (2008). finance for all? policies and pitfalls in expanding access. world bank. 51 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: zainol abidin, j., abdullah, n. a. h., & khaw, k. l. h. (2021). bankruptcy prediction: smes in the hospitality industry. international journal of banking and finance, 16(2), 51-80. https://doi.org/10.32890/ ijbf2021.16.2.3 bankruptcy prediction: smes in the hospitality industry 1juraini zainol abidin, 2nur adiana hiau abdullah & 3karren lee-hwei khaw 1&2school of economics, finance and banking universiti utara malaysia, malaysia 3faculty of business and accountancy university of malaya, malaysia 1corresponding author: jurainizainolabidin@yahoo.com.my received: 17/2/2021 revised: 22/3/2021 accepted: 22/3/2021 published: 15/6/2021 abstract the objectives of this study are to predict bankruptcy risk among smes in the hospitality industry for a three-year horizon period and to investigate the factors that are significant in determining bankruptcy. the contribution of smes in the hospitality industry is essential as businesses in the hospitality industry are dominated by sme operators. however, the failure rate among smes is relatively high and almost 50 percent of hospitality establishments do not survive beyond five years of operation. the stepwise logistic model was employed to determine significant predictors that could predict bankruptcy for the period of one year, two years and three 52 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 years before bankruptcy. return on assets and firm age were found to be significant in all periods while other variables were identified to be important at a specific period prior to bankruptcy. in addition to return on assets and firm age, debt ratio and total assets turnover were found to be significant predictors of bankruptcy one-year prior to bankruptcy. however, in the two years prior to bankruptcy, debt ratio and total assets turnover were no longer important but current ratio, ownership concentration and gender diversity were found to be significant. as for the three years prior to bankruptcy, additional variables namely debt-to-equity ratio and board size were found to be significant, but ownership concentration and gender diversity ceased to be important. the findings of this study contribute to the limited literature in predicting the bankruptcy risk of small firms for a threeyear horizon period by providing empirical evidence from smes in the hospitality industry of malaysia. keywords: bankruptcy, hospitality industry, logistic regression, prediction, smes. jel classification: g30, g33. introduction since the outbreak of the coronavirus disease 2019 (covid-19) at the end of 2019, industries have encountered severe challenges, and this scenario is even more challenging for the hospitality industry (hao et al., 2020). malaysia is no exception as the outbreak of covid-19 has adversely affected the hospitality industry and this industry continues to struggle with the impact of covid-19 (foo et al., 2020). as the hospitality industry is the backbone of tourism in malaysia, the future of this industry is at risk of failure (luk, 2020). the small and medium-sized enterprises (smes) operating in the hospitality industry seem to be more affected than their larger counterparts (luk, 2020). this is significant because businesses in the tourism industry are dominated by sme operators (set et al., 2012). smes contribute significantly to malaysia’s economy in the short and long term (yusoff et al., 2018). they represent the vast majority of the business population accounting for 98.5 percent of total enterprises 53 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 and providing 65.3 percent of total employment (department of statistics, 2017). the smes also contribute 38.9 percent to the national gross domestic product (gdp) and 17.9 percent to total exports in 2019 (department of statistics, 2020). in terms of distribution by sub-industry, there are 169,278 smes and 889,619 employees in the food and beverage (f&b), and accommodation business. in addition, the contribution of the tourism industry to malaysia’s gdp is significant, the third highest after manufacturing and commodities (hirschmann, 2020). the gross value added of tourism industries (gvati) recorded a contribution of 15.9 percent to gdp, amounting to rm240.2 billion in 2019 (department of statistics, 2020). both f&b and accommodation contributed 29.8 percent to the total gvati, the second highest after retail trade (department of statistics, 2020). since smes are vital to the nation’s economy, the government continues to support the smes with large budget allocations every year. nonetheless, failure rates among smes are relatively high (kalemliozcan et al., 2020). for example, in australia, an average of 64 percent of small businesses survived over a four-year period from 2014 to 2018 (australian small business and family enterprise ombudsman, 2019). approximately 20 percent of u.s. small businesses fail within the first year, and around 50 percent of them fail within five years, and the rates are seen to be consistent over time (mcintyre, 2020). for the restaurant business, about 70 percent of them remain in business after two years in operation and 50 percent make it through their fifth year (mcintyre, 2020). similarly in malaysia, the smes are unable to sustain their businesses with approximately 60 percent of them failing (chong, 2012), while almost 50 percent of hospitality establishments do not survive beyond five years of operation (abod, 2017). due to this, many studies have been conducted to examine the causes of business failure. since the work of edmister (1972), numerous failure prediction models focusing on smes and utilizing financial, non-financial and governance indicators have been proposed (altman et al., 2010; donato & nieddu, 2020; keasey & watson, 1987; ma’aji et al., 2019). nevertheless, there is a lack of studies focusing on smes in the hospitality industry (pacheco, 2015; zainol abidin et al., 2020). most of the published works on hospitality failure focused on public-listed firms (barreda et al., 2017; fernández-gámez et al., 2016; gemar et 54 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 al., 2019; kim & gu, 2006; park & hancer, 2012; youn & gu, 2010a; 2010b; zhai et al., 2015). similarly, in the context of malaysian literature, the majority of research on corporate failures has focused on public-listed firms (abdullah, 2020; abdullah & ahmad, 2005; alifiah & tahir, 2018; low et al., 2001; yasser & mamun, 2015). however, there are relatively limited studies looking into the smes and most of the researches predict financially distressed smes in the manufacturing sector (abdullah, ahmad et al., 2016, abdullah et al., 2016, abdullah et al., 2019; ma’aji et al., 2018, 2019). in many studies, predictions of business failures were mainly based on short-term analyses, i.e. one year prior to failure. nevertheless, preceding studies indicate that failure is a continuous process and the symptoms of a firm’s failure emerge several years before the actual event (hossari, 2007). this provides a basis that business failure can be predicted for a longer horizon. in light of the recent global pandemic and the uncertainty of doing business, it is of the utmost concern for a firm to predict the risk of business failure a few years ahead. therefore, entrepreneurs can take proactive action to reduce the risk of failure if bankruptcy can be predicted with reasonable accuracy. furthermore, the smes will have sufficient time to prepare for the crisis, so that strategic measures can be taken for a feasible recovery. the purpose of this paper is to examine the factors that could predict the risk of bankruptcy among smes in the hospitality industry in the three years, two years and one year prior to bankruptcy. financial, non-financial and governance variables are utilized as potential bankruptcy predictors and the model accuracy rate is examined using logistic regression (logistic). thus, this study has several contributions to the research on bankruptcy prediction. firstly, we contribute to a somewhat scarce research on bankruptcy prediction for smes in the hospitality industry. secondly, we predict bankruptcy with a longer prediction period of up to three years prior to bankruptcy, and thirdly, we utilize financial, non-financial and governance variables to find predictors that best discriminate between bankrupt and healthy smes in the hospitality industry. this paper is structured as follows: section 2 discusses the literature review, section 3 describes the methodology, section 4 presents the empirical results, and the conclusions are presented in section 5. 55 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 literature review the first study to model small business failure was conducted by edmister (1972). using the multiple discriminant analysis (mda) technique, he examined a sample of 21 financially distressed and 21 healthy us small firms for the period between 1954 and 1969. utilizing19 financial ratios, the results revealed that quick ratio, inventory to sales, equity to sales, fund flow to current liabilities, current liabilities to equity and working capital to sales were the significant predictor variables with an overall classification rate of 93 percent. using a sample of uk small firms, keasey and watson (1987) analyzed a sample of 73 failed and 73 healthy firms from various industries for the period between 1970 and 1983. employing a logistic model, they selected 28 financial indicators and non-financial indicators to predict small enterprise failure. for the financial variables, the logistic model selected pre-tax profit to total assets, profit before interest to total debt, quick ratio, total debt to total assets and fixed assets to total assets as significant determinants of failure. for the non-financial variables, the logistic model identified three-year average accounts submission lag, going concern qualification, audit to submission lag, bank floating charges and the number of directors as important indicators of failure. they reported that the model performed better when using both financial and non-financial information with a total classification rate of 82.2 percent in the testing sample. later, cressy (1992) expanded the work of edmister (1972), and keasey and watson (1987) by demonstrating that a five-year lag structure using financial ratios was able to generate better models. using a logistic model and a sample of 636 small uk firms from different industries, results showed that net profits to total assets, current assets to total assets, net profit relative to total debts and current assets to current liabilities were significant determinants of bankruptcy. in addition, only net profits to total assets were found to be significant in every prior year samples. hence, the author suggested that profitability should be regarded as the important determinant of small firm failure. furthermore, he argued that it was necessary to utilize several years’ data using financial ratios to generate a better classification rate on small firm solvency. 56 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 while cressy (1992) suggested that profitability was an important determinant of failure for uk’s small firms, yazdanfar and nilsson (2008) identified debt ratio as a significant bankruptcy predictor for swedish smes. they constructed two bankruptcy prediction models for a three-year lag utilizing the mda and logistic. using a matched pair’s sample of 1991 bankrupt and 1991 non-bankrupt firms in various industries, the mda model identified quick ratio, debt ratio and profitability ratio as important predictor variables for all periods, while accounts receivable to total assets was found to be significant in the two years and three years before bankruptcy. the logistic model selected only one indicator, i.e. debt ratio as a significant bankruptcy indicator for all periods, while quick ratio, return on assets and accounts receivable to total assets were found significant either in one year, two years or three years before the actual bankruptcy event. the accuracy rates of both the mda and logistic models decreased when the predictive horizon increased. the performance of the logistic model was slightly better than the mda model with an accuracy rate for one year, two years and three years before bankruptcy, at 83.5 percent, 80.5 percent and 77.8 percent, respectively as compared to the mda model at 82.8 percent, 78.1 percent and 75.4 percent, respectively. in line with the above study, altman et al. (2015) constructed a prediction model for a longer period, i.e. up to a 10-year horizon period. using a sample of finnish smes, three logistic models were constructed, i.e. financial model, non-financial model and a combined model (financial and non-financial). they reported that the performance of the combined model was better than the financial and non-financial models. equity to total assets and firm size were the important predictors of bankruptcy in most of the prediction periods. this confirmed that both indicators were the core of the financial dimensions in the long run. the area under the receiver operating characteristic (roc) curve of the model decreased from 0.9482 in the first year to 0.7300 in 10th year in the testing sample. similarly, klepac and hampel (2018) suggested that the model classification rate decreases with the shortened distance to bankruptcy. they developed a bankruptcy prediction model for a three-year period using a sample of manufacturing smes in 28 european union countries. a sample of 170 smes that were declared bankrupt in 2014 57 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 and 830 active smes were utilized for the analysis. the decision tree model identified solvency ratio (net income + depreciation/liabilities) as a significant predictor of bankruptcy for all periods, while interest cover, liquidity ratio and return on assets were significant determinants of bankruptcy at a certain period prior to bankruptcy. the overall model classification rates for the one year, two years and three years prior to bankruptcy was 95.3 percent, 84.8 percent and 80.3 percent, respectively in the testing sample. consistent with findings by yazdanfar and nilsson (2008), abdullah et al. (2019) found that debt ratio was an important indicator of failure for manufacturing smes in malaysia. by employing a logistic model, they developed a four-year prediction model using financial and non-financial variables. the sample consisted of 278, 234, 162 and 58 matched pairs of distress and active smes over the period between 2000 and 2010. debt ratio was found to have significant discriminating power for all periods, while some of the predictors, such as firm size, age, earnings before interest and tax to total assets, short-term liabilities to total liabilities, current ratio, net income to share capital and sales to total assets were found to be significant at a certain period prior to distress. the classification rates of the model for the one year, two years, three years and four years prior to distress was 90 percent, 87.5 percent, 75 percent and 66.5 percent, respectively in the holdout sample. similar to yazdanfar and nilsson (2008), altman et al. (2015), and klepac and hampel (2018), findings of their study revealed that the model prediction accuracy decreased as the period prior to the distress situation increased. furthermore, they concluded that the sign of a firm in financial distress could be detected as early as four years before the actual event occurred. continuing with the same stream of research, papana and spyridou (2020) analyzed a sample of bankrupt and healthy smes in greece for one year, two years and three years prior to bankruptcy. they developed four models, namely linear discriminant analysis (lda), logistic, decision tree and neural network (nn) using financial ratios. findings showed that all the four models identified either profitability or liquidity ratios as significant determinants of bankruptcy. thus, the authors suggested that the smes in greece should be cautious with their liquidity level and the productive use of assets to generate revenue. in terms of the models’ performance, the lda performed 58 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 slightly better than the rest of the models. the overall accuracy rates of the models for the respective one year, two years and three years before bankruptcy in the testing sample were (1) lda – 70.8 percent, 63 percent, 70.5 percent, (2) logistics – 65.8 percent, 62.5 percent, 68 percent, (3) decision tree – 62.5 percent, 62 percent, 61 percent, (4) nn – 70 percent, 65.7 percent, 65.5 percent. thus far, existing empirical studies on hospitality failure have focused on public-listed firms (gemar et al., 2019; gu, 2002; kim, 2011, 2018; kim & upneja, 2014). a limited number of published works have analyzed the likelihood of sme failure in the hospitality industry. pacheco (2015) employed a logistic model to predict bankruptcy using a sample of 25 failed and 460 active smes in the restaurant and accommodation business, and a set of financial ratios as input variables. the results indicated that debt to total assets and equity to total assets were related to the likelihood of failure among smes in portugal. the model classification rate was only 67.8 percent. similarly, zainol abidin et al. (2020) examined the probability of failure among malaysian smes in the restaurant and accommodation business. using financial, non-financial and governance information, they developed two models, i.e. logistic and artificial neural network (ann). the logistic model identified return on assets and board size as significant indicators of failure, while the ann model identified current ratio, debt-to-equity ratio, return on sales, return on assets and board size as significant. since both models suggested board size as an important predictor of failure, this indicated that a firm’s governance was also important for business survival. the overall classification rate of the logistic and ann models in the holdout sample were 80 percent and 92 percent, respectively. in the context of malaysian literature, most of the studies model sme failure in the manufacturing sector (abdullah, ahmad et al., 2016, abdullah et al., 2016, abdullah et al., 2019; ma’aji et al. 2018, 2019) and the only study that has conducted a failure prediction investigation of smes in the hospitality industry was for a single period only, i.e. two years prior to failure (zainol abidin et al., 2020). therefore this study is intended to fill the gap by predicting the failure of smes in the hospitality industry for up to three years in advance. moreover, this study utilized additional governance variables that have not been used in those studies, namely ownership concentration 59 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 related to sme that is 100 percent owned by a holding company and shareholders owned equal shares in the company. a model that can predict financially distressed smes for a period up to three years prior to distress is essential so that the model developed for smes in the hospitality industry can be useful to smes to help in sustaining their businesses. additionally, earlier studies had suggested that ownership concentration was significant in predicting the failure of smes (ciampi, 2015; abdullah et al., 2016; ma’aji et al. 2018). since these additional variables are common in the sme ownership structure, these variables are expected to be significant predictors of failure for smes. methodology sample and data the sample firms consisted of bankrupt and non-bankrupt smes in the hospitality industry. the list of smes with malaysia standard industrial classification (msic) codes of 5510 (short-term accommodation activities) and 5610 (food and beverage activities) was retrieved from the companies commission of malaysia (ccm) database for the period between 2000 and 2016. the ccm database provides information such as company profile, balance sheets and income statements. the samples were chosen based on the sme’s definition endorsed by the national sme development council. smes are defined as having annual sales of up to rm20 million (sme corporation, 2013), while bankrupt smes are defined as those being wound up by a court order or creditor’s request under section 218 (1) (e) and (2) of the companies act 1965. subsequently, after going through the smes’ financial data and other relevant information, subject to data availability, 634 bankrupt smes were identified. the bankrupt smes were matched with non-bankrupt smes on the basis of same firm size (total assets within the range of 10 percent) and same sub-industry. matching sample was required as there could be significant differences between two groups if healthy firms were to be selected at random (jones, 1987). the total smes sampled were 1,268, representing 350 companies (40 accommodation and 310 f&b) one year prior to bankruptcy, 444 firms 60 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 (60 accommodation and 384 f&b) two years prior to bankruptcy and 474 firms (54 accommodation and 420 f&b) three years prior to bankruptcy. as observed, the more we moved closer to the bankruptcy date, the less number of samples of bankrupt companies as most of them were unable to submit financial reports, which resulted in a smaller sample for the year closer to bankruptcy. in line with previous studies, 70 percent of the total sample was utilized as the training sample and the remaining 30 percent was retained as the holdout sample to test the models’ performance (cultrera & brédart, 2016; ptak-chmielewska, 2019). variables the logistic model employed a dichotomous dependent variable for bankruptcy prediction. the dependent variable took the value of one if the sme was bankrupt and zero if the sme was not bankrupt. as there is a lack of established theory that discusses the appropriate predictors of business failure, the independent variables in this study were selected based on previous empirical studies on sme failure and on the availability of the data (balcaen & ooghe, 2006; chancharat, 2011). the analysis considered three categories of input variables namely financial, non-financial and governance as potential indicators of business failure. table 1 presents the list of variables and their descriptions. the selected variables included eight financial variables that were grouped into five categories namely liquidity, leverage, profitability, efficiency and firm size. (1) liquidity measures the ability of a company to meet its short-term debt obligations as they fall due. it is measured using the current ratio (abdullah et al., 2019; cultrera & brédart, 2016; klepac & hampel, 2018; zainol abidin et al., 2020). the higher the ratio, the more liquid the company as it has enough resources to pay off its debt commitments, hence the possibility of failure is low. (2) leverage was represented by debt ratio and debt-to-equity ratio. a higher leverage ratio indicates a higher level of indebtedness that can lead to the risk of default and bankruptcy (abdullah et al., 2019; pacheco, 2015; yazdanfar & nilsson, 2008). (3) profitability measures the ability of a company to generate profit relative to revenue, assets and shareholders’ equity. it is measured using return on assets, return on equity and return on sales (abdullah 61 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 table 1 description of financial, non-financial and governance variables variable category description current ratio financial current assets to current liabilities debt ratio financial total liabilities to total assets debt-to-equity ratio financial total liabilities to total equity return on assets financial net income to total assets return on equity financial net income to total equity return on sales financial net income to sales total assets turnover financial sales to total assets size financial natural logarithm of total assets age nonfinancial natural logarithm of firm age in years ownerc1 governance a dummy variable with a value of 1 if a company is 100 percent owned by a holding company, otherwise 0. ownerc2 governance a dummy variable with a value of 1 if shareholders own equal shares, otherwise 0. ownerc3 governance a dummy variable with a value of 1 if one or more shareholders hold more than 25 percent of the company’s outstanding shares, otherwise 0. ownerc4 governance a dummy variable with a value of 1 if one shareholder holds more than 50 percent of the company’s outstanding shares, otherwise 0. board size governance number of directors gender diversity governance a dummy variable with a value of 1 if there is at least a female director in the boardroom, otherwise 0. 62 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 et al., 2019; klepac & hampel, 2018; zainol abidin et al., 2020). a higher ratio means the business is performing well by generating revenue and profits hence signifying lower bankruptcy probabilities. (4) efficiency was represented by the total assets turnover. a higher ratio implies the ability of a firm to employ its assets effectively to generate sales (abdullah et al., 2019; terdpaopong & mihret, 2011). hence, a negative relationship between efficiency and business failure. (5) size refers to the company size and is measured by the natural logarithm of total assets (abdullah et al., 2019; altman et al., 2010, 2015; back, 2005; ma’aji et al., 2019). as for non-financial variables, age is the company age and is measured by the natural logarithm of company age in years (altman et al., 2010, 2015; ma’aji et al., 2019). with regard to the governance variables, ownerc1 was represented by a dummy variable that took the value of one if a company is 100 percent owned by a holding company, otherwise zero. ownerc2 was a dummy variable that took the value of one if shareholders owned equal shares, otherwise zero. ownerc3 was a dummy variable that took the value of one if one or more shareholders owned more than 25 percent of the firm’s outstanding shares, otherwise zero (abdullah et al., 2016; ma’aji et al., 2019). ownerc4 was a dummy variable that took the value of one if a shareholder owned more than 50 percent of the firm’s outstanding shares; otherwise zero (ciampi, 2015). board size indicated the number of directors in the company (abdullah et al., 2016 ciampi, 2015; ma’aji et al., 2019) and gender diversity constituted a dummy variable that took the value of one if there was at least a woman director on board, otherwise zero (abdullah et al., 2016; ma’aji et al., 2019). method various techniques are available to construct models that can predict business failure. among the most popular statistical methods that are widely used in the corporate failure studies is logistic (shi & li, 2019). shi and li’s (2019) review of 321 papers found that the most frequently used model is logistic, which represented 38.3 percent of the total sample. the logistic model does not require the independent variables to be multivariate normal or groups to have equal covariance matrixes (balcaen & ooghe, 2006). it fits well into the failure 63 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 prediction problem characteristics, where the dependent variable is binary (bankrupt/non-bankrupt), the groups are discrete, identifiable and non-overlapping (ohlson, 1980). the logistic model yields a score between zero and one, which conveniently gives the probability of failure. furthermore, the estimated coefficients were interpreted separately and provided the importance of each of the independent variables in explaining the estimated probability of failure. the coefficient was estimated using the maximum likelihood approach. a cut-off value of 0.5 was utilized to differentiate between bankrupt and non-bankrupt smes in this study. a company was classified as bankrupt if the calculated probability was more than 0.5, otherwise as non-bankrupt. to examine which variables influence the occurrence of bankruptcy, a logistic model is estimated as follows: (1) where p represents the bankruptcy probability, represents the model parameter estimates, and xn represents the input variables. the logistic model was estimated using a forward stepwise method to identify the most significant factors that could predict bankrupt and non-bankrupt smes in the hospitality industry. results descriptive statistics and correlations the descriptive statistics of the independent variables mean difference used to estimate the logistic model is presented in table 2. the results of the financial variables revealed that current ratio, debt ratio, return on assets, return on sales and total assets turnover had a significant mean difference between the bankrupt and non-bankrupt smes throughout one year, two years and three years prior to bankruptcy. however, debt-to-equity ratio was found to be significant one year and two years prior to bankruptcy. as for non-financial variables, the bankrupt smes were significantly younger than the healthy smes throughout the one year, two years and three years prior to bankruptcy. the average age of bankrupt smes was seven to nine is the company age and is measured by the natural logarithm of company age in years (altman et al., 2010, 2015; ma’aji et al., 2019). with regard to the governance variables, ownerc1 was represented by a dummy variable that took the value of one if a company is 100 percent owned by a holding company, otherwise zero. ownerc2 was a dummy variable that took the value of one if shareholders owned equal shares, otherwise zero. ownerc3 was a dummy variable that took the value of one if one or more shareholders owned more than 25 percent of the firm’s outstanding shares, otherwise zero (abdullah et al., 2016; ma’aji et al., 2019). ownerc4 was a dummy variable that took the value of one if a shareholder owned more than 50 percent of the firm’s outstanding shares; otherwise zero (ciampi, 2015). board size indicated the number of directors in the company (abdullah et al., 2016 ciampi, 2015; ma’aji et al., 2019) and gender diversity constituted a dummy variable that took the value of one if there was at least a woman director on board, otherwise zero (abdullah et al., 2016; ma’aji et al., 2019). method various techniques are available to construct models that can predict business failure. among the most popular statistical methods that are widely used in the corporate failure studies is logistic (shi & li, 2019). shi and li’s (2019) review of 321 papers found that the most frequently used model is logistic, which represented 38.3 percent of the total sample. the logistic model does not require the independent variables to be multivariate normal or groups to have equal covariance matrixes (balcaen & ooghe, 2006). it fits well into the failure prediction problem characteristics, where the dependent variable is binary (bankrupt/non-bankrupt), the groups are discrete, identifiable and non-overlapping (ohlson, 1980). the logistic model yields a score between zero and one, which conveniently gives the probability of failure. furthermore, the estimated coefficients were interpreted separately and provided the importance of each of the independent variables in explaining the estimated probability of failure. the coefficient was estimated using the maximum likelihood approach. a cut-off value of 0.5 was utilized to differentiate between bankrupt and non-bankrupt smes in this study. a company was classified as bankrupt if the calculated probability was more than 0.5, otherwise as non-bankrupt. to examine which variables influence the occurrence of bankruptcy, a logistic model is estimated as follows: 𝑃𝑃(𝑌𝑌 = 1) = 1 1 + 𝑒𝑒−(𝛼𝛼+𝛽𝛽1𝑋𝑋1+𝛽𝛽2𝑋𝑋2+⋯+𝛽𝛽𝑛𝑛𝑋𝑋𝑛𝑛) = 1 1 + 𝑒𝑒−𝑦𝑦 (1) where p represents the bankruptcy probability, 𝛽𝛽𝑛𝑛 represents the model parameter estimates, and 𝑋𝑋𝑛𝑛 represents the input variables. the logistic model was estimated using a forward stepwise method to identify the most significant factors that could predict bankrupt and non-bankrupt smes in the hospitality industry. results descriptive statistics and correlations the descriptive statistics of the independent variables mean difference used to estimate the logistic model is presented in table 2. the results of the financial variables revealed that current ratio, debt ratio, return is the company age and is measured by the natural logarithm of company age in years (altman et al., 2010, 2015; ma’aji et al., 2019). with regard to the governance variables, ownerc1 was represented by a dummy variable that took the value of one if a company is 100 percent owned by a holding company, otherwise zero. ownerc2 was a dummy variable that took the value of one if shareholders owned equal shares, otherwise zero. ownerc3 was a dummy variable that took the value of one if one or more shareholders owned more than 25 percent of the firm’s outstanding shares, otherwise zero (abdullah et al., 2016; ma’aji et al., 2019). ownerc4 was a dummy variable that took the value of one if a shareholder owned more than 50 percent of the firm’s outstanding shares; otherwise zero (ciampi, 2015). board size indicated the number of directors in the company (abdullah et al., 2016 ciampi, 2015; ma’aji et al., 2019) and gender diversity constituted a dummy variable that took the value of one if there was at least a woman director on board, otherwise zero (abdullah et al., 2016; ma’aji et al., 2019). method various techniques are available to construct models that can predict business failure. among the most popular statistical methods that are widely used in the corporate failure studies is logistic (shi & li, 2019). shi and li’s (2019) review of 321 papers found that the most frequently used model is logistic, which represented 38.3 percent of the total sample. the logistic model does not require the independent variables to be multivariate normal or groups to have equal covariance matrixes (balcaen & ooghe, 2006). it fits well into the failure prediction problem characteristics, where the dependent variable is binary (bankrupt/non-bankrupt), the groups are discrete, identifiable and non-overlapping (ohlson, 1980). the logistic model yields a score between zero and one, which conveniently gives the probability of failure. furthermore, the estimated coefficients were interpreted separately and provided the importance of each of the independent variables in explaining the estimated probability of failure. the coefficient was estimated using the maximum likelihood approach. a cut-off value of 0.5 was utilized to differentiate between bankrupt and non-bankrupt smes in this study. a company was classified as bankrupt if the calculated probability was more than 0.5, otherwise as non-bankrupt. to examine which variables influence the occurrence of bankruptcy, a logistic model is estimated as follows: 𝑃𝑃(𝑌𝑌 = 1) = 1 1 + 𝑒𝑒−(𝛼𝛼+𝛽𝛽1𝑋𝑋1+𝛽𝛽2𝑋𝑋2+⋯+𝛽𝛽𝑛𝑛𝑋𝑋𝑛𝑛) = 1 1 + 𝑒𝑒−𝑦𝑦 (1) where p represents the bankruptcy probability, 𝛽𝛽𝑛𝑛 represents the model parameter estimates, and 𝑋𝑋𝑛𝑛 represents the input variables. the logistic model was estimated using a forward stepwise method to identify the most significant factors that could predict bankrupt and non-bankrupt smes in the hospitality industry. results descriptive statistics and correlations the descriptive statistics of the independent variables mean difference used to estimate the logistic model is presented in table 2. the results of the financial variables revealed that current ratio, debt ratio, return 64 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 years while the non-bankrupt smes was 12 to 16 years. overall, the findings showed that bankrupt smes were significantly younger, highly leveraged, less liquid, less profitable and less efficient. with regard to governance variables, the non-bankrupt smes had significantly more members in the boardroom compared to the bankrupt smes, two years and three years prior to bankruptcy. it appeared that the non-bankrupt smes had three directors sitting on the board, whereas the bankrupt smes had only two directors sitting on the board. ownership concentration and gender diversity were found to be statistically insignificant for all periods except for gender diversity in the one year prior to bankruptcy, the non-bankrupt smes had more female directors in the boardroom compared to the bankrupt smes. on average, 39 percent of non-bankrupt smes and 32 percent of bankrupt smes had female directors in the boardroom. table 2 descriptive statistics and mean difference test no. variable sample size year prior to bankruptcy bankrupt non-bankrupt bankrupt nonbankruptmean sd mean sd 1 current ratio 350 1 0.5145 0.8090 4.4635 4.7141 -3.9491*** 444 2 0.4981 0.4576 4.9656 9.3740 -4.4675*** 474 3 1.1898 1.9228 2.5336 3.3145 -1.3438*** 2 debt ratio 350 1 2.3210 3.6002 0.3033 0.1731 2.0176*** 444 2 1.6772 2.8302 0.5277 1.0050 1.1496*** 474 3 1.2560 1.2537 0.6856 0.9159 0.5704*** 3 debttoequity ratio 350 1 -0.7437 8.0683 0.5464 0.4862 -1.2900** 444 2 -3.0672 11.7251 0.7078 1.7739 -3.7750*** 474 3 -0.8301 14.5717 0.7963 4.3061 -1.6264 4 return on assets 350 1 -0.5715 1.7564 0.1558 0.2313 -0.7273*** 444 2 -0.3786 0.8150 0.1671 0.2694 -0.5457*** 474 3 -0.1741 0.8352 0.1550 0.1485 -0.3292*** 5 return on equity 350 1 0.2524 2.0217 0.2341 0.3686 0.0183 444 2 0.3650 2.1834 0.2279 0.3161 0.1370 474 3 0.3588 2.0109 0.2374 0.7765 0.1214 (continued) 65 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 no. variable sample size year prior to bankruptcy bankrupt non-bankrupt bankrupt nonbankrupt 6 return on sales 350 1 -0.2743 0.5111 0.0595 0.1027 -0.3337*** 444 2 -0.2501 0.4144 0.0972 0.5230 -0.3472*** 474 3 -0.2642 1.2488 0.0886 0.5037 -0.3527*** 7 total assets turnover 350 1 2.9180 2.5807 4.5193 3.5653 -1.6013*** 444 2 2.4920 2.6575 4.0796 3.9298 -1.5876*** 474 3 2.2908 2.3938 4.0682 3.6267 -1.7774*** 8 size 350 1 12.5180 1.1682 12.5239 1.1559 -0.0059 444 2 12.7718 1.1349 12.7671 1.1365 0.0048 474 3 12.8632 1.1768 12.8640 1.1832 -0.0008 9 age 350 1 1.8797 0.5450 2.3951 0.6071 -0.5154*** 444 2 1.8833 0.5944 2.4791 0.5704 -0.5958*** 474 3 1.9918 0.5603 2.6210 0.4524 -0.6292*** 10 ownerc 350 1 0.5748 0.2213 0.6022 0.2251 -0.0274 444 2 0.5742 0.2329 0.5879 0.2352 -0.0136 474 3 0.5931 0.2266 0.6009 0.2377 -0.0078 11 board size 350 1 2.63 1.007 2.58 0.853 0.0514 444 2 2.68 1.007 2.89 1.005 -0.2072** 474 3 2.65 0.995 3.08 1.319 -0.4219*** 12 gender diversity 350 1 0.3247 0.2779 0.3866 0.2701 -0.0619** 444 2 0.3223 0.2962 0.3480 0.2911 -0.0258 474 3 0.3402 0.2932 0.3749 0.2766 -0.0347 note: *, **and *** significant at 10%, 5% and 1%, respectively. pearson correlation matrix1 between the independent variables were implemented for one year, two years and three years prior to bankruptcy. the findings revealed that the correlations among the variables were relatively low except for debt ratio and return on assets, and ownerc1 and ownerc3, one year prior to bankruptcy and debtto-equity ratio and return on equity, two years prior to bankruptcy. although the low correlations indicated that the dataset did not suffer from multicollinearity problems, variance inflation factor (vif) test was conducted to cross-check the findings. table 3 presents the vif values of each variable for one year, two years and three years prior to bankruptcy, and the findings suggested that multicollinearity was not a problem in this study. 66 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 81–110 table 3 variance inflation factor no. independent variable vif 1 year prior 2 years prior 3 years prior 1 current ratio 1.137 1.078 1.198 2 debt ratio 3.251 1.448 1.277 3 debt-to-equity ratio 1.400 1.702 1.172 4 return on assets 3.089 1.674 1.539 5 return on equity 1.442 1.695 1.223 6 return on sales 1.249 1.286 1.481 7 total assets turnover 1.345 1.403 1.385 8 size 1.512 1.417 1.573 9 age 1.177 1.169 1.141 10 ownerc1 1.116 1.089 1.179 11 ownerc2 1.707 1.626 1.427 12 ownerc3 1.301 1.296 1.367 13 ownerc4 1.702 1.639 1.576 14 board size 1.193 1.293 1.384 15 gender diversity 1.124 1.029 1.086 logistic regression model table 4 presents the results of the three estimated logistic models one year, two years and three years prior to bankruptcy. the hosmer and lemeshow test for all the three models showed that the p-value was insignificant at the 0.05 level suggesting that the models adequately fitted the data. the logistic models identified return on assets and firm age as significant predictors throughout the one year, two years and three years prior to bankruptcy indicating that these were the most influential variables in predicting bankruptcy among smes in the hospitality industry. a significant negative coefficient for return on assets indicated that achieving a high level of profit lowered the likelihood of bankruptcy. when firms recorded sufficient profits, they were able to retain part of the profits and reinvest it for future growth. 67 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 the findings was consistent with cressy’s (1992) as he found return on assets to be an important determinant of bankruptcy throughout a five-year horizon period among small firms in the uk. this also corresponded with findings in previous studies (abdullah et al., 2019; altman et al., 2010; cultrera & brédart, 2016; klepac & hampel, 2018; ma’aji et al., 2019; williams, 2014). the results also yielded a significant negative coefficient between firm age and bankruptcy suggesting that the longer a firm existed, the higher the chances of survival. the resource-based view of the firm argued that older firms had larger resources than younger firms, hence the probability of failure among older firms were lower (williams, 2014). due to this, it was much easier for older firms to deal with unforeseen expenses and operational problems (williams, 2014). furthermore, older firms had a better understanding of their business environment which would enable them to effectively manage highly competitive business environment (ucbasaran et al., 2013). in addition, altman et al. (2015) explained that a young firm had a high risk of failure which diminished over time as the firm aged. this could be seen as the firm progressed from micro, small, medium to being a large, established firm which may take place over a period of years. therefore, an increase in the firm’s years of business operations decreases the probability of bankruptcy. altman et al. (2010), cultrera and brédart (2016), ma’aji et al. (2019) reported similar findings using sme samples. in addition to return on assets and firm age, the logistic model identified debt ratio and total assets turnover as important predictors of bankruptcy, one year before bankruptcy. a significant positive coefficient for debt ratio suggests that the higher the debt level, the higher the chances of a firm going into bankruptcy. too much debt increases the firm’s financial risks, thus increasing its bankruptcy risk and offsetting the tax savings benefit of debt interest (hirshleifer, 1966). the findings of this study concurred with studies by abdullah et al. (2019), altman (1968), pacheco (2015), yazdanfar and nilsson (2008) and youn and gu (2010a). shane (1996) further explained that younger firms tended to borrow more because their owners had limited resources and this could result in large amounts of debt outstanding. consequently, the inability of owners to fulfil their debt commitments could drive firms to financial distress. 68 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 the negative coefficient of total assets turnover indicated that smes with lower efficiency were more likely to fail. it seemed to suggest that bankrupt smes failed to use their assets efficiently and to manage their operations effectively. this inefficiency would make them suffer from negative earnings and in turn would affect their operating cash flow. as a result, they could face difficulties meeting their financial commitments. the failure to meet these commitments could drive the company to financial distress and ultimately to bankruptcy. the results of this study was consistent with terdpaopong and mihret’s (2011) which recorded a negative relationship; despite this, it contradicted the findings of abdullah et al. (2019), four years prior to failure which indicated a positive sign. a possible explanation for a positive sign could be that a company with high total assets turnover but recorded lower profits could indicate high sales volume without good control over its costs, thus increasing the probability of failure. two years prior to bankruptcy, debt ratio and total assets turnover were no longer important and the logistic model identified current ratio, ownerc1 and gender diversity as significant indicators of bankruptcy. the negative coefficient of current ratio suggested that less liquid smes were more likely to fail (altman & sabato 2007; brédart, 2014; klepac & hampel, 2018; lugovskaya, 2010; terdpaopong & mihret, 2011, wellalage & locke, 2012). smes that had low levels of liquidity may have limited cash flows for meeting their working capital requirements and debt obligations. hence, there was a higher probability that these smes could default their financial commitments and eventually be forced into bankruptcy. as for the ownerc1, a positive coefficient revealed that smes that were 100 percent owned by a holding company were more likely to fail. the possible reason was that the decision-making process could be slow due to multiple management levels. major decisions must often go through various chains of command within the parent company bureaucracy before any action can be taken. hence, in a competitive business environment, the ability of a company to make speedy decisions is critical for business survival. the negative coefficient of gender diversity indicated that more female directors in the boardroom were more likely related to survival among smes in the hospitality industry. abdullah et al. (2016) found 69 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 ta bl e 4 st ep w is e lo gi st ic r eg re ss io n m od el v ar ia bl e c at eg or y 1 ye ar p ri or 2 ye ar s pr io r 3 ye ar s pr io r c oe ffi ci en t c ha ng e in -2 l og l ik el ih oo d c oe ffi ci en t c ha ng e in -2 l og l ik el ih oo d c oe ffi ci en t c ha ng e in -2 l og l ik el ih oo d c ur re nt ra tio fi na nc ia l -4 .8 29 80 .8 05 -0 .2 12 8. 08 1 (0 .0 00 )* ** (0 .0 04 )* ** d eb t r at io fi na nc ia l 6. 94 1 72 .2 82 1. 02 2 33 .4 53 (0 .0 00 )* ** (0 .0 00 )* ** d eb tto -e qu ity r at io fi na nc ia l -0 .0 23 2. 95 4 (0 .0 86 )* r et ur n on a ss et s fi na nc ia l -1 4. 08 0 39 .0 48 -3 9. 12 1 76 .6 80 -0 .3 78 3. 49 4 (0 .0 00 )* ** (0 .0 00 )* ** (0 .0 62 )* to ta l a ss et s tu rn ov er fi na nc ia l -0 .5 67 14 .3 03 -0 .4 36 54 .9 24 (0 .0 00 )* ** (0 .0 00 )* ** a ge n on -fi na nc ia l -0 .8 67 4. 08 8 -1 .1 56 3. 19 3 -2 .6 38 86 .2 50 (0 .0 43 )* * (0 .0 74 )* (0 .0 00 )* ** (c on tin ue d) 70 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 v ar ia bl e c at eg or y 1 ye ar p ri or 2 ye ar s pr io r 3 ye ar s pr io r c oe ffi ci en t c ha ng e in -2 l og l ik el ih oo d c oe ffi ci en t c ha ng e in -2 l og l ik el ih oo d c oe ffi ci en t c ha ng e in -2 l og l ik el ih oo d su bs id ia ry g ov er na nc e 8. 47 3 10 .4 11 (0 .0 01 )* ** g en de r d iv er si ty g ov er na nc e -3 .9 84 13 .1 21 -0 .9 18 6. 62 6 (0 .0 00 )* ** (0 .0 10 )* ** b oa rd s iz e g ov er na nc e -0 .3 42 6. 11 3 (0 .0 13 )* * h os m er & l em es ho w te st 12 .5 58 1. 04 0 14 .1 60 (0 .1 28 ) (0 .9 98 ) (0 .0 78 ) c on st an t 0. 03 3 12 .6 45 8. 53 5 (0 .9 78 ) (0 .0 00 ) (0 .0 00 ) n ot e: * , * *a nd * ** s ig ni fic an t a t 1 0% , 5 % a nd 1 % , r es pe ct iv el y. 71 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 that manufacturing smes with female managing directors were less likely to fail than those having male counterparts. previous studies suggested that firm performance improved with the presence of female directors in the boardroom (campbell & minguez-vera, 2008; julizaerma & mohamad sori, 2012; post & byron, 2015). the possible explanation could be that more diverse board members could bring their own personal background to their board position as their experience is more extensive and comprehensive. bassett-jones (2005) argued that diverse perspectives could influence board members to critically analyse complex problems and develop creative and innovative solutions. robinson and dechant (1997) further suggested that female directors performed better than their male counterparts in group problem solving and decision-making that required discussion and consensus, thus this could lead to board effectiveness. three years prior to bankruptcy, additional variables namely debtto-equity ratio and board size were established as significant, but ownerc1 and gender diversity were no longer important. the positive coefficient of debt-to-equity ratio implied that highly leveraged smes were more likely to fail. generally, smes rely heavily on debt to finance their business operations (altman, 2010). hence, a firm has to carry a bigger burden when interest payment takes a significant portion of the firm’s profit and this may affect its operating profits. a highly leveraged firm would be generating insufficient profit and cash flows to cover its debt obligations, thus was more likely to face bankruptcy risk. the findings concurred with those of back et al. (1996), ciampi (2015), kim (2011), and terdpaopong and mihret (2011). the negative coefficient of board size suggested that more directors in the boardroom increased the likelihood of sme survival (abdullah et al., 2016; zainol abidin et al., 2020; keasey & watson, 1987). the result was consistent with that of ma’aji et al. (2019), who concluded that the probability of failure among smes was lower when the board size was larger due to increased oversight and expertise. drawing from the expertise of its many members, the firm was expected to achieve better performance. further, larger boardroom members would provide better perspectives and ideas that could lead to more in-depth and thorough consideration of issues. farag and mallin (2017) suggested that the quality of decisions made by firms with larger boards were more superior. 72 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 table 5 presents the classification results of the logistic models one year, two years and three years before bankruptcy. one year prior to bankruptcy, the model correctly classified 93.5 percent of the bankrupt smes and 99.1 percent of the non-bankrupt smes in the estimation sample and 94.2 percent of the bankrupt smes and 93.1 percent of the non-bankrupt smes in the holdout sample. this resulted in an overall classification rate of 96.3 percent and 93.6 percent in the estimation and holdout samples, respectively. the model seemed to perform better in predicting the bankrupt smes than the non-bankrupt smes in the holdout sample. furthermore, the model performed well in predicting bankruptcy two years prior with an overall classification rate of 98.1 percent in the estimation sample and 95.5 percent in the holdout sample. however, the accuracy rate of the model decreased when predicting bankruptcy three years before the bankruptcy event. the overall accuracy rate was 85.0 percent and 76.4 percent in the estimation and holdout samples, respectively. the results showed that an increase in the bankruptcy prediction period resulted in a decrease in the prediction model accuracy rate. similar results were reported by abdullah et al. (2019), klepac and hampel (2018), and yazdanfar and nilsson (2008). table 5 classification accuracy rate percentage correctly classified 1 year prior 2 years prior 3 years prior estimation sample holdout sample estimation sample holdout sample estimation sample holdout sample bankrupt 93.5 94.2 98.1 95.5 83.3 76.8 nonbankrupt 99.1 93.1 98.1 95.6 86.7 76.1 overall 96.3 93.6 98.1 95.5 85.0 76.4 observation 240 110 310 134 334 140 conclusion the study used empirical data of smes in the hospitality industry and assessed the prediction accuracy of logistic models using a set 73 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 of financial, non-financial and governance variables on a threeyear horizon period. the findings revealed that return on assets and firm age was consistently significant throughout the one year, two years and three years prior to bankruptcy. smes that recorded sufficient profits were able to retain part of their profits and over the years they would have larger resources that would enable them to handle financial crisis more easily, hence increasing their chances of survival. as for other indicators, they were only found to be significant at a certain period before bankruptcy. for the one year prior to bankruptcy, debt ratio and total assets turnover were found to be significant predictors of bankruptcy, while current ratio, ownerc1 and gender diversity were found to be significant in the two years prior to bankruptcy. other variables were found to be significant in three years prior to bankruptcy namely current ratio, debt ratio, debtto-equity ratio, total assets turnover, gender diversity and board size. in addition, governance variables seemed to be significant during the early periods of bankruptcy. the results showed that having female directors and more members in the boardroom during the early periods could contribute to the success of a business. furthermore, as smes moved closer to bankruptcy, the classification accuracy rate of the model increased with a less number but significant variables which were identified to predict bankruptcy. the findings of this study also showed that the bankruptcy risk of smes could be detected as early as three years in advance. the logistic model emphasized the importance of return on assets and firm age to predict bankruptcy risk of smes in the hospitality industry as these variables have always been significant in discriminating between bankrupt and non-bankrupt smes for all periods. achieving a sufficiently high level of profit is crucial in sustaining long run business growth. furthermore, younger companies have a lower chance of survival as they have limited cash reserves for their business operations. therefore, an effective failure prediction model could reduce economic losses for the affected parties by providing signals that would enable them to take preventive measures in possible adverse situations. lenders could use the model to assess the risk of loan defaulters, while creditors could use the results of this study as a tool to evaluate the creditworthiness of their potential debtors. investors could use the 74 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 model to assess the financial health of a company before investing, thus minimizing their investment risks. for the regulators and government agencies that are managing the smes, improved policies are required such as providing special funds for young companies, debt limit, profit margin, total assets turnover and setting the minimum number of board of directors. as in other industries, there must be a guideline for loan disbursement to smes in order for the government not to lose its investments and to improve the hospitality industry. a limitation of this study is that other variables which could impact the smes such as business location and external factors were not considered as these could also contribute in analyzing the bankruptcy risk of smes in the hospitality industry. acknowledgment this research received no specific grant from any funding agency in the public. endnote 1 the results of pearson correlation matrix for one year, two years and three years prior to bankruptcy are available upon request. references abdullah, n. a. h., & ahmad, a. h. (2005). detecting financial distress. international journal of management studies, 12(1), 77–97. abdullah, n. a. h., ahmad, a. h., zainudin, n., & md rus, r. (2016). modelling small and medium-sized enterprises failure in malaysia. international journal of entrepreneurship and small business, 28(1), 101–116. abdullah, n. a. h., ma’aji, m. m., & khaw, k. l. h. (2016). the value of governance variables in predicting financial distress among small and medium-sized enterprises in malaysia. asian academy of management journal of accounting & finance, 12(suppl. 1), 77–91. 75 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 abdullah, n. a. h., ahmad, a. h., zainudin, n., & md rus, r. (2019). predicting financially distressed small-and mediumsized enterprises in malaysia. global business review, 20(3), 627–639. abdullah, m. a. (2020). identifying the determinants of financial distress for public listed companies in malaysia. journal pengurusan (ukm journal of management), 59. https:// ejournal.ukm.my/pengurusan/article/view/35017 abod, s. g. (2017, september). surviving and thriving in the food & beverage industry. bizpulse, 18. https://www.smebank.com. my/images/pdf/bizpulse/sme-bank-bizpulse-issue-18.pdf alifiah, m. n., & tahir, m. s. (2018). predicting financial distress companies in the manufacturing and non-manufacturing sectors in malaysia using macroeconomic variables. management science letters, 8, 593–604. altman, e. i. (1968). financial ratios, discriminant analysis and the prediction of corporate bankruptcy. journal of finance, 22, 589-610. altman, e. i., & sabato, g. (2007). modelling credit risk for smes: evidence from the u.s. market. abacus, 43(3), 332–357. altman, e. i., sabato, g., & wilson, n. (2010). the value of nonfinancial information in small and medium-sized enterprise risk management. journal of credit risk, 6(2), 1–33. altman, e. i., iwanicz-drozdowska, m., laitinen, e. k., & suvas, a. (2015). financial and non-financial variables as long-horizon predictors of bankruptcy (september 5, 2015). https://ssrn.com/ abstract=2669668 australian small business and family enterprise ombudsman. (2019). small business counts. commonwealth of australia. back, b., laitinen, t., sere, k., & van wezel, m. (1996). choosing bankruptcy predictors using discriminant analysis, logit analysis, and genetic algorithms. (technical report no. 40). turku centre for computer science. back, p. (2005). explaining financial difficulties based on previous payment behavior, management background variables and financial ratios. european accounting review, 14(4), 839–868. balcaen, s., & ooghe, h. (2006). 35 years of studies on business failure: an overview of the classic statistical methodologies and their related problems. the british accounting review, 38(1), 63–93. 76 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 barreda, a. a., kageyama, y., singh, d., & zubieta, s. (2017). hospitality bankruptcy in united states of america: a multiple discriminant analysis-logit model comparison. journal of quality assurance in hospitality & tourism, 18(1), 86–106. bassett-jones, n. (2005). the paradox of diversity management, creativity and innovation. creativity and innovation management, 14(2), 169–175. brédart, x. (2014). bankruptcy prediction model using neural networks. accounting and finance research, 3(2), 124–128. campbell, k., & minguez vera, a. (2008). gender diversity in the boardroom and firm financial performance. journal of business ethics, 83(3), 435–451. chancharat, n. (2011). smes failure prediction: literature review. abac journal, 31(3), 70–81. chong, w. y. (2012). critical success factors for small and medium enterprises: perceptions of entrepreneurs in urban malaysia. journal of business and policy research, 7(4), 204–215. ciampi, f. (2015). corporate governance characteristics and default prediction modeling for small enterprises. an empirical analysis of italian firms. journal of business research, 68(5), 1012–1025. cressy, r. c. (1992). u.k. small firm bankruptcy prediction: a logit analysis of financial trend-, industry-, and macro-effects. journal of small business finance, 1(3), 233–253. cultrera, l., & brédart, x. (2016). bankruptcy prediction: the case of belgian smes. review of accounting and finance, 15(1), 101–119. department of statistics. (2017). economic census 2016: profile of small and medium enterprises. department of statistics malaysia, putrajaya. department of statistics. (2020). small and medium enterprises (smes) performance 2019. department of statistics malaysia, putrajaya. department of statistics. (2020). tourism satellite account 2019. department of statistics malaysia, putrajaya. donato, f., & nieddu, l. (2020). corporate failure: bankruptcy prediction for italian smes based on a longitudinal case study from 2000 to 2011. corporate ownership & control, 17(3), 27-33. 77 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 edmister, r. (1972). an empirical test of financial ratio analysis for small business failure prediction. journal of financial and quantitative analysis, 7(2), 1477–1493. farag, h., & mallin, c. (2017). board diversity and financial fragility: evidence from european banks. international review of financial analysis, 49(c), 98–112. fernández-gámez, m. á., cisneros-ruiz, a. j., & callejon-gil, a. (2016). applying a probabilistic neural network to hotel bankruptcy prediction. tourism & management studies, 12(1), 40–52. foo, l. p., chin., m. y., tan, k. l., & phuah, k. t. (2020). the impact of covid-19 on tourism industry in malaysia. current issues in tourism. https://doi.org/10.1080/13683500.2020.1777951 gemar, g. p., soler, i., & guzman-parra, v. f. (2019). predicting bankruptcy in resort hotels: a survival analysis. international journal of contemporary hospitality management, 31(4), 1546–1566. gu, z. (2002). analyzing bankruptcy in the restaurant industry: a multiple discriminant model. international journal of hospitality management, 21(1), 25–42. hao, f., xiao, q., & chon, k. (2020). covid-19 and china’s hotel industry: impacts, a disaster management framework, and post-pandemic agenda. international journal of hospitality management, 90, 1–11. hirschmann, r. (2020, dec 8). travel and tourism in malaysia statistics & facts. statista. https://www.statista.com/ topics/5741/travel-and-tourism-in-malaysia/ hirshleifer, j. (1966). investment decision under uncertainty: applications of the state-preference approach. the quarterly journal of economics, 80(2), 252–277. hossari, g. (2007). benchmarking new statistical techniques in rationbased modelling of corporate collapse. international review of business research papers, 3(3), 141–161. jones, f. l. (1987). current techniques in bankruptcy prediction. journal of accounting literature, 6, 131-164. julizaerma, m. k., & sori, z. m. (2012). gender diversity in the boardroom and firm performance of malaysian public listed companies. procedia social and behavioral sciences, 65, 1077–1085. 78 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 kalemli-ozcan, s., gourinchas, p. o., penciakova, v., & sander, n. (2020). covid-19 and sme failures. (imf working paper no. wp/20/207). international monetary fund. https://www.imf. org/en/publications/wp/issues/2020/09/25/covid-19-andsme-failures-49753 keasey, k., & watson, r. (1987). non-financial symptoms and the prediction of small company failure: a test of argentina’s hypotheses. journal of business finance & accounting, 14(3), 335–354. kim, h., & gu, z. (2006). a logistic regression analysis for predicting bankruptcy in the hospitality industry. the journal of hospitality financial management, 14(1), 17–34. kim, s. y. (2011). prediction of hotel bankruptcy using support vector machine, artificial neural network, logistic regression, and multivariate discriminant analysis. the service industries journal, 31(3), 441–468. kim, s. y., & upneja, a. (2014). predicting restaurant financial distress using decision tree and adaboosted decision tree models. economic modelling, 36(c), 354–362. kim, s. y. (2018). predicting hospitality financial distress with ensemble models: the case of us hotels, restaurants, and amusement and recreation. service business, 12(3), 483–503. klepac, v., & hampel, d. (2018). predicting bankruptcy of manufacturing companies in eu. ekonomie a management, 21(1), 159–174. low, s.-w., mat nor, f., & yatim, p. (2001). predicting corporate financial distress using the logit model: the case of malaysia. asian academy of management journal, 6(1), 49–61. lugovskaya l. (2010). predicting default of russian smes on the basis of financial and non-financial variables. journal of financial services marketing, 14, 301–313. luk., t. m. (2020, october 21). budget 2021: our wish for the sme hospitality industry. businesstoday. https://www. businesstoday.com.my/2020/10/21/budget-2021-our-wish-forthe-sme-hospitality-industry/ ma’aji, m. m., abdullah, n. a. h., & khaw, k. l. h. (2018). predicting financial distress among smes in malaysia. european scientific journal, 14(7), 91–102. ma’aji, m. m., abdullah, n. a. h., & khaw, k. l. h. (2019). financial distress among smes in malaysia: an early warning signal. international journal of business and society, 20(2), 775–792. 79 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 mcintyre, g. (2020, november 20). what percentage of small businesses fail? (and other need-to-know stats). fundera. https://www.fundera.com/blog/what-percentage-of-smallbusinesses-fail. ohlson, j. a. (1980). financial ratios and the probabilistic prediction of bankruptcy. journal of accounting research, 109–131. pacheco, l. (2015). smes probability of default: the case of the hospitality sector. tourism & management studies, 11(1), 153– 159. papana, a., & spyridou, a. (2020). bankruptcy prediction: the case of the greek market. forecasting, 2(4), 505–525. park, s. s., & hancer, m. (2012). a comparative study of logit and artificial neural networks in predicting bankruptcy in the hospitality industry. tourism economics, 18(2), 311–338. post, c., & byron, k. (2015). women on boards and firm financial performance: a meta-analysis. the academy of management journal, 58(5), 1546–1571. ptak-chmielewska, a. (2019). predicting micro-enterprise failures using data mining techniques. journal of risk financial management, 12(1), 1–17. robinson, g., & dechant, k. (1997). building a business case for diversity. the academy of management executive, 11(3), 21– 31. set, k., mahmood, a., & agbola, f. (2012). developing small and medium-sized tourism enterprises (tsmes): the malaysian experience. in n. che ha, & a. mahmood (eds.), malaysian smes in the new economy (pp. 220–237). sme corporation. (2013). guideline for new sme definition. https:// www.smecorp.gov.my/images/pdf/guideline_for_new_sme_ definition_7jan2014.pdf. shane, s. (1996). hybrid organizational arrangements and their implications for company growth and survival: a study of new franchisors. academy of management journal, 39(1), 216–234. shi, y., & li, x. (2019). an overview of bankruptcy prediction models for corporate firms: a systematic literature review. intangible capital, 15(2), 114–127. terdpaopong, k., & mihret, d. g. (2011). modelling sme credit risk: a thai empirical evidence. small enterprise research, 18(1), 63–79. 80 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 51–80 ucbasaran, d., shepherd, d. a, lockett, a., & lyon., s. j. (2013). life after business failure: the process and consequences of business failure for entrepreneurs. journal of management, 39(1), 163–202. wellalage, n. h., & locke, s. (2012). factors affecting the probability of sme bankruptcy: a case study on new zealand unlisted firms. business journal for entrepreneurs (quarterly). 1–17. williams, d. a. (2014). resources and business failure in smes: does size matter? journal of business and management, 20(2), 89–102. yasser, q. r., & mamun, a. (2015). corporate failure prediction of public listed companies in malaysia. european researcher, 91(2), 114–126. yazdanfar, d., & nilsson, m. (2008, november 5–7). the bankruptcy determinants of swedish smes. proceedings of isbe international entrepreneurship conference, pp. 1–14, belfast, n. ireland. http://www.diva-portal.org/smash/get/ diva2:573128/fulltext01.pdf youn, h., & gu, z. (2010a). predict us restaurant firm failures: the artificial neural network model versus logistic regression model. tourism and hospitality research, 10(3), 171–187. youn, h., & gu, z. (2010b). predicting korean lodging firm failures: an artificial neural network model along with a logistic regression model. international journal of hospitality management, 29(1), 120–127. yusoff, t., wahab, s. a., ahmad s. a., lariff, a. s. a, osman, s. i. w., zawawi1, n. f. m., & fazal, s. a. (2018). sustainable growth in smes: a review from the malaysian perspective. journal of management and sustainability, 8(3), 43–54. zainol abidin, j., abdullah, n. a. h., & khaw, k. l. h. (2020). predicting smes failure: logistic regression vs artificial neural network models. capital markets review, 28(2), 29–41. zhai, s-s., choi, j-g., & kwansa, f. (2015). a financial ratio-based predicting model for hotel business failure. global business & finance review, 20(1), 71–86. 55 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 how to cite this article: ozili, p. k. (2021). bank profitability determinants: comparing the united states, nigeria and south africa. international journal of banking and finance, 16(1), 55-78. bank profitability determinants: comparing the united states, nigeria and south africa peterson k. ozili central bank of nigeria, abuja, nigeria corresponding author: petersonkitakogelu@yahoo.com received: 8/11/2020 revised: 21/11/2020 accepted: 30/11/2020 published: 30/1/2021 abstract this study investigates the determinants of banking sector profitability in south africa, nigeria and the united states. the findings reveal that cost efficiency, the size of non-performing loans and overhead cost to total asset ratio are significant determinants of the banking sector profitability. in the comparative analysis, the findings from south africa show that the cost efficiency ratio, overhead cost to total asset ratio and non-performing loans are significant determinants of the banking sector profitability. in the united states, capital adequacy ratio and the size of non-performing loans are significant determinants of its banking sector profitability. in nigeria, the overhead cost to total asset ratio and cost efficiency ratio are significant determinants of the banking sector profitability. the descriptive analysis reveal that bank net interest margin and return on asset are higher in nigeria and lowest in the united states which suggests that the nigerian banking sector is more profitable than the us banking sector. return on equity is higher in south africa and lowest in the united states. http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance https://doi.org/10.32890/ijbf2021.16.1.4 56 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 keywords: banks, profitability, non-performing loans, efficiency, nigeria, south africa, united states. jel classification: g21, g28 introduction this paper investigates the determinants of the banking sector profitability in nigeria, south africa and the united states. banking sector profitability is an important indicator of a stable financial sector. country-specific differences can affect the level of bank profit in unique ways. these differences may be amplified by differences in the level of financial development and the level of country development especially the differences in developed countries, emerging countries and developing countries. the focus on nigeria, south africa and the united states is due to the differences in the level of the financial sector and country development in the three countries. the literature shows that the level of financial (sector) development significantly affects bank profitability (see demirgüç-kunt & huizinga, 2000). the united states has a high level of financial development followed by south africa while nigeria has the lowest according to the world bank’s global financial development indicators. more so, the united states is a market-based economy where only 15 percent of total credit in the us financial system are supplied by banks. the remaining 85 percent of total credit are supplied by other financial institutions. in contrast, the financial system in nigeria is dominated by banks where 64 percent of total credit supply is provided by the banking sector. the implication is that a significant drop in credit supply by nigerian banks will significantly affect nigeria’s financial system. south africa, on the other hand, is a bank-based economy where banks control about 52 percent of total credit supply which leaves room for other capital market lenders to offer loans to corporate borrowers. arguably, the financial development differences in nigeria, south africa and the united states may explain the differences in bank profitability in the three countries. many studies focussed on bank profitability determinants in several contexts (e.g. athanasoglou et al. (2008); borio et al. (2017); zheng et al. (2017); boungou (2019); ali and puah (2018); batten and vo (2019) and huang (2020)). but studies that explicitly compare bank 57 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 profitability determinants between countries are scarce in the prior literature. this study contributes to the existing literature by exploring the determinants of banking sector profitability across countries. secondly, this study is related to the bank stability literature as banking sector profitability is an important predictor of bank stability. also, this study shows whether country development differences have a direct effect on bank profitability determinants. the rest of the paper is structured as follows. section 2 presents the related literature on bank profitability. section 3 discusses the research design. section 4 presents the empirical findings, while section 5 reports the conclusion. literature review several studies examined the determinants of bank profitability. bougatef (2017) found that a higher level of corruption is associated with higher bank profitability in tunisia. ozili (2017) investigated bank profitability determinants amongst african banks, and found that factors such as size of the bank, regulatory capital ratio and provisions for loan loss are significant determinants of the return on assets of listed banks in africa. borio et al. (2017) investigated the impact of monetary policy on bank profitability in 14 major advanced economies from 1995 to 2012. they found that the level of the short-term rate has a positive impact on bank profitability measured as return on assets. zheng et al. (2017) examined the effect of capital requirements on the profitability of banks in bangladesh from 2000 to 2015. they found that higher regulatory capital ratios increased the profitability of banks in bangladesh. their results remained the same when the authors used equity to total assets ratio as an alternative measure of bank capital. bonaccorsi di patti and palazzo (2018) investigated the effects of macroeconomic factors on the profitability of banks in the european union (eu), and found that growth in gdp and loan growth influence the profitability of eu banks. ozili (2015) investigated the determinants of bank profitability in nigeria, and found that non-performing loans, size of the bank and cost efficiency significantly affect the profitability of banks in nigeria. hesse and poghosyan (2016) analysed the effects of oil price shocks on bank profitability for 145 banks in 11 oil-exporting mena 58 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 countries from 1994 to 2008. they found that oil price shocks have an indirect impact on bank profitability, and the indirect impact is channelled through country-specific macroeconomic and institutional variables. ammar and boughrara (2019) investigated the effects of revenue diversification on bank profitability in 14 middle east and north african (mena) countries from 1990 to 2011. they found that revenue diversification leads to higher bank profitability. bouzgarrou et al. (2018) examined the profitability of domestic banks and foreign banks prior to the 2008 financial crisis and during the financial crisis. they found that foreign banks are more profitable compared with domestic banks during the 2008 financial crisis. bolarinwa et al. (2019) examined the determinants of bank profitability in nigeria. they examined the effects of bank size, deposit growth, credit risk, capital ratio and cost efficiency on commercial bank profitability. they analysed 15 commercial banks from 2005 to 2015, and found that cost efficiency is a determinant of bank profitability in nigeria. the current study is different from bolarinwa et al. (2019). they used bank level data while the current study examines bank profitability determinants using industry data for a longer period of 1996 to 2017. overall, the above studies, in their analyses, did not compare developed countries with emerging countries and developing countries. the current study adds to the literature by explicitly comparing the profitability determinants of banks in a major developed country (the united states), emerging country (south africa) and a developing country (nigeria). data and methodology data country-level bank information was collected from the global financial development indicators while macroeconomic data was collected from the world development indicators in the world bank databank. data was extracted for three countries with unique attributes: nigeria (representing a major developing country), south africa (representing a major emerging country) and the united states (representing a major developed country). the sample period spans 22 years from 1996 to 2017. the sample period is sufficient to cover two economic cycles. 59 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 variable justification π is the dependent variable, representing a vector of bank profitability variables. the vector variable consists of the net income margin (nim), return on asset before tax (roabt) and return on equity before tax (roebt) variables. the nim, roabt and roebt variables are widely used in the literature to measure bank profitability (ben naceur & goaied, 2008; borio et al., 2017; and ozili & uadiale, 2017). bank concentration (cn) variable represents bank concentration. high bank concentration signifies greater market power for banks. such banks enjoy oligopolistic advantage in the credit market, which puts them in a position to charge high interest rates on loans and high fees for offering non-interest services to bank customers, thereby increasing their profit levels (huang, 2020; ozili & uadiale, 2017). thus, the relationship between cn and π is predicted to be positive. cap variable represents the capital adequacy ratio. ideally, banks that have high capital ratio are more likely to engage in risky lending because they have the equity capital needed to absorb unexpected losses that may arise from risky lending (batten & vo, 2019; hallunovi & berdo, 2018). the higher the risk, the higher the return or profit. thus, the relationship between cap and π is predicted to be positive. eff variable represents the cost efficiency ratio. efficient banks tend to have a low cost-to-income ratio because they are able to minimise cost and maximise income, and as a result, such banks are able to generate high profit levels (bitar et al., 2018). thus, the relationship between eff and π is predicted to be negative. opta variable is the overhead cost to total asset ratio. banks that have high overhead cost will have lower profit levels due to rising overhead expenses (serwadda, 2018). thus, the relationship between opta and π is predicted to be negative. npl variable is non-performing loans to gross loan ratio. ideally, banks that have high problem loans will have lower net interest income and reduced net profit (panta, 2018). thus, the relationship between npl and π is predicted to be negative. inf variable is the inflation rate. high inflation can make banks increase the price of loans and increase the fee charged for offering 60 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 non-interest activities to bank customers (bouzgarrou et al., 2018), thus leading to high bank profitability. thus, the relationship between inf and π is predicted to be positive. gdpr variable represents the growth in gdp. banks tend to report high profit levels in times of economic prosperity. this is because a large number of debtors are able to repay their loan during periods of economic prosperity, thereby contributing to high interest income for banks. in contrast, banks often report low profit levels in recessionary periods because a large number of debtors may default on their loan repayment, thereby contributing to reduced interest income for banks (kohlscheen et al., 2018). thus, the relationship between gdpr and π is predicted to be positive. model specification the model used to analyse the determinants of bank profitability (π) is similar to the models used in prior studies such as huang (2020); ozili and uadiale (2017) and borio et al. (2017). π = f (bank specific determinants, macroeconomic determinants). the econometric model is specified below as: (1) π = vector of dependent variables, namely, roabt, roebt and nim. where: cap = bank capital to total assets (%); cn = bank concentration (%); eff = bank cost to income ratio (%); nim = bank net interest margin (%); npl = bank non-performing loans to gross loans (%); opta = bank overhead costs to total assets (%); roabt = bank return on assets (before tax); roebt = bank return on equity (before tax); inf = inflation, consumer prices (annual %); gdpr = gdp growth (annual %). estimation procedure robust least squares estimation technique is used to estimate the model. the robust least square is a regression method that is less 4 cap variable represents the capital adequacy ratio. ideally, banks that have high capital ratio are more likely to engage in risky lending because they have the equity capital needed to absorb unexpected losses that may arise from risky lending (batten & vo, 2019; hallunovi & berdo, 2018). the higher the risk, the higher the return or profit. thus, the relationship between cap and π is predicted to be positive. eff variable represents the cost efficiency ratio. efficient banks tend to have a low cost-to-income ratio because they are able to minimise cost and maximise income, and as a result, such banks are able to generate high profit levels (bitar et al., 2018). thus, the relationship between eff and π is predicted to be negative. opta variable is the overhead cost to total asset ratio. banks that have high overhead cost will have lower profit levels due to rising overhead expenses (serwadda, 2018). thus, the relationship between opta and π is predicted to be negative. npl variable is non-performing loans to gross loan ratio. ideally, banks that have high problem loans will have lower net interest income and reduced net profit (panta, 2018). thus, the relationship between npl and π is predicted to be negative. inf variable is the inflation rate. high inflation can make banks increase the price of loans and increase the fee charged for offering non-interest activities to bank customers (bouzgarrou et al., 2018), thus leading to high bank profitability. thus, the relationship between inf and π is predicted to be positive. gdpr variable represents the growth in gdp. banks tend to report high profit levels in times of economic prosperity. this is because a large number of debtors are able to repay their loan during periods of economic prosperity, thereby contributing to high interest income for banks. in contrast, banks often report low profit levels in recessionary periods because a large number of debtors may default on their loan repayment, thereby contributing to reduced interest income for banks (kohlscheen et al., 2018). thus, the relationship between gdpr and π is predicted to be positive. model specification the model used to analyse the determinants of bank profitability (π) is similar to the models used in prior studies such as huang (2020); ozili and uadiale (2017) and borio et al. (2017). π = f (bank specific determinants, macroeconomic determinants). the econometric model is specified below as: 𝜋𝜋𝜋𝜋𝜋𝜋 = 𝑐𝑐 + 𝛽𝛽1𝐶𝐶𝐶𝐶𝜋𝜋𝜋𝜋 + 𝛽𝛽2𝐶𝐶𝐶𝐶𝐶𝐶𝜋𝜋𝜋𝜋 + 𝛽𝛽3𝐸𝐸𝐸𝐸𝐸𝐸𝜋𝜋𝜋𝜋 + 𝛽𝛽4𝑂𝑂𝐶𝐶𝑂𝑂𝐶𝐶𝜋𝜋𝜋𝜋 + 𝛽𝛽5𝐶𝐶𝐶𝐶𝑁𝑁𝜋𝜋𝜋𝜋 + 𝛽𝛽6 𝐼𝐼𝐶𝐶𝐸𝐸𝜋𝜋𝜋𝜋 + 𝛽𝛽7𝐺𝐺𝐺𝐺𝐶𝐶𝐺𝐺𝜋𝜋𝜋𝜋 + 𝑒𝑒 (1) π = vector of dependent variables, namely, roabt, roebt and nim. where: cap = bank capital to total assets (%); cn = bank concentration (%); eff = bank cost to income ratio (%); nim = bank net interest margin (%); npl = bank non-performing loans to gross loans (%); opta = bank 61 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 sensitive to outliers in the data. the robust least square estimation is the m-estimation based on huber (1973). the m-estimation addresses outliers in the dependent variable when the value of the dependent variable differs significantly from the regression line. eviews 11 was used to estimate the model. the robust least squares estimation method is a superior estimation method compared to the ordinary least squares estimators which is sensitive to outliers in the data. results and discussion summary of descriptive statistics the descriptive statistics result is presented in table 1. the mean value of cn is 49 for the full sample. cn is the highest in south africa and lowest in the united states. this suggests that the us banking sector is less concentrated compared with the banking sector in south africa and nigeria. the cap variable, on average, is 9.63 for the full sample. cap is much higher in united states and lowest in south africa. this suggests that the us banking sector is well-capitalised compared with the banking sectors of nigeria and south africa. the eff variable, on average, is 59.8 for the full sample. eff is the highest in nigeria and lowest in south africa. this suggests that nigeria’s banking sector is largely cost inefficient as indicated by its high cost-to-income ratio. south africa has a lower cost-to-income ratio. the mean value of opta is 4.78 for the full sample. opta is much higher in nigeria and lowest in the united states. this suggests that the nigerian banking sector has a high overhead cost to total asset ratio. the united states has a low opta which indicates that the us has a low overhead cost to total asset. the mean value of npl is 6.75 for the full sample. npl is much higher in nigeria and lower in the united states. this suggests that the us banking sector has high loan quality compared to nigeria and south africa. the macroeconomic variables, inf and gdpr, are 6.75 and 3.56, respectively for the full sample. inf and gdpr are lower in the united states and much higher in nigeria. this suggests that the us experience greater macroeconomic stability compared to nigeria. for the profitability variables, roabt and nim are higher in nigeria and lowest in the united states. this suggest that the nigerian banking sector is more profitable than the us banking sector. roebt is higher in south africa and lowest in the united states. 62 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 ta bl e 1 d es cr ip tiv e st at is tic s v ar ia bl es c n c a p e ff o pt a n pl in f g d pr r o a b t r o e b t n im % % % % % % % % % % fu ll sa m pl e m ea n 4 9. 3 9 .6 3 5 9. 8 4 .7 8 6 .5 3 6 .7 5 3 .5 6 1 .6 1 1 8. 66 5 .3 7 m ed ia n 3 8. 7 9 .3 5 8. 85 3. 36 3. 35 5 .6 9 2 .9 9 1 .7 5 1 9. 11 3 .8 1 m ax im um 9 9. 5 1 7. 7 8 1. 37 1 8. 36 3 7. 30 2 9. 26 1 5. 32 4 .1 3 4 3. 11 1 6. 08 m in im um 2 0. 2 1 .4 9 4 7. 68 0. 75 0. 70 -0 .6 9 -2 .5 3 -1 5. 09 0 .6 4 1 .4 7 st . d ev 2 3. 5 2 .7 3 5. 91 3. 12 7. 97 5 .4 22 2 .7 6 2 .2 5 8 .8 6 3 .0 3 o bs er va tio n 6 4 5 6 6 6 6 6 5 6 6 6 6 6 6 6 6 6 6 6 n ig er ia m ea n 4 0. 89 1 0. 42 6 3. 22 7 .8 6 1 4. 82 1 2. 37 5 .4 2 1 .6 9 1 9. 56 8 .7 3 m ed ia n 3 9. 11 1 0. 41 6 2. 11 6 .9 7 1 6. 11 1 1. 89 5 .9 8 2 .4 9 1 9. 63 7 .6 1 m ax im um 7 1. 08 1 7. 70 8 1. 37 1 8. 36 3 7. 30 2 9. 26 1 5. 32 4 .1 2 4 3. 11 1 6. 08 m in im um 2 2. 28 1 .4 9 4 9. 17 3 .4 8 2 .9 5 5 .3 8 -1 .6 1 -1 5. 09 0 .6 4 5 .6 0 st . d ev 1 4. 36 3 .9 7 7 .2 9 3 .3 2 9 .6 6 5 .3 2 3 .4 9 3 .8 6 8 .9 5 2 .5 4 o bs er va tio n 2 2 1 8 2 2 2 2 1 8 2 2 2 2 2 2 2 2 2 2 (c on tin ue d) 63 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 v ar ia bl es c n c a p e ff o pt a n pl in f g d pr r o a b t r o e b t n im % % % % % % % % % % so ut h a fr ic a m ea n 8 0. 17 7 .7 9 5 6. 80 3 .4 1 3 .3 7 5 .6 9 2 .8 0 1 .6 0 2 0. 38 3 .7 8 m ed ia n 7 7. 59 7 .9 1 5 7. 17 3 .0 5 3 .1 8 5 .7 1 2 .8 2 1 .5 1 2 0. 38 3 .3 6 m ax im um 9 9. 53 9 .3 0 7 1. 73 1 1. 45 5 .9 0 1 0. 05 5 .6 1 4 .1 3 4 2. 63 1 1. 66 m in im um 7 5. 14 5 .7 0 4 7. 68 0 .7 5 1 .1 0 -0 .6 9 -1 .5 3 0 .3 3 3 .0 0 1 .4 7 st . d ev 6 .2 5 0 .8 4 5 .3 9 2 .0 3 1 .3 9 2 .3 4 1 .7 6 0 .7 3 1 0. 29 2 .0 7 o bs er va tio n 2 0 1 8 2 2 2 2 1 8 2 2 2 2 2 2 2 2 2 2 u ni te d st at es m ea n 2 9. 76 1 0. 56 5 9. 48 3 .0 9 1 .9 2 2 .1 8 2 .4 5 1 .5 3 1 6. 03 3 .6 2 m ed ia n 3 3. 42 1 0. 40 5 9. 43 2 .9 1 1 .3 6 2 .2 3 2 .5 1 1 .6 2 1 5. 93 3 .5 2 m ax im um 3 6. 13 1 2. 73 6 3. 64 3 .9 9 5 .0 0 3 .8 4 4 .7 5 2 .1 9 2 6. 33 4 .3 2 m in im um 2 0. 18 8 .4 0 5 5. 98 2 .5 2 0 .7 0 -0 .3 5 -2 .5 3 0 .1 2 1 .6 1 3 .0 9 st . d ev 6 .1 1 1 .4 5 1 .9 9 0 .4 3 1 .2 8 1 .0 4 1 .6 6 0 .5 3 6 .7 8 0 .3 3 o bs er va tio n 2 2 2 0 2 2 2 2 2 0 2 2 2 2 2 2 2 2 2 2 64 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 correlation analysis in the full sample, roabt is positive and correlated with cap and gdpr, and negatively correlated with cn, eff, opta, npl and inf. the roebt profitability variable is positive and correlated with cn, opta, npl, inf and gdpr, and negatively correlated with cap and eff. the nim profitability variable is positive and correlated with cap, eff, opta, npl, inf and gdpr, and negatively correlated with cn. the nim and opta variables are highly correlated at 91.6 percent. the npl and opta variables are highly correlated at 81.3 percent. the nim and inf variables are highly correlated at 71.6 percent (see appendix a1). in the nigerian sub-sample, roabt is positive and correlated with cap and eff, and negatively correlated with cn, opta, npl, inf and gdpr. the roebt profitability variable is positive and correlated with cap, opta, npl, inf and gdpr, and negatively correlated with cn and eff. the nim profitability variable is positive and correlated with opta, npl, inf and gdpr, and negatively correlated with cn, cap and eff. the nim and opta variables are highly correlated at 75.1 percent (see appendix a2). in the united states sub-sample, roabt is positive and correlated with opta, inf and gdpr, and negatively correlated with cn, cap, eff and npl. the roebt profitability variable is positive and correlated with opta, inf and gdpr, and negatively correlated with cn, cap, eff and npl. the nim profitability variable is positive and correlated with opta and gdpr, and negatively correlated with cn, cap, eff and npl. the roebt and cn variables are highly correlated at 81.3 percent. the nim and cn variables are highly correlated at 74.8 percent. the opta and cn variables are highly correlated at 91.2 percent (refer appendix a3). in the south africa sub-sample, roabt is positive and correlated with cn, cap, opta and gdpr, and negatively correlated with eff, npl and inf. the roebt profitability variable is positive and correlated with cn, eff, opta and gdpr, and negatively correlated with cap, npl and inf. the nim profitability variable is positive and correlated with cn, cap, opta, npl and gdpr, and negatively correlated with eff and inf. the opta and cn variables are highly correlated at 90.7 percent. the inf and eff variables are highly correlated at 74.4 percent. the inf and opta variables are highly correlated at 82.1 percent (see appendix a4). 65 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 regression results pooled sample analysis the result is presented in table 2. the opta coefficient is significant and positively related to the roabt, roebt and nim variables. this indicates that higher overhead cost leads to higher profitability. the npl coefficient is significant and negatively related to the roabt, roebt and nim variables. table 2 robust least square regression – full sample 1 2 3 roabt roebt nim coefficient (t-values) coefficient (t-values) coefficient (t-values) c 2.257*** (3.20) 65.212*** (8.02) 5.355*** (4.04) cn -0.001 (-0.23) 0.056** (2.02) -0.019*** (-4.14) cap 0.049** (2.39) -0.976*** (-4.16) 0.059 (1.55) eff -0.037*** (-3.27) -0.855*** (-6.63) -0.093*** (-4.41) opta 0.305*** (6.58) 2.705*** (5.07) 1.231*** (14.15) npl -0.047*** (-3.69) -0.552*** (-3.79) -0.065*** (-2.76) inf 0.015 (0.92) -0.240 (-1.26) 0.101*** (3.27) gdpr 0.067*** (2.78) 1.278*** (4.62) -0.019 (-0.42) r-square 46.02 43.62 63.22 adjusted r-square 37.63 34.85 57.50 observations (after adjustment) 53 53 53 this indicates that higher non-performing loans lead to reduced profitability. the eff coefficient is significant and negatively related to the roabt, roebt and nim variables. this indicates that a 66 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 high cost-to-income ratio has a negative impact on profitability. the inf coefficient is significant and negatively related to nim but insignificantly related to roabt and roebt. this suggest that higher inflation leads to higher net interest margin. the gdpr coefficient is significant and positively related to roabt and roebt but insignificantly related to nim. this suggests that economic booms are associated with higher operating profit and higher profit to equity shareholders. the cap coefficient is significant and positively related to roabt and negatively related to roebt. this suggests that higher capital levels lead to higher operating profit and lower profit to equity shareholders. the cn coefficient is significant and positively related to roebt and negatively related to nim. this suggests that high bank concentration leads to higher profit to equity shareholders and lower net interest margin. overall, the results indicate that the most consistent determinants of bank profitability in the full sample result (in columns 1 to 3) are the efficiency ratio, non-performing loans, and overhead cost to total asset ratio. nigeria the result is presented in table 3. opta coefficient is positive and significantly related to the roabt, roebt and nim variables. this indicates that higher overhead cost leads to greater bank profitability. npl coefficient is negative for roabt, roebt and nim. the npl coefficient is negative and significantly related to nim, which indicates that higher non-performing loans lead to lower net interest margin. the eff coefficient is negative and is significantly related to roabt, roebt and nim. this indicates that there is an inverse relationship between the cost-to-income ratio and profitability. the gdpr coefficient is positive and significantly related to roebt, which indicates that the nigerian banking sector generates higher profit to equity shareholders during economic booms and vice versa. the cn coefficient report mixed signs in columns 1 to 3. cn coefficient is significant and positively related to roabt and negatively to nim. this indicates that high bank concentration is associated with high operating profit and low net interest margin. the inf and cap coefficients are not significant in columns 1 to 3. overall, the results suggest that the most consistent determinants of bank profitability in nigeria (in columns 1 to 3) are overhead cost to total asset ratio, and the cost efficiency ratio. 67 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 table 3 robust least square regression – nigeria banking sector 1 2 3 roabt roebt nim coefficient (t-values) coefficient (t-values) coefficient (t-values) c 3.315 (1.49) 46.483*** (3.30) 11.320*** (5.95) cn 0.038** (2.44) -0.008 (-0.08) -0.039*** (-2.92) cap -0.085 (-1.29) -0.225 (-0.54) -0.074 (-1.31) eff -0.063* (-1.80) -0.653*** (-2.92) -0.104*** (-3.44) opta 0.307*** (2.88) 2.219*** (3.28) 1.037*** (11.33) npl -0.033 (-1.28) -0.255 (-1.56) -0.107*** (-4.84) inf 0.059 (1.30) -0.060 (-0.21) -0.021 (-0.52) gdpr 0.072 (1.38) 0.854*** (2.59) 0.065 (1.45) r-square 45.83 59.81 62.72 adjusted r-square 7.92 31.68 36.62 observations (after adjustment) 18 18 18 south africa table 4 reports the result for south africa. opta coefficient is positive and significantly related to the roabt, roebt and nim variables. this indicates that higher overhead costs lead to higher profitability. the npl coefficient is negative in columns 1 to 3, and the npl coefficient is significant and negatively related to roabt and roebt, which indicates that higher non-performing loans lead to reduced operating profit and lower profit to equity shareholders. the cap coefficient is significant and positively related to roabt and nim, and negatively related with roebt. this suggests that higher capital ratios lead to higher operating profit, higher net interest margin and lower profit to equity shareholders in the south african banking sector. the inf coefficient reports a positive sign in columns 1 to 3. 68 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 the inf coefficient is significant and positively related to roabt, which indicates that the south african banking sector generates higher operating profit during high inflation. the cn, gdpr, eff, cap coefficients report mixed signs in columns 1 to 3. overall, the findings suggest that the most consistent determinants of bank profitability in south africa (in columns 1 to 3) are the capital adequacy ratio and the overhead cost to total asset ratio. table 4 robust least square regression – south africa banking sector 1 2 3 roabt roebt nim coefficient (t--values) coefficient (t-values) coefficient (t-values) c 0.706 (0.89) 74.546 (0.86) 1.264 (0.32) cn -0.011 (-1.61) -0.793 (-1.07) 0.023 (0.68) cap 0.067** (2.55) -7.442*** (-2.61) 0.294** (2.27) eff -0.005 (-0.82) 0.474 (0.78) -0.067** (-2.38) opta 0.408*** (6.38) 13.334* (1.91) 0.598* (1.87) npl -0.073*** (-5.20) -2.607* (-1.69) -0.038 (-0.54) inf 0.056*** (3.95) 0.780 (0.50) 0.008 (0.11) gdpr 0.063*** (6.82) 0.879 (0.87) -0.049 (-1.08) r-square 65.14 68.14 64.52 adjusted r-square 30.26 36.28 29.04 observations (after adjustment) 15 15 15 united states table 5 reports the results for the united states. opta coefficient is positive in columns 1 to 3. opta coefficient is significant and 69 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 positively related to roebt, which suggests that higher overhead cost leads to higher profit to equity shareholders in the united states banking sector. the cap coefficient is positively related to the roabt, roebt and nim variables. cap coefficient is significant and positively related to roabt and roebt which indicates that higher capital levels lead to higher operating profit and higher profit to equity shareholders in the united states banking sector. the npl coefficient is significant and negatively related with roabt and roebt but insignificantly related to nim. table 5 robust least square regression – united states banking sector 1 2 3 roabt roebt nim coefficient (t-values) coefficient (t-values) coefficient (t-values) c -1.134 (-0.35) -13.436 (-0.48) 2.549 (0.99) cn -0.046 (-1.15) -0.501 (-1.47) -0.030 (-0.96) cap 0.352** (2.45) 2.789*** (2.29) 0.105 (0.92) eff -0.017 (-0.39) -0.106 (-0.29) -0.015 (-0.45) opta 0.594 (1.37) 8.467** (2.30) 0.519 (1.52) npl -0.358*** (-3.62) -3.375*** (-3.99) 0.059 (0.75) inf 0.047 (0.84) -0.206 (-0.43) -0.011 (-0.25) gdpr 0.051 (0.83) 0.509 (0.97) 0.036 (0.74) r-square 75.38 84.11 75.80 adjusted r-square 61.02 74.85 61.69 observations (after adjustment) 20 20 20 this suggests that higher non-performing loans lead to reduced operating profit and lower profit to equity shareholders. the cn, eff and gdpr coefficients are negatively related to the roabt, 70 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 roebt and nim variables, but are not significant. inf coefficient is insignificant and reports mixed signs in columns 1 to 3. overall, the results suggest that the most consistent determinants of bank profitability in the united states in columns 1 to 3 (excluding the nim model in column 3) are capital adequacy ratio and size of non-performing loans. also, the profitability determinants are not significantly related to nim ratio in the us banking sector. conclusion and policy recommendations this paper examined the banking sector profitability determinants in south africa, nigeria and the united states. the findings reveal that the determinants of banking sector profitability in the full sample analysis are the efficiency ratio, non-performing loans, and overhead cost to total asset ratio. the comparative analyses show that efficiency ratio, overhead cost to total asset ratio and non-performing loans are significant determinants of banking sector profitability in south africa. in nigeria, the significant determinants of bank profitability are the overhead cost to total asset ratio and the efficiency ratio. in the united states, the determinants of bank profitability are capital adequacy ratio and non-performing loans. the implication of the results is that the determinants of bank profitability differ across countries. these differences may be explained by multiple factors, for example, differences in the nature of banking systems, differences in financial sector development and differences in banking regulation and supervision. some policy recommendations include the following. one, the findings show that nigeria has a higher cost-to-income ratio compared to the united states which imply that the nigerian banking sector is less efficient. bank supervisors in nigeria should ensure that banks operate more efficiently despite being profitable. two, the findings show that the banking sectors of nigeria and south africa have high non-performing loans compared to the united states. bank regulators in nigeria and south africa should issue strict policies against rising non-performing loans in banks while their bank supervisors should also ensure that banks improve their loan screening process and credit risk management system in order to reduce the size of nonperforming loans in banks. finally, the banking sectors of nigeria and 71 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 south africa are more concentrated than that of the united states. policy makers should issue policies that encourage the entry of nonbank financial intermediaries in the financial sector to reduce the dominance of banks in the credit market and in the financial sector. one limitation of the study is the small number of countries examined. using a larger sample of countries to analyse the determinants of bank profitability may provide greater insights on the factors that affect bank profitability across countries. future studies can analyse the determinants of banking sector profitability using a larger country sample. secondly, future studies can employ several robust statistical techniques to estimate the determinants of bank profitability. also, future studies can examine the impact of institutional factors on banking sector profitability. such studies should make a comparison between several countries. finally, it would be interesting to investigate the regional differences in bank profitability determinants. such studies can compare the profitability determinants of banks in the g7 countries versus banks in the euro sector, or compare the profitability determinants of banks in the mena and ecowas countries. acknowledgment this research received no specific grant from any funding agency. references ali, m., & puah, c. h. (2018). the internal determinants of bank profitability and stability: an insight from banking sector of pakistan. management research review, 42(2). https://doi. org/10.1108/mrr-04-2017-0103 ammar, n., & boughrara, a. (2019). the impact of revenue diversification on bank profitability and risk: evidence from mena banking industry. macroeconomics and finance in emerging market economies, 12(1), 36-70. athanasoglou, p. p., brissimis, s. n., & delis, m. d. (2008). bankspecific, industry-specific and macroeconomic determinants of bank profitability. journal of international financial markets, institutions and money, 18(2), 121-136. 72 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 batten, j., & vo, x. v. (2019). determinants of bank profitability— evidence from vietnam. emerging markets finance and trade, 55(6), 1417-1428. ben naceur, s., & goaied, m. (2008). the determinants of commercial bank interest margin and profitability: evidence from tunisia. frontiers in finance and economics, 5(1), 106-130. bitar, m., pukthuanthong, k., & walker, t. (2018). the effect of capital ratios on the risk, efficiency and profitability of banks: evidence from oecd countries. journal of international financial markets, institutions and money, 53, 227-262. bolarinwa, s. t., obembe, o. b., & olaniyi, c. (2019). re-examining the determinants of bank profitability in nigeria. journal of economic studies, 46(3), 633-651. bonaccorsi di patti, e., & palazzo, f. (2018). bank profitability and macroeconomic conditions: are business models different? bank of italy occasional paper, (436). borio, c., gambacorta, l., & hofmann, b. (2017). the influence of monetary policy on bank profitability. international finance, 20(1), 48-63. bougatef, k. (2017). determinants of bank profitability in tunisia: does corruption matter? journal of money laundering control, 20(1), 70-78. boungou, w. (2019). negative interest rates, bank profitability and risk-taking. ofce working paper. https://scholar.google.com bouzgarrou, h., jouida, s., & louhichi, w. (2018). bank profitability during and before the financial crisis: domestic versus foreign banks. research in international business and finance, 44, 26-39. demirgüç-kunt, a., & huizinga, h. (2000). financial structure and bank profitability. policy research working paper (no. 2430). world bank, washington, dc. hallunovi, a., & berdo, m. (2018). the relationship between risk management and profitability of commercial banks in albania. asian themes in social sciences research, 1(2), 44-49. hesse, h., & poghosyan, t. (2016). oil prices and bank profitability: evidence from major oil-exporting countries in the middle east and north africa. in financial deepening and post-crisis development in emerging markets (pp. 247-270). new york: palgrave macmillan. huang, q. (2020). ownership concentration and bank profitability in china. economics letters, 196, 109525. huber (1973). robust regression: asymptotics, conjectures and monte carlo. the annals of statistics, 1(5), 799-821. 73 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 kohlscheen, e., murcia pabón, a., & contreras, j. (2018). determinants of bank profitability in emerging markets. bis working papers, no 686. ozili, p. k. (2015). determinants of bank profitability and basel capital regulation: empirical evidence from nigeria. research journal of finance and accounting, 6(2), 124-131. ozili, p. k., & uadiale, o. (2017). ownership concentration and bank profitability. future business journal, 3(2), 159-171. ozili, p. k. (2017). bank profitability and capital regulation: evidence from listed and non-listed banks in africa. journal of african business, 18(2), 143-168. panta, b. (2018). non-performing loans and bank profitability: study of joint venture banks in nepal. international journal of sciences: basic and applied research (ijsbar), 42, 151-16. serwadda, i. (2018). determinants of commercial banks’ profitability. evidence from hungary. acta universitatis agriculturae et silviculturae mendelianae brunensis, 66(5), 1325-1335. zheng, c., rahman, m. m., begum, m., & ashraf, b. n. (2017). capital regulation, the cost of financial intermediation and bank profitability: evidence from bangladesh. journal of risk and financial management, 10(2), 9. https://doi.org/10.3390/ jrfm10020009 74 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 a p p e n d ix a 1: c or re la tio n – fu ll sa m pl e pva lu es a re re po rt ed in p ar en th es es v ar ia bl e c n c a p e ff o pt a n pl in f g d pr r o a b t r o e b t n im c n 1. 00 0 --c a p -0 .2 60 * 1. 00 0 (0 .0 6) --e ff -0 .1 66 -0 .0 81 1. 00 0 (0 .2 3) (0 .5 6) --o pt a -0 .2 04 -0 .0 11 0. 49 6* ** 1. 00 0 (0 .1 4) (0 .9 3) (0 .0 0) --n pl -0 .1 55 -0 .2 23 0. 46 6* ** 0. 81 3* ** 1. 00 0 (0 .2 6) (0 .1 1) (0 .0 0) (0 .0 0) --in f -0 .0 35 -0 .0 93 0. 24 0* 0. 68 0* ** 0. 68 3* ** 1. 00 0 (0 .7 9) (0 .5 0) (0 .0 8) (0 .0 0) (0 .0 0) --g d pr -0 .0 68 0. 00 1 0. 49 8* ** 0. 63 5* ** 0. 44 1* ** 0. 46 6* ** 1. 00 0 (0 .6 2) (0 .9 9) (0 .0 0) (0 .0 0) (0 .0 0) (0 .0 0) --r o a b t -0 .0 88 0. 28 9* * -0 .4 77 ** * -0 .0 83 -0 .3 51 ** -0 .0 14 0. 01 3 1. 00 0 (0 .5 2) (0 .0 3) (0 .0 0) (0 .5 5) (0 .0 1) (0 .9 1) 0. 92 --r o e b t 0. 20 0 -0 .2 03 -0 .1 39 0. 28 6* * 0. 04 12 0. 14 1 0. 39 1* ** 0. 37 5* ** 1. 00 0 (0 .1 5) (0 .1 4) (0 .3 1) (0 .0 4) (0 .7 6) (0 .3 1) (0 .0 0) (0 .0 0) --n im -0 .3 01 ** 0. 13 9 0. 30 7* * 0. 91 6* ** 0. 68 7* ** 0. 71 6* ** 0. 60 5* ** 0. 14 6 0. 20 8 1. 00 0 (0 .0 3) (0 .3 1) (0 .0 2) (0 .0 0) (0 .0 0) (0 .0 0) (0 .0 0) (0 .2 9) (0 .1 3) -- 75 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 a 2: c or re la tio n – n ig er ia pva lu es a re re po rt ed in p ar en th es es v ar ia bl e c n c a p e ff o pt a n pl in f g d pr r o a b t r o e b t n im c n 1. 00 0 --c a p 0. 35 1 1. 00 0 (0 .1 5) --e ff 0. 36 4 -0 .4 36 * 1. 00 0 (0 .1 3) (0 .0 7) --o pt a -0 .3 36 -0 .3 03 0. 27 7 1. 00 0 (0 .1 7) (0 .2 2) (0 .2 6) --n pl -0 .2 24 -0 .6 19 * 0. 36 5 0. 63 5* * 1. 00 0 (0 .3 6) (0 .0 1) (0 .1 3) (0 .0 0) --in f -0 .2 82 -0 .2 63 0. 06 2 0. 20 4 0. 27 7 1. 00 0 (0 .2 5) (0 .2 9) (0 .8 1) (0 .4 1) (0 .2 6) --g d pr 0. 03 2 -0 .0 69 0. 43 8* 0. 41 7* 0. 19 9 0. 16 9 1. 00 0 (0 .8 9) (0 .7 8) (0 .0 6) (0 .0 8) (0 .4 2) (0 .5 0) --r o a b t -0 .3 17 0. 39 8* 0. 62 9* * -0 .2 17 -0 .5 23 ** -0 .0 47 -0 .0 64 1. 00 0 (0 .1 9) (0 .1 0) (0 .0 1) (0 .3 8) (0 .0 2) (0 .8 5) (0 .7 9) --r o e b t -0 .4 64 ** 0. 09 31 -0 .4 15 * 0. 48 3* * 0. 14 5 0. 31 2 0. 28 1 0. 47 3 1. 00 0 (0 .0 5) (0 .7 1) (0 .0 9) (0 .0 4) (0 .5 6) (0 .2 1) (0 .2 5) (0 .0 4) --n im -0 .6 63 ** * -0 .0 94 -0 .1 52 0. 75 1* ** 0. 20 8 0. 19 8 0. 41 8* 0. 24 9 0. 65 7* * 1. 00 0 (0 .0 0) (0 .7 1) (0 .5 5) (0 .0 0) (0 .4 0) (0 .4 3) (0 .0 8) (0 .3 1) (0 .0 0) --76 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 a 3: c or re la tio n – u ni te d st at es pva lu es a re re po rt ed in p ar en th es es v ar ia bl e c n c a p e ff o pt a n pl in f g d pr r o a b t r o e b t n im c n 1. 00 0 --c a p 0. 86 4* ** 1. 00 0 (0 .0 0) --e ff 0. 19 2 -0 .1 19 1. 00 0 (0 .4 1) (0 .6 1) --o pt a -0 .9 12 ** * -0 .7 96 ** * 0. 03 5 1. 00 0 (0 .0 0) (0 .0 0) (0 .8 8) --n pl 0. 53 6* * 0. 63 1* ** 0. 18 4 -0 .3 24 1. 00 0 (0 .0 1) (0 .0 0) (0 .4 3) (0 .1 6) --in f -0 .2 07 -0 .4 51 ** 0. 30 3 0. 23 2 -0 .3 26 1. 00 0 (0 .3 8) (0 .0 4) (0 .1 9) (0 .3 2) (0 .1 5) --g d pr -0 .5 04 ** -0 .4 23 * 0. 01 9 0. 48 7* * -0 .6 68 ** * 0. 29 3 1. 00 0 (0 .0 2) (0 .0 6) (0 .9 3) (0 .0 2) (0 .0 0) (0 .2 1) --r o a b t -0 .6 31 ** * -0 .4 59 ** -0 .3 76 0. 49 5* * -0 .8 01 ** * 0. 17 2 0. 80 6* ** 1. 00 0 (0 .0 0) (0 .0 4) (0 .1 0) (0 .0 2) (0 .0 0) (0 .4 6) (0 .0 0) --r o e b t -0 .8 13 ** * -0 .6 60 ** * -0 .2 84 0. 71 1* ** -0 .7 87 ** * 0. 21 1 0. 79 9* ** 0. 95 7* ** 1. 00 0 (0 .0 0) (0 .0 0) (0 .2 2) (0 .0 0) (0 .0 0) (0 .3 7) (0 .0 0) (0 .0 0) --n im -0 .7 48 ** * -0 .4 40 * -0 .2 10 0. 81 7* ** -0 .1 09 -0 .0 42 0. 44 0* 0. 49 8* * 0. 62 9* ** 1. 00 0 (0 .0 0) (0 .0 5) (0 .3 7) (0 .0 0) (0 .6 4) (0 .8 6) (0 .0 5) (0 .0 3) (0 .0 0) -- 77 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 a 4: c or re la tio n – so ut h a fr ic a pva lu es a re re po rt ed in p ar en th es es v ar ia bl e c n c a p e ff o pt a n pl in f g d pr r o a b t r o e b t n im c n 1. 00 0 --c a p 0. 19 7 1. 00 0 (0 .4 8) --e ff 0. 30 7 0. 16 5 1. 00 0 (0 .2 6) (0 .5 5) --o pt a 0. 90 7* ** 0. 46 3* 0. 49 8* 1. 00 0 (0 .0 0) (0 .0 8) (0 .0 5) --n pl -0 .1 63 -0 .4 49 * -0 .5 11 * -0 .3 78 1. 00 0 (0 .5 6) (0 .0 9) (0 .0 5) (0 .1 6) --in f -0 .6 51 ** * -0 .4 79 * -0 .7 44 ** * -0 .8 21 ** 0. 41 6 1. 00 0 (0 .0 0) (0 .0 7) (0 .0 0) (0 .0 0) (0 .1 2) --g d pr 0. 26 8 0. 25 2 0. 37 4 0. 37 8 -0 .6 60 ** * -0 .4 87 * 1. 00 0 (0 .3 3) (0 .3 6) (0 .1 6) (0 .1 6) (0 .0 0) (0 .0 6) --r o a b t 0. 52 1* * 0. 28 3 -0 .1 51 0. 56 5* * -0 .1 14 -0 .2 95 0. 28 1. 00 0 (0 .0 4) (0 .3 0) (0 .5 9) (0 .0 2) (0 .6 8) (0 .2 8) (0 .2 9) --r o e b t 0. 47 0* -0 .0 08 0. 67 7* ** 0. 58 1* * -0 .6 86 ** * -0 .5 48 ** 0. 63 6* * 0. 22 3 1. 00 0 (0 .0 7) (0 .9 7) (0 .0 0) (0 .0 2) (0 .0 0) (0 .0 3) (0 .0 1) (0 .4 2) --n im 0. 60 4* * 0. 49 5* -0 .2 83 0. 61 7* * 0. 02 2 -0 .2 96 0. 05 7 0. 85 2* ** -0 .0 71 1. 00 0 (0 .0 1) (0 .0 6) (0 .3 1) (0 .0 1) (0 .9 3) (0 .2 8) (0 .8 3) (0 .0 0) (0 .7 9) --78 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 55–78 a5: variable source and description variable variable description source cap bank capital to total assets (%) global financial development indicators, world bank cn bank concentration (%) global financial development indicators, world bank eff bank cost to income ratio (%) global financial development indicators, world bank nim bank net interest margin (%) global financial development indicators, world bank npl bank non-performing loans to gross loans (%) global financial development indicators, world bank opta bank overhead costs to total assets (%) global financial development indicators, world bank roabt bank return on assets (%, before tax) global financial development indicators, world bank roebt bank return on equity (%, before tax) global financial development indicators, world bank inf inflation, consumer prices (annual %) imf gdpr gdp growth (annual %) imf 57 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 how to cite this article: bashir, r., & azeez, a. a. (2022). risk management practices of islamic and conventional banks of pakistan: a comparative study. international journal of banking and finance, 17(2), 57-90. https://doi. org/10.32890/ ijbf2022.17.2.3 risk management practices of islamic and conventional banks of pakistan: a comparative study 1rubina bashir & 2a. a. azeez 1faculty of graduate studies, university of colombo, sri lanka 2faculty of management and finance, university of colombo, sri lanka 1corresponding author: rubinabashir23@gmail.com received: 21/2/2021 revised: 23/6/2021 accepted: 9/7/2021 published: 27/6/2022 abstract this paper is the report of a study that has investigated the impact of the risk management process, including liquidity and operational risk, on the risk management practices of islamic banks and conventional banks in pakistan. the study compared the risk management practices of both types of banks. data was collected through the 200 selfadministered questionnaires distributed to senior managers and risk officials of both types of banks. the data was analyzed using basic descriptive statistics, one-way anova and multiple regression analysis. the empirical results showed that risk assessment and analysis, risk monitoring and liquidity risk analysis were the most influential determinants of the risk management practices of conventional banks, whereas risk identification, risk assessment and http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 58 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 analysis, credit risk analysis, liquidity risk analysis and operational risk analysis had influenced the risk management practices of islamic banks. overall, islamic banks were better in terms of understanding risk management, risk identification, risk assessment and analysis, risk monitoring and operational risk analysis, while conventional banks were ahead of islamic banks in credit risk analysis, liquidity risk analysis, and risk management practices. in the context of the emerging economy of pakistan, the study outcomes can be helpful for investors, potential and existing customers to make prudent investment decisions. findings are also helpful for strategy managers and regulators in policy formulation, elevation, and implementation of risk management regulations. keywords: islamic banks, conventional banks, risk management process, risk management practices, liquidity risk analysis, operational risk analysis, pakistan. jel classification: g32, g21. introduction risk has always been an important concern for all business enterprises, especially for financial institutions. banking is more sensitive to risk because banks operate on the funds of depositors and each transaction creates a risk (bessis, 2011). at the same time, cutthroat competition in the industry calls for effective implementation of risk management practices (bulbul et al., 2019). therefore, risk management is a vital component of a bank’s success (de angelo & stulz, 2015). increased market volatility, development of new products and derivatives, increased cost of risk management, sophisticated information technology systems and global financial crisis are the challenges faced by modern-day banks. to address these issues, an elaborated and specialized risk management framework is required because these variations call for the constant upgrading of the risk management framework (abu hussain & al ajmi, 2012). the five-step risk management process empirically tested by al tamimi and almazrooei (2007) and hassan (2009) has not been sufficient to dictate effective risk management practices for the globalized banking system because it included only credit risk analysis along with the understanding of risk, risk identification, risk assessment and 59 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 analysis, and risk monitoring. basel i focused on credit risk analysis and management because at that time it was considered that the key risk for banks is the credit risk, but in the modern age liquidity risk and operational risk are more or equally significant risks for banks (abu hussain & al-ajmi, 2012; raza bilal et al., 2013). basel ii and basel iii have also endorsed these findings by highlighting the operational and liquidity risk as key risks for banks and formulating the regulations to manage them effectively. mismanagement of liquidity is the key reason which worsen the situation during a financial crisis. liquidity risk management must be an integral part of the overall risk management framework and the governance of banks (basel committee on banking supervision, 2008). operational risk is in the limelight because operational risk events are increasing with the development of financial institutions (chernobai et al., 2018) and had resulted in huge losses (neifar & jarboui, 2018). risk management has become even more important for lower middle-income countries like pakistan, where the empirical literature on risk management practices of banks is limited (shafiq & nasr, 2010). pakistan is an emerging economy, and banks are the major players in the financial sector of the country (aurangzeb, 2012). the banking system of pakistan, like in other parts of the world, is dual. the islamic banks and conventional banks are functioning side by side, whereby islamic banks hold an asset base share of 15 percent and a 17 percent share of deposits (state bank of pakistan, 2019). the world portfolio share of the deposit was 1.75 percent (islamic financial services board [ifsb], 2015), which was relatively low compared to malaysia and the middle east. still, being an islamic country with 98 percent of the population being muslim, islamic banks have a great potential for growth. the islamic banks of pakistan are in their evolutionary phase and facing various challenges and problems, i.e., the same liquidity policy is being implemented on both islamic banks and conventional banks, which is not fair, and a separate policy is required to serve the unique liquidity issues of islamic banks. though both types of banks perform two basic functions, i.e., fund mobilization and utility services, still both are quite different from each other (nasser & muhammad, 2013) in terms of the execution and operationalization of various products and services. unlike conventional banks, all transactions of islamic banks are riba-free 60 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 (based upon profit and loss sharing) and developed according to shariah laws. being conceptually different from each other, both banking systems face risks while executing various transactions and dealing with different types of products and services. however, the risk exposure of islamic banks is two-fold because they are bound to follow both the general regulations of conventional banks and shariah laws. these facts make it worth to study and compare the risk management process and practices of these two unique, but correlated banking systems operating in pakistan in order to understand their strengths and weaknesses and unfold the competitive edge they have over each other. moreover, there are three main reasons to compare the islamic banks and conventional banks of pakistan, and these are:1) the banking industry is showing good growth; 2) islamic banks performed well during the financial crisis (ouerghi, 2014) and have gained more acceptance among muslims; and 3) islamic banks follow the same accounting standards and basic banking regulations used by conventional banks (khan et al., 2017). though the current study is an extension of empirical researches on risk management practices of pakistani banks (shafiq & nasr, 2010; khalid & amjad, 2012; nazir et al., 2012; shafique et al., 2013, raza bilal et al., 2013 and rehman et al., 2017), it nevertheless adds to the literature by looking into two new dimensions in the risk management process of banks i.e., liquidity risk analysis (lra) and operational risk analysis (ora), in addition to the existing fivestep (understanding of risk, risk identification, risk assessment and analysis, risk monitoring and credit risk analysis) risk management process. thus so far, there has been no research which has included the liquidity and operational risk in the risk management process, despite the availability of solid evidence that these two risks are more, or equally important as credit risk. the prime purpose of the present research is to make a significant contribution to the existing model of risk management framework of banks by studying the risk management process (rmp) of islamic banks and conventional banks of pakistan using the following seven aspects of the risk management process: understanding of risk, risk identification, risk assessment and analysis, risk monitoring, credit risk analysis, liquidity risk analysis, and operational risk analysis. more specifically, this study was aims to investigate the impact of the risk management process, including liquidity and operational 61 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 risk on the risk management practices of islamic banks, as well as conventional banks, and to compare the risk management practices of islamic banks and conventional banks of pakistan. the remainder of the paper is organized as follows. section 2 reviews the related literature on risk management practices and presents the hypotheses. section 3 describes the data and method of analysis. section 4 comprises results and discussions on the preliminary and inferential analyses of the study. section 5 concludes the paper. review of literature and hypothesis development agency theory highlights the existence of risk in an organization (jensen & meckling, 1976), while institutional theory emphasizes the formulation and implementation of the risk management process and practices to mitigate these risks (hudin & hamid, 2014). an agency problem arises due to a conflict of interest between agents (managers) and principals (stakeholders), which in turn exposes the organizations to various types of risks because managers tend to misuse their authority for their personal interest (jensen & meckling, 1976). to this end, institutional theory suggests that a standard set of rules are mandatory to formulate an efficient risk management structure which according to tolbert and zucker (1983) can be achieved through institutionalization, as “the process through which components of formal structure become widely accepted, as both appropriate and necessary, and serve to legitimate organizations” (pp. 5). it has been pointed out that the implementation of risk management practices depends upon the firm size, technology and environment (collier & wood, 2011). the environment includes government policies which will provide the key motivation for the enactment of the risk management practices. meanwhile, the execution of risk management is linked to innovation (new products and services), communication, time, and the social system (hudin & hamid, 2014). hence, the formulation and implementation of an effective risk management framework are equally necessary for islamic banks and conventional banks (van greuning & iqbal, 2008; hassan, 2009) to address the agency problem and to protect the interest of all stakeholders of the firm including the creditors, employees, 62 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 customers, etc. the customers are the most important stakeholders, especially those in the in-service sector (banking). they should have trust and confidence in the firm, but financial distress and bankruptcy can shake this trust. risk management helps to overcome these issues and increase the customer’s confidence, leading to enhanced business value (klimczak, 2007). risk management is a stepwise process. according to bessis (2011), “it operates through three lines of defense, i.e., 1) lines of business. 2) function of the enterprise including risk management, compliance, finance, human resource and legal. 3) corporate audit” (pp 9). initially, the risk management process was empirically tested by hassan (2009) and altamimi and al-mazrooei (2007) for the islamic banks and conventional banks of brunei darussalam and the uae, respectively. they reported that risk management practices (rmps) are the function of the risk management process (rmp), whereby the rmp has five steps, i.e., understanding risk and risk management (urrm), risk identification (ri), risk analysis and assessment (raa), risk monitoring (rm) and credit risk analysis (cra). results revealed that the ri and the raa could influence the rmps more, compared to the other aspects of the rmp in both conventional banks and islamic banks. in addition, hassan (2009) further reported that islamic banking was a little bit better at risk management than conventional banks. since then, a significant number of researches have been conducted to study risk management practices in islamic and conventional banks operating in various parts of the world, including pakistan. a comprehensive comparative study by abu hussain and al-ajmi (2012) on the rmps of islamic banks and conventional banks of bahrain concluded that all aspects of the rmp were significantly related to the rmps of islamic banks and commercial banks. while islamic banks and conventional banks were somewhat efficient in the raa, the rmps, the rm and the ri, determinants of the quality of the rmps are the raa, the ri, the raa, and the cra. although islamic banks and conventional banks were similar in the ri, the rm, the raa, and the cra, they were substantially variant in terms of the urrm. mohad arrifin and kassim (2011) have suggested that the islamic banks of malaysia are efficient in the rmps, but room for improvement is still there. hassan (2011) pointed out that the islamic banks and conventional banks of five middle eastern countries were well aware of the importance of risk and its management. moreover, 63 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 the banks which were part of the sample of the present study dealt effectively with the ri, the raa and the rm, and the management of the different types of risks. sleimi (2020) suggested that all the five aspects of the risk management process had a significant relationship with the risk management practices of jordanian banks and this led to their better performance. raza bilal et al. (2013) studied basel iii’s risk management practices and their implementation in the banks of pakistan, bahrain, and the uae. their findings showed that the rmu, the raa, the ri and the cra had a significant relationship with the rmps of islamic banks and conventional banks of bahrain. they also found that the rmu, the ior and the raa had a significant influence on the risk management practices of banks in the uae, and in the banks of pakistan all features of the risk management process had a significant association with the rmps. similarly, muhammad et al. (2018) suggested that the urm, the ri, the raa, the rm and the cra were determinants of the risk management practices of the commercial banks of pakistan. as a matter of fact, the raa and the rm were the most influential components. while, the study by khalid and amjad (2012) concluded that all aspects of the risk management process were positively related to the rmps in the islamic banks of pakistan. however, the rm and the urm were found to have influenced the rmps the most. nazir et al. (2012) have commented that credit risk analysis, risk monitoring and understanding of risk were the most important aspects of the rmp and have had a significant effect on the rmps of islamic banks of pakistan. they also argued that there is a substantial difference between rmps of islamic banks and conventional banks. meanwhile, shafique et al. (2013) concluded that there is no variance among the risk management practices of islamic banks and conventional banks of pakistan. this difference shows that compliance levels at various points of time and local regulatory frameworks play a significant role in the rmps of banks. the above literature suggests that the risk management process has been studied based on five steps. hence, review of literature in the field shows that, there has been no significant extension or modification of these steps, except in the research work of rehman et al. (2017), which included the liquidity risk analysis in addition to the basic five step risk management process. they concluded that the ri, the raa, and the cra have a significant relation with the rmps of islamic banks while, the urm, the lra and the cra are the significant 64 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 determinants of the rmps of conventional banks in pakistan. it is true that there is a significant variance in islamic banks and conventional banks in terms of the ri, the rmps and the liquidity risk analysis. moreover, conventional banks are efficient in liquidity analysis and islamic banks are efficient in the ri and the rmps. yet another stream of studies has suggested that liquidity risk and operational risk are the most significant risks confronted by banks, in addition to credit risk (hassan, 2009; abu hussain & al-ajmi, 2012; raza bilal et al., 2013; shafiquet al., 2013; al-ali & naysary, 2014). on the one hand, islamic banks face credit risk, liquidity risk, foreign exchange risk, operational risk and shariah risk (ariffin et al., 2009; rehman et al., 2017) and conventional banks on the other, face credit risk, liquidity risk, interest rate risk, foreign exchange risk, and operational risk (shafique & nsar, 2010; alam & maskujama, 2011; wood & kellman, 2013). basel ii (2004), basel iii (2010) and ifsb (2005) also suggest that banks should manage four major risks, i.e., credit, liquidity, operational, and market risk. the importance of liquidity risk was further underscored after the financial crisis which occurred from 2007 until 2009. even banks with high capital faced liquidity issues during the crisis (jenkinson, 2008). that is why the main focus of basel iii (2010) is liquidity management (giordana & schumacher, 2013). in basel iii, the liquidity issue has been addressed with the introduction of the liquidity coverage ratio (lcr) and high-quality liquid assets (hqla). liquidity is vital for islamic banks, just as it is for the other types of banks (bello et al., 2017). islamic banks do not suffer liquidity issues due to liquidity coverage; these banks face this issue because there are only a few options/ instruments available as per shariah laws to improve liquidity (archer & karim, 2013). the comparative studies of the performance of the islamic banks and conventional banks conducted by jaffar and manarvi (2011) and kassim and abdulle (2012) have suggested that islamic banks are more liquid than conventional banks due to a limited number of islamic investment instruments. similarly, akhter et al. (2011) and rehman et al. (2017) concluded that the conventional banks of pakistan managed their liquidity in a more effective manner than the islamic banks. operational risk identification and management have gained crucial importance during the last decade because of the immense losses suffered by financial institutions due to excessive operational risk. for example, bernard l. madoff investment securities (usd 17 billion), 65 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 general society (euro 6.3 billion), rabobank (usd 1 billion) and fondiaria-sai (252 million euros) had suffered losses during the year 2013 as a result of fraudulent business activities. these losses are a warning for financial institutions and it has become essential for banks to identify, monitor, and manage operational risks effectively (neifar & jarboui, 2018). operational risk management is crucial for banks because of the following reasons: 1) operational risks result in big losses; 2) operational risks arise through internal sources, which are due to poor internal control; 3) the poor management of operational risk is a reflection of the fact that the other areas of risk management system are also weak (chernobai et al., 2018). furthermore, raza bilal et al. (2013) has concluded that the most important risk confronting banks is the operational risk. in pakistan, both types of banks have experienced a severe cybersecurity breach in october 2017 and suffered a loss of almost 6 million usd. this incident shows that the information systems of banks need to be improved, which comes under operational risk. in addition, a unique operational risk face by islamic banks, is the shariah non-compliance risk, i.e., the probability of non-compliance to shariah rules and principles in the bank’s operations; or risk related to islamic banks fiduciary obligations as in mudarib (entrepreneur) toward fund providers in a two-tier mudarabah contract. in the case of carelessness and unethical conduct of mudarib, the bank becomes liable to return funds to the original fund provider. so, operational risk management is even more crucial for islamic banks. based on the foregoing discussions, it can be argued that the agency issue calls for the implementation of standardized risk management practices in banks. while the risk management process is not limited to credit risk only, it should also include liquidity and operational risks, which are the most significant risks for islamic banks and conventional banks. to this end, the risk management process and the practices of both types of banks are different from each other in some parts of the world while, in some countries, they are the same in light of this, it can be hypothesized that: h1 : the risk management practices of islamic banks and conventional banks are determined by the risk management process. h2 : the risk management process and the risk management practices of islamic banks and conventional banks are significantly different from each other. 66 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 research methodology data the primary data was collected through a self-administered survey questionnaire because information required on the various aspects of the risk management process and practices is generally not published in the annual reports of banks, or any other reports published by central banks (abu hussain & al-ajmi, 2012). in addition, researchers can explain the research and motivate the respondents, and the respondents can fill in the questionnaire at their convenience (sekaran & bougie, 2016). the questionnaire used in the current study was adopted from the ones used in the studies by altamimi and mazrooei (2007) and hassan (2009) to measure the risk management process and practices. the constructs used in the questionnaire have been adopted because they have already been piloted and validated (bryman & bell, 2012). these pre-validated and well-reputed constructs make reliable measures that lead to good results (sekaran & bougie, 2016). the constructs have already been used by raza bilal et al. (2013) and khalid and amjad (2012) in their researches to study the risk management process and practices in pakistani banks. the existing model of risk management process has been extended by adding two new dimensions, i.e., operational risk analysis and liquidity risk analysis. constructs of these variables were adopted from rehman et al. (2017), the principles of liquidity risk management (bsbs, 2008) and the principles of operational risk developed by the bcbc (2011) and the sbp (2003), respectively. construct contents were further validated through consultation with practitioners and academicians, as suggested by devellis (1991) and practiced by al –tamimi and almazrooei (2007). the first draft of the questionnaire was sent to three academics in the field and five chief risk officers of various banks to assess the sufficiency and validity of the construct. a final draft was developed accordingly by deleting and rearranging some questions. the study questionnaire was divided into two main sections. the first section included information about respondents (bank employees) and banks. the second section measured the risk management process and practices of banks. the risk management process was measured in terms of seven aspects, i.e., understanding and risk management, risk identification, risk assessment and analysis, risk monitoring, credit risk analysis, liquidity risk analysis, and operational risk analysis. these variables were measured through 55 close-ended questions. 67 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 more specifically, risk management practices were measured through six close-ended questions. these questions were gauged through a seven-point interval scale (1, being strongly disagree to 7, being strongly agree). past literature has shown that many significant studies, i.e., hussain and al-ajmi (2012), rehman et al. (2017), al tamimi and mazrooei (2007) and khalid and amjad, (2012) have used the seven-point likert scale. respondents were asked to gauge the application of each item according to the risk management procedures and existing practices of their banks. population and sample the population of this study consisted of all commercial banks operating in pakistan, which included the two major categories of banks: islamic banks and conventional banks. both types of banks have various products and services which are different from each other, yet they own a large market share and face some common and some specialized types of risk. that is why these two types of banks have been defined as the population of research to draw a comparison of the risk management framework. at the time of the study there were 21 islamic banks operating in pakistan, of which four were full-fledged islamic banks and another 17 conventional banks were running islamic banking windows. there were 21 conventional banks operating in pakistan. therefore, in total, 41 islamic banks and conventional banks were identified as the population of this study. a stratified random sampling technique was used for sample selection. out of which eight islamic banks (four full-fledged islamic banks and four islamic banking standalone branches of conventional banks) and seven conventional banks were selected as the study sample based on market share, branch network, portfolio diversification and time, cost, and resource limitation. in modern banking, a risk management framework operates through three lines of defense. therefore, risk management is not limited to the risk management department only (kpmg, 2009). hence, data was collected from the top and middle management of all departments and units of the banks, i.e., risk management departments, trade units, credit departments, treasuries, regional offices, head offices, main branches, and corporate branches. two hundred and fifty questionnaires were self-administered, of which 220 were returned, but 20 of these, being incomplete were not included in analysis. the response rate was 80 percent. 68 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 method of data analysis data reliability and multicollinearity among explanatory variables were evaluated through cronbach’s alpha and the vif (variance inflation factor) and correlation matrix. descriptive statistics were used to understand and compare the usefulness of the risk management framework of islamic banks and conventional banks. finally, hypotheses were tested through multiple regression analysis and one-way anova. empirical results and discussion first of all, data reliability and internal consistency of variables were established through the cronbach alpha. the coefficient equal to or above 0.70 means that the construct is reliable (nunnally, 1978). the coefficient for most of the variables was more than 0.70, and these included the understanding and management of risk, risk assessment and analysis, risk management practices, liquidity risk analysis and operational risk analysis. however, the alpha for risk identification was 0.52, but the overall alpha for all the seven dimensions (urm. ri, raa, rm, cra, ora, lra) of the risk management process was 0.885. therefore, a lower value for risk identification was acceptable. these results showed that responses among the various items of all the variables were consistent with each other at an acceptable range. table 1 reliability measures of study variables variables cronbach alpha items urm 0.767 8 ri 0.520 5 raa 0.730 7 rm 0.801 6 rmps 0.851 6 cra 0.703 7 lra 0.834 9 ora 0.891 13 note. urm= understanding of risk management, ri= risk identification, raa, risk assessment and analysis, rm= risk monitoring, rmps= risk management practices, cra= credit risk analysis, lra= liquidity risk analysis, and ora= operational risk analysis. 69 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 descriptive analysis understanding of risk management table 2 presents the results of understanding risk, which reveals that the average mean for all eight questions was 5.62 and 5.87 for conventional and islamic banks, respectively. it indicates that islamic banks can manage their risk more efficiently and have a better understanding of risk than conventional banks. khalid and amjad (2012), hassan and al-ajmi (2012) and hassan (2009) also concluded the same. table 2 responses to questions about the urm questions conventional banks islamic banks no. mean sd mean sd 1 there is mutual understanding of risk management throughout the bank. 5.13 1.55 5.82 0.64 2 obligation for management of risk is undoubtedly formulated and understood across the bank. 5.26 1.45 5.80 0.75 3 accountability for management of risk is clearly designed and understood. 5.34 1.14 5.92 0.61 4 risk management is important for the success and performance of the bank. 6.20 1.21 6.07 0.69 6 your bank aims to enlarge the application of innovative risk management techniques. 5.69 1.18 5.78 0.86 7 your bank emphasizes the constant examination of assessment of risk management procedures. 5.96 1.00 5.92 0.69 8 use of risk management practices decrease the expenses or anticipated losses. 5.90 1.18 5.85 0.87 average 5.62 5.87 70 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 moreover, the staff of conventional banks were fully cognizant of the importance of the management of risk, but uniformity in the management of risk and its understanding was still lacking. while, the employees of islamic banks were not convinced that the implementation of sophisticated risk management techniques was the only important objective to improve the management of risk. conventional banks need to develop a better understanding of risk and risk management throughout all their management levels while, islamic banks are required to formulate more sophisticated and state of the art risk management technique to keep pace with the competitive financial markets. risk identification it is the initial stage of risk management and all the other phases of the rmp follows from risk identification. accurate risk identification leads to efficient risk management. responses to statements of risk identification (see table 3) show that the lowest mean value for both islamic banks and conventional banks was from statement 2, i.e., 3.99 (conventional banks) and 4.04 (islamic banks). table 3 responses to questions about the ri questions conventional banks islamic banks no. mean sd mean sd 1 your bank carries out thorough and methodical documentation of its risks. 5.65 1.18 5.91 0.66 2 it is tough for the bank to rank its major risks. 3.99 1.76 4.04 1.40 3 variations in risk are predicted and recognized along with the bank’s role and responsibilities. 5.38 1.27 5.81 0.72 4 your bank is cognizant of the strengths and weaknesses of the risk management framework of their competitors. 5.67 0.96 5.73 0.92 5 investment prospects are recognized through the systematic implementation of procedures. 5.43 1.53 5.82 0.70 average 5.22 5.58 71 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 it suggests that both types of banks could prioritize their important risks easily; the same was reported by hassan and alajmi (2012). whereas, the highest mean value (5.82) was from statement number 5 about islamic banks. this indicates that they have efficient procedures for risk identification. in conventional banks the highest means is from statement 4 (5.67), showing that conventional banks were well aware of the strengths and weaknesses of their competitors. overall, the average means of all the six statements for islamic banks (5.58) and conventional banks (5.22) indicated that islamic banks were slightly ahead of conventional banks in the identification of risks. hence, islamic banks have been able to recognize the latent risk to attain future objectives. this observation is consistent with the findings in khalid and amjad (2012). it also becomes obvious that a better understanding of risk leads to better risk identification. risk assessment and analysis the raa refers to risk quantification and is an important step. it helps to plan strategies to mitigate the risks. table 4 shows that the average mean of the seven statements is 5.60 and 5.84 for conventional banks and islamic banks, respectively. it can be inferred that islamic banks are better than conventional banks in terms of risk assessment and analysis. these results are consistent with the studies by hussain and al ajmi (2012), khalid and amjad (2012) and hasaan (2009). the highest means value for conventional banks (6.15) and islamic banks (5.92) relates to statement 7, which shows that both types of banks use risk prioritizing treatment for risk analysis. according to statement 3, conventional banks (mean =5.47) are less likely to use qualitative methods of risk analysis than islamic banks (mean=5.76). similarly, statement 2 indicates that islamic banks are less likely to use quantitative methods to analyze risk because they face risks like legal risks and shariah compliance risks which cannot be quantified. moreover, islamic banks are good at risk prioritizing and the active management of selected risks than conventional banks. risk monitoring results of responses to the rm show (see table 5) that the highest mean value for conventional banks was from statement 1 (5.97) while, 72 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 the mean value for this statement was lower (5.80) in islamic banks. it indicates that risk monitoring has been better implemented in routine reporting systems in conventional banks compared to islamic banks. table 4 responses to questions about the raa no. questions conventional banks islamic banks mean sd mean sd 1 chances of occurrence of various risks are evaluated by your bank. 5.48 1.23 5.87 0.65 2 bank uses quantitative methods for risk assessment. 5.50 1.10 5.73 0.66 3 bank uses qualitative methods for risk assessment. 5.42 1.47 5.76 0.82 4 bank investigates and assesses the opportunities available to attain its goals. 5.57 1.29 5.90 0.70 5 your bank’s stance on the analysis of risk comprises estimation of costs and benefits of dealing with risks. 5.56 1.10 5.79 0.80 6 analysis of risk embraces the ranking of risks and choosing those which require vigorous management. 5.57 1.05 5.91 0.68 7 analysis of risk includes the ranking of risk mitigation techniques to address scarce resources for implementation. 6.15 6.05 5.92 0.73 average 5.60 5.84 the highest mean value for islamic banks was 5.91 for statement 3, showing that reporting and communication process plays a significant role in risk monitoring. the results indicate that the control level of risk faced by banks was better in islamic banks than in conventional banks. overall, the mean value of all the six statements of risk monitoring was 5.79 for conventional banks and 5.85 for islamic banks. these results show that both types of banks were quite efficient in monitoring their risks, but the islamic banks were a little better than conventional banks in risk monitoring. such findings appear to be consistent with those in hussain and alajmi (2012). 73 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 table 5 responses to questions about the rm no. questions conventional banks islamic banks mean sd mean sd 1 efficacy of risk management is monitored as a vital part of regular management reporting. 5.97 1.06 5.80 0.80 2 the control level of bank is suitable for the risk exposure. 5.76 1.00 5.83 0.70 3 reporting and communication process of your bank is helpful for active risk mitigation. 5.79 0.99 5.91 0.77 4 banks’s response to risk includes an evaluation of the effectiveness of the existing controls and risk management responses. 5.79 1.01 5.88 0.59 5 identified risks are addressed by the bank through the application of decided strategies. 5.79 1.04 5.79 0.66 6 the bank’s stance to risk consists of an estimation of the expenses and paybacks of tackling the risks. 5.64 1.06 5.87 0.75 average 5.79 5.85 credit risk analysis credit risk is considered to be one of the most important risk. table 6 shows that the average mean value for all the seven questions was 5.94 and 5.85 for conventional banks and islamic banks, respectively. this indicates that conventional banks were slightly better than islamic banks in credit risk management. these results are similar to the findings in rehman et al. (2017), but contrary to those in hussain and alajmi (2012) because the islamic banks of bahrain were much more established and advanced than the islamic banks in pakistan. the mean values for all the statements were higher for conventional banks except for statements number 5 and 6. these values show that islamic banks were more concerned and conscious about a collateral taking policy and strictly follow the internal external rules. such findings are consistent with the observations in khalid and amjad 74 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 (2012) because islamic banks have unique credit products and a distinct set of rules which cannot be violated. however, conventional banks are quite particular about the classification of their borrowers and perform vigilant credit worthiness analysis, conduct better 7 cs analysis, and are less prone to give loans to the defaulters. table 6 responses to questions about the cra no. questions conventional banks islamic banks mean sd mean sd 1 credit worthiness analysis is conducted by the bank prior to the approval of loans. 6.19 0.97 5.86 0.85 2 the bank will commence explicit analysis like client’s character, capacity, collateral, capital and condition prior to loan approval. 6.32 0.82 5.99 0.75 3 the bank classifies its debtors based upon risk factor. 6.75 1.10 5.92 0.61 4 it is indispensable to acquire adequate collateral from the small debtors. 5.78 1.29 5.84 0.66 5 it is the bank’s strategy to take collaterals for the disbursement of all loans. 5.35 1.50 6.00 0.65 6 the bank prefers to take collaterals against some loans instead of all loans. 5.28 1.44 5.56 1.10 7 the level of loan given to default must be decreased. 5.94 1.13 5.76 0.97 average 5.94 5.85 liquidity risk analysis liquidity mismanagement is considered to be one of the major reasons for the financial crisis which occurred from 2007 until 2009. in this study, liquidity risk analysis has been included as an explanatory variable in the risk management process of banks. results are as shown in table 7. overall, the liquidity management of conventional banks was better with an average mean of 5.95. islamic banks 75 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 showed a lower average mean value of 5.89. this was because the islamic banks had fewer options for liquidity management. most of the liquidity management products and transactions available in the market were against shariah law and therefore, islamic banks could not use these products and transactions. similarly, investment options were also limited for islamic banks. as a result, islamic banks faced liquidity surpluses which led to idle funds and reduced the profitability of the banks. table 7 responses to questions about the lra no. questions conventional banks islamic banks mean sd mean sd 1 liquidity is the major predictor of the effectiveness of the banking sector. 6.25 0.89 5.78 0.81 2 bank management pays special attention to external and internal features causing liquidity risk while developing a liquidity management strategy. 5.91 0.95 5.86 0.75 3 procedures of the bank explain the overall plan about liquidity. 5.86 1.05 5.93 0.61 4 strategy is adequately elastic to cope with routine liquidity pressures. 5.80 0.94 5.94 0.60 5 bod and senior management review the liquidity policy regularly. 6.52 1.09 5.86 0.75 6 it is obligatory on the part of the almc to review and recommend a liquidity policy. 5.86 0.96 5.95 0.69 7 banks have recognized the ways and means to fulfill their funding needs. 5.89 0.98 6.03 0.67 8 stress testing and scenario analysis play a central role in the liquidity risk management framework of the bank. 5.76 1.14 5.84 0.72 9 the bank implements stress test according to the value at risk (var) technique to manage market risk. 5.73 1.16 5.80 0.74 average 5.95 5.89 76 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 table 8 responses to questions about the ora no. questions conventional banks islamic banks mean sd mean sd 1 bank follows the rule of tone at the top 5.64 1.20 5.76 0.73 2 governance of operational risk is wholly cohesive with overall risk management of the bank. 5.96 0.82 5.84 0.75 3 the bank has a system to identify and manage the inherent risk associated with products, services and business unit level. 5.72 0.98 5.89 0.78 4 operational risk management framework is chosen by the bank as per type, magnitude, intricacy and risk portfolio of the bank. 5.87 0.99 5.85 0.64 5 orm policies, processes and procedures are reviewed and updated periodically. 5.87 0.88 5.89 0.65 7 the bank sets and reviews the operational risk limit and tolerance and communicate it to respective members of the bank. 5.69 0.96 5.89 0.79 8 the bank warrants that operational risk for every new product, activity, procedure and system is fully evaluated through an approval procedure. 5.71 1.28 5.92 0.73 9 operational risk portfolio and substantial exposure to losses is regularly monitored by the bank. 6.00 0.84 6.47 0.73 10 the bank has a reporting mechanism at the board, senior management and business line level that aids the active management of risk. 5.90 1.11 5.92 0.71 11 the bank has robust settings to mitigate and transfer operational risk. 5.86 1.09 5.87 0.68 12 the bank has a contingency plan to function on a constant basis and bound losses on the occurrence of several business interruptions. 5.76 1.12 5.91 0.57 13 the bank strictly follows kyc and antimoney laundering rules as an ongoing practice. 6.06 1.09 6.01 0.80 average 5.92 5.98 77 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 operational risk analysis table 8 represents responses to operational risk analysis. it is another significant risk for both islamic banks and conventional banks. overall results show that both types of banks were almost equally efficient in operational risk management, but islamic banks (5.98) were slightly better than conventional banks (5.92). abdullah et al. (2011), ifsb (2005) and sundarajan (2007) also suggested that islamic banks must be more vigilant in operational risk management because the operational risk exposure of islamic banks is more than conventional banks. furthermore, this is because the operational risk will also include shariah non-compliance risk, legal and reputational risk (ifsb, 2005). the highest mean value (6.06) of statement 13 indicates that conventional banks strictly follow the ‘know your policy’ (kyc) and anti-money laundering rules in the course of routine business to avoid operational risks in the long run. similarly, the highest mean value of statement 9 (6.47) in islamic banks is the result of the monitoring the banks carry out on the operational risk profile on a regular basis. islamic banks also materialize the losses incurred more effectively than conventional banks. risk management practices the formulation of effective procedures and frameworks for risk management based upon a good understanding of risk ensures efficient risk management in a bank. it is the implementation of these policies and procedures by the staff at all levels which creates a difference, i.e., effective risk management practices. the results shown in table 9 reveal that there was not a big difference between both banking systems in terms of their risk management practices. conventional banks (5.88) were slightly better than islamic banks (5.85) in the implementation of risk management practices. these results are similar to those in the studies by khalid and amjad (2012) and raza bilal et al. (2013). it can be postulated that islamic banks pay more attention to the arrangement of training sessions for the employees, so as to improve the quality of risk management at all levels. conventional banks in pakistan need to pay more attention to on job training, by holding seminars and workshops to equip the employees with state-of-the-art knowledge and skills to deal with the various types of risk management tools. as for islamic banks, they need to improve their organizational performance review and feedback for better management of the various types of risks. 78 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 table 9 responses to questions about the rmps no. questions conventional banks islamic banks mean sd mean sd 1 overall performance of the bank is evaluated by senior management on a regular basis to manage various risk. 5.98 0.86 5.87 0.76 2 the bank has an extremely efficient incessant evaluation/ response on risk practices. 6.35 1.06 5.71 0.87 3 risk management dealings and policies are recorded by the bank, which is helpful for the staff in managing risk. 5.79 1.11 5.86 0.73 4 the bank’s policy emphasizes training programs in the field of risk management. 5.64 1.06 5.80 0.70 5 the bank encourages the employment of extremely competent people in risk management. 5.60 1.13 5.94 0.74 6 effective risk management is an important goal of the bank. 5.90 0.99 5.90 0.72 average 5.88 5.85 hypothesis testing table 10 presents the results of the regression analysis conducted for the conventional banks of pakistan. seven explanatory variables including, the urm, the ri, the raa, the rm, the cra, the lra and the ora have been used to study the variation in the rmps. results reveal that the tolerance coefficient of variables was between 0.389 to 0.574 and vif value also ranged between 1.742 to 2.572 which indicates that there is no multicollinearity problem among the explanatory variables. the data was considered normal, as 57.9 percent of the variance in the risk management practices of the conventional banks of pakistan could be explained by the independent variables. the f (18.085) was significant (p =0.000) at 1 percent. it means the overall model was a good fit. the estimated coefficient ( ) for the 79 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 raa (0.374), the rm (0.295) and the lra (0.342) was significantly related to the rmps at 10 percent and 1 percent, respectively. the urm (0.336), the ri (0.111) and the cra (0.054) were positively related to the rmps, but were insignificant. the ora (-0.253) was also insignificant and negative. it is obvious from the results that risk assessment and analysis (raa), risk monitoring (rm) and liquidity risk analysis (lra) are the most important variables that can influence the risk management practices of conventional banks. these results are partially similar to the ones in raza bilal et al (2013) because these researchers suggested that the urm, the ri, the raa, the rm and the cra were determinants of the rmps in pakistani banks. this difference can be attributed to the introduction of new risk management policies and their implementation phases that vary from time to time. table 10 regression analysis of the rmp and the rmps of conventional banks anova statistics collinearity statistics estimates se stand tstat. sig (p) tolerance vif (constant) -1.170 1.438 -0.814 0.418 urm 0.336 0.470 0.064 0.715 0.476 0.574 1.742 ri 0.111 0.194 0.054 0.572 0.569 0.509 1.965 raa 0.372 0.204 0.196 1.822 0.072*** 0.395 2.530 rm 0.295 0.101 0.296 2.922 0.004* 0.446 2.243 cra 0.054 0.145 0.040 0.372 0.711 0.389 2.572 lra 0.342 0.109 0.326 3.143 0.002* 0.426 2.345 ora -0.0253 0.566 -0.047 -0.447 0.656 0.414 2.415 r .0761 r2 0.579 adj. r2 0.547 fstatistics 18.085* note. * indicates significance at 1percent, ***indicates significance at 10 percent, dependent variable: rmps = risk management practices, independent variables: urm = understanding of risk management, ri = risk identification, raa, risk assessment and analysis, rm = risk monitoring, cra = credit risk analysis, lra = liquidity risk analysis, and ora = operational risk analysis. ß ß 80 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 table 11 shows the regression results of islamic banks. the statistical results show that d-w was 2.08, collinearity tolerance ranged between 0.311 to 0.680 and the vif value of all variables was less than 3. these results show that the data was normal and free from autocorrelation and the multicollinearity problem. the regression results indicate that the 70.06 percent variation in the rmps of islamic banks is explained by the urm, the ri, the raa, the rm, the cra, the lra and the ora, where f=31.523 was significant at 1 percent (p = 0.00). among the seven explanatory variables, the ri ( = 0.257, p = 0.00), the raa ( = 0.349, p = 0.00), the cra ( = 0.252, p = 0.01), the lra ( = 0.431, p = 0.00) and the ora ( = 0.486, p = 0.00) were significant and positively related to the rmps. the urm was however, insignificantly related to the risk management practices of the islamic banks of pakistan. table 11 regression analysis of the rmp and the rmps of islamic banks anova statistics collinearity statistics estimates se stand. tstat. sig (p) tolerance vif (constant) -1.058 0.500 -2.114 0.037 urm 0.024 0.104 0.019 0.227 0.821 0.457 2.187 ri 0.254 0.070 0.249 3.636 0.000* 0.680 1.470 raa -0.355 0.096 -0.324 -3.695 0.000* 0.417 2.400 rm 0.111 0.095 0.100 1.162 0.248 0.431 2.321 cra 0.247 0.096 0.216 2.571 0.012* 0.455 2.200 lra 0.432 0.107 0.336 4.051 0.000* 0.464 2.154 ora 0.478 0.129 0.378 3.716 0.000* 0.311 1.994 r 0.840 r2 0.706 adj. r2 0.684 f statistics 31.546* note. * indicates significance at 1percent, dependent variable: rmps = risk management practices, independent variables: urm= understanding of risk management, ri= risk identification, raa, risk assessment and analysis, rm= risk monitoring, cra= credit risk analysis, lra= liquidity risk analysis, and ora= operational risk analysis. ß ß ß ß ß ß 81 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 these results are consistent with the findings of khalid and amjad (2012). moreover, credit risk analysis and operational risk analysis are also important factors in addition to the raa, the rm and the lra which shows that islamic banks must pay more attention to credit and operational risk analysis. islamic banks have to follow strict credit and operational risk management policies due to their double supervisory structure, i.e., sbp regulation and shariah supervision board. hence, h1 of the study is accepted as the risk management practices of conventional banks and islamic banks of pakistan, as they were determined by the urm, the ri, the raa, the rm, the cra, the lra and the ora. the second hypothesis of the study was tested by a one-way anova. table 12 shows the results of the anova. results of the risk identification, risk assessment and analysis, risk monitoring, credit analysis, liquidity risk analysis and operational risk analysis were found to be not significantly different in the islamic banks and conventional banks of pakistan. however, the understanding of risk management (p= 0.007) and risk identification (p=0.006) were significantly different in islamic banks and conventional banks, a result which is partially consistent with the findings of hussain and alajmi (2012) and al tamimi and almazrooei (2007). the rmps of islamic banks and conventional banks was also not different from each other. there was no difference in most aspects of the risk management framework of both types of banks because both follow the same banking regulations under similar circumstances, i.e., a “level playing field” (imf, 2006). however, the difference in the understanding of risk is because of the different types of products and services which have exposed the banks to various unique and common risks. that is why the risk identification of both types of banks was significantly different from each other. nevertheless, the exposure of islamic banks was higher because they faced unique risks like the shariah compliance risk (comes under operational risk) which led to reputational and displaced commercial risk, i.e., profit sharing between bank and investment account holders where the bank shared the risk with the borrower (ariffin et al., 2009). on the other hand, conventional banks are less exposed to the risk because their financing products are fully backed by collaterals, which transfer all the risk to the client. 82 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 table12 analysis of variance of the risk management framework of islamic banks and conventional banks of pakistan ss df ms f sig (p) urm between groups 3.063 1 3.063 7.371 0.007* within groups 82.270 198 0.416 total 85.333 199 ri between groups 2.832 1 2.832 7.832 0.006* within groups 71.598 198 0.362 total 74.430 199 raa between groups 0.041 1 0.041 0.215 0.643 within groups 37.509 198 0.189 total 37.550 199 rm between groups 0.161 1 0.161 0.409 0.523 within groups 77.794 198 0.393 total 77.955 199 rmps between groups 0.361 1 0.361 0.870 0.352 within groups 82.219 198 0.415 total 82.580 199 cra between groups 0.605 1 0.605 2.341 0.128 within groups 51.174 198 0.258 total 51.779 199 lra between groups 0.007 1 0.007 0.022 0.882 within groups 66.636 198 0.337 total 66.644 199 ora between groups 0.204 1 0.204 0.595 0.442 within groups 67.862 198 0.343 total 68.066 199 note. * indicates significance at 1percent, urm = understanding of risk management, ri= risk identification, raa, risk assessment and analysis, rm= risk monitoring, rmps = risk management practices, cra= credit risk analysis, lra= liquidity risk analysis, and ora= operational risk analys 83 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 conclusion and implications this study has identified the impact of the risk management process, including the liquidity and operational risk on the risk management practices of islamic banks, as well as conventional banks. it then compared the risk management practices of islamic banks and conventional banks of pakistan. for this purpose, data was collected through the 200 questionnaires distributed to the selected respondents from the islamic banks and conventional banks. the data obtained was analyzed using basic descriptive statistics, one-way anova and multiple regression analysis. the descriptive analysis has revealed that both islamic banks and conventional banks understood that risk management is one of the most important and significant objectives of banks. islamic banks were a little better than conventional banks in terms of the policies, procedures, monitoring and feedback related to operational risk management. the liquidity management of conventional banks was better than islamic banks because islamic banks had fewer options for liquidity management. conventional banks were managing their liquidity risk in a slightly better way because all the loans were fully backed by the borrower’s assets (primary security), which transferred the whole risk to the customer. islamic banks however, had to share the risk of an asset with their clients because of shariah laws. for instance, the mudaraba and musharakah contracts. islamic banks are better at risk monitoring and risk assessment than conventional banks. islamic banking is an emerging sector in pakistan and doing well to earn a reasonable market share and competitive edge over conventional banks. overall, islamic banks are better than conventional banks in terms of understanding risk and risk management, risk identification, risk assessment and analysis, risk monitoring and operational risk analysis. on the other hand, conventional banks are ahead of islamic banks in credit risk analysis, liquidity risk analysis and risk management practices. the empirical comparison of risk management practices of islamic banks and conventional banks of pakistan carried out in this study has revealed that risk assessment and analysis (raa), risk monitoring (rm) and liquidity risk analysis (lra) were the most influential determinants of risk management practices of conventional banks. on 84 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 the other hand, in islamic banks the ri, the raa, the cra, the lra and the ora were found to have influenced their risk management practices. outcomes of the current study showed that liquidity risk analysis was a significant determinant of risk management practices in both islamic banks and conventional banks. moreover, operational risk analysis had a significant impact on the rmps in islamic banks only, suggesting that the operational risk was higher in islamic banks as compared to conventional banks, which was due to the inclusion of shariah non-compliance risk, in addition to other common categories of operational risk. theoretically, the results of this study can be seen as an endorsement of the belief that the various types of risk faced by banks due to the agency problem are resolved through the implementation of the standard risk management process that helps to maintain “institutionalization” in the banking sector. this in turn, will support effective formulation, implementation and monitoring by the regulators. the outcomes of this research have implications for strategy managers, potential and existing customers, investors and regulators. investors can decide based on the fact that unsystematic risk is higher in the stocks of islamic banks, while depositors must be aware that islamic banks are exposed to more risks and which can reduce their returns. similarly, debtors have to pay more profit (interest) due to the sharing of asset risk by the bank. nevertheless, an islamic bank cannot charge a higher profit (interest) to compete with their counterparts, as this action can reduce their profitability. islamic banks use the market interest rate as a benchmark and apply them according to the relevant shariah contract, which cannot be changed over the period of the contract. simultaneously, islamic banks are under pressure to pay profit as per market rate even when banks are not earning enough. banks do so because they want to retain their investment account holders to be competitive and liquid enough. when the actual profit is not enough to pay profit as per market rate, banks will reduce their profit share and pay more profit, which is more than the agreed ratio, to investment account holders. hence, the profitability of banks is reduced. the study of liquidity and operational risk, as part of risk management process shows that liquidity risk analysis is one of the important determinants of the rmps of islamic banks and conventional 85 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 banks, while the operational risk significantly influences the rmps of islamic banks. it is an important finding for the top management of banks to formulate more specific and effective strategies to maximize the efficiency of the risk management framework. these results are also an endorsement of the importance of the implementation of basel ii and iii, in true spirit. this study is also subject to some limitations. during the data collection period, the banking industry was going through some major changes in managerial structure and supervision policies initiated by the central bank. this was the result of the drastic changes occurring in the political scenario, which had led to the economic downfall in pakistan. this could have affected the responses to the study questionnaire to a certain extent. moreover, since this was a survey study where respondents were asked to share their own views and perceptions, these views could sometimes be contrary to their actions. moreover, it is suggested that more researches should be conducted to understand the rmps in other segments of the financial sector like in smes, the development financial institutions and insurance companies. the rmps cannot be implemented effectively in the absence of good governance. therefore, it needs to be tested empirically in future studies. the compliance level of banks with the internal and external policies is also a potential research area. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. references abu hussain, h., & al-ajmi, j. (2012). risk management practices of conventional and islamic banks in bahrain. the journal of risk finance, 13(3), 215-239. abdullah, m., shahimi, s., & ghafar ismail, a. (2011). operational risk in islamic banks: examination of issues. qualitative research in financial markets, 3(2), 131-151. 86 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 akhtar, m., ali, k., & sadaqat, s., (2011). liquidity risk management: a comparative study between conventional and islamic banks of pakistan. interdisciplinary journal of research in business, 1(1), 35-44. al ali, h., & naysary, b. (2014). risk management practices in islamic banks in kuwait. journal of islamic banking and finance, 2(1), 123-148. alam, m. z., & musukujjaman, m. (2011). risk management practices: a critical diagnosis of some selected commercial banks in bangladesh. journal of business and technology (dhaka), 6(1), 15-35. al-tamimi, h., & al-mazrooei m. (2007). banks risk management: a comparison study of uae national and foreign banks. the journal of risk finance, 8(4), 394-409. archer, s., & karim, r. a. a. (2013). insights and thought leadership. insights and thought leadership, 3(november), 11-20. ariffin, n. m., archer, s., & karim, r. a. a. (2009). risks in islamic banks: evidence from empirical research. journal of banking regulation, 10(2), 153-163. aurangzeb, d. (2012). contributions of banking sector in economic growth: a case of pakistan. economics and finance review, 2(6), 45-54. basel committee on banking supervision, (2008). principles of sound liquidity risk management and supervision. bis paper, bank of international settlement, basel, switzerland. https:// www.bis.org/publ/bcbs144.pdf. basel committee on banking supervision (2011). principles of sound management of operational risk. bis, basel, switzerland. https://www.bis.org/publ/bcbs195.html. bello, n., hasan, a., & saiti, b. (2017). the mitigation of liquidity risk in islamic banking operations. https://businessperspectives. org/images/pdf/applications/publishing/templates/article/ assets/9186/bbs_2017_03cont_bello.pdf. bessis, j. (2011). risk management in banking (3rd edn.). hoboken, nj, usa: wiley p. 841. bryman, a., & bell, e. (2011). ethics in business research. business research methods, 7(5), 23-56. bülbül, d., hakenes, h., & lambert, c. (2019). what influences banks’ choice of credit risk management practices? theory and evidence. journal of financial stability, 40, 1-14. 87 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 chernobai, a., ozdagli, a. k., & wang, j. (2018). business complexity and risk management: evidence from operational risk events in us bank holding companies. available at ssrn 2736509. collier, p. m., & woods, m. (2011). a comparison of the local authority adoption of risk management in england and australia. australian accounting review, 21(2), 111-123. deangelo, h., & stulz, r. m. (2015). liquid-claim production, risk management, and bank capital structure: why high leverage is optimal for banks. journal of financial economics, 116(2), 219-236. devellis, r. f. (1991). applied social research methods series, vol. 26. scale development: theory and applications. thousand oaks, ca, us: sage publications. giordana, g. a., & schumacher, i. (2013). bank liquidity risk and monetary policy: empirical evidence on the impact of basel iii liquidity standards. international review of applied economics, 27(5), 633 655. hassan, a., 2009. risk management practices of islamic banks of brunei darussalam. the journal of risk finance, 10(1), pp. 23-37. hassan, w. m. (2011). risk management practices: a comparative analysis between islamic banks and conventional banks in the middle east. international journal of academic research, 3(3), 288-295. hudin, n. s., & hamid, a. b. a. (2014). drivers to the implementation of risk management practices: a conceptual framework. journal of advanced management science, 2(3), 163-169. islamic financial services board. (2015). core principles for islamic finance regulation (banking segment) (cpifr). kula lumpur, malaysia. islamic financial services board. (2005). guiding principles of risk management for institutions (other than insurance institutions) offering only islamic financial services. kula lumpur, malaysia. http://www.ifsb.org/standard/ifsb1.pdf. jaffar, m., & manarvi, i. (2011). performance comparison of islamic and conventional banks in pakistan. global journal of management and business research, 11(1). jenkinson, n. (2008). strengthening regimes for controlling liquidity risk: some lessons from the recent turmoil. quarterly bulletin, bank of england, 48(2), 223 228. 88 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 jensen, m. c., & meckling, w. h. (1976). theory of the firm: managerial behavior, agency costs and ownership structure. journal of financial economics, 3(4), 305-360. kassim, s. h., & abdulle, m. y. (2012). impact of global financial crisis on the performance of islamic and conventional banks: empirical evidence from malaysia. journal of islamic economics, banking and finance, 113(470), 1-12. khalid, s., & amjad, s. (2012). risk management practices in islamic banks of pakistan. the journal of risk finance, 13(2), 148159. khan, i., khan, m., & tahir, m. (2017). performance comparison of islamic and conventional banks: empirical evidence from pakistan. international journal of islamic and middle eastern finance and management, 10(3), 419-433. khan, t., & ahmed, h. (2001). risk management: an analysis of issues in islamic financial industry (occasional papers) (no. 91). the islamic research and teaching institute (irti). klimczak, k. m. (2007). risk management theory: a comprehensive empirical assessment. https://mpra.ub.uni-muenchen. de/4241/1/mpra_paper_4241.pdf. kpmg international. (2009). never again? risk management in banking beyond the credit crisis. kpmg international, berne. mohad ariffin, n., & kassim, s. h. (2011, december). risk management practices and financial performance of islamic banks: malaysian evidence. in 8th international conference on islamic economics and finance (pp. 19-21). muhammad, b., khan, s., & xu, y. (2018). understanding risk management practices in commercial banks: the case of the emerging market. risk governance and control: financial markets & institutions, 8(2), 54-62. nasser, s. a. s. a., & muhammed, d. j. (2013). introduction to history of islamic banking in malaysia, humanomics, 29(2), 80-87. nazir, m. s., daniel, a., & nawaz, m. m. (2012). risk management practices: a comparison of conventional and islamic banks in pakistan. american journal of scientific research, 68(68), 114-122. neifar, s., & jarboui, a. (2018). corporate governance and operational risk voluntary disclosure: evidence from islamic banks. research in international business and finance, 46, 43-54. 89 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 nunnally, j. c. (1978). psychometric theory. mcgraw-hill. ouerghi, f. (2014). are islamic banks more resilient to global financial crisis than conventional banks? asian economic and financial review, 4(7), 941-955. raza bilal, a., bt. abu talib, n., & noor azli ali khan, m. (2013). remodeling of risk management in banking: evidence from the sub-continent and gulf. the journal of risk finance, 14(5), 468-489. rehman, a. a., benamraoui, a., & dad, a. m. (2017). a comparative study of islamic and conventional banks’ risk management practices: empirical evidence from pakistan. journal of banking regulation, 1-14. sekaran, u., & bougie, r. (2016). research methods for business: a skill building approach. john wiley & sons. sbp (2018). sharaih governance framework for islamic banking institution. islamic banking department, sbp, karachi, pakistan. shafiq, a., & nasr, m. (2010). risk management practices followed by the commercial banks in pakistan. international review of business research papers, 6(2), 308-325. shafique, o., hussain, n., & taimoor hassan, m. (2013). differences in the risk management practices of islamic versus conventional financial institutions in pakistan: an empirical study. the journal of risk finance, 14(2), 179-196. sleimi, m. (2020). effects of risk management practices on banks’ performance: an empirical study of the jordanian banks. management science letters, 10(2), 489-496. state bank of pakistan (2019), islamic banking bulletin. islamic banking department, sbp. karachi. https://dnb.sbp.org.pk/ ibd/bulletin/2019/dec.pdf. state bank of pakistan. (2003). risk management guidelines for commercial banks & dfis. karachi. http://www.sbp.org.pk/ riskmgm.pdf. sundararajan, v. (2007). risk measurement and disclosure in islamic finance and the implications of profit-sharing investment accounts. islamic economics and finance, 121. tolbert, p. s., & zucker, l. g. (1983). institutional sources of change in the formal structure of organizations: the diffusion of civil service reform, 1880-1935. 90 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 57–90 van, greuning, h., & iqbal, z. (2008). risk analysis for islamic banks. world bank publications, herndon, va, usa. http:// www.iefpedia.com/english/wp. wood, a., & kellman, a. (2013). risk management practices by barbadian banks. international journal of business and social research, 3(5), 22-33. 1 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: bashir, r., & regupathi, a. (2022). aggregate and disaggregate measures of operating and non-operating working capital influence on firm performance: evidence from malaysia. international journal of banking and finance, 17(1), 1-26. https://doi. org/10.32890/ ijbf2022.17.1.1 aggregate and disaggregate measures of operating and non-operating working capital influence on firm performance: evidence from malaysia 1rabia bashir & 2angappan regupathi department of finance school of economics finance and banking universiti utara malaysia 1corresponding author: rabiauum@gmail.com received: 10/12/2020 revised: 1/4/2021 accepted: 22/4/2021 published: 2/12/2021 abstract the study is aimed at investigating the following issues: firstly, whether the different types of working capital, namely operating and non-operating working capital influence the short-term (return on assets) and long-term (tobin’s q) firm performance differently, and secondly whether the different measures of operating working capital, namely disaggregated and aggregated (cash conversion cycle) operating working capital, influence the short-term (return on assets) and long-term (tobin’s q) firm performance differently. it uses the panel data of 208 listed non-financial firms in malaysia covering the period from 2013 to 2017, and the data has been sourced from datastream. it employs the panel corrected standard errors regression model. the study has found that quicker sale of inventory increased 2 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 both the short-term and long-term performance of the firm. likewise, faster collection of receivables increased the long-term, but not shortterm, performance. however, prompter payment of payables increased both the short-term and long-term performance. the study has also found that the disaggregated working capital measures – inventory, receivables, and payables contributed to a more nuanced influence of working capital on performance, compared to the aggregated working capital. the study has provided novel evidence that– higher nonoperating working capital increased firm performance. keywords: working capital, short-term firm performance, long-term firm performance, pcse regression, malaysian firms. jel classification: g3. introduction the literature on corporate finance has given a greater emphasis and focus on long-term financial decisions. however, the short-term resources (i.e., current assets) and short-term obligations (i.e., current liabilities) are also important components of total assets, and it is vital too that they be given due scrutiny. the management of current assets and current liabilities is known as working capital management. the size of current assets and current liabilities varies from firm to firm. expansion of a firm’s operations could induce an increase in working capital investment (padachi, 2006). additional investment in the working capital is expected to have positive effects for the firms with low working capital levels, however, it may also have adverse effects and lead to the loss of shareholder value for firms with already high working capital levels. working capital management has always been important to the liquidity and profitability of businesses (deloof, 2003). a conservative working capital policy would try to minimize financial distress by maintaining more than sufficient working capital investment, whereas an aggressive working capital policy would try to maximize profitability by retaining only the minimum working capital investment (kieschnick et al., 2013). conservative management of inventory would entail reducing the ordering cost, as well as eliminating the risk of stock-outs and cost of forgoing purchase discounts (corsten & gruen, 2004). allowing the customers to pay late might increase 3 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 sales and improve customer relationship (summers & wilson, 2003), although it might also increase the risk of bad debts. however, aggressive inventory management would involve reducing holdings costs and the risk of obsolescence. a longer payable period might help to improve short-term liquidity, but simultaneously, it might also endanger the relationship with suppliers. working capital management has remained a problem for many manufacturing firms across the globe. however, the malaysian economy showed strong signs of recovery in 2010 from the global economic crisis (watanabe et al., 2013)2013. because of the increased competition to enhance performance among firms in developing nations, challenges for malaysian firms are increasing. a market research report by the pwc (2018) on the state of working capital investment of listed malaysian firms across 14 sectors comprising 424 firms, revealed that up to rm 110 billion cash was tied up in their working capital. this implied that some of this could be released by improving the working capital performance. generally, studies such as those by mohamad and saad (2010), ng et al. (2017), zariyawati et al. (2017), alarussi and alhaderi (2018), al-mawsheki et al. (2019) and sim et al. (2019) have investigated the effect of working capital management on the profitability of malaysian firms. however, this present study differs from the aforementioned studies in several aspects. first, it has divided working capital into two primary groups – operating and non-operating working capital management. operating working capital includes current asset and current liability accounts that change spontaneously with the firm’s operating scale. these accounts include inventory, receivables, and payables. non-operating working capital includes current asset and liability accounts that do not change spontaneously with the firm’s operating scale, such as loans receivable, idle equipment, unused land or outdated machinery. these should include all other current asset and liability accounts (le, 2019). as the two groups of working capital are of different natures, and more importantly, their influence on firm performance might be different, they are measured separately. the measure of non-operating working capital used is other net working capital (onwc), derived by taking the sum of all non-operating current assets and then deducting from it the sum of all non-operating current liabilities (expressed as a percentage of total assets). second, this present study has further conceived using two alternative measures of operating working capital – disaggregate and aggregate 4 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 – separately. three disaggregate measures of operating working capital are used in the present study, each of which was derived using each of the three operating working capital component accounts, i.e., inventory, receivables, and payables (zariyawati et al., 2017). the measures for these three operating working capital components are inventory conversion period (icp), average collection period (acp), and average payment period (app), respectively. as an alternative, the study has also used the aggregate measure of operating working capital, which is the cash conversion cycle (ccc), and this is measured as the summation of icp and acp minus app (vural et al., 2012). the use of these disaggregate and aggregate measures of operating working capital has allowed the study to examine which of the two measures could explain firm performance better. third, the study has also used the following two measures of firm performance, return on asset (roa) and tobin’s q (tq). in this context, roa is a short-term accounting-based measure, whereas tq is a market-based, and arguably, a long-term measure of a firm’s performance (masadeh et al., 2015; smirlock et al., 1984). although accounting-based data has been useful in capturing firm performance (giner & reverte, 2006), investors and economists have been more concerned with market-based performance measure (ullmann, 1985; woo et al., 1992). market-based measure reflect the market (investors’) perception of the expected future performance of the firms (dubofsky & varadarajan, 1987; wisner & eakins, 1994), and is less influenced by company-specific financial reporting rules and managerial manipulation (mcguire et al., 1988). in an attempt to resolve these issues, various studies have found tq to be a more accurate measure of firm performance, compared to the use of the accounting-based firm performance measures (wolfe & sauaia, 2005). according to barney (2007), tq has various advantages over the accounting-based firm performance measures since tq does not rely only on profits, which are framed by accounting rules. moreover, tq, it has been argued, is seen as a long-term firm performance measure (dwivedi & jain, 2005) and is based on current information that reflects the present value of future cash flows (ganguli & agrawal, 2009; wahla & hussain, 2012). a tq value higher than 1 would mean that the firm has created more value by using invested resources (li et al., 2004), has better investment opportunities (lang et al., 1989), and has a higher growth potential (brainard & tobin, 1968). while the value lower than 1 indicates that the firm’s market value is less than the book value of the assets, which means that the market value 5 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 of the firm’s assets, based on investors’ assessment, is less than the investment made (at cost) in the firm’s assets. even though the data of accounting return has been criticized as not sufficiently representing genuine economic value, not the least, because of its possibility of being manipulated by managers (johnson & kaplan, 1987), it has still been widely accepted and applied in many contexts (maines & wahlen, 2006) by various users in decision making (giner & reverte, 1999; truica & trandafir, 2009). therefore, this present study has used both the accounting-based and market-based measures of firm performance and has defined them as short-term performance (roa) and long-term performance (tq) respectively (dwivedi & jain, 2005; rafindadi & bello, 2019). the rest of the article has been organized into five sections as follows: section 2 provides a brief review of previous studies. section 3 describes the data source and the research methodology. in section 4, the results of analysis are discussed. finally, section 5 provides the conclusion. literature review there is a growing body of literature analyzing the working capital management and firm performance relationship in different countries, for example studies like vătavu (2014) in romania; adekola et al. (2017) in nigeria; adam et al. (2017) in ghana; tahir and anuar (2015) in pakistan; afrifa and padachi (2016) in the uk; and bhatia and srivastava (2016) in india. the findings of these studies highlighted a number of variables that have had a significant impact on a firm’s performance. the prevalent variables used to explain firm performance were namely, inventory conversion period, average collection period, average payment period, cash conversion cycle, liquidity, firm size, leverage, and sales growth. however, the findings of these studies were rather divergent, primarily due to the underlying differences in the period of study, performance measures used, and the use of different sets of firm-specific and country-specific explanatory factors. afza and adnan (2007) investigated the relationship between aggressive and conservative working capital policies and their relationship with firm performance for 17 industrial groups within a sample of 263 firms listed on the karachi stock exchange for the period 1998-2003. the negative relationship between aggressive working capital financing 6 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 and investment policies and firm performance suggested that firms yielded lower performance when they had followed aggressive working capital financing and investment policies. abuzayed (2012) investigated the relationship between working capital management and firm profitability for 52 firms listed on the amman stock exchange from 2000 to 2008. results of the study found that gross profit was positively correlated with cash conversion cycle and average collection period, but negatively correlated with inventory conversion period. bhatia and srivastava (2016) conducted a study on 179 nonfinancial firms listed on the bombay stock exchange and determined that by decreasing the cash conversion cycle days, average inventory days, and average receivable period, and by increasing the average payment period, firms can increase profitability. zhang et al. (2017) conducted a study on 140 smes listed on the london stock exchange between 2008 and 2016. findings revealed that longer cash conversion cycle, longer average payment period, longer average receivable period, and longer average inventory days increased firm profitability. in another study on indonesian firms, kusuma and bachtiar (2018) reported that shorter cash conversion cycle, higher inventory turnover, and shorter average payment period increased firm performance. compared to the numerous studies mentioned above on working capital and firm performance in developed and other developing countries, there have been relatively, few studies carried out in malaysia. for example, mohamad and saad (2010) examined the influence of working capital management on profitability for 172 firms listed on bursa malaysia from 2003 to 2007. their study employed tq, return on invested capital and roa as the proxies for firm profitability. correlation and multiple regression analyses indicated that working capital measures, namely cash conversion cycle, current asset to current liabilities ratio, current asset to total asset ratio, current liabilities to total asset ratio, and total debt to total asset ratio significantly influenced the three measures of profitability. zariyawati et al. (2017) investigated the relationships between cash conversion cycle, inventory turnover in days, average collection period, average payment period and firm profitability for 30 large and 70 small firms listed on bursa malaysia from 2009 to 2013. roa was employed as a proxy of firm profitability. the hausman specification resulted in the adoption of the random effect method, which suggested 7 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 that large firms could increase their profitability by increasing the average payment period, while small firms could increase their profitability by reducing the average collection period and inventory days. alarussi and alhaderi (2018) examined the relationship between working capital and profitability of 120 malaysian listed firms from 2012 to 2014. they selected return on equity (roe) and earning per share (eps) to measure firm profitability. their study provided evidence that working capital components differently influenced different profitability measures. in the regression results, they found that working capital significantly and positively influenced eps, but not roe. recently, wong et al. (2019) investigated the impact of working capital on sme’s profitability from 2005 to 2012. their findings revealed that malaysian smes created value by adopting efficient working capital practices. researchers have also employed different measures of working capital and their efforts have provided mixed results. for example, studies by akbar (2014), gul et al. (2013), makori and jagongo (2013), and rehman and anjum (2013), showed that cash conversion, inventory turnover ratio and average payment period significantly influenced firm profitability. in contrast, some studies showed no relationship between the mentioned measures and firm profitability (see arshad & gondal, 2013 and manzoor, 2013). previous studies such as the work by kusuma and bachtiar (2018), rehman and anjum (2013) and wuryani (2015) used the aggregate measure of net working capital ratio that included inventory, accounts receivable, and accounts payables along with its component measures of icp, acp and app in the same model. it is argued here that this practice was inappropriate because then working capital would have been measured multiple times leading to multicollinearity arising from the inter-dependent ‘independent’ variables. therefore, the present study has avoided this problem by ensuring that no two related variables or measures of working capital are included in the same model. crucially, the previous studies have not identified these following areas of concern: (1) whether the disaggregate measures of operating working capital are, or the aggregate measure of operating working capital is, more useful in explaining firm performance, (2) whether non-operating working capital would influence firm performance differently than operating working capital, and (3) whether working capital would affect short-term performance (roa), long-term performance (tq), or both. this present study has tried to fill in the aforementioned knowledge gaps. 8 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 methodology this study has used secondary data which were retrieved from the thompson reuters financial and economic database, datastream. the sample consisted of non-financial firms listed on bursa malaysia for the period 2013-2017. a total of 776 non-financial firms were listed on bursa malaysia in all sectors (bursa malaysia, 2020). this study has focused only on three sectors that had more than 100 firms each, and they are as follows: (a) industrial products and services, (b) consumer products and services, and (c) construction. only these three sectors have been selected as only these had more than 100 firms each, and possessed full data for the five-year time frame of the study, that is, from 2013-2017. table 1 frequency of firms by sector sectors number of listed firms available in datastream sampled firms industrial products & services 220 96 84 consumer products & services 179 115 89 construction 111 74 35 510 255 208 the datastream database categorized sectors differently from bursa malaysia. the present study has combined the industrial engineering and industrial metals and mining sectors (as categorized by datastream) into the industrial products and services sector. similarly, the food producers and household goods and home construction sectors have been combined and re-categorized as the consumer products and services sector. only data for 255 firms was available throughout the years 2013 to 2017 in datastream. however, 47 of these firms were removed from the study sample because their data seemed to lack credibility, e.g., they had very high acp, icp, app, or ccc. as a result, the study was left with a sample of 208 firms. table 1 shows the number of firms in these three sectors, while table 2 lists the definitions of the dependent, independent and control variables used in the study. 9 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 table 2 definition of variables variable label name definition dependent variable performance roa return on assets net profit total assets tq tobin’s q (market value of equity + book value of liability) book value of total assets independent variables working capital icp inventory conversion period inventory x 365 net sales acp average collection period accounts receivable x 365 net sales app average payment period accounts payable x 365 net sales ccc cash conversion cycle icp+acp-app onwc other net working capital ratio other current assets-other current liabilities total assets where other current assets mean current assets minus accounts receivable, while other current liabilities means current liabilities minus accounts payable. control variables fs firm size natural logarithm of total assets dr debt ratio total liabilities total assets the analysis of the data was carried out in two stages. the first stage explored the descriptive statistics for all variables, and these included their mean, minimum, maximum, standard deviation, skewness, and kurtosis. in the second stage, the relationship between the working capital measures and firm performance was examined using linear regression with the panel-corrected standard errors method. 10 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 analysis and discussion of results this section presents the analysis and discussion. table 3 shows the key descriptive statistics of the variables used in the study for the 208 firms from 2013 to 2017. icp is the average number of days a firm takes to sell its inventory. a lower icp reveals that the firm is taking a shorter time to turn its inventory into sales. acp represents the average number of days a firm takes to collect its receivables from its customers. in other words, acp indicates the efficiency in collecting the receivables. similarly, app is the average number of days that payment for a purchase is outstanding, which indicates how many days, on average, a firm takes to pay off its suppliers. app may also indicate the credit terms granted by the suppliers. a longer app may indicate a longer credit term granted by suppliers. onwc indicates the proportion of other net working capital (other than inventory, receivables, and payables, over total assets) a firm has used. ccc is icp plus acp minus app, and indicates the average number of days between a firm paying its suppliers and the firm collecting from its customers. a shorter ccc is broadly taken to reflect a more efficient management of working capital. it also reflects less amount of inventory and receivables that are not financed by payables. in fact, a negative ccc implies that a firm’s payables are more than sufficient to finance its inventory and receivables. in other words, it means that the firm collects from the sale of inventories, and the collection of receivables, earlier than it pays for its payables. table 3 descriptive statistics n mean minimum maximum std. dev. skewness kurtosis roa 1040 0.047 -0.572 0.607 0.114 -1.170 9.490 tq 1040 1.055 0.545 2.733 0.461 1.837 6.419 icp 1040 114.250 5.000 512.000 106.665 2.003 7.150 acp 1040 113.263 4.000 676.000 95.103 2.084 8.794 app 1040 79.970 10.000 321.000 71.244 1.629 5.352 ccc 1040 148.754 -657.000 1077.000 150.821 1.501 9.433 onwc 1040 0.106 -0.505 0.765 0.202 0.037 3.043 fs 1040 13.065 10.015 17.593 1.402 0.475 3.253 dr 1040 0.413 0.025 0.936 0.187 0.121 2.389 11 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 the mean values indicate that the sample firms, on average took 114 days to sell their inventory, 113 days to collect receivables, and 80 days to pay for payables. the mean ccc value indicates that the average time between paying the payables and collecting from sales was 149 days. the mean value of onwc implies that the excess of other current assets (other than inventory and receivables) over other current liabilities (other than payables) was, on average, equivalent to 10.6 percent of total assets. among control variables, the average firm size (total assets) of the sample firms was rm1.542 billion (not shown in table 3), but which is reflected by the mean of the natural of total assets of these firms, 13.065. in the sample, only 26 percent of the firms had assets of more than rm1 billion. finally, the debt ratio shows that firms were on average using debt to finance 41 percent of the assets. table 4 frequency distribution of roa and tq range frequency percent cumulative percent roa less than -2.50% 63 6.1 6.1 -2.51% to 0% 132 12.7 18.8 0.01% to 2.50% 591 56.8 75.6 more than 2.50% 254 24.4 100.0 total 1040 100 tq less than 0.75 248 23.8 23.8 0.76 to 1.00 385 37.0 60.9 1.01 to 1.25 389 37.4 98.3 more than 1.25 18 1.7 100.0 total 1040 100 while table 3 provides the mean values of both roa and tq, table 4 provides the frequency distribution across the roa and tq categories. the roa distribution seems normal; only 18.8 percent of the firms had a negative roa, while 56.8 percent had roa from 0.01 percent to 2.50 percent. table 3 shows that the average roa was 4.7 percent over the five-year period under investigation. although table 3 shows that the average tq was slightly above 1, table 4 shows that 61 percent of the firms had tq values of less than 1 and only 39 12 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 percent had values higher than 1, meaning that the market or investors seemed to perceive that for the majority of the firms, the firms’ assets were worth less then compared to the cost of investments made in those assets before then. table 5 shows the frequency distribution for the icp, acp, app, and ccc categories. the frequency distributions for the first three have been grouped into the following five categories – less than 30 days, 31 to 90 days, 91 to 180 days, 181 to 360 days, and more than 360 days. table 5 frequency distribution of icp, acp, app and ccc range frequency percent cumulative percent icp less than 30 days 149 14.3 14.3 31 to 90 days 428 41.2 55.5 91 to 180 days 282 27.1 82.6 181 to 360 days 134 12.9 95.5 more than 360 days 47 4.5 100 total 1040 100 acp less than 30 days 100 9.6 9.6 31 to 90 days 456 43.8 53.5 91 to 180 days 312 30.0 83.5 181 to 360 days 144 13.8 97.3 more than 360 days 28 2.7 100.0 total 1040 100 app less than 30 days 275 26.4 26.4 31 to 90 days 447 43 69.4 91 to 180 days 216 20.8 90.2 181 to 360 days 86 8.3 98.5 more than 360 days 16 1.5 100 total 1040 100 ccc less than 0 days 141 13.6 13.6 1 to 90 days 237 22.8 36.3 91 to 180 days 351 33.8 70.1 181 to 360 days 243 23.4 93.5 more than 360 days 68 6.5 100 total 1040 100 13 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 the data indicated that the (mode) majority of malaysian firms generally took between 31 to 90 days to sell their inventory (41.2%), to collect receivables (43.8%), and to pay payables (43%). however, for 33.8% (mode) of the firms, their cash conversion cycle (ccc), i.e., the delay between paying payables and collecting receivables, was between 91 to 180 days. table 6 presents the frequency distribution of the onwc variable. the frequency distribution is grouped into the following four categories – less than zero, zero to 0.25, 0.26 to 0.50 and more than 0.50. table 6 shows that onwc in 44.7 percent of the firms ranged between zero to 0.25. in 31.1 percent of the firms, onwc was less than zero, while in only 2.7 percent of the firms, it was more than 0.50. these seemed to imply that in almost half of the firms, excess current assets over current liabilities, not including inventory, receivables, and payables, accounted for between 0 and 25 percent of the firms’ total assets. table 6 frequency distribution of onwc range frequency percent cumulative percent onwc less than 0.00 323 31.1 31.1 0.00 to 0.25 465 44.7 75.8 0.26 to 0.50 224 21.5 97.3 more than 0.50 28 2.7 100 total 1040 100 linear regression with panel corrected standard errors as the literature review has revealed, there are different measures of working capital and firm performance. the present study has used four empirical models in order to examine whether the disaggregate measures of operating working capital (icp, acp, and app) affect, or the aggregate measure (ccc) of operating working capital affects, a firm’s performance more, both in the short-term (roa) and long-term (tq). the regression models are specified as follows: 14 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 model 1 – disaggregate operating working capital shortterm performance (1) model 2 – aggregate operating working capital short-term performance (2) model 3 – disaggregate operating working capital long-term performance (3) model 4 – aggregate operating working capital short-term performance (4) where α0 is the constant, and β1, β2, β3, β4, β5, and β6 are regression coefficients, while is the error term. the time-series data exhibited autocorrelations, while the crosssection data displayed heteroscedasticity. as a result, the standard errors estimated by the ordinary least squares (ols) method would be considered to be incorrect (bailey & katz, 2011). table 7 exhibits the results of heteroscedasticity and cross-sectional dependence. the cross-sectional dependence was estimated using the pesaran test, friedman test and frees test. the pesaran and friedman tests showed conflicting results, and therefore, the present study had to rely on the frees test. the results indicated the presence of heteroscedasticity and autocorrelation and correlation among cross-sections. although the generalized least squares (gls) method (parks, 1967) is considered theoretically superior to the ols, it is generally only applicable to time series cross-sectional data (beck & katz, 1995). a practical and better alternative is the feasible gls (fgls). however, fgls tends to underestimate the precise variability of the estimator, if the time period is less than the cross-sectional units (beck & katz, 1995; jönsson, 2005). to avoid these problems, the panel corrected standard errors (pcse) method, which yields robust covariances has been suggested in the analysis of time-series cross-sectional data (beck & katz, 1995). when computing the standard errors and the variance– covariance estimates, pcse removes the heteroscedastic and crosssectional dependence problems across panels (moundigbaye et al., 2018). 10 frequency distribution of onwc range frequency percent cumulative percent onwc less than 0.00 323 31.1 31.1 0.00 to 0.25 465 44.7 75.8 0.26 to 0.50 224 21.5 97.3 more than 0.50 28 2.7 100 total 1040 100 linear regression with panel corrected standard errors as the literature review has revealed, there are different measures of working capital and firm performance. the present study has used four empirical models in order to examine whether the disaggregate measures of operating working capital (icp, acp, and app) affect, or the aggregate measure (ccc) of operating working capital affects, a firm’s performance more, both in the short-term (roa) and long-term (tq). the regression models are specified as follows: model 1 – disaggregate operating working capital short term performance 𝑅𝑅𝑅𝑅𝐴𝐴𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐴𝐴𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐴𝐴𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (1) model 2 – aggregate operating working capital short-term performance 𝑅𝑅𝑅𝑅𝐴𝐴𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (2) model 3 – disaggregate operating working capital long-term performance 𝑇𝑇𝑄𝑄𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐴𝐴𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐴𝐴𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (3) model 4 – aggregate operating working capital short-term performance 𝑇𝑇𝑄𝑄𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (4) where α0 is the constant, and β1, β2, β3, β4, β5, and β6 are regression coefficients, while ɛ𝑖𝑖,𝑡𝑡 is the error term. the time-series data exhibited autocorrelations, while the cross-section data displayed heteroscedasticity. as a result, the standard errors estimated by the ordinary least squares (ols) method would be considered to be incorrect (bailey & katz, 2011). table 7 exhibits the results of heteroscedasticity and cross-sectional dependence. the cross-sectional dependence was estimated using the pesaran test, friedman test and frees test. the pesaran and friedman tests showed conflicting results, and therefore, the present study had to rely on the frees test. the results indicated the presence of heteroscedasticity and autocorrelation and correlation among cross-sections. table 7 heteroscedasticity and correlation tests 10 frequency distribution of onwc range frequency percent cumulative percent onwc less than 0.00 323 31.1 31.1 0.00 to 0.25 465 44.7 75.8 0.26 to 0.50 224 21.5 97.3 more than 0.50 28 2.7 100 total 1040 100 linear regression with panel corrected standard errors as the literature review has revealed, there are different measures of working capital and firm performance. the present study has used four empirical models in order to examine whether the disaggregate measures of operating working capital (icp, acp, and app) affect, or the aggregate measure (ccc) of operating working capital affects, a firm’s performance more, both in the short-term (roa) and long-term (tq). the regression models are specified as follows: model 1 – disaggregate operating working capital short term performance 𝑅𝑅𝑅𝑅𝐴𝐴𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐴𝐴𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐴𝐴𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (1) model 2 – aggregate operating working capital short-term performance 𝑅𝑅𝑅𝑅𝐴𝐴𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (2) model 3 – disaggregate operating working capital long-term performance 𝑇𝑇𝑄𝑄𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐴𝐴𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐴𝐴𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (3) model 4 – aggregate operating working capital short-term performance 𝑇𝑇𝑄𝑄𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (4) where α0 is the constant, and β1, β2, β3, β4, β5, and β6 are regression coefficients, while ɛ𝑖𝑖,𝑡𝑡 is the error term. the time-series data exhibited autocorrelations, while the cross-section data displayed heteroscedasticity. as a result, the standard errors estimated by the ordinary least squares (ols) method would be considered to be incorrect (bailey & katz, 2011). table 7 exhibits the results of heteroscedasticity and cross-sectional dependence. the cross-sectional dependence was estimated using the pesaran test, friedman test and frees test. the pesaran and friedman tests showed conflicting results, and therefore, the present study had to rely on the frees test. the results indicated the presence of heteroscedasticity and autocorrelation and correlation among cross-sections. table 7 heteroscedasticity and correlation tests 10 frequency distribution of onwc range frequency percent cumulative percent onwc less than 0.00 323 31.1 31.1 0.00 to 0.25 465 44.7 75.8 0.26 to 0.50 224 21.5 97.3 more than 0.50 28 2.7 100 total 1040 100 linear regression with panel corrected standard errors as the literature review has revealed, there are different measures of working capital and firm performance. the present study has used four empirical models in order to examine whether the disaggregate measures of operating working capital (icp, acp, and app) affect, or the aggregate measure (ccc) of operating working capital affects, a firm’s performance more, both in the short-term (roa) and long-term (tq). the regression models are specified as follows: model 1 – disaggregate operating working capital short term performance 𝑅𝑅𝑅𝑅𝐴𝐴𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐴𝐴𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐴𝐴𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (1) model 2 – aggregate operating working capital short-term performance 𝑅𝑅𝑅𝑅𝐴𝐴𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (2) model 3 – disaggregate operating working capital long-term performance 𝑇𝑇𝑄𝑄𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐴𝐴𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐴𝐴𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (3) model 4 – aggregate operating working capital short-term performance 𝑇𝑇𝑄𝑄𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (4) where α0 is the constant, and β1, β2, β3, β4, β5, and β6 are regression coefficients, while ɛ𝑖𝑖,𝑡𝑡 is the error term. the time-series data exhibited autocorrelations, while the cross-section data displayed heteroscedasticity. as a result, the standard errors estimated by the ordinary least squares (ols) method would be considered to be incorrect (bailey & katz, 2011). table 7 exhibits the results of heteroscedasticity and cross-sectional dependence. the cross-sectional dependence was estimated using the pesaran test, friedman test and frees test. the pesaran and friedman tests showed conflicting results, and therefore, the present study had to rely on the frees test. the results indicated the presence of heteroscedasticity and autocorrelation and correlation among cross-sections. table 7 heteroscedasticity and correlation tests 10 frequency distribution of onwc range frequency percent cumulative percent onwc less than 0.00 323 31.1 31.1 0.00 to 0.25 465 44.7 75.8 0.26 to 0.50 224 21.5 97.3 more than 0.50 28 2.7 100 total 1040 100 linear regression with panel corrected standard errors as the literature review has revealed, there are different measures of working capital and firm performance. the present study has used four empirical models in order to examine whether the disaggregate measures of operating working capital (icp, acp, and app) affect, or the aggregate measure (ccc) of operating working capital affects, a firm’s performance more, both in the short-term (roa) and long-term (tq). the regression models are specified as follows: model 1 – disaggregate operating working capital short term performance 𝑅𝑅𝑅𝑅𝐴𝐴𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐴𝐴𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐴𝐴𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (1) model 2 – aggregate operating working capital short-term performance 𝑅𝑅𝑅𝑅𝐴𝐴𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (2) model 3 – disaggregate operating working capital long-term performance 𝑇𝑇𝑄𝑄𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝐴𝐴𝐼𝐼𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐴𝐴𝑃𝑃𝑃𝑃𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽6𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (3) model 4 – aggregate operating working capital short-term performance 𝑇𝑇𝑄𝑄𝑖𝑖,𝑡𝑡 = 𝛼𝛼0 + 𝛽𝛽1𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽2𝑅𝑅𝑂𝑂𝑂𝑂𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐹𝐹𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐷𝐷𝑅𝑅𝑖𝑖,𝑡𝑡 + ɛ𝑖𝑖,𝑡𝑡 (4) where α0 is the constant, and β1, β2, β3, β4, β5, and β6 are regression coefficients, while ɛ𝑖𝑖,𝑡𝑡 is the error term. the time-series data exhibited autocorrelations, while the cross-section data displayed heteroscedasticity. as a result, the standard errors estimated by the ordinary least squares (ols) method would be considered to be incorrect (bailey & katz, 2011). table 7 exhibits the results of heteroscedasticity and cross-sectional dependence. the cross-sectional dependence was estimated using the pesaran test, friedman test and frees test. the pesaran and friedman tests showed conflicting results, and therefore, the present study had to rely on the frees test. the results indicated the presence of heteroscedasticity and autocorrelation and correlation among cross-sections. table 7 heteroscedasticity and correlation tests 15 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 ta bl e 7 h et er os ce da st ic ity a nd c or re la tio n te st s m od el 1 m od el 2 m od el 3 m od el 4 b re us ch -p ag an / c oo kw ei sb er g te st fo r he te ro sc ed as tic ity ch i2 (1 ) = 3 2. 17 pr ob > c hi 2 = 0. 00 00 ch i2 (1 ) = 1 7. 79 pr ob > c hi 2 = 0. 00 00 ch i2 (1 ) = 1 27 3. 37 pr ob > c hi 2 = 0. 00 00 ch i2 (1 ) = 1 29 8. 39 pr ob > c hi 2 = 0. 00 00 m od ifi ed w al d te st fo r gr ou pw is e he te ro sc ed as tic ity ch i2 (2 08 ) = 2 .7 e+ 06 pr ob >c hi 2 = 0 .0 00 0 ch i2 (2 08 ) = 1 .4 e+ 06 pr ob >c hi 2 = 0 .0 00 0 ch i2 (2 08 ) = 1 .2 e+ 06 pr ob >c hi 2 = 0. 00 00 ch i2 (2 08 ) = 1 .1 e+ 06 pr ob >c hi 2 = 0. 00 00 p es ar an ’s te st o f c ro ss -s ec tio na l d ep en de nc e pr ob >c hi 2 = 0 .0 00 pr ob >c hi 2 = 0 .0 00 pr ob >c hi 2 = 0. 00 0 pr ob >c hi 2 = 0. 00 0 f ri ed m an ’s te st o f c ro ss -s ec tio na l d ep en de nc e pr ob >c hi 2 = 1 .0 00 pr ob >c hi 2 = 1. 00 0 pr ob >c hi 2 = 1. 00 0 pr ob >c hi 2 = 1. 00 0 f re es ’ t es t o f c ro ss -s ec tio na l i nd ep en de nc e pr ob >c hi 2 = 3 .9 79 pr ob >c hi 2 = 4 .9 49 pr ob >c hi 2 = 11 .1 09 pr ob >c hi 2 = 10 .6 35 16 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 ta bl e 8 li ne ar r eg re ss io n w ith p an el c or re ct ed s ta nd ar d e rr or s (p c se ) r o a t q m od el 1 m od el 2 m od el 3 m od el 4 c oe ff . pva lu e c oe ff . pva lu e c oe ff . pva lu e c oe ff . pva lu e c -0 .1 64 ** * 0. 00 0 -0 .1 54 ** * 0. 00 0 -0 .3 74 ** 0. 02 7 -0 .2 96 ** 0. 04 5 ic p -0 .0 72 ** * 0. 00 0 -0 .6 89 0. 00 0 a c p 0 .0 15 0. 46 9 -0 .1 84 ** * 0. 00 4 a pp -0 .1 16 ** * 0. 00 4 -0 .3 49 ** * 0. 00 4 c c c -0 .0 41 ** * 0. 00 0 -0 .5 07 ** * 0. 00 0 o n w c 0 .2 62 ** * 0. 00 0 0 .2 12 ** * 0. 00 0 1 .7 98 ** * 0. 00 0 1 .5 21 ** * 0. 00 0 fs 0 .0 15 ** * 0. 00 0 0 .0 15 ** * 0. 00 0 0 .0 76 ** * 0. 00 0 0 .0 77 ** * 0. 00 0 d r 0 .0 38 0. 32 6 -0 .0 26 0. 37 2 1 .4 15 ** * 0. 00 0 1 .0 67 ** * 0. 00 0 a dj us te d r 2 0. 20 3 0. 17 8 0. 12 7 0. 11 4 n ot e. * si gn ifi ca nt a t a = 0 .1 0; * *s ig ni fic an t a t a = 0 .0 5; * ** si gn ifi ca nt a t a = 0 .0 1. 17 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 table 8 presents the results of the linear regression using panel corrected standard errors method for all the four models. results reveal that both the disaggregate and aggregate measures of operating working capital were relevant and affected both the short-term as well as long-term firm performance. models 1 and 2 show that icp, app, and ccc negatively influenced roa. this means, first of all, the lower the icp, i.e., the sooner inventory was sold, the higher was the roa, i.e., the higher the firm performance. this result supported the findings in the studies by arunkumar and ramanan (2013) and makori and jagongo (2013). however, this result contradicted the findings in studies by arshad and gondal (2013), gul et al. (2013), and jakpar et al. (2017) which found a positive impact of icp on firm performance. second, the lower the app, i.e., the sooner suppliers were paid, the higher the roa, i.e., the higher the firm performance. this result supported the findings of makarani and bineshian (2013) and manzoor (2013). however, this finding contradicted the findings of studies by gul et al. (2013) and kusuma and bachtiar (2018). third, acp was not significant, indicating that the collection of receivables did not appear to influence short-term firm performance. this result is in contrast to the findings in jakpar et al. (2017). fourth, the lower the ccc, i.e., the shorter the period between a firm paying its suppliers and collecting from its customers, the higher the firm performance. this result supported the findings in mohamad and saad (2010), but contradicted the findings in jakpar et al. (2017). fifth, about nonoperating working capital, the higher the onwc, i.e., the greater a firm’s proportion of other (excluding inventory, receivables, and payables) net working capital over total assets, the higher the firm’s performance. there have been no comparable findings from previous studies regarding this issue. models 3 and 4 show that all the three disaggregate measures (icp, acp, and app) and the aggregate measure (ccc) of operating working capital were significant in influencing tq. the coefficients of icp, acp, app, and ccc were all negative, implying that the quicker conversion of inventory to sales, prompter collection of receivables, faster payment of payables, and the shorter the period between paying suppliers and collecting from customers, each of these resulted in higher long-term firm performance. the findings of the negative influence of icp and acp on firm performance were supported by the findings in bhatia and srivastava (2016). however, these findings 18 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 contradicted the findings of the study by nurein et al. (2015). the finding of the negative influence of app on firm performance was at odds with the studies by bhatia and srivastava (2016) and nurein et al. (2015), who found a positive impact on firm performance (tq). the negative influence of ccc on firm performance was supported in studies by bhatia and srivastava (2016), nurein et al. (2015) and singh et al. (2017). however, contradictory findings were obtained in a study by vural et al. (2012), which found a positive and significant impact on firm performance (tq). as with the previous models, onwc also positively influenced tq in model 3 and model 4. this variable has not been used widely in previous studies. the intent in the present study was to model the onwc as a separate variable, and not consolidate it with the other working capital accounts. this was done to see if non-operating working capital would have a different influence on firm performance, compared to operating working capital. interestingly, onwc, the measure of non-operating working capital, positively influenced firm performance, whereas measures of operating working capital negatively influenced firm performance. although less inventory, receivables (only for long-term firm performance), and even payables seemed to increase firm performance, in contrast, more other working capital accounts appeared to increase firm performance. this also holds an important implication for research into the influence of the working capital on firm performance – all working capital accounts should not be consolidated into an aggregate sum, e.g., net working capital expressed as a percentage of total assets, as this will lump all the operating working capital accounts (inventory, receivables, and payables) with the non-operating working capital accounts. had this been done, the nuanced influences of different working capital components on firm performance found here would have remained uncovered. less operating working capital (inventory, receivables, and payables), but more non-operating working capital, seemed to increase firm performance. additional results show the impact of two control variables, namely firm size and debt ratio, on firm performance. firm size has been found to be positive and significant in all four models, while debt ratio was positive and significant in the third and fourth models. these findings seemed to imply that firm size influenced both the short-term 19 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 (roa) and long-term firm performance (tq). in other words, larger firms showed better performance, whereas debt ratio only influenced long-term firm performance (tq). this meant that higher financial leverage (use of debt) increased long-term firm performance. another notable finding was the slight difference in the adjusted r-square among the four models, in particular, between model 1 and model 2, and also between model 3 and model 4. the adjusted r-square was greater for model 1 (than model 2), and model 3 (than model 4), indicating that the use of the three disaggregate measures of operating working capital was slightly better than the use of the single aggregate measure of operating working capital in explaining firm performance, both for the short-term and long-term. conclusion the aim of this study was to provide recent empirical evidence about the working capital management practices of malaysian firms and the relationship between working capital and firm performance. this study used the panel data set of non-financial firms listed on bursa malaysia. from a total of 776 non-financial manufacturing firms listed on bursa malaysia, 208 firms were selected because they had complete data, i.e., throughout the period of study from 2013 to 2017 in datastream. although a handful of studies have been published about the relationship between working capital management and firm performance in malaysia, most of these had used the ols estimation, a method in which the standard errors were likely to be incorrect. therefore, the present study has tried to estimate this relationship using the more robust four panel corrected standard error (pcse) models. this study has established that operating and non-operating working capital have different impacts on firm performance. more specifically, less operating working capital, but more non-operating working capital, increased firm performance. in relation to operating working capital, the present study has also found that the set of disaggregate measures, which were derived from the three operating working capital components, namely inventory, receivables, and payables, were slightly better than the single aggregate measure, which was derived by consolidating or netting the three operating working 20 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 capital components, in explaining firm performance. furthermore, the study found that the set of operating working capital components that influenced short-term firm performance was different from the set of operating working capital components that influenced long-term firm performance. only two operating working capital components, i.e., inventory and payables, negatively influenced short-term firm performance, but all three operating working capital components, i.e., inventory, receivables, and payables, negatively influenced long-term firm performance. all the above findings seemed to imply that receivables (or collection of receivables), unlike inventory (or sale of inventory) and payables (or payment of payables), were not critical to a firm’s short-term performance, although they were critical, along with inventory and payables, to a firm’s long-term performance. more specifically, quicker sale of inventory and prompter payment of payables would improve both short-term and long-term firm performance. in contrast, earlier collection of receivables would improve only long-term firm performance. overall, these results have reinforced the conventional ideas that quicker inventory sale and earlier receivables collection could enhance a firm’s performance. nonetheless, the results also revealed a couple of counter-intuitive ideas. first, prompter, rather than later, payment of payables would raise a firm’s performance. second, the use of more, rather than less, non-operating working capital would improve a firm’s performance. however, the findings from the present study have been limited to the firms in only three non-financial sectors, as such it is recommended that future research should explore a wider range of other sectors. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for profit sectors. references abuzayed, b. (2012). working capital management and firms’ performance in emerging markets: the case of jordan. international journal of managerial finance, 8(2), 155-179. 21 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 adam, a. m., quansah, e., & kawor, s. (2017). working capital management policies and returns of listed manufacturing firms in ghana. scientific annals of economics and business, 64(2), 255-269. adekola, a., samy, m., & knight, d. (2017). efficient working capital management as the tool for driving profitability and liquidity: a correlation analysis of nigerian companies. international journal of business and globalisation, 18(2), 251-275. afrifa, g. a., & padachi, k. (2016). working capital level influence on sme profitability. journal of small business and enterprise development, 23(1), 44-63. afza, t., & adnan, s. (2007). determinants of corporate cash holdings: a case study of pakistan. paper presented at the proceedings of singapore economic review conference (serc). akbar, a. (2014). working capital management and corporate performance: evidences from textile sector of china. european academic research, 2(9), 11440-11456. al-mawsheki, r. m. s. a., ahmad, n. b., & nordin, n. b. (2019). the effects of efficient working capital management and working capital policies on firm performance: evidence from malaysian manufacturing firms. international journal of academic research in accounting, finance management sciences, 9(3), 59-69. alarussi, a. s., & alhaderi, s. m. (2018). factors affecting profitability in malaysia. journal of economic studies, 45(3), 442-458. arshad, z., & gondal, m. y. (2013). impact of working capital management on profitability a case of the pakistan cement industry. interdisciplinary journal of contemporary research in business, 5(2), 384-390. arunkumar, o., & ramanan, t. r. (2013). working capital management and profitability: a sensitivity analysis. international journal of research and development, 2(1), 52-58. bailey, d., & katz, j. n. (2011). implementing panel corrected standard errors in r: the pcse package. journal of statistical software, 42(cs1), 1-11 barney, j. (2007). gaining and sustaining competitive advantage. new jersey 07458. new jersey: pearson international edition. beck, n., & katz, j. n. (1995). what to do (and not to do) with timeseries cross-section data. american political science review, 89(3), 634-647. bhatia, s., & srivastava, a. (2016). working capital management and firm performance in emerging economies: evidence from india. management and labour studies, 41(2), 71-87. 22 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 brainard, w. c., & tobin, j. (1968). pitfalls in financial model building. the american economic review, 58(2), 99-122. bursa malaysia. (2020). https://www.bursamalaysia.com/trade/ trading_resources/ listing_directory /main_market corsten, d., & gruen, t. w. (2004). stock-outs cause walkouts. harvard business review, 82(5), 26-28. deloof, m. (2003). does working capital management affect profitability of belgian firms? journal of business finance & accounting, 30(3-4), 573-588. dubofsky, p., & varadarajan, p. r. (1987). diversification and measures of performance: additional empirical evidence. academy of management journal, 30(3), 597-608. dwivedi, n., & jain, a. k. (2005). corporate governance and performance of indian firms: the effect of board size and ownership. employee responsibilities and rights journal, 17(3), 161-172. ganguli, s. k., & agrawal, s. (2009). ownership structure and firm performance: an empirical study on listed mid-cap indian companies. iup journal of applied finance, 15(12), 37-52. giner, b., & reverte, c. (1999). the value relevance of earnings disaggregation provided in the spanish profit and loss account. european accounting review, 8(4), 609-629. giner, b., & reverte, c. (2006). the risk-relevance of accounting data: evidence from the spanish stock market. journal of international financial management & accounting, 17(3), 175-207. gul, s., khan, m. b., raheman, s., khan, m. t., khan, m., & khan, w. (2013). working capital management and performance of sme sector. european journal of business and management, 5(1), 60-68. jakpar, s., tinggi, m., siang, t., johari, a., myint, k., & sadique, m. (2017). working capital management and profitability: evidence from manufacturing sector in malaysia. journal of business & financial affairs, 6(2), 1-9. johnson, t. h., & kaplan, r. s. (1987). relevance lost: the rise and fall of management accounting. harvard business school press. jönsson, k. (2005). cross-sectional dependency and size distortion in a small-sample homogeneous panel data unit root test. oxford bulletin of economics statistics, 67(3), 369-392. kieschnick, r., laplante, m., & moussawi, r. (2013). working capital management and shareholders’ wealth. review of finance, 17(5), 1827-1852. 23 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 kusuma, h., & bachtiar, a. d. (2018). working capital management and corporate performance: evidence from indonesia. journal of management and business administration. central europe, 26(2), 76-88. lang, l. h., stulz, r., & walkling, r. a. (1989). managerial performance, tobin’s q, and the gains from successful tender offers. journal of financial economics, 24(1), 137-154. le, b. (2019). working capital management and firm’s valuation, profitability and risk. international journal of managerial finance, 15(2), 191-204. li, m. z., oum, t. h., & zhang, y. (2004). tobin’sq and airline performances. public works management & policy, 9(1), 51-65. maines, l. a., & wahlen, j. m. (2006). the nature of accounting information reliability: inferences from archival and experimental research. accounting horizons, 20(4), 399-425. makarani, k., & bineshian, z. (2013). an empirical study on the relationship between working capital management and profitability: a case study of mehregan sangesar company. management science letters, 3(3), 771-776. makori, d. m., & jagongo, a. (2013). working capital management and firm profitability: empirical evidence from manufacturing and construction firms listed on nairobi securities exchange, kenya. international journal of accounting and taxation, 1(1), 1-14. manzoor, h. (2013). working capital management and profitability: evidence from cement sector of pakistan, listed on karachi stock exchange. journal of business administration and management sciences research, 2(10), 215-223. masadeh, r., tayeh, m., al-jarrah, i. m., & tarhini, a. (2015). accounting vs. market-based measures of firm performance related to information technology investments. international review of social sciences and humanities, 9(1), 129-145. mcguire, j. b., sundgren, a., & schneeweis, t. (1988). corporate social responsibility and firm financial performance. academy of management journal, 31(4), 854-872. mohamad, n., & saad, n. b. m. (2010). working capital management: the effect of market valuation and profitability in malaysia. international journal of business and management, 5(11), 140-147. 24 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 moundigbaye, m., rea, w. s., & reed, w. r. (2018). which panel data estimator should i use? a corrigendum and extension. economics: the open-access. open-assessment e-journal, 12(4), 1-31. ng, h., chen, y., san ong, t., & teh, b. h. (2017). the impact of working capital management on firm’s profitability: evidence from malaysian listed manufacturing firms. international journal of economics financial issues, 7(3), 662-670. nurein, s. a., din, m. s., & mohd rus, r. (2015). nonlinear relationship between investment in working capital and performance: innovative evidence from food and beverages industry. european journal of economics, finance and administrative sciences, 77, 64-72. padachi, k. (2006). trends in working capital management and its impact on firms’ performance: an analysis of mauritian small manufacturing firms. international review of business research papers, 2(2), 45-58. pwc (2018). disrupt or be disrupted: 2018 malaysia working capital study. https://www.pwc.com/my/en /assets/publications/2018/ deals/pwc-my-wcm-study-2018.pdf accessed on 22.08.2020 parks, r. w. (1967). efficient estimation of a system of regression equations when disturbances are both serially and contemporaneously correlated. journal of the american statistical association, 62(318), 500-509. rafindadi, a. a., & bello, a. (2019). is dividend payment of any influence to corporate performance in nigeria? empirical evidence from panel cointegration. international journal of economics and financial issues, 9(2), 48-58. rehman, m. u., & anjum, n. (2013). determination of the impact of working capital management on profitability: an empirical study from the cement sector in pakistan. asian economic and financial review, 3(3), 319-332. sim, s. l., ali, a., & ming, w. w. p. (2019). working capital management and firm performance: an empirical study for malaysian public listed companies in property industry. borneo journal of social science & humanities, 1(2), 1-15. singh, h. p., kumar, s., & colombage, s. (2017). working capital management and firm profitability: a meta-analysis. qualitative research in financial markets, 9(1), 34-47. 25 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 smirlock, m., gilligan, t., & marshall, w. (1984). tobin’s q and the structure-performance relationship. the american economic review, 74(5), 1051-1060. summers, b., & wilson, n. (2003). trade credit and customer relationships. managerial and decision economics, 24(6-7), 439-455. tahir, m., & anuar, m. b. a. (2015). the determinants of working capital management and firms performance of textile sector in pakistan. quality & quantity, 50(2), 605-618. truica, l. d., & trandafir, r. (2009). review on financial risk procedures for assessing companies regard to young customers. paper presented at the proceedings of fikusz’09. ullmann, a. a. (1985). data in search of a theory: a critical examination of the relationships among social performance, social disclosure, and economic performance of us firms. academy of management review, 10(3), 540-557. vătavu, s. (2014). the determinants of profitability in companies listed on the bucharest stock exchange. annals of the university of petrosani economics, 14(1), 329-338. vural, g., sökmen, a. g., & çetenak, e. h. (2012). affects of working capital management on firm’s performance: evidence from turkey. international journal of economics and financial issues, 2(4), 488-495. wahla, k. s., syed zulfiqar ali, & hussain, z. (2012). impact of ownership structure on firm performance evidence from non-financial listed companies at karachi stock exchange. international research journal of finance and economics (84), 7-13. watanabe, a., xu, y., yao, t., & yu, t. (2013). the asset growth effect: insights from international equity markets. journal of financial economics, 108(2), 529-563. wisner, j. d., & eakins, s. g. (1994). a performance assessment of the us baldrige quality award winners. international journal of quality & reliability management, 11(2), 8-25. wolfe, j., & sauaia, a. c. a. (2005). the tobin q as a business game performance indicator. simulation & gaming, 36(2), 238-249. wong, w.-c., sabki, s., regupathi, a., salim, s., & na’im, s. m. (2019). malaysian sme’s liability strucutre and its impact on profitability and growth. asian academy of management journal of accounting finance, 15(2), 77-94 26 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 1–26 woo, c. y., willard, g. e., & daellenbach, u. s. (1992). spin-off performance: a case of overstated expectations? strategic management journal, 13(6), 433-447. wuryani, e. (2015). analysis working capital on company profitability. australian journal of basic and applied sciences, 9(7), 86-88. zariyawati, m., hirnissa, m., & diana-rose, f. (2017). working capital management and firm performance of small and large firms in malaysia. journal of global business and social entrepreneurship (gbse), 3(7), 166-177. zhang, x., chen, s., & yu, s. (2017). trends in working capital management and its impact on firms’ performance: an analysis of smes. research on modern higher education, 3, 47-54. 21 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 how to cite this article: bani atta, a. a., & marzuki, a. (2021). selectivity and market timing ability of mutual fund houses in emerging countries. international journal of banking and finance, 16(1), 21-42. https://doi.org/10.32890/ijbf2021.16.1.2 selectivity and market timing ability of mutual fund houses in emerging countries 1anas ahmad bani atta & 2ainulashikin marzuki faculty of economics and muamalat universiti sains islam malaysia, malaysia 2corresponding author: ainulashikin@usim.edu.my baniata_anas@hotmail.com received: 27/7/2020 revised: 28/8/2020 accepted: 8/9/2020 published: 30/1/2021 abstract the paper investigates the selectivity and market timing ability of fund houses in emerging countries. the study uses comprehensive performance models on fund houses from four emerging countries. data span is from 2007 to 2018. findings indicate that fund managers benefit from the common facilities provided by the fund houses like market research, diversification and investment opportunity. fund houses showed good selectivity skills but poor market timing ability. the possible reason is that fund houses manage large and different types of funds. this resulted in more complex management processes and thus reduced the ability to track the fluctuations in the market. the findings are important for investors as they are able to allocate their resources more effectively to funds that are best managed by fund houses while for managers, they are able to position themselves relative to their competing peers. http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 22 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 keywords: mutual fund, fund house, selectivity ability, market timing ability, emerging countries. jel classification: g11, g14, g17, g20, g40 introduction fund houses act as financial intermediaries offering a variety of mutual funds under their common brand name and via common marketing and distribution channels (bani atta & marzuki, 2019). a fund house is a group of funds that is managed by an asset management company (amc) (iqbal, aleemi, zeeshan, & tariq, 2019). the study of mutual fund at the fund house level (instead of at individual fund) is important for several reasons (nanda, wang, & zheng, 2004). firstly, a fund house structure adds economies of scale to the distribution, servicing, and funds promotion. secondly, a fund house has more flexibility in reallocating its human resources and other capital in response to market opportunities compared with standalone funds. thirdly, the reputation of a fund house will help reassure investors about the investment managers’ selection and monitoring. the importance of fund house study is evident with increasing reports that provide classification of fund houses in order to create the most effective investment data and research for investors (examples are morningstar and barron’s). according to the reports of morningstar and barron, it is possible to use the fund house performance to provide investors a ranking of the fund houses. they began reporting on the classification of fund houses using the performance of the fund house (weighted average performance of all funds in the family) to create the most successful investment data and research for investors. the aim of these reports are, firstly, to make it easier for investors to find data about fund houses. secondly, to remove the information asymmetry and check the statistics provided by the fund houses. thirdly, provide investors more confidence in the fund house they choose (laske, 2019). globally, the value of asset under management (aum) of the mutual fund industry was estimated to be worth approximately usd 79.2 23 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 trillion in 2017. the value marked a 12 percent growth from usd 71 trillion in 2016, and it is expected to continue to grow in the future. it is forecasted that the global aum will triple its value by 2025 (fages et al., 2018). a mutual fund is an investment instrument available in many countries across the world. it is essentially an investment basket that collects funds from investors and allocates them into various securities, typically capital and money market instruments. mutual funds offer a distinctive advantage over retail investors, as it affords the individual investor the opportunity to invest in a diversified basket of securities without the burden of information collection, administration, and other costs (bani atta & marzuki, 2019). equity funds make up almost half of these mutual funds. the second and third largest are fixed income and real estate and private equity funds, each constituting respectively 17 and 12 percent of the total. the prevalence of fund houses underlies the motivation of this study. investigating the performance of member funds allows one to identify funds with good performance and then determine amongst them the star fund(s). because fund houses manage a range of funds, each with their respective strategies, measuring the overall performance of the fund house is altogether a different matter (gasper & massa, 2006). pertinent to the investment decision of the fund are the attributes of the fund house and reputation of the fund house manager. good reputation emerging from the positive performance of member funds is valuable because it signals the skills of their managers (adrianto, chen, & how, 2019). thus, the paper addresses the issue of whether fund houses can outperform market benchmarks and if they portray market timing abilities. this study seeks to contribute in the mutual fund literature by investigating the performance at the fund house level. most of the previous studies focussed on performance at the individual fund level. it is argued that members of a fund house cannot be treated like standalone funds due to the fact that most of these funds work under the management of the fund house. the performance at the fund house level is important due to most investors using a top-down approach. that means investors firstly choose the fund house and they then choose the funds they will invest in. 24 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 literature review investment in mutual funds has been rising rapidly in the developed and emerging markets. fund performance estimation is a mechanical part of investment management and will include investor input for decision-making purposes. the management seeks to exploit potential market inefficiencies with the goal of optimising returns and mitigating risks through various strategies, such as stock-picking (selectivity of securities) and market timing (price anticipation). this active strategy of management seeks to outperform the market, taking competitive positions towards a benchmark. several studies have been attempted in the past to investigate the fund’s performance, timing ability, and fund selectiveness. in the literature, extensive research has been done on this topic in the general context and in the developed financial markets. every mutual fund is managed by a management company called “fund houses”. a fund house offers different types of funds to cater to specific objectives of every investor, allowing them to diversify their investments within the same fund house. fund houses may take on different strategies to attract investment. malhotra and mcleod (1997) concluded that larger houses enjoy economies of scale and, thus, lower the expense ratio and perform better. this is because houses learn from experience and they operate more efficiently over time. research by dowen and mann (2007) concluded with the same results. several studies examined the fund houses’ behaviour and strategies (khorana & servaes, 1999; zhao, 2004; massa 2003; guedj & papastaikoudi, 2004) and several of them analysed the significance of fund house members on the mutual funds (elton, gruber, & green, 2007). khorana and servaes (1999) provided evidence that fund houses issue new funds when the possibility to generate more income is substantial. fund houses attempt to offer more choices to existing investors by launching new funds and promote their visibility by highlighting some of their existing good-performing funds. elton et al. (2007) investigated the risk effect on mutual fund investors which arises from fund houses’ membership. they studied the impact of risk related to limiting mutual fund investments to one fund house. they used monthly funds returns from 1998 until 2002, and analysed the mutual fund house’s impact on investor risk. the results revealed that 25 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 funds with the same goal are more correlated within the fund houses than between other fund houses. the increased correlation is due to the tendency of funds within a fund house to hold similar stocks and have similar exposure to total risk factors. then, they postulated that confining investment to one fund house leads to a greater total portfolio risk than diversifying across different houses. massa (2003), guedj and papastaikoudi (2004) and gasper and massa (2006) examined how houses shift performance between their funds. they showed that fund houses pass resources between member funds within the fund house to favour those funds that were likely to increase the total fund house values. massa (2003) examined how fund houses play a role in determining between-fund competition through either category proliferation of fund strategies. fund heterogeneity correlates with between-fund competition between and within houses. after examining more than 18,000 american mutual funds from 1992 to 2000, the author found that the category proliferation strategy positively correlates with fund differentiation. no relationship was found between the proliferation strategy and fund performance, indicating that a fund is independent of its within-fund house peers. fund house performance also affects their constituents. analysing us funds, guedj and papastaikoudi (2004) discovered the persistent performance of member funds within their houses. this persistent excess performance is linked to the number of funds in the fund house, which can be interpreted as a measure of autonomy that the fund house exercises in allocating resources unevenly amongst its members. this finding is congruent with the view that houses allocate resources to its members based on their performance, not needs. this result is supported by gasper and massa (2006), who investigated whether fund houses strategically transfer performance to members that are more likely to improve the fund houses’ overall return. to do this, they used a sample of usa funds from 1991 to 2001. they discovered that high-value funds, that is, funds with high fees or historically good performers, achieve their superior performance at the expense of low-value funds. these results highlighted how the fund house organisation generates distortions in delegated asset management. 26 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 clare, o’sullivan, and sherman (2014) took a sample of us and european mutual funds from 1999 to 2009 to investigate the competitive and strategic behaviours of fund house funds and to ascertain whether both factors determine risk-taking and performance persistence. they found no evidence to support a superior performance persistence of fund house funds vis-à-vis non-fund house funds. moreover, based on their historical performance, there is a significant difference between the future performance of fund house and non-fund house funds’ portfolios. there is also compelling evidence that the mid-year ranking of a fund within its own fund house and sector influences its risk-taking for the remainder of the year. fang, peress, and zheng (2014) found the strategies carried out by fund houses to coordinate their fund managers by investigating the relationship between managerial placement strategies and market efficiency. as much as 1,869 us mutual funds in the 1991-2010 decade were made as sample. the authors find that fund houses tend to assign highly-skilled managers to less efficient funds, seeing that such managers have the capability to turn the funds around. fund houses thus intervene in the managers’ duties, and these interventions have the apparent purpose of enhancing the overall value of the fund house instead of maximising investors’ investment value. cici, dahm, and kempf (2018) examined how the efficiency of trading desks operated by mutual fund houses affect portfolio performance and investment behaviour of affiliated funds in the us. the results concluded that by operating more efficient trading desks, trading costs can be reduced, and fund houses can then improve the performance of their funds significantly and enable their funds to trade more and hold less liquid portfolios. aleemi, tariq, and zeesha (2019) examined the effects of fund sizes, mainly the induction of new funds and the increase in existing funds, managed by fund houses on their aum for the mutual fund industry of pakistan. this was for the period between july 2009 and july 2016. the main findings suggested that both existing and new fund sizes have a positive and significant impact on aum. additionally, fund growth is strongly associated with fund house growth. on malaysia, bani atta and marzuki, (2020) investigated the selectivity and timing ability of fund families for the period from 2007 to 2018. they started to compare between islamic mutual funds 27 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 (imfs) and conventional mutual funds (cmfs) within the same family, and then examined the performance at the fund family level. the results indicated that the imfs exhibited some fund selection ability over cmfs. however, both types of funds displayed poor market timing ability. at a fund family level, the results showed the fund families exhibited good fund selection skills but poor market timing ability. the novel result is that the difference in performance between islamic and conventional funds shrunk compared to the results of previous studies. this was due to the common advantages offered by the families for both types of funds. other studies elaborated the behaviour of individual fund managers within fund houses. kempf and ruenzi (2007, 2008) concluded that fund managers contest with other fund managers in the same houses for better ranking. it is more serious in large houses than in the smaller ones. nevertheless, they find that teams in large houses participated in less rivalry. as a summary, the investors appeared to respond asymmetrically to fund performance. well-performing funds drew higher capital inflows as opposed to small outflows of capital in poor-performing funds. this convex relationship means that assets under the management of a fund house are supposed to be higher if it produces a one-star fund and some poorly performing funds than if it has a few average performing funds. this influence induces the fund house strategy of star fund generation. we conclude that there is a clear gap in the studies of fund performance at the fund house level. previous studies focused only on the fund level and the characteristics of fund houses, in addition to investigating the impact of these characteristics on fund performance. this study seeks to bridge this gap by providing new evidence about the performance of fund houses, whether related to fund house managers’ skills and ability, and to houses’ attributes specifically in the emerging countries (saudi arabia, malaysia, indonesia and pakistan). methodology data the main source of mutual fund data is collected from bloomberg. the sample comprises 70 houses. of this, 25, 20, 14 and 11 fund 28 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 houses are domiciled in saudi arabia, malaysia, indonesia, and pakistan, respectively. the total funds in these 70 houses equal to 503 funds. the performance of the fund house equals the weighted average of the performance of all funds in the fund house. the study period is between january 2007 and december 2018, focusing mainly on monthly returns. relevant benchmarks were also collected from bloomberg to compare the performance of the fund house under study. the ftse global islamic index is used for global islamic benchmark for all countries and the ftse all-world index is the most relevant for global funds since it covers the largest market capitalisation of global equity markets (wilson & jones, 2002). the risk-free-rate is the 3-month t-bill rate which is used as a risk-free rate in multiple studies that examine mutual funds’ performance. monthly returns are calculated as follows: (1) where, is the price of an index in period t, is the price of an index in period t-1. selectivity models the selectivity models are one of the most commonly used models for evaluating mutual fund performance. stock selection models used in this study include raw returns and excess returns as well as risk adjusted measures which are sharpe ratio, treynor ratio, jensen alpha, and carhart’s four-factor models. joo and park (2011) and adrianto et al. (2019) calculated the performance of fund houses as the average performance of all funds in the same fund house. in this study, the fund house performance will be calculated as the weighted average of all funds in the fund house using all measurements. raw returns and excess returns raw return is the return for a fund house calculated by the weighted average of raw return of all funds in the fund house. (2) where, is the raw returns for the fund house, the weight of fund i calculated by the tna of fund i divided by 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 29 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 the tna of fund house. n is the number of funds in the fund house. excess returns are measured using the following equation: (3) where, is the raw return of the fund house over the period t, is the risk-free rate of return over the period t. sharpe ratio (1966) sharpe introduced a ratio to rank mutual fund performance by deducting the risk-free rate from fund house returns, divided by the standard deviation of fund house returns. (4) where, is the mean return of the fund house over the period t, is the risk-free rate of return over the period t. is the standard deviation of a fund house mean excess return. the sharpe ratio is widely used as a measure to rank mutual fund performance, especially in recent mutual fund performance studies. treynor ratio (1965) treynor ratio is similar to the sharpe ratio, except it uses the beta as a measure of systematic risk instead of using standard deviation. the treynor ratio is calculated after subtracting the risk-free rate from the fund house return and dividing it by the beta. the beta is considered the systematic risk between the fund house and the market index. the treynor model is defined as (5) where, is the mean return of the fund house over the period t, is the risk-free rate of returns over the period t. is the beta coefficient for the fund house, estimating as follows: (6) beta is a measure of sensitivity between the market and the fund house. it is calculated by dividing the covariance between the fund house and market return divided by the variance of the market return. 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 1determinants of attitude towards zakat on employment income in nigeria: 29-48 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 29 selectivity and market timing ability of mutual fund houses in emerging countries 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡𝑡𝑡 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑡𝑡𝑡𝑡−1 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = ∑ 𝑊𝑊𝑊𝑊 𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝=1 𝑝𝑝𝑝𝑝 ∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑝𝑝𝑝𝑝 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑊𝑊𝑊𝑊𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟𝑟𝑟𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆ℎ𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡− 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 30 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 a higher beta indicates that a fund house is highly associated with the market and is playing a dominant role either with positive or negative returns. moreover, the higher the treynor ratio, the better the ranking and fund house performance; it also may indicate that such a fund house is well-diversified. single factor capm model (jensen, 1968). jensen alpha is the first risk-adjusted return measure used in this section. jensen explains how to measure risk-adjusted abnormal performance in the market by capturing the abnormal excess returns of a fund house using jensen’s alpha. (7) where, is the return on the fund house, is the risk-free rate of return, is the return on the relative market benchmark, measures the sensitivity between the excess return of the market benchmark with the fund house, captures any excess returns above market benchmark, while is the term error. the jensen alpha measures the fund house’s over or underperformance; if positive and significant, then the fund house is over-performing and it indicates that managers earned extra returns on the fund house due to stock selection ability. carhart four-factor model (1997) carhart expanded the fama french three-factor model, taking into consideration momentum factor in addition to size and value factors. the carhart model is defined as follows: (8) where, is the mean return of the fund house over the period t, is the risk-free rate of return over the period t, is the return on the relative market benchmark. is the difference in return between a small-cap portfolio and a large-cap portfolio at time t, while is the difference in return between a portfolio of high-bookto-market stock and a low-book-to market stock at time t. measures the sensitivity between the market and the fund house. if it is positive and significant, then the fund house is highly associated with market 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 31 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 movements, is a coefficient that measures the fund house exposure, if seen positive and significant then the fund house is associated with small-capitalisation stocks. a coefficient that measures the fund house exposure and if positive and significant, then the fund house is exposed to high-book-to-market stocks, measures the selectivity ability where if positive and significant, then the fund house has superior stock selection ability, is the different in return between high minus low momentum (prior one year return) at time t. is the coefficient that measures the fund house’s exposure and if positive and significant, then the fund house is exposed to high momentum. market timing models market timing is measured using the treynor and mazuy (tm) (1966) and henriksson and merton (hm) (1981) approaches. the aim is to investigate whether fund house managers exhibit market timing ability. market timing models identify fund house managers’ ability to develop timing strategies to shift capital between safe and risky securities based on whether the market is expected to do well or bad. over-performing fund houses are able to forecast entry and exit strategies in the market for their funds. treynor & mazuy (tm) model (1966) treynor and mazuy (1966) built a model that recognises good market timing of fund houses. the market timing is cached by the square of market returns. the model is as follows: (9) where, is the mean return of the fund house over the period t, is the risk-free rate of returns over the period t, is the return on the relative market benchmark, measures selectivity ability, is the squared market return, and indicates market timing where if positive and significant, then the fund houses are successful and exposure to the market is increased when markets are doing well. henriksson & merton (hm) model (1981) (10) 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 2 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝜎𝜎𝜎𝜎𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑇𝑇𝑇𝑇𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡−𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 (5) 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝐵𝐵𝐵𝐵𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 (𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚) 𝑉𝑉𝑉𝑉𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑟𝑟𝑟𝑟𝑚𝑚𝑚𝑚 𝐽𝐽𝐽𝐽𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽�𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝜀𝜀𝜀𝜀𝑝𝑝𝑝𝑝,𝑡𝑡𝑡𝑡 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 3determinants of attitude towards zakat on employment income in nigeria: 29-48 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽1�𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡� + 𝛽𝛽𝛽𝛽2𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽3𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 + 𝛽𝛽𝛽𝛽4𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵𝐵𝐵𝑡𝑡𝑡𝑡 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽1 𝛽𝛽𝛽𝛽2 𝛽𝛽𝛽𝛽3 𝑆𝑆𝑆𝑆𝑀𝑀𝑀𝑀𝑆𝑆𝑆𝑆𝑡𝑡𝑡𝑡 𝛽𝛽𝛽𝛽4 3.3 market timing models 𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 )2 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 4 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓)𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝,t 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡2 𝛾𝛾𝛾𝛾𝑝𝑝𝑝𝑝,t 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡2 32 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 where is the mean return of the fund house over the period t, is the risk-free rate of return over the period t, is the return on the relative market benchmark, measures selectivity ability, is the market timing coefficient, is a dummy variable that takes a value of one if the market return is positive and zero otherwise, and is the error term. analysis of results this section provides the results for the analysis of fund houses’ performance which include fund house managers’ selectivity and timing ability performance. descriptive statistics table 1 presents the descriptive statistics of monthly returns of the fund house, market benchmarks, and other risk factors from 2007 to 2018. as shown in table 1, the empirical characteristics of negative skewness, excess kurtosis, and non-normality in most portfolio returns are the dominant features of the data. the mean of fund houses’ returns is positive and equal 0.0920. while the mean returns for islamic and conventional benchmarks are negative and equal -0.0615 and -0.0061, respectively. table 1 descriptive statistics fund house ftse islamic ftse all world smb hml mom tb mean 0.092 -0.062 -0.006 -0.080 0.046 -0.019 0.065 med. 0.132 -0.063 -0.003 0.132 -0.156 0.209 0.063 max. 0.693 0.458 0.117 0.184 0.429 0.209 0.139 min. -0.583 -0.454 -0.235 -0.406 -0.156 -0.925 0.018 sd. 0.037 0.057 0.047 0.281 0.267 0.519 0.030 skew -1.158 -1.982 -0.657 -0.219 0.669 -0.645 0.521 kurt. 25.763 24.796 1.863 -1.881 -1.483 -1.520 1.489 4 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓)𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 4 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓)𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 4 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓)𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 4 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓)𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 4 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓)𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 4 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓)𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 4 the international journal of banking and finance, vol. 13. number 1, 2017: 29-48 𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆 = 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 = 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 + 𝛽𝛽𝛽𝛽𝑝𝑝𝑝𝑝𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 + 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝(𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓 − 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓)𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 + 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓ℎ 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓ℎ,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝑅𝑅𝑅𝑅𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 𝛼𝛼𝛼𝛼𝑝𝑝𝑝𝑝 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝑡𝑡𝑡𝑡 𝜀𝜀𝜀𝜀𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓,𝑡𝑡𝑡𝑡 33 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 table 2 displays the pearson correlation coefficients of the fund house, and the risk factors, and also below the correlation coefficients are p-value for tests of significance of coefficients. this is to check if the problem of multicollinearity exists. fund house returns indicate a low correlation with returns of both markets’ benchmarks of ftse global islamic and ftse all-world. similarly, the market indices show low correlation with each other, with significantly positive correlation coefficients. none of the variables or independent variables are highly correlated. the highest reported figure is 30.46 percent, the correlation between ftse islamic returns and treasury bill rate. hence, the estimation is less likely to suffer the multicollinearity problem. table 2 correlation matrix fund house ftse (i) ftse (aw) smb hml mom tb fund house 1.000 ftse (i) -.030.002 1.000 ftse (aw) -.083 .000 .085 .000 1.000 smb .002.041 .001 .055 .149 .000 1.000 hml .007.047 .051 .000 .128 .000 .064 .000 1.000 mom -.010.029 -.068 .000 -.161 .000 -.055 .000 -.418 .000 1.000 tb .044.000 -.305 .000 -.109 .000 -.007 .439 -.025 .011 .015 .138 1.000 selectivity skills models this section reports the finding of the empirical analysis of six performance measures to gauge the performance of the fund house comparative to benchmarks. the performance measures used are: raw returns, excess returns, the sharpe ratio, the treynor ratio, the one factor model (jensen’s alpha), and the carhart’s four-factor model. 34 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 raw return, excess return, the sharpe ratio, and the treynor ratio table 3 presents monthly means for raw returns and excess returns, the sharpe ratio and treynor ratio for the fund house. the fund houses’ performance are compared using two market indices. the mean raw returns for fund houses are 0.07 percent per month, and for islamic and conventional benchmarks are 0.003 and 0.002 percent per month, respectively. however, after taking into account the risk-free rate, the mean excess returns of fund houses remain positive at 0.009 percent per month, but the mean excess returns for both market benchmarks which are islamic and conventional are negative at -0.061 and -0.082, respectively. both the mean returns and mean excess returns for fund houses are higher than both the market benchmarks. although the fund houses’ returns exceed the returns of the market benchmarks, the returns are less volatile. in addition, the beta of fund houses is lower than that of market beta (1.000). this gives an initial indicator of the attractiveness of fund houses that yield higher returns but lower risks, both total and systematic. this could be due to the diversification which are provided by the houses. the sharpe ratio represents the calculation of mean excess returns relative to the total risk indicated by the standard deviation. this measure gives the unit of return earned while taking an additional unit of total risk. the results indicate that investment in fund houses earn a sharpe ratio of 0.39 percent per month, but the sharpe ratio for both market benchmarks is negative -1.08 and -1.34, respectively. that means the investment in fund houses is better than investment in both market benchmarks islamic and conventional. the treynor ratio refers to the calculation of mean excess returns relative to the systemic risk posed by beta. the treynor ratio shows positive results using both market benchmarks. that means the fund houses show better performance and are well diversified. the findings of these essential performance models are used to conclude that the fund house outperforms both the market benchmarks when relative performance measures are used. this is justified by the increased diversification provided by houses which in turn offers advantages of improving the overall performance of fund houses. this corresponds to the modern portfolio theory (mpt) which suggests that the risk-reduction advantages associated with maintaining a diversified portfolio of assets are that it maximises the anticipated 35 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 return based on a given degree of market risk. although the relative performance measures are commonly used in practice, these measures can only be used in ranking funds in relation to their peers in that they provide no fundamental statistical or economic meaning. therefore, the next sections provide the results of the single and multi-factor models. these performance models, based on the capm, give an estimate of the intercept (alpha), which refers to fund performance in relation to the benchmark return. one factor model (jensen, 1968) table 4 provides the results for the analysis of performance employing the one-factor model (jensen, 1968) using both market benchmarks, ftse global islamic and ftse all-world. alpha indicates the monthly abnormal returns of the fund houses. as shown, the alpha estimate for fund houses is positive and equals 0.19 percent per month for ftse global islamic and 0.20 percent per month for ftse allworld. that indicates that the fund houses outperformed both market benchmarks. this result is also in line with the results of the traditional measures in the previous section. table 3 mean raw returns, mean excess returns, sharpe ratio, and treynor ratios fund house ftse islamic ftse all world panel a: mean raw, mean excess return, and sharpe ratio mean raw returns 0.0740 0.0032 0.0023 mean excess return 0.0093 -0.0615 -0.0824 std. dev 0.0317 0.0569 0.0465 sharpe ratio 0.3936 -1.0812 -1.3424 panel b: beta and treynor ratio using ftse islamic as benchmark beta 0.1307 1.0000 -----treynor 0.0711 ----------panel c: beta and treynor ratio using ftse all world as benchmark beta 0.1166 -----1.0000 treynor 0.0797 ----------36 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 the adjusted r2 for fund houses are 81 percent and 85 percent for ftse global islamic and ftse all-world, respectively. the high adjusted r 2 also indicates that the fund managers pursue a passive approach by watching the market closely, but are unable to perform well. the results also show the alpha using a conventional benchmark is better than the alpha using the islamic benchmark, that due to some restrictions are placed on islamic investments that may limit performance. table 4 one factor model (jensen, 1968) ftse global islamic ftse all world alpha beta adj. r2 alpha beta adj. r2 coeff 0.1942 -0.2435 0.81 0.2025 -0.1067 0.85 std.err 0.0072 0.0801 ----0.0077 0.0916 ----p-value 0.0002 0.0023 ----0.0001 0.2441 ----source: jensen (1968). four-factor model (carhart, 1997) for the four-factor model, the factor returns for the four-factor models are not easily available, hence the researcher used the data made available at the fama and french websites, which in turn employs the ftse all-world database to construct monthly for the four-factor model. table v provides the results of the four-factor model for fund houses using ftse all-world as a market benchmark. as shown in table 5, the four-factor alpha for fund houses was 0.20 percent and significant, which indicates that on average, the fund houses are able to outperform the four-factor benchmarks. in addition, the results indicate that the fund houses exhibit lower risks with beta of -0.14. this result is similar to the results using the one-factor model. in terms of size preference, fund houses prefer small stocks than big stocks, given the smb factor loading is -0.03 and statistically significant. the hml factor for fund houses is -0.05 and statistically significant, suggesting a preference for growth-to-value stock. the 37 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 mom factor for fund houses is not significant. the fund houses display a relative preference for small-cap and growth-oriented stock. in conclusion, the preference of fund houses for smaller cap and lower beta results in the significant out-performance of fund houses over the four-factor benchmarks. fund houses also display a preference for the growth-to-value stock. finally, the mom factor for fund houses is not significant. fund houses’ superior performance can be attributed to that the fund houses remove unsystematic risks by diversification, which means the houses make funds work within these houses comprising only market risks (systematic risks). this should make for a stronger relationship, if there is one, in any subsequent time-series regression, especially if there is time-series autocorrelation. the constant or error term in the subsequent regression helps to remove any remaining unsystematic risk. table 5 carhart’s four factor model coef std.err p-value alpha 0.2011 -0.0079 0.0001 market -0.1400 0.0974 0.0505 smb -0.0336 0.0023 0.0488 hml -0.0538 0.0022 0.0015 mom -0.0022 0.0017 0.2042 adj. r2 ----0.88 ---- market timing models table 6 presents the results for the analysis of security selection and market timing ability using the treynor-mazuy model (tm) and hendrickson-merton model (hm) using ordinary least square (ols), for fund houses. in the panel “a”, according to the treynor-mazuy model (tm), fund houses have good selectivity skills but appear to have poor market timing ability irrespective of the benchmarks using 38 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 both market benchmarks “ftse global islamic and ftse all-world”. alpha is positive and equals to 0.195 and 0.193, respectively, and gamma is negative and equals to -0.12 and -0.34, respectively. in the panel “b”, according to the hendrickson-merton model (hm), fund houses have good selectivity skills but appear poor in market timing ability irrespective of the benchmarks using both market benchmarks “ftse global islamic and ftse all-world”. alpha is positive and equals to 0.19 and 0.20, respectively, and gamma is negative and equals to -0.08 and -0.30, respectively. table 6 market timing models: treynor-mazuy model and hendricksonmerton model ftse global islamic ftse allworld alpha gamma\ delta adj. r2 alpha gamma\ delta adj. r2 panel a: market timing measure; treynor-mazuy model coeff 0.1953 -0.1292 0.74 0.1933 -0.3466 0.92 std.err 0.0075 0.2547 ----0.0082 0.0870 ----p-value 0.1914 0.6118 ----0.5611 0.0006 ----panel b: market timing measure; hendrickson-merton model coeff 0.1935 -0.0807 0.76 0.1986 -0.3068 0.79 std.err 0.0072 0.1190 ----0.0079 0.1609 ----p-value 0.0002 0.4980 ----0.0001 0.0311 ----overall, both market timing ability measures (treynor-mazuy model (tm) and hendrickson-merton model (hm)) provide similar results, where there is strong evidence that fund house managers have good selectivity skills and this results in supporting results of the one-factor and four-factor models. this is due to the benefits provided by the advantages of fund houses like diversification and more investment opportunities. however, fund houses have weak market timing ability. the possible reason is that the fund houses contain large and different types of funds, and thus the management process becomes more difficult. this may reduce the ability and efficiency of managers. 39 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 conclusion this study seeks to contribute by providing new evidence about the mutual fund performance at the fund house level. firstly, the study concluded that managers benefit from the advantages provided by the fund house like diversification and more investment opportunity. so, fund house managers show good selectivity skills. at the same time, fund house managers show poor market timing ability. the possible reason is that the fund houses contain large and different types of funds, and thus the management process becomes more difficult. this may reduce the ability and efficiency of managers to track the fluctuations and constant movements in the market. the results are useful for both investors and managers. managers should take the requisite decision or changes to make themselves more efficient in comparison with their competing colleagues. the investors can more effectively allocate their money to better controlled funds. in addition, the results help investors make the correct investment decision, since most of the investors use the top-down approach. the results are also important to academics and regulators because they provide an overview of the mutual fund industry generally, and fund houses specifically. from the results of this study, there are two recommendations that must be considered. firstly, due to limited evidence about performance at the fund house level, it is important to increase the focus upon the fund house level because the advantages of fund houses may lead to improved performance as the results showed. it is then highly recommended to extend the focus to characteristics of these houses and their effects on houses and funds’ performance. secondly, since most of the previous studies focussed on developed countries like the usa and uk, then it is highly recommended to academics and researchers to extend such studies to other emerging countries like the middle-east and south-asia countries. acknowledgment this research received no specific grant from any funding agency. 40 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 references adrianto, f., chen, e., & how, j. c. y. (2019). spillover effects in sri fund families. https://ssrn.com abstract=3311876.doihttp:// dx.doi.org/10.2139/ssrn.3311879. aleemi, a. r., tariq, m., & zeeshan, m. (2019). funds induction, flows and family growth: the dynamics of asset management companies. forthcoming in pakistan business review, 20(4). 926-940. bani atta, a. a. b., & marzuki, a. (2019). the impact of funds and fund family characteristics on fund performance: evidence from malaysia. journal of wealth management & financial planning, 6(june), 3–23. bani atta, a. a. b., & marzuki, a. (2019). the determinants of islamic mutual fund flows: evidence from malaysia. international journal of advanced research in economics and finance, 1(1), 10–21. bani atta, a. a. b., & marzuki, a. (2020). islamic versus conventional funds within the family: selectivity skills and market timing ability. journal of islamic monetary economics and finance, 6(2), 1-24. carhart, m. m. (1997). on persistence in mutual fund performance. the journal of finance, 52(1), 57–82. cici, g., dahm, l. k., & kempf, a. (2018). trading efficiency of fund families: impact on fund performance and investment behaviour. journal of banking and finance, 88, 1–14. clare, a., o’sullivan, n., & sherman, m. (2014). family status and mutual fund performance. journal of asset management, 15(3), 163–175. dowen, r. j., & mann, t. (2007). do investors benefit from 12b-1 fees? american journal of business, 22(1), 21–30. elton, e. j., gruber, m. j., & green, t. c. (2007). the impact of mutual fund family membership on investor risk. journal of financial and quantitative analysis, 42(2), 257–277. fages, r., beardsley, b., brömstrup, i., donnadieu, h., macé, b., pardasani. (2018). global asset management 2018: the digital metamorphosis. fang, l. h., peress, j., & zheng, l. (2014). does media coverage of stocks affect mutual funds’ trading and performance? review of financial studies, 27(12), 3441–3466. 41 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 gaspar, j., & massa, m. (2006). favouritism in mutual fund families? evidence on strategic cross-fund subsidisation. the journal of finance, 61(1), 73-104. guedj, i., & papastaikoudi, j. (2003). can mutual fund families affect the performance of their funds? https://ssrn.com/ abstract=467282. henriksson, r. d., & merton, r. c. (1981). on market timing and investment performance. ii. statistical procedures for evaluating forecasting skills. journal of business, 54(4), 513–533. doi: 10.1086/296144. jensen, m. c. (1968). the performance of mutual funds in the period 1945–1964. the journal of finance, 23(2), 389–416. joo, h. k., & park, y. k. (2011). contribution of star funds to fund families: an empirical analysis of the korean fund market. asia-pacific journal of financial studies, 40(5), 731–762. kempf, a., & ruenzi, s. (2007). tournaments in mutual-fund families. review of financial studies, 21(2), 1013–1036. kempf, a., & ruenzi, s. (2008). family matters: rankings within fund families and fund inflows. journal of business finance and accounting, 35(1–2), 177–199. khorana, a., & servaes, h. (1999). the determinants of mutual fund starts. the review of financial studies, 12(5), 1043–1074. laske, m. (2019). morningstar fund family 150 research on the 150 largest u.s. fund families. morningstar research services report. https:// morning star direct. morningstar. com/ clientcomm/due diligencereports/fundfamily150.pdf. malhotra, d. k., & mcleod, r. w. (1997). an empirical analysis of mutual fund expenses. journal of financial research, 20(2), 175–190. massa, m. (2003). how do family strategies affect fund performance? when performance-maximisation is not the only game in town. journal of financial economics, 67(2), 249–304. nanda, v., wang, z. j., & zheng, l. (2004). family values and the star phenomenon: strategies of mutual fund families. the review of financial studies, 17(3), 667-698. sharpe, w. f. (1966). mutual fund performance. the journal of business, 39(1), 119–138. treynor & mazuy. (1966). can mutual funds outguess the market? harvard business review, 44(4), 131–136. 42 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 21–42 treynor, j. l. (1965). how to rate management of investment funds. harvard business review, 43(1), 63–75. wilson, j. w., & jones, c. p. (2002). an analysis of the s & p 500 index and cowles’s extensions: price indexes and stock returns. 1870_1999. the journal of business, 75(3), 505–533. zhao, x. (2004). why are some mutual funds closed to new investors? journal of banking and finance, 28(8), 1867–1887. 81 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: chatzinas, g., & papadopoulos s. (2022). determinants of the capital structure of european firms: the moderating role of efsf/esm financial assistance programs. international journal of banking and finance, 17(1), 81-108.https://doi. org/10.32890/ ijbf2022.17.1.4 determinants of the capital structure of european firms: the moderating role of efsf/esm financial assistance programs 1georgios chatzinas & 2simeon papadopoulos department of accounting and finance university of macedonia, thessaloniki, greece 1corresponding author: giorgosxatz@yahoo.com received: 22/3/2021 revised: 8/5/2021 accepted: 8/5/2021 published: 2/1/2021 abstract the present study has investigated the moderating effect of the european financial stability facility (efsf) / european stability mechanism (esm) support to the firms’ indebtness. using dynamic panel data, three models were estimated and aimed at the determination of the way that efsf/esm financial assistance programs could influence the impact of five firm-specific characteristics, namely growth, profitability, size, tangibility and non-debt tax shield on the capital structure of european firms. data from 2,086 firms for the period 2003 – 2016 were used, and two dummy variables; one for the efsf/esm support period and one for any kind of economic crisis period were formed. the results indicated that pecking order prevailed over trade-off theory. economic crises did not affect severely the firm-characteristics’ effects, but the efsf/esm programs influence 82 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 appeared in three cases. during the period of efsf/esm assistance, profitability’s negative effect on long-term debt ratio disappeared and on total debt ratio strengthened, growth’s positive impact on total debt ratio diminished and non-debt tax shield acquired positive influence on total debt ratio. these changes might be explained by the increased levels of tax rates and decreased levels of uncertainty that the efsf/ esm programs caused, as well as by the reluctance of lenders to provide new funds. keywords: capital structure, trade-off, pecking order, efsf/esm financial programs, eurozone. jel classification: g32, g38. introduction the capital structure of firms is affected by economic conditions as indicated by the majority of recent researches that investigated the impact of the international financial crisis of 2007-2009 on the financial leverage of firms (harisson & widjaja, 2014; demirguckunt et al., 2015; banerjee, 2017; chatzinas & papadopoulos, 2018). in the european union, however, the international financial crisis was converted to debt crisis. dealing with the consequences of the crisis led to the establishment of the european financial stability facility (efsf) and its successor, the european stability mechanism (esm). these two mechanisms were aimed at providing financial support to country-members which were facing high fiscal deficits and high debt ratios. the support was accompanied by strict and painful austerity measures. so, the efsf/esm were understood as having resulted in a decrease of the country’s economic uncertainty due to the fact that the facility guaranteed the country’s financial needs and an increase in the tax burden in the context of a stricter fiscal policy. however, at least according to the present available knowledge, previous studies which examined the firms’ capital structure of the countries supported by these facilities, were in the same context with the rest of the firms which had been ignoring a possible facilities’ impact. to address the above problem, the present study has focused on the impact of the efsf/esm support on the effect that firm-specific 83 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 characteristics have on firms’ capital structure. to the best of our knowledge, this study is the first of its kind. more specifically, the present study has compared the effect of five firm-specific factors, namely growth, profitability, size, tangibility, non-debt tax shield that have driven capital structure between normal economic conditions and the economic conditions that were formed by the efsf/esm financial assistance programs. the determination of these effects should reveal useful information on the way that the efsf/esm changes a firm’s procedure on financing decisions, as well as whether the trade-off or pecking order theory drives a firm’s capital structure. the sample consisted of listed, non-financial and non-utilities firms operating in ten countries, members of the eurozone (austria, cyprus, estonia, france, germany, greece, ireland, italy, portugal and spain) for the period from 2003 to 2016. the results were quite interesting since they supported the view that a firm’s capital structure behavior was interpreted mostly by the pecking order, regardless of the occurrence of an economic crisis. conversely, the efsf/esm financial assistance programs seemed to change the magnitude of the effect of firmspecific capital structure determinants in the following three cases: profitability, growth and non-debt tax shield (ndts). the remainder of the paper is organized as follows: in section ii, the trade-off and pecking order theories will be briefly, but comprehensively presented. in line with the predictions of these theories, the research hypotheses of the present study were developed. the research methods (sample, data, variables, and statistical techniques) will be described in section iii, while section iv presents the results of statistical analysis that will be discussed in section v. finally, section vi sums up the conclusions and recommendations for future research. the section also includes a discussion of the limitations of the research. literature review researchers and financial professionals have focused on a firm’s capital structure for at least 60 years. the impetus for the formation of modern capital structure theories is to be found in the theorems of modigliani and miller (1958; 1963). based on these, various theories of capital structure have been proposed. the trade-off and pecking order theory were the two main and competing ones (mihalca & antal, 84 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 2009). the trade-off theory supported the existence of an optimal capital structure that minimized the cost of capital and maximized a firm’s value (smart et al., 2004). particularly, the value of a firm depended positively on the amount of its debt due to the tax savings it created (modigliani & miller, 1963). however, the level of debt tax saving would be affected by the tax rates of individuals (miller, 1977), while it decreased in case the company already had levels of high nondebt tax shield (ndts), such as depreciations (deangelo & masulis, 1980). in addition, positive dependence was justified by the mitigation of costs arising from owners-managers conflicts (jensen & meckling, 1976), while negative dependence might be explained by an increase of financial distress costs (wruck, 1990) and owners-lenders conflicts (jensen & meckling, 1976). on the other hand, no optimal capital structure would exist, according to the pecking order theory, which stated that a firm’s capital structure was the outcome of managers’ past financing decisions (shyam-sunder & myers, 1999). managers, in particular, preferred internal funds to external funds based on capital source costs, and when external funds were needed, managers tended to prefer debt to equity (grinblatt & titman, 2002). besides cost, the pecking order hypothesis seemed to face a firm’s asymmetry information problems (sergiescu & vaidean, 2014). many researchers have attempted to determine which of the two theories has been more explanatory and has a higher interpretation power regarding a firm’s capital structure behavior. the most recent researches have implied that the two theories were not mutually exclusive, but complemented each other (chatzinas & papadopoulos, 2018; banga & gupta, 2017; mc namara et al., 2017; ohman & yazdanfar, 2017; serrasqueiro & caetano, 2015; burgstaller & wagner, 2015). the international financial crisis gave a new impetus to the investigation of capital structure theories, since scholars had the opportunity to investigate how economic crises affected the claims of the theories (harisson & widjaja, 2014). thus, after 2007, many studies focused on the impact of a financial crisis on a firm’s indebtness (harisson & widjaja, 2014; banerjee, 2017; chatzinas & papadopoulos, 2018). demirguc-kunt et al. (2020) provided evidence that a financial crisis negatively affected debt on a firm’s balance sheets even in economies that did not suffer from the consequences of the international financial crisis. this impact was higher for small and medium enterprises (smes) and large non-listed companies. in 85 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 the same context, d’amato (2020) drew the same conclusion, but he claimed that short-term debt was affected more severely. on the other hand, migliori et al. (2018) did not find evidence that the international financial crisis influenced the way managers made their financing decisions. moreover, they concluded that neither capital structure theory had been able to interpret these decisions, regardless of the economic conditions. similar conclusions were drawn by many other researches (adair & adaskou, 2017; zhang & mirza, 2015; danso & adomako, 2014). in the european union, the global financial crisis was transformed into a debt crisis that threatened the eurozone’s stability. to face these consequences, four economies, namely cyprus, ireland, greece, and portugal became financially supported by programs developed, supported and monitored by the european financial stability fund (efsf) and the european stability mechanism (esm). in all four cases, the financial support was accompanied by strict austerity measures and structural reforms (singala & kumar, 2012). in the pre-crisis period, each country’s firm capital structure had been investigated. in the case of cypriot firms, the limited number of studies published, agreed that the trade-off theory was prevailing (mokhova & zinecker, 2013; machielsen, 2012). on the other hand, greek (noulas & genimakis, 2011; daskalakis & psillaki, 2008, 2009; eriotis et al., 2007) and portuguese (nunes & serrasqueiro, 2017; pacheco, 2016) financing decisions of firms seemed to be mostly driven by the pecking order theory. finally, there had been ambiguous evidence on irish firm capital structure behavior, since some researchers had concluded that the trade-off theory was prevailing, while others disagreed with this conclusion (mac an bhaird & lucey, 2010; bancel & mittoo, 2004). as mentioned above, the international financial crisis gave a new impetus to the investigation of capital structure theories. many researchers tried to approach the factors that could affect a firm’s financing decisions and the way a firm change during a crisis (harisson & widjaja, 2014; banerjee, 2017; chatzinas & papadopoulos, 2018; demirguc-kunt et al., 2020; d’amato, 2020; migliori et al., 2018; adair & adaskou, 2017; zhang & mirza, 2015; danso & adomako, 2014). however, the case of a firm’s financing decision in cyprus, greece, ireland and portugal would be different because of the interacting effect that the efsf/esm financial supporting programs might have had. 86 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 as a result, the aim of the present study has been two-fold. first, the determinants of capital structure and their direct effect were investigated. second, the moderating effect that the efsf/esm programs might impose on the above direct effects was also investigated. more specifically, the identified five firm-specific characteristics were used as determinants of capital structure, while two country-specific characteristics were used as control variables. the former were, on the one hand, growth, size, profitability, tangibility and ndts. the latter control variables were, on the other hand, tax rates and country risk. the effect of each of these firm-characteristics, as well as the way that the efsf/esm programs were anticipated to influence the variables will be discussed below. according to the trade-off theory, profitability, size, tangibility and tax rates will be positively correlated to debt level, while growth, ndts and country risk will be negatively correlated. higher profitability leads to higher taxes, thus motivating firms to increase their debt (ebrahim et al., 2014; rahim et al., 2020). similarly, larger firms and firms with more tangible assets would face a lower level of financial distress risk due to the higher level of diversification and lower cost of information asymmetry, which in turn, would encourage them to expand their debt levels (chen, 2004; cortez & susanto, 2012; chin & zakaria, 2018). finally, as a result of the positive relationship between tax rates and debt tax shield, there would be a positive relationship between tax rates and financial leverage (modigliani & miller, 1963). on the other hand, growth opportunities and country risk might be interpreted as in a higher level of anticipated financial distress cost (stulz, 1990; demirguc-kunt et al., 2015), as well as leading to a decrease in a firm’s financial leverage. on the other hand, according to the pecking order assumption, growth, tangibility, tax rates and country risk were positively correlated to debt level, while profitability and size were negatively correlated. moreover, it is worth mentioning that the ndts has no anticipated impact on financial leverage (noulas & genimakis, 2011). higher growth might be interpreted as higher anticipated future profitability that would motivate firms to increase the present levels of debt (frank & goyal, 2003). similarly, tangibility would reduce the cost of informational asymmetry and assure the lenders in case of a bankruptcy, allowing debt to increase (ebrahim et al., 2014). finally, an increase in tax 87 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 rates and in country risk could deteriorate the internally generated cash flows and lead to an increase in financial leverage (cetorelli & goldberg, 2011). on the other hand, profitability and size, due to the higher levels of cash flows, would decrease the need for external funds, thus reducing a firm’s debt (jong et al., 2008; myers, 2001). table 1 summarizes the above direct effects of the examined firmand country-specific characteristics on a firm’s leverage according to the trade-off and pecking order theories. table 1 synopsis of the direct effect of firmand country-specific characteristics on a firm’s capital structure according to the tradeoff and pecking order theories trade-off pecking order growth + profitability + size + tangibility + + non-debt tax shield none tax rates + + country’s risk + the objective of an esm loan is “to assist esm members in significant need of financing and which have lost access to the market, either because they cannot find lenders or because the financing costs would adversely impact the sustainability of public finances”. as a result, a country that enters an esm financial assistance program is a country that will face significant barriers in raising capital from the markets. in entering an esm program, the country’s financial needs are immunized for a long-term period at the expense of strict austerity measures and structural reforms. as a result, eu-members accepting an esm financial assistance program face two direct economic condition changes. the first one is a decrease of uncertainty; country risk decreases, since the esm ensures the country’s financial needs. the second one is an increase in tax burden, in the context of a stricter fiscal policy. the resultant interacting/moderating effect of the efsf/esm programs will have a direct impact on the firm-specific 88 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 determinants of capital structure and will be further discussed below. according to the trade-off theory, growth expresses anticipated levels of financial distress risk. as a result, the decreased levels of uncertainty are expected to mediate growth’s impact. on the other hand, the bigger the size and the higher the tangibility of firms are, the lower the levels of financial distress risk will be. so, a positive effect of the esm programs on these determinants’ effect on leverage is expected, due to the decreased level of uncertainty, too. simultaneously, the increase in tax rates increases the non-debt tax shield, resulting in the amplification of its effect, while the firms are expected to take advantage of the new tax conditions, and intensifying the positive relationship between profitability and tax rates. consequently, in accordance with the trade-off theory, the following hypotheses were proposed for the present study: to-rh1:the negative effect of growth on a firm’s financial leverage decreases due to the lower level of uncertainty (alternatively, a positive change in the negative effect of growth on a firm’s financial leverage is anticipated). to-rh2:the positive effect of profitability on a firm’s financial leverage increases due to the higher levels of tax rates (alternatively, a positive change in the positive effect of profitability on a firm’s financial leverage is anticipated). to-rh3:the positive effect of size on a firm’s leverage increases due to the lower levels of uncertainty (alternatively, a positive change in the positive effect of size on a firm’s financial leverage is anticipated). to-rh4: the positive effect of tangibility on a firm’s leverage increases due to the lower levels of uncertainty (alternatively, a positive change in the positive effect of tangibility on a firm’s financial leverage is anticipated). to-rh5:the negative effect of ndts on a firm’s financial leverage increases due to the higher levels of tax rates (alternatively, a negative change in the negative effect of ndts on a firm’s financial leverage is anticipated). 89 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 on the other hand, the pecking order theory postulated that growth has the potential for higher future profitability. the decrease in the level of uncertainty increases the probability of higher future growth, resulting in an amplification of growth’s direct impact. moreover, the lower level of uncertainty negatively influences the level of informational asymmetry. so, tangibility’s direct impact on a firm’s financial leverage is anticipated to increase. profitability and size reduce the external financial needs of the firms. however, the increase of tax levels is anticipated to increase the need for external financing, intensifying the impact of these two firm-specific factors. finally, no impact of the ndts on a firm’s leverage is anticipated in normal economic conditions. consequently, in accordance with the pecking order hypothesis, the following hypotheses have been proposed for the present study: po-rh1:the positive effect of growth on a firm’s financial leverage increases due to the lower level of uncertainty (alternatively, a positive change in the positive effect of growth on a firm’s financial leverage is anticipated). po-rh2:the negative effect of profitability on a firm’s financial leverage increases due to the higher level of tax rates (alternatively, a negative change in the negative effect of profitability on a firm’s financial leverage is anticipated). po-rh3: the negative effect of size on a firm’s leverage increases due to the lower levels of uncertainty (alternatively, a negative change in the negative effect of size on a firm’s financial leverage is anticipated). po-rh4:the positive effect of tangibility on a firm’s leverage increases due to the lower levels of uncertainty (alternatively, a positive change in the positive effect of tangibility on a firm’s financial leverage is anticipated). table 2 summarizes the above changes of direct effects of the examined firm -specific characteristics under the efsf/esm financial assistance programs, on a firm’s leverage according to the trade-off and pecking order theories. 90 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 table 2 synopsis of the change of direct effect of firm -specific characteristics under the efsf/esm financial assistance programs on a firm’s capital structure according to the trade-off and pecking order theories change of effect of: trade-off pecking order growth + + profitability + size + tangibility + + non-debt tax shield none methodology bloomberg’s database was used and accounting data on cypriot, irish, greek, and portuguese firms was retrieved from 2003 to 2016. to distinguish between the effect of the efsf/esm programs and the effect of the international financial crisis, data from french, german, italian and spanish firms were collected. in order to mitigate the potential bias due to country size, estonian and austrian firms were then added. finally, firms of the financial sector and utility firms were excluded from the sample. the final sample consisted of 2,086 firms and was winsorized at the 1 percent level. table 3 presents the variables used in the analysis carried out in the present study. these variables have already been discussed above, except for two cases. in particular, these two variables refer to the two dummy variables that were formed to distinguish between the effects of a general economic crisis and the economic conditions that might be attributed to the efsf/esm financial support. 91 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 ta bl e 3 va ri ab le s of th e p re se nt s tu dy v ar ia bl e sy m bo l e xp re ss ed b y: l ev el so ur ce pr ev io us r es ea rc he s fi na nc ia l le ve ra ge fl (1 ) l on gte rm d eb t t o to ta l a ss et s ra tio fi rm b lo om be rg d at ab as e an d o w n pr oc es si ng m or ad i & p au le t ( 20 19 ); d as ka la ki s et a l. (2 01 7) ; d eg ry se e t a l. (2 01 2) (2 ) t ot al d eb t t o to ta l a ss et s ra tio fi rm l ia ng e t a l.( 20 20 ); d oa n et al . ( 20 19 ); d as ka la ki s et a l. (2 01 7) ; d eg ry se e t a l.( 20 12 ) g ro w th g r pe rc en ta ge c ha ng e of to ta l a ss et s fi rm n gu ye n et a l. (2 02 0) ; m or ad i & p au le t ( 20 19 ); d eg ry se e t al . ( 20 12 ) pr ofi ta bi lit y pr o pe ra tin g in co m e to to ta l a ss et s fi rm l ia ng e t a l. (2 02 0) ; d om ne s et a l. (2 01 9) ; b al lio s et a l. (2 01 6) ; g au d et a l. (2 00 5 si ze si n at ur al lo ga ri th m o f t ot al a ss et s fi rm v o (2 01 7) , d an so & a do m ak o (2 01 4) ; m or ad i & pa ul et (2 01 9) ; h al l e t a l. (2 00 4 (c on tin ue d) 92 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 v ar ia bl e sy m bo l e xp re ss ed b y: l ev el so ur ce pr ev io us r es ea rc he s ta ng ib ili ty ta su m o f p pe a nd in ve nt or y to to ta l a ss et s fi rm k re m p et a l. (1 99 9) ; v o, (2 01 7) ; a da ir & a da sk ou , (2 01 7) ; g ha zu an i, (2 01 3) n d t s n d t s d ep re ci at io ns to to ta l a ss et s fi rm m or ad i & p au le t ( 20 19 ); d as ka la ki s et a l. (2 01 7) ; c or te z & s uz an to (2 01 2) ta x ra te s tr su m o f t ax ra te s an d so ci al s ec ur ity c os t a s a pe rc en ta ge o f a fi rm ’s re ve nu e c ou nt ry e ur os ta t d at ab as e c ou nt ry ri sk c r a c ou nt ry ’s c re di t r at in g, w hi ch w as ca lc ul at ed a s th e av er ag e of th e an nu al cr ed it ra tin gs o f s ta nd ar d an d po or ’s , f itc h an d m oo dy ’s fo r e ac h co un tr y. c ou nt ry st an da rd a nd po or ’s , f itc h an d m oo dy ’s w eb si te s an d o w n pr oc es si ng e fs f/ e sm d um m y d it is e qu al to 1 fo r t he y ea rs a ft er th e in tr od uc tio n of th e pr og ra m s, ot he rw is e is e qu al to 0 c ou nt ry o w n pr oc es si ng a ny k in d of ec on om ic cr is is d um m y d c e qu al to 1 w he n bo th th e an nu al e co no m ic gr ow th ra te is n eg at iv e an d th e e fs f/ e sm du m m y is e qu al to z er o, a nd 0 o th er w is e c ou nt ry e ur os ta t a nd o w n pr oc es si ng 93 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. (1) (2) (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the [9] based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀1,𝑖𝑖,𝑡𝑡 (1) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀2,𝑖𝑖,𝑡𝑡 (2) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏1𝑁𝑁 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑁𝑁 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏3𝑁𝑁 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝑁𝑁 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑁𝑁 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝑁𝑁 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝑁𝑁 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀3,𝑖𝑖,𝑡𝑡 (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results of the statistical analysis table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. [9] based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀1,𝑖𝑖,𝑡𝑡 (1) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀2,𝑖𝑖,𝑡𝑡 (2) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏1𝑁𝑁 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑁𝑁 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏3𝑁𝑁 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝑁𝑁 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑁𝑁 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝑁𝑁 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝑁𝑁 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀3,𝑖𝑖,𝑡𝑡 (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results of the statistical analysis table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. [9] based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀1,𝑖𝑖,𝑡𝑡 (1) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀2,𝑖𝑖,𝑡𝑡 (2) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏1𝑁𝑁 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑁𝑁 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏3𝑁𝑁 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝑁𝑁 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑁𝑁 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝑁𝑁 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝑁𝑁 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀3,𝑖𝑖,𝑡𝑡 (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results of the statistical analysis table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. [9] based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀1,𝑖𝑖,𝑡𝑡 (1) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀2,𝑖𝑖,𝑡𝑡 (2) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏1𝑁𝑁 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑁𝑁 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏3𝑁𝑁 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝑁𝑁 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑁𝑁 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝑁𝑁 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝑁𝑁 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀3,𝑖𝑖,𝑡𝑡 (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results of the statistical analysis table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. [9] based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀1,𝑖𝑖,𝑡𝑡 (1) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀2,𝑖𝑖,𝑡𝑡 (2) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏1𝑁𝑁 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑁𝑁 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏3𝑁𝑁 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝑁𝑁 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑁𝑁 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝑁𝑁 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝑁𝑁 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀3,𝑖𝑖,𝑡𝑡 (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results of the statistical analysis table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. [9] based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀1,𝑖𝑖,𝑡𝑡 (1) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀2,𝑖𝑖,𝑡𝑡 (2) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏1𝑁𝑁 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑁𝑁 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏3𝑁𝑁 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝑁𝑁 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑁𝑁 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝑁𝑁 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝑁𝑁 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀3,𝑖𝑖,𝑡𝑡 (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results of the statistical analysis table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. [9] based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀1,𝑖𝑖,𝑡𝑡 (1) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀2,𝑖𝑖,𝑡𝑡 (2) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏1𝑁𝑁 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑁𝑁 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏3𝑁𝑁 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝑁𝑁 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑁𝑁 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝑁𝑁 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝑁𝑁 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀3,𝑖𝑖,𝑡𝑡 (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results of the statistical analysis table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. [9] based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀1,𝑖𝑖,𝑡𝑡 (1) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀2,𝑖𝑖,𝑡𝑡 (2) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏1𝑁𝑁 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑁𝑁 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏3𝑁𝑁 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝑁𝑁 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑁𝑁 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝑁𝑁 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝑁𝑁 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀3,𝑖𝑖,𝑡𝑡 (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results of the statistical analysis table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. [9] based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀1,𝑖𝑖,𝑡𝑡 (1) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀2,𝑖𝑖,𝑡𝑡 (2) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏1𝑁𝑁 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑁𝑁 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏3𝑁𝑁 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝑁𝑁 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑁𝑁 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝑁𝑁 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝑁𝑁 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀3,𝑖𝑖,𝑡𝑡 (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results of the statistical analysis table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. [9] based on the above specified variables, the following regressions were estimated. the first regression assessed the impact of each of the independent variables, without taking into account the special economic conditions formed by the efsf/esm financial programs, or by any type of crisis. the second estimates the respective impact, considering only the special economic conditions as a result of the efsf/esm financial programs. finally, the third one estimated the respective impact, taking into account the special economic conditions due to both efsf/esm financial programs and by any type of economic crisis. 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀1,𝑖𝑖,𝑡𝑡 (1) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀2,𝑖𝑖,𝑡𝑡 (2) 𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + ∑ 𝜈𝜈𝑗𝑗𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡−𝑗𝑗𝑛𝑛𝑗𝑗=1 + 𝑎𝑎1𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎2𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎3𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎4𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑎𝑎5𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑎𝑎6𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑎𝑎7𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏1𝑁𝑁 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑁𝑁 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏3𝑁𝑁 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝑁𝑁 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑁𝑁 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝑁𝑁 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝑁𝑁 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐1𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐺𝐺𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐2𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑃𝑃𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐3𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐4𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝑐𝑐5𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝑇𝑇𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝑐𝑐6𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝑇𝑇𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝑐𝑐7𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖 ∙ 𝐶𝐶𝐺𝐺𝑖𝑖,𝑡𝑡 + 𝜀𝜀3,𝑖𝑖,𝑡𝑡 (3) where i refers to each firm, t to each year and ε the residuals. the above three dynamic panel data models were estimated using the generalized method of moments (gmm) for endogeneity problems (omitted variables, confounding factors, simultaneity, and measurement errors) to be faced (roodman, 2009). system gmm was preferred to difference gmm, since panel data comprised a large n and small t dataset (arellano & bover, 1995), while the number of lagged dependent variables was selected so that the models satisfied both the test for instrument validity (sargan/hansen) and the test for second-order serial correlation (arellano & bond, 1991). simultaneously, wald tests were also executed to test for the sign of coefficients of the variables (ghazouani, 2013; degryse et al., 2012). results of the statistical analysis table 4 shows the descriptive statistics of the variables used in the present study (arithmetic mean and standard deviation). all the leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. 94 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 leverage ratios increased during the efsf/esm financial supporting period, while the same change was observed for tangibility and country risk. on the other hand, growth, profitability and tax rates decreased. table 4 descriptive statistics for the research variables total sample d = 0 (no efsf/esm support) d = 1 (during efsf/esm support) n mean std. dev. n mean std. dev. n mean std. dev. fl1 26,999 0.1429 0.1529 25,187 0.1414 0.1511 1,812 0.1647 0.1747 fl2 26,999 0.2448 0.2135 25,187 0.2364 0.2043 1,812 0.3624 0.2904 gr 26,999 0.0624 0.3185 25,187 0.0712 0.3221 1,812 -0.0599 0.2320 pr 26,999 0.0226 0.1529 25,187 0.0236 0.1555 1,812 0.0089 0.1097 si 26,999 5.4844 2.2691 25,187 5.4948 2.2847 1,812 5.3399 2.0345 ta 26,999 0.3476 0.2416 25,187 0.3413 0.2388 1,812 0.4360 0.2624 ndts 26,999 0.0393 0.0341 25,187 0.0398 0.0345 1,812 0.0323 0.0278 tr 26,999 0.5952 0.1357 25,187 0.6071 0.1284 1,812 0.4298 0.1268 cr 26,999 3.3272 4.0860 25,187 2.5069 2.5667 1,812 14.7287 4.2248 note. fl1 long-term debt to total asset, fl2 total debt to total assets, gr growth, pr profitability, si size, ta tangibility, ndts non-debt tax shield, cr country risk, tr tax rates. table 5 presents the estimated three regressions using long-term debt to total assets ratio as the dependent variable. in cases of all models, tangibility (a4>0, p<0.01) and growth (a1>0, p<0.01) positively influenced leverage, while profitability exerted a negative impact (a2<0, p<0.01). out of the country-characteristics, the impact of the tax rates (a7>0, p<0.01) was positive. the estimation of model (2) shows that the efsf/esm support increased the effect of profitability on a firm’s leverage (b2>0, p<0.10) when no other crisis was taken into account; while model (3) indicates that there was no interaction of the efsf/esm and firm-specific characteristics when crises were taken into account. regarding the country-specific characteristics, the impact of country risk decreased (b7<0, p<0.10) and the tax rates increased (b6>0, p<0.10), in both model (2) and model (3). 95 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 table 5 estimation of the models with unbalanced panel data: the longterm debt to total assets ratio as the dependent variable variable fl1: long-term debt to total assets model (1) (2) (3) gr 0.0333*** 0.0298*** 0.0350*** pr -0.0800*** -0.0925*** -0.0881*** si 0.0044*** 0.0043*** 0.0031*** ta 0.1377*** 0.1358*** 0.1312*** ndts 0.1405*** 0.1201*** 0.0848*** tr 0.4149*** 0.2711*** 0.2717*** cr -0.0005*** -0.0006*** -0.0002*** d x gr 0.0450*** 0.0396*** d x pr 0.1520*** 0.1439*** d x si -0.0027*** -0.0017*** d x ta 0.0200*** 0.0215*** d x ndts 0.0313*** 0.0837*** d x tr 0.1556*** 0.1464*** d x cr -0.0034*** -0.0038*** difc x gr -0.0235 *** difc x pr -0.0069 *** difc x si 0.0022 *** difc x ta 0.0014 *** difc x ndts 0.0828 *** difc x tr -0.0209 *** difc x cr -0.0007 *** year 2009 -0.0022*** -0.0026*** 0.0049*** year 2010 -0.0039*** -0.0044*** 0.0005*** year 2011 -0.0051*** -0.0058*** -0.0012*** year 2012 -0.0051*** -0.0071*** -0.0026*** year 2013 -0.0002*** -0.0030*** 0.0013*** year 2014 0.0042*** 0.0014*** 0.0062*** year 2015 0.0083*** 0.0054*** 0.0100*** year 2016 0.0046*** 0.0008*** 0.0052*** constant -0.2840*** -0.1989*** -0.1948*** (continued) 96 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 variable fl1: long-term debt to total assets model (1) (2) (3) no of firms 2,083 no of observations 13,786 ar(2) : p-value 0.1993 0.1799 0.1494 sargen test : p-value 0.1355 0.1250 0.1009 * p < 0,10, ** p < 0,05, *** p < 0,01 note. 1. fl1 long-term debt to total asset, gr growth, pr profitability, si size, ta tangibility, ndts non-debt tax shield, cr country’s risk, tr tax rates, d dummy variable for years of efsf/esm financial support, difs dummy variable for years of other type of economic crisis. note 2. the estimated impact of lags of the dependent variable is omitted. table 6 estimation of the models with unbalanced panel data: the total debt to total assets ratio as the dependent variable variable fl2: total debt to total assets model (1) (2) (3) gr 0.0473*** 0.0494*** 0.0491*** pr -0.2377*** -0.2257*** -0.2165*** si -0.0179*** -0.0182*** -0.0193*** ta 0.2241*** 0.2217*** 0.2252*** ndts 0.1462*** 0.1303*** 0.0984*** tr 0.2476*** 0.3038*** 0.3124*** cr 0.0016*** 0.0022*** 0.0026*** d x gr -0.0142*** -0.0123*** d x pr -0.1875*** -0.2075*** d x si 0.0011*** 0.0025*** d x ta 0.0106*** -0.0037*** d x ndts 0.3588*** 0.4396*** d x tr -0.0753*** -0.0837*** d x cr 0.0002*** -0.0002*** difc x gr -0.0006 *** difc x pr -0.0380 *** (continued) 97 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 variable fl2: total debt to total assets model (1) (2) (3) difc x si 0.0022 *** difc x ta -0.0144 *** difc x ndts 0.1333 *** difc x tr -0.0153 *** difc x cr -0.0007 *** year 2009 -0.0226*** -0.0225*** 0.0247*** year 2010 -0.0261*** -0.0262*** -0.0017*** year 2011 -0.0165*** -0.0163*** 0.0077*** year 2012 -0.0215*** -0.0215*** 0.0028*** year 2013 -0.0227*** -0.0224*** 0.0019*** year 2014 -0.0161*** -0.0157*** 0.0083*** year 2015 -0.0125*** -0.0120*** 0.0119*** year 2016 -0.0199*** -0.0189*** 0.0050*** constant -0.0788*** -0.1098*** -0.1329*** no of firms 2,083 no of observations 13,786 ar(2) : p-value 0.5470 0.5819 0.5636 sargen test : p-value 0.2996 0.2918 0.2875 * p < 0,10, ** p < 0,05, *** p < 0,01 note. 1. fl2 total debt to total asset, gr growth, pr profitability, si size, ta tangibility, ndts non-debt tax shield, cr country’s risk, tr tax rates, d dummy variable for years of efsf/esm financial support, difs dummy variable for years of other type of economic crisis. note. 2. the estimated impact of lags of the dependent variable is omitted. table 6 shows the estimations of the regression with total debt to total assets ratio as the dependent variable. the results were similar to those presented in table 5 with the following differences. firstly, firm size seemed to influence negatively the leverage ratio of total debt to total assets (b3<0, p<0.05). secondly, considering the efsf/esm period and/or crisis period, the country risk had influenced a firm’s leverage in a positive way in the normal economic conditions period (a6>0, p<0.05). third, there was a positive impact on the ndts during the efsf/esm financial assistance period [b5>0, p<0.10]. fourth, the effect of profitability during the efsf/esm period seemed to become more negative (b2<0.20, p<0.10). 98 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 table 7 wald tests for the statistical significance of the independent variables model (1) (2) (3) period total normal economic conditions during efsf/esm support normal economic conditions during crisis during efsf/esm support d = 0 d = 1 d = dc = 0 dc = 1 d = 1 fl1: long-term debt to total assets gr + + + + + pr si ta + + + + + + ndts tr + + + + + + cr fl2: total debt to total assets gr + + + + pr si ta + + + + + + ndts + + tr + + + + + + cr + + + + note. fl1 long-term debt to total asset fl2 total debt to total asset, gr growth, pr profitability, si size, ta tangibility, ndts non-debt tax shield, cr country’s risk, tr tax rates, d dummy variable for years of efsf/esm financial support, difs dummy variable for years of other type of economic crisis. finally, table 7 shows the results of wald tests on the statistical significance of the coefficients. according to table 7, after the acceptance of an efsf/esm financial assistance program, profitability had a negative impact on the long-term leverage, but its effect on the total debt leverage nullifies. in addition, the impact of profitability was intensified (became more negative) regarding total debt to total 99 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 assets ratio, but weakened (became more positive) in the case of longterm debt to total assets ratio. the negative effect of size appeared only in the case of total debt ratio, which remained strong during the efsf/esm support period. tangibility positively influenced both the leverage ratios regardless of economic conditions. regarding total debt to total assets ratio, the positive effect of growth lost its statistical significance and the impact of the ndts became positive during the period of efsf/esm financial assistance. with regard to country-specific characteristics, country risk had no effect on the longterm leverage, but there was a positive effect on the total debt ratio. however, during the efsf/esm support, the effect of country risk on long-term leverage transforms to negative and its effect on total debt ratio was nullified. the impact of tax rates remained positive regardless of the referring leverage ratio, or the referring economic conditions. finally, it is worth mentioning that only one difference was noticed when normal economic conditions were compared to the crisis period; the positive impact of growth on long-term leverage ratio was diminished. discussion the statistical analysis presented in the foregoing section seemed to suggest that the pecking order theoretical framework has a significantly higher interpretation power than the trade-off perspective of things, when it comes down to explaining a firm’s financing decisions. however, there were also findings that neither theory could explain (wong, 2018), such as the positive impact of the ndts during the efsf/esm support period, and findings that only trade-off predicts, such as the positive change of the effect of profitability. so, it was considered safe to conclude that financial leverage could not be interpreted exclusively by the trade-off or pecking order theory, but both theories were necessary (chatzinas & papadopoulos, 2018; banga & gupta, 2017; mc namara et al., 2017; ohman & yazdanfar, 2017; serrasqueiro & caetano, 2015; burgstaller & wagner, 2015). more specifically, it was found that growth and tangibility positively influenced the financial leverage of firms, but the impact from profitability and size had proved to be negative. however, the present study has provided evidence that the debt level would increase when the growth prospects increased (frank & goyal, 2003) and when the tangibility increased due to the lower levels of financial distress costs 100 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 (frank & goyal, 2003; ebrahim et al., 2014). on the other hand, the financial leverage would decrease when the profitability increased due to the higher amounts of available internal funds (myers, 2001), and when the size of a firm would increase due to the lower levels of risk (gaud et al., 2007). the above impacts have also been observed in periods of crisis, with only one exception; the effect of growth on the long-term leverage would be nullified during a period of crisis. on the other hand, some effects would be diminished or strengthened in the case of efsf/esm support periods. what could be the reasons for this result? during a crisis, an increase of uncertainty was observed, resulting in a rise of the cost of both debt and equity (cetorelli & goldberg, 2011). however, an efsf/esm program would reduce uncertainty, as it would see the implementation of strict fiscal policy measures (increase of taxes and/ or government spending cuts). as a result, the factors that inhibited the above effects should be sought as a result of these changes. a probable reason could be a firm’s need for reorganization. an efsf/esm program would incorporate measures that could affect a variety of macro and micro economic variables through the implementation of structural reforms. except for interest rates and tax rates, other parameters such as the disposable income, consumption and consumption habits, investment procedures, and savings were also influenced by government policies. in this rapidly changing environment, the management of a firm would be responsible for the firm’s survival. appropriate adjustments in prices, costs, wages and expenses should be decided and implemented. as a result, the managers would have to reorganize many aspects of the firm’s function. under the above assumptions, each change that the efsf/esm program triggered will be discussed below. firstly, during the period of efsf/esm assistance, the negative effect of profitability on the long-term leverage would disappear and the negative effect on total debt leverage would strengthen. alternatively, the efsf/esm financial assistance could negatively change the effect of profitability on the total debt leverage, which would be consistent with po-rh2, but positively change the respective effect on the long-term leverage. previous researchers have found that an increase in profitability 101 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 seemed to lead firms to increase the long-term debt, but to reduce the short term debt (vo, 2017). the interaction effect of the efsf/esm might be explained under this view. the more profitable a firm was, the more it would reduce its short-term debt because of the tax rates increase. this reaction of firms was also proposed by d’amato (2020). however, the same adjustment was not likely to be implemented for the long-term debt usually used for long-term investments, since the efsf/esm program had reduced the level of uncertainty. secondly, the positive impact of growth on the total debt ratio, which is in accordance with the pecking order theory, will diminish when an economy enters an efsf/esm financial assistance program. the same change is not observed in case of the long-term debt ratio. moradi and paulet (2019) observed that shareholders during the international financial crisis seemed to be obliged to finance their firms’ growth due to the cost of capital increase in the banking sector, and the negative economic prospects. this may be also the case in the context of the efsf/esm programs. since the borrowing interest rates of the banking sector are expected to slowly adjust, firms with high growth opportunities seem to face problems to get access to short-term funds (inadequate funds, expensive interest rate) and therefore, they turn to self-financing. thirdly, the non-debt tax shield will have a positive effect on the total debt to total assets ratio during efsf/esm assistance, a result that has been anticipated by the no capital structure theory. previous researches have also faced this conflicting result and two explanations have been provided. either the firms’ profitability was so high that the full exploitation of debtand non-debt tax shields was justified or, due to high tax rates, the firms had found new ways to decrease their tax burden (moradi & paulet, 2019; moradi & paulet, 2015). given that during an efsf/esm program, a firm’s profitability was not expected to be significantly high, it was safe enough to suppose that the firm increased its depreciations and amortization, as a means of reacting to the increase in taxes. with regard to the two macroeconomic variables, the impact of country risks and tax rates on the leverage will be mostly positive. hence, the present study has been able to provide empirical evidence that as a country’s financial conditions deteriorated, a firm’s financial leverage 102 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 would increase, just as the pecking order theory had predicted. on the other hand, the rationale behind the positive influence of tax rates on a firm’s optimal capital structure might be explained by both the pecking order and trade-off theories. conclusion the present study examined five firm-specific capital structure determinants, namely growth, profitability, size, tangibility and tax deductibility in ten eurozone-member countries in order to study the impact of the efsf/esm financial supporting programs on a firm’s financial leverage. the results showed that the pecking order theory prevails. however, a detailed examination of the evidence indicates that despite the pecking order’s prevalence, trade-off is also necessary for a full comprehension of a firm’s financing decisions. a crisis in general, changes the magnitude, but not the sign of the effect that a firm’s characteristics have on its financial leverage. on the other hand, when the efsf/esm support programs were implemented, changes in effects of three firm-characteristics were observed. during the period of the efsf/esm assistance, the negative effect of profitability on long-term leverage disappeared, and on total debt leverage strengthened, the positive impact of growth on total debt ratio diminished and the ndts had a positive influence on the total debt ratio. these changes might be explained by the fact that the firms were motivated to proceed to re-organization due to the following three environmental factors: increased levels of tax rates, decreased levels of uncertainty, and the reluctance of lenders to provide new funds. the above analysis was based on data from the stock exchange of listed european firms. as a result, the study has focused on large firms. a similar study for small and medium sized enterprises would be interesting and worthwhile. finally, an interesting practice can be recommended for policy makers. to the best of our knowledge, it is the first time that the effect of financial distress has been distinguished from the effect of tax rates. as a result, policy makers should make every effort to provide low tax rates in the event of rising interest rates. 103 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for profit sectors. references adair, p., & adaskou, m. (2017). the capital structure of french smes and impact of the financial crisis: a dynamic panel data analysis (2002-2010). https://hal-upec-upem.archivesouvertes.fr/hal-01667313/document arellano, m., & bond, s. (1991). some tests of specification for panel data: monte carlo evidence and an application to employment equations. the review of economic studies, 58(2), 277-297. arellano, m., & bover, o. (1995). another look at the instrumental variable estimation of error-components models. journal of econometrics, 68(1), 29-51. bancel, f., & mittoo, u. r. (2004). cross-country determinants of capital structure choice: a survey of european firms. financial management, 33(4), 103-132. banerjee, a. (2017). capital structure and its determinants during the pre and post period of recession: pecking order vs. trade off theory. indian journal of science, 11(1), 44-58. banga, c., & gupta, a. (2017). effect of firm characteristics on capital structure decisions of indian smes. international journal of applied business and economic research, 15(10), 281-301. burgstaller, j., & wagner, e. (2015). how do family ownership and founder management affect capital structure decisions and adjustment of smes? the journal of risk finance, 16(1), 73-101. cetorelli, n., & goldberg, l. s. (2011). global banks and international shock transmission: evidence from the crisis, imf economic review, 59(1), 41–76. chatzinas, g., & papadopoulos, s., (2018). trade-off vs pecking order theory: evidence from greek firms in a period of debt crisis. international journal of banking, accounting and finance, 9(2), 170-191. chen, j. j. (2004). determinants of capital structure of chinese-listed companies. journal of business research, 57(12), 1341-1351. chin, k. y., & zakaria, z. (2018). board characteristics and capital structure of malaysia consumer products sector. international journal of banking and finance, 13(1), 71-94. 104 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 cortez, m. a., & suzanto, s. (2012). the determinants of corporate capital structure: evidence from japanese manufacturing companies. journal of international business research, 11(3), 121-134. d’ amato, a. (2020). capital structure, debt maturity, and financial crisis: empirical evidence from smes. small business economics, 55, 919-941. danso, a., & adomako, s. (2014). the financing behaviour of firms and financial crisis. managerial finance, 40(12), 1159-1174. daskalakis, n., & psillaki, m. (2008). do country or firm factors explain capital structure? evidence from smes in france and greece. applied financial economics, 18(2), 87-97. daskalakis, n., & psillaki, m. (2009). are the determinants of capital structure country or firm specific? small business economics, 33(3), 319-333. daskalakis, n., balios, d., & dalla, v. (2017). the behaviour of smes’ capital structure determinants in different macroeconomic states. journal of corporate finance, 46(10), 248-260. deangelo, h., & masulis, r. w. (1980). optimal capital structure under corporate and personal taxation. journal of financial economics, 8(1), 3-29. degryse, h., de goeij, p., & kappert, p. (2012). the impact of firm and industry characteristics on small firms’ capital structure. small business economics, 38(4), 431-447. demirguc-kunt, a., marinez-peria, s.m., & tressel, t. (2020). the global financial crisis and the capital structure of firms: was the impact more severe among smes and non-listed firms? journal of corporate finance, 60, 101514, https://doi.org/10.1016/j. jcorpfin.2019.101514 demirguc-kunt, a., martinez-peria, m. s., & tressel, t. (2015). the impact of the global financial crisis on firms capital structure. the world bank. https://elibrary.worldbank.org/doi/ abs/10.1596/1813-9450-7522 doan, t. t. t. (2019). firm-specific factors as determinants of capital structure: evidence from an emerging country. international journal of economics and management systems, 4, 298-304. ebrahim, m. s., girma, s., shah, m. e., & williams, j. (2014). dynamic capital structure and political patronage: the case of malaysia. international review of financial analysis, 31, 117-128. 105 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 eriotis, n., vasiliou, d., & ventoura-neokosmidi, z. (2007). how firm characteristics affect capital structure: an empirical study. managerial finance, 33(5), 321-331. european stability mechanism (esm): https://www.esm.europa.eu/ assistance/lending-toolkit#lending_toolkit frank, m. z., & goyal, v. k., (2003). testing the pecking order theory of capital structure. journal of financial economics, 67(2), 217-248. gaud, p., hoesli, m., & bender, a. (2007). debt-equity choice in europe. international review of financial analysis, 16(3), 201-222. gaud, p., jani, e., hoesli, m., & bender, a. (2005). the capital structure of swiss companies: an empirical analysis using dynamic panel data. european financial management, 11(1), 51-69. ghazouani, t. (2013). the capital structure through the trade-off theory: evidence from tunisian firm. international journal of economics and financial issues, 3(3), 625-636. grinblatt, m., & titman, s. (2002). financial market and corporate strategy (2nd ed.). the mcgraw-hill companies. hall, g. c., hutchinson, p. j., & michaelas, n. (2004). determinants of capital structures of european smes. journal of business finance and accounting, 31(5/6), 711-728. harrison, b., & widjaja, t. w. (2014). the determinants of capital structure: comparison between before and after financial crisis. economic issues, 19(2), 55-82. jensen, m., & meckling, n. (1976). the theory of the firm: managerial behavior, agency cost and ownership structure. journal of financial economics, 3(10), 305-360. jong, a., kabir, r., & nguyen, t. t. (2008). capital structure around the world: the roles of firmand countryspecific determinants. journal of banking and finance, 32(9), 1954-1969. kremp, e., stoss, e., & gerdesmeier, d. (1999). estimation of a debt function: evidence from french and german firm panel data. in: a. sauve, & m. scheuer (eds.), corporate finance in germany and france: a joint research project of deutsche bundesbank and the banque de france, www.ssrn.com liang, c. c., liu, y., troy, c., & chen, w. w. (2020). firm characteristics and capital structure: evidence from asean-4 economies. advances in pacific basin business, economics and finance, 8(9), 149-162. 106 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 mac an bhaird, c., & lucey, b. (2010). determinants of capital structure in irish smes. small business economics, 35(3), 357375. machielsen, b. (2012). empirical evidence on the existence of a pecking order: a study about whether the pecking order theory is an accurate means to describe the incremental financing practices by firm in the european union. bachelor thesis in the area of business administration of twente univesity, https:// essay.utwente.nl/63784/1/bsc_bas_machielsen_s1131044.pdf mc namara, murro, p., & o’donohoe, s. (2017). countries lending infrastructure and capital structure determination: the case of european smes. journal of corporate finance, 43(4), 122-138. migliori, s., maturo, f., & paolone, f. (2018). capital structure determinants in family firms: an empirical analysis in context of crisis. international business research, 11(4), 65-83. mihalca, g., & antal, r. (2009). an empirical investigation of the trade-off and pecking order hypotheses on romanian market. in asmda. proceedings of the international conference applied stochastic models and data analysis (vol. 13, pp. 109114). vilnius: gediminas technical university, department of construction economics & property. miller, e. m. (1977). risk, uncertainty, and divergence of opinion. the journal of finance, 32(4), 1151-1168. modigliani, f., & miller, m. (1958). the cost of capital, corporation finance and theory of investment. american economic review, 48(3), 261-297. modigliani, f. & miller, m. (1963). taxes and the cost of capital: a correction. american economic review, 53(3), 433-443. mokhova, n., & zinecker, m. (2013). the determinants of capital structure: the evidence from the european union. acta universitatis agriculturae et silviculturae mendelianae brunensis, 61(7), 2533-2545. moradi, a., & paulet, e. (2015). a causal loop analysis of the austerity policy adopted to address the euro crisis-effects and side effects. international journal of applied decision sciences, 8(1), 1-20. moradi, a., & paulet, e. (2019). the firm-specific determinants of capital structure–an empirical analysis of firms before and during the euro crisis. research in international business and finance, 47(1), 150-161. 107 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 myers, s. c. (2001). capital structure. journal of economics perspectives, 15(2), 81-102. nguyen, h. m., vuong, t. h. g., nguyen, t. h., wu, y. c., & wong, w. k. (2020). sustainability of both pecking order and trade-off theories in chinese manufacturing firms. sustainability, 12(9), 3883. noulas, a., & genimakis, g. (2011). the determinants of capital structure choice: evidence from greek listed companies. applied financial economics, 21(6), 379-387. nunes, p. m., & serrasqueiro, z. (2017). short-term debt and longterm debt determinants in small and medium-sized hospitality firms. tourism economics, 23(3), 543-560. öhman, p., & yazdanfar, d. (2017). short-and long-term debt determinants in swedish smes. review of accounting and finance, 16(1), 106-124. pacheco, l. (2016). capital structure and internationalization: the case of portuguese industrial smes. research in international business and finance, 38(9), 531-545. rahim, n., nor, f. m., ramli, n., & marzuki, a. (2020). determinants of capital structure for malaysian shariah-compliant firms: the impact of revised screening methodology. international journal of banking and finance, 14, 45-74. roodman, d. (2009). a note on the theme of too many instruments. oxford bulleting of economics and statistics, 71(1), 135-158. sergiescu, l., & vaidean, v.-l. (2014). determinant factors of the capital structure of a firm – an empirical analysis. procedia economics and finance, 15, 1447-1457. serrasqueiro, z., & caetano, a. (2015). trade-off theory versus pecking order theory: capital structure decisions in a peripheral region of portugal. journal of business economics and management, 16(2), 445-466. shyam-sunder, l., & myers, s. c. (1999). testing static tradeoff against pecking order models of capital structure. journal of financial economics, 51(2), 219-244. singala, s., & kumar, r. (2012). the global financial crises with a focus on the european sovereign debt crisis. asci journal of management, 42(1), 20-36. smart, b. s., megginson, l. w., & gitman, j. l. (2004). corporate finance. thomson south-western. 108 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 81–108 stulz, r. (1990). managerial discretion and optimal financing policies. journal of financial economics, 26(1), 3-27. vo, x. v. (2017). determinants of capital structure in emerging markets: evidence from vietnam. research in international business and finance, 40(4), 105-113. wong, w. c. (2018). debt conservatism and debt-equity choices: evidence from reits’ unused debt capacity. international journal of banking and finance, 13(1), 1-28. wruck, k. h. (1990). financial distress, reorganization, and organizational efficiency. journal of financial economics, 27(2), 419-444. zhang, x., & mirza, s. s. (2015). determinants of capital structure of firms in pre-post financial crisis: evidence from china. research journal of finance and accounting, 6(12), 33-51. determinants of capital structure for malaysian shariah-compliant firms: the impact of revised screening methodology norfhadzilahwati rahim* fauzias mat nor nurainna ramli ainulashikin marzuki faculty of economics and muamalat, universiti sains islam malaysia, malaysia *corresponding author: fhadzilahwati@raudah.usim.edu.my _______________________________________________________________ a r t i c l e i n f o _____________________________ article history: received 10 april 2019 revised 15 august 2019 accepted 29 august 2019 published 6 january 2020 ____________________________ keywords: capital structure, malaysian public-listed shariah-compliant firms, panel data analysis. jel codes: g32. a b s t r a c t _________________________________ this study investigates two main objectives. firstly, the determinants of capital structure were examined for each sector among malaysian shariah-compliant firms, and whether the inclusion of islamic debt (leverage 1 and leverage 2) has led to different results due to changes in the screening methodology. secondly, this paper analyzes the target capital structure and speed of adjustment for both before and after the revised screening methodology. this study employs panel data analysis by using generalized method of moment (gmm). the sample consists of 192 shariah-compliant companies in malaysia during the period of 1999 to 2017. the results demonstrated that the firm has target capital structure and identified specific determinants that have affected the capital structure of the international journal of banking and finance, vol. 14, 2018-2019: 45-74 45 46 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 shariah-compliant firms in malaysia. moreover, the findings have also revealed certain implications toward large firms. large firms tend to generate more income and profit, however at the same time, these firms require more debt to support investment activities. hence, with regards to profitability, this study identified a negative relationship between profitability and leverage for shariah-compliant firms for all sectors. shariah-compliant firms with high profitability will use a lower leverage in their financial activities. thus, the results strongly support the pecking order theory. other than that, this study found that the lagged dependent variable (lagged leverage 1 and leverage 2) presented a positive significance, and concluded that the speed of adjustment takes approximately 2 years. this suggests that the shariah-compliant firms close approximately by 30% to 70% of the gap between current and target capital structure within one and two years. furthermore, the findings on the target leverage level imply that after the revised screening methodology was introduced in november 2013, the speed of adjustment became faster than before the implementation of the new screening methodology. thus, it is important for management to maintain the target leverage during financial decision making, which in turn strengthens the firm’s shariahcompliant financial stability and sustainability, and continue to remain listed as shariah-compliant securities. this paper provides an overview of capital structure behaviour in malaysia. determinants of capital structure for malaysian shariah-compliant firms: 45-74 47 1. introduction in the arena of corporate finance, capital structure is a very important aspect of a company’s investment choice as this will affect the company’s value and financial risk. this topic is one of the most debated issue in the finance literature. many researchers have conducted studies on its theories, and obtained empirical evidences on the factors that have influenced the choice of capital structure, optimum capital structure and the influence of capital structure on the value of the firm. based on the pecking order theory, a company is required to fund the financial asset by retaining earnings at the beginning. however, if the company faces financial problems, it must raise funds on debt, followed by equity. however, the use of debt is limited as companies might face bankruptcy that would affect the long term prosperity of the company, which in turn relates to how managers deal and operate the company’s capital structure (johnson & soenen, 2003). capital structure decisions are influenced by firm-specific factors (haron, 2016). the examples for firm-specific factors are profitability, business risk, firm size, liquidity and share price performance. in addition to the capital structure decision, the speed of adjustment occur when firms have either a higher or lower debt, or when the firm face a financial surplus or deficit (byoun, 2008). thus, firms will move towards the target capital structure when companies experience a financial deficit or surplus. by identifying the factors that influence capital structure decision, companies would be able to move towards value maximization that would further strengthen the company’s financial stability. in 2013, the securities commission malaysia (sc) revised the screening methodology for the operating, financing and investing criteria of the listed firms. the new criteria added into the shariah screening methodology is financial ratio benchmarks. it is based on two ratios; cash over total assets and debt over total assets. the limit set for the financial ratio benchmarks is 33 percent, whereby a firm is required to have 33 percent of its holdings of cash or cash equivalents in conventional deposits or accounts, and conventional debts or borrowings for companies listed as shariah-compliant firms. due to this revision, the number of listed shariah-compliant firms were reduced to 653 firms (71.4 percent) out of the total 914 firms from the listed securities on the bursa malaysia. the main reason for the reduced listed shariah-compliant companies is due to the firms having a higher level of conventional debt which exceeded the 33 percent limit that was set by the sc. therefore, in order to remain listed as a shariahcompliant company, the alternative option is to take up islamic debt after hitting the 33 percent limit. in other words, by managing financial leverage, companies can enhance the shariah-compliant financial stability while remaining listed as shariah-compliant securities. furthermore, the identification of the determinants of capital structure in malaysian shariah-compliant firms have continued to spark debates, and is due to the lack of previous empirical studies that have not provided substantial evidences with regards to the impacts of these factors on capital structure. it 48 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 is important to measure the determinants of capital structure as it could help managers to conduct risk management efficiently. therefore, this study examines two objectives; firstly, to investigate the determinants of capital structure for each sector among malaysian shariah-compliant firms, and whether the inclusion of islamic debt (leverage 1 and leverage 2) has led to different results due to changes in the screening methodology. secondly, this study examines the target capital structure and speed of adjustment for both before and after the revised screening methodology. this study employs a quantitative approach to collect data from the thomson reuters database. in this study, 192 shariah-compliant companies in malaysia were selected from 1999 to 2017. this study applies panel data analysis by using generalized method of moment (gmm). the panel data analysis is examined using stata software. this paper is organized as follows; the next session presents a literature review on the revised shariah screening methodology, capital structure and speed of adjustment based on previous research. the following section is the results of the data and model of analysis. the subsequent section presents a discussion on the findings of the research analysis. the last section summarizes the conclusion and discussion of the research. 2. literature review 2.1 the impact of the revised screening methodology shariah-compliant firms in malaysia are screened based on the benchmarks that have been specified by the sc. the screening methodology was revised by the sc in 2013. the revision not only included operating activities, but also the financing and investing aspects of the listed firms’ business operations. the financial ratios were added benchmark indicators in the screening process during the revision of the methodology for firms to be listed as shariah-compliant. in light of these changes, relevant questions were raised on the effectiveness of the revised screening methodology (zainudin et al., 2014; najib et al., 2014; muhammad, 2015; suffian et al. (2015); md. hashim et al., 2017). zainudin et al. (2014) revealed that companies with high levels of conventional debt were considerably affected as there were no screening criteria that were based on the total conventional debt of company. moreover, muhammad (2015) suggested that after the introduction of the revised screening methodology, the status of firms were affected due to (i) the companies’ mixed activities that were previously assessed under the 5, 10, 20, and 25 percent benchmarks, and were subsequently revised to reassess the current firms activities under the 5 and 20 percent benchmarks, and (ii) companies with high conventional debt were affected as there were no previous screening that was based on the total conventional debt of the companies. furthermore, a review of the study conducted by suffian et al. (2015) indicated that some of the shariah-compliant firms had a high risk of having determinants of capital structure for malaysian shariah-compliant firms: 45-74 49 larger debts. consequently, these results suggest that shariah-compliant firms should minimise their risk and uncertainties. with regards to the shariah screening criteria, najib et al. (2014) concluded that the inclusion of financial ratio benchmarks and restructuring of the business activity benchmarks would enhance the robustness of the screening methodology for the listed shariahcompliant companies. at the same time, the robustness of the screening methodology would be able to stimulate competitiveness within the malaysian islamic equity market and islamic fund management industries. in accordance with the study conducted by md. hashim et al. (2017), the formulas that were applied for the financial ratios in the shariah-compliance criteria are taken based on the best practice approach, which may or may not concur with most of the current leading shariah stock screening providers. 2.2 the determinants of capital structure in relation to financial leverage, the existing literature on capital structure are extensive, and mainly highlights the factors that influence capital structure. the causes of capital structure have been studied extensively (al-najjar, 2011; alyahyaee et al., 2013; antoniou et al., 2002; bhattacharjee et al., 2010; bouraoui and li, 2014; brendea, 2014; drobetz et al., 2013; getzmann et al., 2014; haron et al., 2013; haron, 2016; haron & ibrahim, 2011). in the analysis of capital structure, al-najjar (2011) suggested that capital structure choices are influenced by institutional ownership, profitability, business risk, asset tangibility, asset liquidity, market-to-book, and firm size. additionally, it has been concluded that the factors influencing capital structure choice are consistent in both developed and developing countries. moreover, the study found that jordanian firms have target capital structure ratios, and were adjusted relatively quickly to the target capital structure. the study carried out by antoniou et al. (2002) demonstrated that leverage ratio is positively correlated to the size of the firm. however, the leverage ratio is inversely affected by the market-to-book ratio, term-structure of interest rates, and share price performance. the results indicated different directions and degrees of leverage that have been influenced by fixed-assets ratio, equity market premium, profitability, and effective tax rates. the firm’s capital structure were found to be close to the target, with the results showing different speeds in all three countries when the firm’s debt ratios were adjusted. the literature further concluded that the capital structure decision is not only based on its own characteristics, but is also influenced by the firm’s operating traditions and environment. similarly, haron and ibrahim (2011) have concurred with previous literature that the capital structure decision is not only the product of the firm’s own characteristics, but is also a result of the institutional environment and corporate governance in which the firm operates. based on econometric analysis, the growth of firms is positively related to the firm’s leverage ratio, particularly for it firms (bhattacharjee et al., 2010). the authors found that the creditor’s rights, maintenance of legal reserves and law 50 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 enforcement, director’s rights on borrowing, and risk assessments are important determining factors for the capital structure decision of a firm. previous studies on target capital structure identified profitability, firm’s size, and asset tangibility as the most important determining factors for target capital structure. conversely, researchers have reported that ownership structure had no significant effect on the target capital structure for romanian firms. furthermore, it is observed that the adjustment speed of size is high, and suggested that it is costly to firms when they deviate from the target capital structure. based on previous studies, determinants for capital structure decision can be divided into two components; common and industry-based components. in asia, profitability and tangibility of assets are categorized under common determinants, while industrial median, size, and non-debt tax shield are categorized under industry-based components. the results revealed significant evidences that suggest large asian companies pursue target capital structures similarly to those in the u.s. and europe (getzmann et al., 2014). in a study conducted by haron (2016), target capital structure and capital structure decision were influenced by firm-specific factors observed within indonesian firms. some of the examples of firm-specific factors are profitability, business risk, firm size, liquidity, and share price performance. in addition, the identification of the factors that influence target capital structure and capital structure decision will be able to guide firms toward value maximization. 2.3 target capital structure on speed of adjustment various studies have investigated target capital structure on speed of adjustment (abdeljawad et al., 2013; arioglu & tuan, 2014; baxamusa & jalal, 2014; bonaimé et al., 2014; byoun, 2008; chan & chang, 2008; deangelo et al., 2011; drobetz et al., 2013; drobetz & wanzenried, 2006; haron et al., 2013; haron, 2014; hovakimian & li, 2011). in a study on target capital structure, abdeljawad et al. (2013) found that over-leveraged firms have shown faster adjustment compared to under-leveraged firms. additionally, firms that are far from the target capital structure have a faster adjustment than those close to the target. this is further supported by haron et al. (2013), whereby a negative relationship between the speed of adjustment and distance from target leverage was identified. hence, there is fast adjustment for firms that are far from the target capital structure. the results are consistent with the dynamic trade-off theory. the researchers have also observed that the firms in malaysia are under-adjusted because the required adjustment will be below the target within a year. moreover, the study suggested that firm-specific factors (firm size and profitability) have significantly affected the speed of adjustment for firms in malaysia. the same authors obtained similar results on the speed of adjustment and target capital structure for firms in thailand. haron (2014) proposed that property firms do practice target capital structure that influences firm characteristics such as profitability, liquidity, asset determinants of capital structure for malaysian shariah-compliant firms: 45-74 51 structure, firm size, non-debt tax shield, and growth opportunity. property firms are partially adjusted from time to time and supports the dynamic trade-off theory, while the capital structure decisions of property firms are influenced by the pecking order and market timing theories. with regards to the malaysian shariah-compliant securities, haron and ibrahim (2012) have observed that there exists target capital structure for shariah-compliant firms in malaysia, for both firm specific and country-specific factors that have played important roles and are related to the theory (based on table 1). the authors have proposed that firms would need to readjust with a faster speed of adjustment in order for firms to be under-adjusted. however, such rapid adjustment towards target leverage suggests the existence of the dynamic trade-off theory. other than that, arioglu and tuan (2014) have concluded that the speed of adjustment is estimated to be approximately 29%. the results for the adjustment speed are in line with the prediction made from the trade-off theory, and have suggested that firms must follow their target capital structures when their leverage ratios deviate from their targets. in a separate study on target capital structure, baxamusa and jalal (2014) found that an increase in one percent of a country’s protestant religiosity can decrease 0.4 percent of leverage and frequent debt issuance. the study indicates that religiosity could significantly affect the firms’ adjustment speed toward the target capital structure. the study noted that the differences in leverage within the u.s. are similar to the behavior of firms in many catholic and protestant countries outside of the u.s. numerous studies have been conducted to investigate market reactions on share repurchases and capital structure policy. to further investigate the relationship between stock returns to repurchase announcements and capital structure policy, bonaimé et al. (2014) carried out a series of experiments. the authors evaluated the benefits of the trade-off theory under levered firms on share repurchase as the repurchased equity enabled firms to move towards the optimal debt ratio. on the other hand, the market timing theory benefited undervalued firms that took advantage of the mispricing opportunities from repurchased equities. capital structure adjustments that require the repurchasing of overvalued stock have been shown to be less beneficial and more costly. in a study conducted by chan and chang (2008), firm-specific stock return variation is a significant adjustment towards target capital structure. this adjustment supports the trade-off theory for financing decisions in taiwanese firms. furthermore, high or low firm-specific stock return variation supports the pecking order and market timing theory. the results suggested that firmspecific stock variation provided an understanding to capital structure decisions. corporate financing decisions are mostly associated with the efficiency of capital markets. according to byoun (2008), the speed of adjustment happens when firms have a higher or lower debt, or when firms face a financial surplus or deficit. thus, the study concluded that firms will move to target capital structures when they experience financial deficits or surplus. 52 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 table 1. predicted signs and supporting theories for each explanatory variables no. variables predicted sign supporting theory 1. non-debt tax shield (-) (+) static trade-off 2. tangibility (-) (+) agency theory static trade-off 3. profitability (-) (+) pecking order static trade-off 4. business risk (-) static trade-off 5. firm size (-) static trade-off 6. growth opportunities (-) (+) agency theory pecking order 7. liquidity (-) pecking order 8. share price performance (-) market timing source: haron and ibrahim (2012) 3. data and model of analysis the sample for this study comprises of 192 shariah-compliant companies that were listed in bursa malaysia from 1999 to 2017. out of a total of 1087 shariahcompliant firms that were listed between the years 2007 and 2016, only 200 (including financial sector) shariah-compliant firms were consistently listed from 2007 to 2016. after the arrangement, 192 firms were then selected to be included into the sample for this study as the data on these firms were fully documented (refer table 2). the data was sourced from the thomson reuters database that recorded the company’s profile on a annual basis. listed companies from the financial sector were excluded from the sample (refer table 2) due to its exclusive features in financial statements and business activities (ali, ibrahim, mohammad, zain, & alwi, 2009). table 2. structure of the panel data no. sector no. of records on shariahcompliant firms maintain for the 10 years from 2007 until 2016 no. of observation 1. consumer products 35 665 2. industrial products 59 1121 3. construction 19 361 4. trading services 35 665 (continued) determinants of capital structure for malaysian shariah-compliant firms: 45-74 53 no. sector no. of records on shariahcompliant firms maintain for the 10 years from 2007 until 2016 no. of observation 5. properties 16 304 6. plantation 18 342 7. technology 7 133 8. infrastructure 3 57 total 192 3648 table 3 presents the variables used in this study with the measurement for each variable. in this study, leverage 1 (total conventional debt to total assets) and leverage 2 (total debt to total assets) were applied as dependent variables. this study used two measurements of leverage, where leverage 1 is defined as total conventional debt to total assets and leverage 2 is the total debt to total assets. the main objective for the calculation of leverage 1 is due to the 33 percent threshold for financial ratio benchmarks that was set by the securities commission malaysia which consists of debt to total assets, where debt only includes conventional debt. therefore, the calculation of leverage 1 in this study consists of only conventional debt while leverage 2 includes both islamic debt and conventional debt. independent variables consist of non-debt tax shield, tangibility, profitability, business risk, firm size, growth opportunities, liquidity, and share price performance. the explanatory variables are firm-specific, and are in line with previous studies (haron et al., 2013; haron, 2016; haron & ibrahim, 2011). table 3. variables used in the measure of leverage and its explanatory variables no. variable measurement leverage: 1. leverage 1 (lev1) total conventional debt over total assets 2. leverage 2 (lev2) total debt over total assets explanatory variable: 1. non-debt tax shield (ndts) annual depreciation expenses over total assets 2. tangibility (tang) net fixed assets over total assets 3. profitability (profit) ebit over total assets 4. business risk (risk) yearly change of ebit 5. firm size (size) natural logarithm of total assets 6. growth opportunities (growth) market value of equity to book value of equity 7. liquidity (liquidity) current assets over current liabilities 8. share price performance (spp) first difference of the year end share price 54 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 this study employs the panel data method to examine the existence of target capital structure and identify the factors that affect the capital structure for shariah-compliant firms for each sector. this method allows the elimination of unobservable heterogeneity for each observation in the sample of the study. panel data analysis was carried out by using generalised method of moment (gmm). the regression models for the panel data must comply with some assumptions; either it is linear, unbiased, lag structure or contained important results prior to testing the model. to examine the determinants of capital structure, this research used the following model: levi,t = αi,t + β1ndtsi,t + β2tangi,t + β3profiti,t + β4riski,t + β5sizei,t + β6growthi,t + β7liquidityi,t + β8sppi,t + εi (1) where levi,t is the leverage ratio of firm i at time t; ndtsi,t is a non-debt tax shield of firm i at time t; tangi,t is the tangibility of firm i at time t; profiti,t is the profitability of firm i at time t; riski,t represents a business risk of firm i at time t; sizei,t is the firm size of firm i at time t; growthi,t is the growth opportunities of firm i at time t; liquidityi,t is the liquidity of firm i at time t; and sppi,t is the share price performance of firm i at time t. furthermore, the main advantage of using this model is that it allows for the relationship between non-debt tax shield, tangibility, profitability, business risk, firm size, growth opportunities, liquidity, share price performance, and capital structure to be dynamic in nature. this study uses the gmm estimator proposed by arellano & bond (1991) to investigate the effect of non-debt tax shield, tangibility, profitability, business risk, firm size, growth opportunities, liquidity, and share price performance on capital structure. thus, the dynamic regression model that incorporated the panels containing many firms and a small number of time periods is presented as follows: levi,t levi,t-1 = δ (lev*i,t levi,t-1) (2) therefore, the new dynamic regression model is as follows: lev*i,t = δα + (1 – δ) levi,t-1 + δβ1ndtsi,t + δβ2tangi,t + δβ3profiti,t + δβ4riski,t + δβ5 sizei,t + δβ6growthi,t + δβ7liquidityi,t + δβ8sppi,t + λi + ηt + εit (3) whereby, δ is the adjustment speed that represents the magnitude of adjustment from actual to target capital structure. the δ is between 0 and 1. if δi,t = 0, there is no adjustment to the target capital structure. however, when δi,t < 1, an adjustment is required to attain the target, while if δi,t >1, this indicates that the firms are over-adjusted, and therefore requires the necessary adjustments which determinants of capital structure for malaysian shariah-compliant firms: 45-74 55 may still not be enough to reach the optimal level. to solve the heterogeneity bias, error terms are denoted as λi and ηt, representing the unobserved individual specific effect (such as firm and time). 4. empirical results this section presents the results of the descriptive statistics (table 4) and estimations for the dynamic regression model using gmm estimations on data sets described above over the period from 1999 to 2017. the results are divided into eight columns, representing the eight sectors (eight columns are as such: column 1 (consumer products); column 2 (industrial products); column 3 (construction); column 4 (trading services); column 5 (properties); column 6 (plantation); column 7 (technology); and column 8 (infrastructure)). table 5 records the gmm results at first different for leverage 1 (total conventional debt to total assets), while table 6 records the gmm results at first different for leverage 2 (total debt and total assets). table 7 depicts the impact of target capital structure maturity in speed of adjustment. table 8 shows the impact on leverage 1 (total conventional debt to total assets) of revised screening methodology. table 9 presents the impact on leverage 2 (total debt to total assets) of revised screening methodology. lastly, table 10 presents the summary of target capital structure maturity in speed of adjustment in implementation of revised screening methodology on nov 2013. 4.1 descriptive statistics table 4 presents the descriptive statistics such as mean, standard deviation, variance, minimum and maximum for each sector, which are consumer products, industrial products, construction, trading & services, properties, plantation, technology and infrastructure. the mean for leverage 1, leverage 2, profitability, size of firm and growth opportunities for infrastructure are 20.12%, 25.82%, 10.37%, 19.46 and 75.53% respectively. these values are higher than those of other industries such as consumer products, industrial products, construction, trading & services, properties, plantation and technology. the technology sector has the lowest mean for both leverage 1 and leverage 2, which is 9.93%. the lowest average profitability and growth for trading & services sector are 4.14% and -515.18% respectively, while the size of firm under consumer products is 15.91. the minimum for leverage 1 of the infrastructure sector is 0% and its maximum is 63.32%, while its standard deviation is at 21.26%. the minimum for leverage 1 of the technology sector is 0% and its maximum is 41.99%, while the standard deviation is 11.46%. this indicates that the leverage 1 for infrastructure is more volatile than technology. on the other hand, the minimum for leverage 2 of 56 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 infrastructure is 0% and its maximum is 66.06%, while the standard deviation is at 23.86%. the minimum for leverage 2 of technology is 0% and its maximum is 41.99%, while the standard deviation is 11.46%. this indicates that leverage 2 for infrastructure is more volatile than technology. additionally, the minimum and maximum profitability for infrastructure are -44.83% and 62.07% respectively, while the minimum and maximum profitability for trading & services are -248.91% and 37.83% respectively. the profitability for infrastructure is more volatile with a standard deviation of 18.36%, compared to trading & services that has a standard deviation of 14.16%. the minimum growth opportunities for infrastructure is 0% with its maximum at 107.784%, while the standard deviation is 201.34%. the growth opportunities for trading & services has a minimum of -36182.71% and a maximum of 182.856%, with a standard deviation of 1403.274%. in contrast to profitability, the growth opportunities for infrastructure is less volatile than trading & services. the minimum and maximum values for size of firm of infrastructure are 0% and 22.52% respectively, while the minimum and maximum values for size of firm of consumer products are 0% and 23.63% respectively. the size of firm for infrastructure is less volatile with a standard deviation of 6.7435, as compared to consumer products that has a standard deviation of 7.5248. the highest average business risk under consumer product is 159.067%, while the lowest average business risk for trading & services is -1571.13%. the minimum business risk for consumer product is -679.857% and the maximum is 4726.667%, while the standard deviation is 186.214%. the business risk for trading & services has a minimum of -1046880% and a maximum of 738.41%, while the standard deviation is at 40596.42%. this indicates that the business risk for consumer product is less volatile than trading & services. the average non-debt tax shield (ndts) for properties is -0.05%, which is higher than those from the other industries (consumer products, industrial products, construction, trading & services, plantation, technology and infrastructure). conversely, the infrastructure sector has the lowest average ndts which is -5.14%. the minimum and maximum ndts for properties are -1.01% and 0% respectively, while the minimum and maximum ndts for infrastructure are -33.13% and 0% respectively. the standard deviation for ndts of properties (0.15%) indicates that this sector is less volatility when compared to infrastructure (7.57%). in addition, the averages for the tangibility and liquidity of plantation are 52.49% and 295.49% respectively, which are lower than the values for properties (tangibility) and infrastructure (liquidity) respectively. the minimum and maximum tangibility values for plantation are 0% and 94.21% respectively, while the minimum and maximum tangibility values for properties are 0% and 70.94% respectively. the standard deviation for tangibility of plantation is 28.03%, while the standard deviation value for properties is 16.79%. this suggests that the tangibility for plantation are more volatile than properties. similarly, the minimum, maximum and standard determinants of capital structure for malaysian shariah-compliant firms: 45-74 57 deviation of liquidity for plantation (0%, 641.519% and 639.24% respectively) imply that this sector is more volatile than infrastructure (0%, 459.09% and 116.13% respectively). finally, the average share price performance (spp) for technology is 18.91%, which is higher than properties with a value of 8.47%. the minimum spp for technology is -82.03%, with a maximum at 384.8% and standard deviation at 74.13%. on the other hand, properties has a minimum spp of -83.96%, a maximum spp of 174.28% and a standard deviation of 43.35%. this indicates that the spp for technology is more volatile than properties. based on the descriptive statistic, the preliminary observation suggests that the infrastructure sector poses higher leverage with higher debt ratios than the other sectors. however, due to the high debt, the profitability, size of firm and business growth are higher and consistent than the other sectors. the results reveal that the large firms tend to raise higher capital with greater business risk which tends to generate larger profitability. additionally, most of the variables indicate that variance is higher within standard deviation, as compared to between standard deviations. this further implies that the time series variation is more dominant than the cross-sectional variation. therefore, the use of the panel estimation method is applicable to this research. in addition, the results for the mean of leverage 1 (total conventional debt to total assets) and leverage 2 (total debt to total assets) for all sectors is less than 33 percent of the financial benchmarks set by the security commission malaysia (sc), and subsequently validates that all firms for each sector has fulfilled the requirements by the sc to be listed as shariah-compliant firm. table 4. descriptive statistics variables mean overall standard deviation between standard deviation within standard deviation minimum maximum 1) consumer products lev1 0.1276 0.1368 0.1040 0.0905 0.0000 0.6191 lev2 0.1299 0.1380 0.1048 0.0914 0.0000 0.6191 ntds -0.0088 0.0183 0.0150 0.0108 -0.0960 0.0000 tang 0.3351 0.2254 0.1452 0.1741 0.0000 0.9377 profit 0.0778 0.0803 0.0478 0.0651 -0.7108 0.3517 risk 0.5364 18.6214 3.9544 18.2083 -67.9857 472.6667 size 15.9067 7.5248 2.8654 6.9738 0.0000 23.6261 growth 0.0586 0.3231 0.1841 0.2672 0.0000 6.8471 liquidity 2.6080 3.0841 2.4004 1.9764 0.0000 23.2557 spp 0.0913 0.3852 0.0805 0.3770 -0.7458 2.5797 (continued) 58 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 variables mean overall standard deviation between standard deviation within standard deviation minimum maximum 2) industrial products lev1 0.1715 0.2021 0.1049 0.1732 0.0000 2.8224 lev2 0.1744 0.2032 0.1065 0.1735 0.0000 2.8224 ntds -0.0062 0.0148 0.0125 0.0081 -0.0918 0.0000 tang 0.3726 0.2199 0.1397 0.1707 0.0000 0.9750 profit 0.0555 0.0821 0.0426 0.0704 -0.6320 0.8501 risk -0.2055 5.3301 1.4787 5.1244 -138.8327 34.4225 size 17.4245 6.6628 2.5109 6.1798 0.0000 23.5299 growth 0.0833 0.5556 0.1379 0.5385 -15.6658 3.0995 liquidity 2.2308 2.5465 1.5273 2.0468 0.0000 26.7954 spp 0.1464 0.7854 0.1421 0.7727 -0.8832 14.6000 3) construction lev1 0.1519 0.1477 0.0977 0.1129 0.0000 0.6844 lev2 0.1574 0.1504 0.1006 0.1141 0.0000 0.6844 ntds -0.0012 0.0042 0.0025 0.0034 -0.0354 0.0000 tang 0.1575 0.1354 0.0976 0.0964 0.0000 0.7997 profit 0.0516 0.0952 0.0419 0.0859 -0.4725 1.1042 risk -0.2751 6.7087 1.5190 6.5433 -119.2114 33.5026 size 17.4614 7.0204 3.0796 6.3464 0.0000 23.7627 growth 0.0438 0.0876 0.0543 0.0698 -0.2506 0.6010 liquidity 1.7516 1.2221 0.8009 0.9402 0.0000 8.5722 spp 0.1061 0.5654 0.1068 0.5558 -0.8317 4.7917 4) trading and services lev1 0.1876 0.2347 0.1071 0.2096 0.0000 3.4515 lev2 0.1969 0.2368 0.1126 0.2092 0.0000 3.4515 ntds -0.0074 0.0196 0.0184 0.0072 -0.1019 0.0000 tang 0.3322 0.2387 0.1700 0.1699 0.0000 0.9260 profit 0.0414 0.1416 0.0656 0.1260 -2.4891 0.3783 risk -157.113 4059.642 931.3978 3954.328 -104688 73.841 size 18.5554 6.7675 2.9176 6.1251 0.0000 25.6129 growth -5.1518 140.3274 32.2427 136.6761 -3618.271 18.2856 liquidity 2.2058 7.4910 2.5230 7.0656 0.0000 154.4811 spp 0.1546 1.9363 0.4246 1.890427 -0.9070 48.2491 (continued) determinants of capital structure for malaysian shariah-compliant firms: 45-74 59 variables mean overall standard deviation between standard deviation within standard deviation minimum maximum 5) properties lev1 0.1836 0.1449 0.0893 0.1162 0.0000 0.7516 lev2 0.1866 0.1461 0.0922 0.1156 0.0000 0.7516 ntds -0.0005 0.0015 0.0012 0.0010 -0.0101 0.0000 tang 0.1255 0.1679 0.1311 0.1096 0.0000 0.7094 profit 0.0440 0.0624 0.0372 0.0509 -0.1654 0.3452 risk 0.0050 4.6680 1.0683 4.5515 -18.3245 70.3846 size 17.8796 6.5083 2.0309 6.2031 0.0000 22.6127 growth 0.6620 3.0017 2.1379 2.1705 0.0000 28.8765 liquidity 2.2508 2.1443 1.4980 1.5771 0.0000 15.2700 spp 0.0847 0.4335 0.0667 0.42870 -0.8396 1.7428 6) plantation lev1 0.1285 0.1567 0.0869 0.1319 0.0000 0.9104 lev2 0.1391 0.1638 0.0923 0.1370 0.0000 0.9104 ntds -0.0023 0.0081 0.0069 0.0044 -0.0489 0.0019 tang 0.5249 0.2803 0.1141 0.2574 0.0000 0.9421 profit 0.0604 0.0936 0.0448 0.0829 -0.9916 0.3595 risk 0.0179 9.3566 2.1168 9.1270 -128.8638 99.8636 size 17.7559 7.17912 2.9795 6.5674 0.0000 23.7693 growth 0.1028 0.3479 0.2445 0.2538 0.0000 3.9792 liquidity 2.9549 6.3924 3.7600 5.2413 0.0000 64.1519 spp 0.1068 0.3467 0.0510 0.3432 -0.6376 1.6452 7) technology lev1 0.0993 0.1146 0.0684 0.0954 0.0000 0.4199 lev2 0.0993 0.1146 0.0684 0.0954 0.0000 0.4199 ntds -0.0170 0.0318 0.0293 0.0164 -0.1113 0.0000 tang 0.3963 0.2400 0.1843 0.1681 0.0000 0.89512 profit 0.0569 0.0909 0.0421 0.0820 -0.3978 0.2719 risk -0.2383 2.9761 0.5339 2.9345 -16.4397 17.772 size 16.7523 7.1421 3.2935 6.4530 0.0000 21.4112 growth 0.1197 0.1667 0.1014 0.1375 0.0000 0.7528 liquidity 1.7608 1.2648 0.6531 1.1097 0.0000 5.9838 spp 0.1891 0.7413 0.1304 0.7313 -0.8203 3.848 (continued) 60 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 variables mean overall standard deviation between standard deviation within standard deviation minimum maximum 8) infrastructure lev1 0.2012 0.2126 0.1509 0.1725 0.0000 0.6332 lev2 0.2582 0.2386 0.2440 0.1285 0.0000 0.6606 ntds -0.0514 0.0757 0.0738 0.0451 -0.3313 0.0000 tang 0.2623 0.2962 0.2909 0.1740 0.0000 0.8291 profit 0.1037 0.1836 0.1527 0.1338 -0.4483 0.6207 risk 0.4184 3.5055 1.0575 3.3956 -14.1735 18.0918 size 19.4600 6.7435 1.0639 6.6863 0.0000 22.5233 growth 0.7553 2.0134 0.8599 1.8847 0.0000 10.7784 liquidity 1.5592 1.1613 0.9932 0.8241 0.0000 4.5909 spp 0.1596 0.3709 0.0883 0.3637 -0.6755 1.1867 4.2 determinants of leverage 1 (total conventional debt to total assets) and leverage 2 (total debt to total assets) from the results of the dynamic regression model, the error terms are assumed to be independent and homoscedastic across companies and over time. specifically, the estimated coefficient of lagged leverage 1 (total conventional debt to total assets) is significant at 1 and 5 percent significance level for all sectors (refer table 5 and 6). these results assert that the model is a dynamic model. moreover, the lagged dependent variables (lagged leverage 1 and leverage 2) have positive significance, with the speed of adjustment deduced to be approximately 2 years. this concludes that shariah-compliant firms close approximately 30% to 70% of the gap between current and target capital structure within one year (refer table 7). therefore, it is beneficial for firms to get closer to the target capital structure in order to grow faster at a maximum rate without changes in financial leverage. using the one-step system gmm results, for ntds, only consumer products, trading & services, and properties (column 1, 4, and 5) are significant at 1 percent and 10 percent significance level. ntds is positively significant at 1 percent for the consumer products sector, however it is negatively significant to the leverage for trading services and properties. the negative effect is due to the static trade-off, whereby firms with a higher ntds than cash flow expected will be able to reduce their debt in the capital structure. the results are similar to both leverage 1 (total conventional debt to total assets) and leverage 2 (total debt to total assets). with regards to tangibility, both leverage 1 (total conventional debt to total assets) and leverage 2 (total debt to total assets) have the same results. nonetheless, tangibility has a positive effect on leverage for consumer products, trading services, and properties. for the technology sector, tangibility has a negative effect on leverage. the positive relationship determinants of capital structure for malaysian shariah-compliant firms: 45-74 61 ta bl e 5. g m m f ir st d if fe re nc e by e ac h se ct or fo r l ev er ag e 1 (t ot al c on ve nt io na l d eb t t o to ta l a ss et s) v a r ia b l e s (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) c on su m er p ro du ct in du st ri al p ro du ct s c on st ru ct io n tr ad in g se rv ic es pr op er tie s pl an ta tio n te ch no lo gy in fr as tr uc tu re l e v 1 (1) 0. 52 3* ** (0 .0 31 5) 0. 58 7* ** (0 .0 22 1) 0. 45 2* ** (0 .0 42 6) 0. 38 2* ** (0 .0 35 4) 0. 46 9* ** (0 .0 45 1) 0. 59 8* ** (0 .0 33 6) 0. 66 0* ** (0 .0 68 5) 0. 56 1* ** (0 .1 13 ) n t d s 1. 75 5* ** (0 .2 93 ) -0 .4 16 (0 .7 70 ) -0 .7 04 (1 .7 36 ) -2 .9 18 ** (1 .3 20 ) -1 1. 03 ** (4 .9 08 ) 1. 89 8 (1 .5 17 ) 0. 48 8 (0 .3 60 ) -0 .3 28 (0 .5 05 ) ta n g 0. 09 23 ** * (0 .0 27 8) 0. 05 35 (0 .0 46 9) 0. 04 60 (0 .0 55 0) 0. 58 0* ** (0 .0 72 2) 0. 15 2* ** (0 .0 46 4) 0. 01 17 (0 .0 37 9) -0 .1 19 ** (0 .0 58 9) -0 .0 41 0 (0 .1 53 ) pr o fi t -0 .3 26 ** * (0 .0 44 6) -0 .7 17 ** * (0 .0 60 7) -0 .1 74 ** * (0 .0 53 6) -0 .5 44 ** * (0 .0 54 2) -0 .3 18 ** * (0 .0 99 1) -0 .4 25 ** * (0 .0 73 6) -0 .1 87 ** * (0 .0 71 1) -0 .1 88 (0 .2 13 ) r is k 0. 00 01 (9 .3 1e -0 5) 0. 00 4* ** (0 .0 00 6) -0 .0 00 7 (0 .0 00 5) -2 .0 8e -0 7 (2 .0 0e -0 6) -0 .0 00 3 (0 .0 00 8) 0. 00 02 (0 .0 00 4) -0 .0 02 2 (0 .0 01 7) -0 .0 05 2 (0 .0 08 0) si z e 0. 00 92 ** * (0 .0 00 8) 0. 01 15 ** * (0 .0 01 3) 0. 00 89 ** * (0 .0 01 0) -0 .0 00 3 (0 .0 01 7) 0. 00 79 ** * (0 .0 01 0) 0. 00 86 ** * (0 .0 01 5) 0. 00 69 ** * (0 .0 01 9) 0. 01 04 ** (0 .0 05 2) g r o w t h -0 .0 01 3 (0 .0 06 7) -0 .0 02 2 (0 .0 22 9) -0 .0 10 8 (0 .0 59 8) -6 .7 3e -0 5 (5 .9 0e -0 5) 0. 00 77 ** * (0 .0 01 9) -0 .0 32 4* * (0 .0 15 6) -0 .0 14 4 (0 .0 41 0) 0. 01 25 (0 .0 12 0) l iq u id it y -0 .0 10 2* ** (0 .0 01 8) -0 .0 18 3* ** (0 .0 02 4) 0. 00 68 (0 .0 06 4) 0. 00 08 (0 .0 01 9) -0 .0 02 5 (0 .0 03 6) -0 .0 01 2* (0 .0 00 7) -0 .0 05 3 (0 .0 07 5) -0 .0 10 1 (0 .0 30 2) sp p -0 .0 06 3 (0 .0 05 1) 0. 00 64 (0 .0 04 4) -0 .0 11 0* (0 .0 06 4) -0 .0 17 9* ** (0 .0 03 4) -0 .0 10 1 (0 .0 09 0) -0 .0 18 1* (0 .0 10 6) -0 .0 20 1* * (0 .0 08 5) -0 .1 14 ** (0 .0 54 1) sa rg an t es t 28 7. 10 0. 00 00 55 3. 87 0. 00 00 31 2. 95 0. 00 00 34 6. 38 0. 00 00 17 5. 58 0. 08 35 22 5. 29 0. 00 01 11 1. 38 0. 15 09 50 .9 2 0. 16 27 n ot es : a ll m od el s ar e es tim at ed u si ng th e a re lla no a nd b on d dy na m ic p an el g m m e st im at io ns (s ta ta x ta bo nd c om m an d) . t he v ar ia bl es a re d efi ne d as fo llo w s: l e v 1 = to ta l c on ve nt io na l d eb t/t ot al a ss et s; n t d s = a nn ua l d ep re ci at io n e xp en se s/ to ta l a ss et s; t a n g = n et f ix ed a ss et s/ to ta l a ss et s; pr o fi t = e b it /t ot al a ss et s; r is k = e b it o ve r t ot al a ss et s; s iz e = n at ur al l og o f to ta l a ss et s; g r o w t h = m ar ke t v al ue o f e qu ity /b oo k v al ue of e qu ity ; l iq u id it y = c ur re nt a ss et s ov er c ur re nt l ia bi lit ie s; s pp = f ir st d if fe re nc e of th e y ea r e nd s ha re p ri ce . f ig ur es in th e pa re nt he se s ar e tst at is tic s. * ** , * *, a nd * in di ca te s ig ni fic an ce a t t he 1 % , 5 % , a nd 1 0% le ve ls , r es pe ct iv el y 62 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 ta bl e 6. g m m f ir st d if fe re nc e by e ac h se ct or fo r l ev er ag e 2 (t ot al d eb t t o to ta l a ss et s) v a r ia b l e s (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) c on su m er p ro du ct in du st ri al p ro du ct s c on st ru ct io n tr ad in g se rv ic es pr op er tie s pl an ta tio n te ch no lo gy in fr as tr uc tu re l e v 2 (1) 0. 54 3* ** (0 .0 30 8) 0. 58 9* ** (0 .0 22 2) 0. 44 5* ** (0 .0 40 6) 0. 38 2* ** (0 .0 35 3) 0. 46 8* ** (0 .0 44 6) 0. 60 5* ** (0 .0 33 0) 0. 66 0* ** (0 .0 68 5) 0. 31 1* * (0 .1 33 ) n t d s 1. 81 6* ** (0 .2 89 ) -0 .2 28 (0 .7 71 ) -0 .4 20 (1 .7 26 ) -2 .6 33 ** (1 .2 96 ) -1 1. 32 ** (4 .8 23 ) 2. 11 3 (1 .4 82 ) 0. 48 8 (0 .3 60 ) -0 .4 35 (0 .3 38 ) ta n g 0. 07 67 ** * (0 .0 27 5) 0. 05 88 (0 .0 46 5) 0. 02 89 (0 .0 52 7) 0. 63 0* ** (0 .0 70 0) 0. 14 6* ** (0 .0 45 5) 0. 01 63 (0 .0 38 4) -0 .1 19 ** (0 .0 58 9) -0 .0 19 5 (0 .1 02 ) pr o fi t -0 .3 16 ** * (0 .0 43 4) -0 .7 19 ** * (0 .0 60 6) -0 .2 02 ** * (0 .0 53 1) -0 .5 25 ** * (0 .0 52 5) -0 .2 90 ** * (0 .0 97 2) -0 .4 21 ** * (0 .0 73 4) -0 .1 87 ** * (0 .0 71 1) -0 .3 29 ** (0 .1 51 ) r is k 0. 00 01 (9 .1 2e -0 5) 0. 00 34 ** * (0 .0 00 6) -0 .0 00 4 (0 .0 00 5) -2 .2 9e -0 7 (1 .9 4e -0 6) -0 .0 00 3 (0 .0 00 8) 0. 00 02 (0 .0 00 4) -0 .0 02 2 (0 .0 01 7) -0 .0 07 4 (0 .0 05 4) si z e 0. 00 96 ** * (0 .0 00 8) 0. 01 14 ** * (0 .0 01 3) 0. 00 95 ** * (0 .0 01 0) -0 .0 00 5 (0 .0 01 7) 0. 00 82 ** * (0 .0 00 9) 0. 00 86 ** * (0 .0 01 5) 0. 00 69 ** * (0 .0 01 9) 0. 00 92 ** * (0 .0 03 4) g r o w t h -0 .0 00 6 (0 .0 06 6) 0. 01 42 (0 .0 23 0) 0. 04 15 (0 .0 58 5) -7 .0 8e -0 5 (5 .7 2e -0 5) 0. 00 77 ** * (0 .0 01 9) -0 .0 02 1 (0 .0 15 5) -0 .0 14 4 (0 .0 41 0) 0. 02 09 ** (0 .0 08 6) l iq u id it y -0 .0 11 3* ** (0 .0 01 8) -0 .0 18 2* ** (0 .0 02 4) 0. 00 23 3 (0 .0 06 4) 0. 00 09 (0 .0 01 8) -0 .0 03 2 (0 .0 03 6) -0 .0 00 7 (0 .0 00 7) -0 .0 05 3 (0 .0 07 5) 0. 01 51 (0 .0 20 2) sp p -0 .0 04 02 (0 .0 04 92 ) 0. 00 70 9 (0 .0 04 36 ) -0 .0 09 82 (0 .0 06 35 ) -0 .0 18 0* ** (0 .0 03 31 ) -0 .0 09 49 (0 .0 08 81 ) -0 .0 20 0* (0 .0 10 4) -0 .0 20 1* * (0 .0 08 49 ) -0 .0 97 1* ** (0 .0 37 5) sa rg an t es t 29 6. 57 47 0. 00 00 56 1. 43 39 0. 00 00 31 1. 11 34 0. 00 00 35 3. 80 9 0. 00 00 17 8. 59 15 0. 06 20 21 8. 35 52 0. 00 03 11 1. 37 62 0. 15 09 63 .5 33 52 0. 01 75 n ot es : a ll m od el s ar e es tim at ed u si ng th e a re lla no a nd b on d dy na m ic p an el g m m e st im at io ns ( st at a xt ab on d co m m an d) . t he v ar ia bl es a re d efi ne d as f ol lo w s: l e v 2 = to ta l d eb t/t ot al a ss et s; n t d s = a nn ua l d ep re ci at io n e xp en se s/ to ta l a ss et s; t a n g = n et f ix ed a ss et s/ to ta l a ss et s; p r o fi t = e b it /t ot al a ss et s; r is k = e b it o ve r to ta l a ss et s; s iz e = n at ur al l og o f to ta l a ss et s; g r o w t h = m ar ke t v al ue o f e qu ity /b oo k v al ue o f e qu ity ; l iq u id it y = c ur re nt a ss et s ov er c ur re nt l ia bi lit ie s; s pp = f ir st d if fe re nc e of th e y ea r e nd s ha re p ri ce . f ig ur es in th e pa re nt he se s ar e tst at is tic s. * ** , * *, a nd * in di ca te s ig ni fic an ce a t t he 1 % , 5 % , a nd 1 0% le ve ls , r es pe ct iv el y. determinants of capital structure for malaysian shariah-compliant firms: 45-74 63 ta bl e 7. im pa ct o f t ar ge t c ap ita l s tr uc tu re m at ur ity in s pe ed a dj us tm en t (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) c on su m er p ro du ct in du st ri al p ro du ct s c on st ru ct io n tr ad in g se rv ic es pr op er tie s pl an ta tio n te ch no lo gy in fr as tru ct ur e l e v 1 (1) 0. 52 3* ** 0. 58 7* ** 0. 45 2* ** 0. 38 2* ** 0. 46 9* ** 0. 59 8* ** 0. 66 0* ** 0. 56 1* ** sp ee d of a dj us tm en t 0. 47 7 0. 41 3 0. 54 8 0. 61 8 0. 53 1 0. 40 2 0. 34 0 0. 43 9 y ea r 2. 10 2. 42 1. 82 1. 62 1. 88 2. 49 2. 94 2. 28 l e v 2 (1) 0. 54 3* ** 0. 58 9* ** 0. 44 5* ** 0. 38 2* ** 0. 46 8* ** 0. 60 5* ** 0. 66 0* ** 0. 31 1* * sp ee d of a dj us tm en t 0. 45 7 0. 41 1 0. 55 5 0. 61 8 0. 53 2 0. 39 5 0. 34 0 0. 68 9 y ea r 2. 19 2. 43 1. 80 1. 62 1. 88 2. 53 2. 94 1. 45 n ot es : * ** , * *, a nd * d en ot e si gn ifi ca nt a t 1 % , 5 % a nd 1 0% le ve ls , r es pe ct iv el y. t he c om pu ta tio n of s pe ed o f a dj us tm en t i s de ri ve d fr om o ne m in us th e co ef fic ie nt s on l e v 1 (1) a nd l e v 2 (1) . 64 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 indicates that the firms must have asset-backed debts, whereby debt must be lower than tangible assets for a firm to operate based on islamic principles. these results comply with the trade-off theory. this study identified a negative relationship between profitability and leverage for shariah-compliant firms for all sectors. the result is analogous to a previous study conducted on malaysian shariah-compliant securities (haron and ibrahim, 2012). this suggests that shariah-compliant firms with high profitability will use lower leverages in their financial activities. similarly, firm size for leverage 1 (total conventional debt to total assets) and leverage 2 (total debt to total assets) have the same results, for which firm size has a positive significant relationship to leverage. the results indicate that bigger firms will tend to generate more income and profit, however these firms would require more debt to support their investment activities. hence, these findings strongly support the pecking order theory. on the other hand, only several sectors have indicated a relationship between growth and leverage (leverage 1 and leverage 2). higher growth causes firms to raise more leverage, and vice versa. similarly, there are some sectors (leverage 1: consumer products, industrial products, and plantation; leverage 2: consumer products, and industrial products) where liquidity has a negative relationship to leverage. this negative relationship supports the pecking order theory, for which liquidity of a firm should be negatively associated with leverage. moreover, the negative relationship between liquidity and leverage shows that firms with highly liquid assets will be able to incur more debt that would enable the firms to continue to run their business. share price performance is found to have a negative impact on leverage which implies that firms will issue equity to debt when the firm’s share price increases. this result supports the market timing theory. 4.3 target capital structure and speed of adjustment, before and after revised screening methodology table 8 records the results of the target capital structure based on two scenarios, where the first scenario is before the revised screening methodology (year 19992013) and the second scenario is after the implementation of the new screening methodology (year 2014-2017). from the year 2014 to 2017, the estimated coefficient for the target leverage 1 (total conventional debt to total assets) is significant at the 1 percent significance level for all sectors. this indicates that the target leverage 1 (total conventional debt to total assets) was present within the malaysian shariah-compliant firms for all sectors. moreover, it was identified that all sectors adjust at a speed of approximately 0.3 to 0.7 to achieve long-term target leverage. the speed of adjustment can be explained as how quickly firms converge to the target capital structure from their current capital structure (haron & ibrahim, 2012). from the results, it is deduced that the speed of adjustment would take approximately 2 years for firms to achieve the target leverage from their current leverage (refer table 10). after the implementation determinants of capital structure for malaysian shariah-compliant firms: 45-74 65 ta bl e 8. im pa ct o n l ev er ag e 1 (t ot al c on ve nt io na l d eb t t o to ta l a ss et s) o f r ev is ed s cr ee ni ng m et ho do lo gy y ea r 1 99 9 – 20 13 v a r ia b l e s (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) c on su m er p ro du ct in du st ri al p ro du ct s c on st ru ct io n tr ad in g se rv ic es pr op er tie s pl an ta tio n te ch no lo gy in fr as tr uc tu re l e v 1 (1) 0. 64 3* ** 0. 57 1* ** 0. 50 5* ** 0. 33 1* ** 0. 47 5* ** 0. 68 0* ** 0. 60 8* ** 0. 57 5* ** (0 .0 48 1) (0 .0 26 2) (0 .0 57 5) (0 .0 40 0) (0 .0 55 2) (0 .0 40 9) (0 .0 79 2) (0 .1 43 ) n t d s 2. 22 0* ** -0 .5 74 -0 .3 58 -0 .8 08 -1 9. 92 ** * -0 .2 76 0. 67 1 -0 .0 22 8 (0 .3 81 ) (1 .0 04 ) (1 .8 35 ) (1 .8 95 ) (6 .1 47 ) (1 .9 88 ) (0 .4 44 ) (0 .6 43 ) ta n g 0. 13 1* ** 0. 00 62 9 0. 08 15 0. 85 8* ** 0. 11 6* 0. 00 26 -0 .1 82 ** * 0. 07 61 (0 .0 42 3) (0 .0 63 6) (0 .0 66 2) (0 .0 95 0) (0 .0 61 1) (0 .0 45 4) (0 .0 68 5) (0 .2 30 ) pr o fi t -0 .4 49 ** * -0 .9 08 ** * -0 .1 51 ** * -0 .5 11 ** * -0 .5 51 ** * -0 .4 09 ** * -0 .2 60 ** * -0 .4 33 * (0 .0 69 4) (0 .0 73 7) (0 .0 56 1) (0 .0 61 4) (0 .1 34 ) (0 .0 81 4) (0 .0 77 4) (0 .2 61 ) r is k 0. 00 21 ** 0. 00 72 ** * -0 .0 00 8 -2 .9 6e -0 7 -0 .0 00 3 0. 00 04 -0 .0 01 3 -0 .0 09 8 (0 .0 00 9) (0 .0 01 3) (0 .0 00 5) (2 .1 2e -0 6) (0 .0 00 9) (0 .0 00 4) (0 .0 01 7) (0 .0 09 5) si z e 0. 00 91 ** * 0. 01 69 ** * 0. 01 06 ** * -0 .0 04 0 0. 01 24 ** * 0. 01 41 ** * 0. 00 79 ** * 0. 14 8 (0 .0 01 2) (0 .0 02 1) (0 .0 01 4) (0 .0 03 3) (0 .0 02 3) (0 .0 02 2) (0 .0 02 3) (0 .1 19 ) g r o w t h 0. 00 53 -0 .0 12 7 -0 .0 00 1 -7 .7 9e -0 5 0. 01 02 ** * -0 .0 38 0* * 0. 01 21 0. 03 52 (0 .0 28 0) (0 .0 26 4) (0 .0 64 4) (6 .3 2e -0 5) (0 .0 02 5) (0 .0 18 5) (0 .0 49 1) (0 .0 28 2) l iq u id it y -0 .0 07 1* ** -0 .0 24 8* ** -0 .0 01 5 0. 00 22 -0 .0 01 2 -0 .0 01 3 0. 00 61 -0 .0 41 3 (0 .0 02 3) (0 .0 03 6) (0 .0 07 8) (0 .0 02 0) (0 .0 04 4) (0 .0 01 1) (0 .0 09 0) (0 .0 38 7) sp p -0 .0 07 9 0. 00 37 -0 .0 12 6* -0 .0 17 7* ** -0 .0 12 8 -0 .0 21 2* -0 .0 20 8* * -0 .0 97 5 (0 .0 05 7) (0 .0 05 9) (0 .0 06 6) (0 .0 03 6) (0 .0 09 9) (0 .0 11 4) (0 .0 09 1) (0 .0 60 3) (c on tin ue d) 66 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 y ea r 2 01 4 – 20 17 l e v 1 (1) 0. 14 4* * 0. 54 5* ** 0. 95 3 0. 56 7* * 1. 80 0 0. 76 7 0. 06 89 ** * 0. 00 00 (0 .0 65 4) (0 .1 44 ) (2 .5 80 ) (0 .2 70 ) (1 .7 35 ) (0 .5 68 ) (0 .0 00 0) (0 .0 00 0) n t d s 4. 95 3* ** 1. 57 1 -2 .9 79 -4 .7 76 ** * -1 0. 22 7. 68 3 0. 00 00 34 .2 5 (1 .4 96 ) (1 .0 14 ) (1 1. 96 ) (1 .3 12 ) (2 1. 54 ) (5 .9 58 ) (0 .0 00 0) (0 .0 00 0) ta n g -0 .1 82 ** * 0. 09 15 -1 .6 48 0. 22 1* 0. 38 4 -0 .4 00 * -0 .9 12 ** * -1 .8 13 (0 .0 64 9) (0 .0 57 5) (1 .6 09 ) (0 .1 30 ) (0 .3 61 ) (0 .2 31 ) (0 .0 00 0) (0 .0 00 0) pr o fi t -1 .3 98 ** * -0 .1 27 -0 .1 86 -0 .0 51 1 0. 33 6 -1 .7 35 * -0 .3 53 ** * 12 .9 0 (0 .1 70 ) (0 .1 51 ) (1 .8 93 ) (0 .3 57 ) (0 .9 19 ) (0 .9 31 ) (0 .0 00 0) (0 .0 00 0) r is k -1 .6 6e -0 5 -0 .0 00 2 -0 .0 07 4 -0 .0 27 0* * 0. 00 02 -0 .0 10 9 0. 11 9* ** (9 .7 9e -0 5) (0 .0 00 5) (0 .0 50 4) (0 .0 13 1) (0 .0 22 7) (0 .0 12 3) (0 .0 00 0) si z e 0. 02 16 ** * 0. 00 63 ** * 0. 02 16 ** * 0. 01 03 ** * -0 .0 04 5 0. 02 25 ** * 0. 03 27 ** * (0 .0 01 8) (0 .0 01 6) (0 .0 04 8) (0 .0 02 9) (0 .0 08 7) (0 .0 07 9) (0 .0 00 0) g r o w t h 0. 00 51 9 0. 02 64 0. 74 9 -0 .1 67 0. 26 3 1. 40 5 -0 .2 57 ** * (0 .0 05 5) (0 .0 67 7) (0 .9 14 ) (0 .1 33 ) (0 .2 90 ) (1 .0 26 ) (0 .0 00 0) l iq u id it y -0 .0 21 4* ** -0 .0 08 3* ** 0. 01 73 -0 .0 56 8* ** 0. 08 25 * -0 .0 02 5 -0 .0 43 7* ** (0 .0 02 8) (0 .0 02 3) (0 .0 57 8) (0 .0 16 2) (0 .0 46 8) (0 .0 01 7) (0 .0 00 0) sp p -0 .0 24 6 0. 01 56 ** -0 .0 54 0 0. 04 06 0. 09 11 0. 22 5 -0 .0 51 5* ** 0. 00 00 (0 .0 31 9) (0 .0 07 6) (0 .1 71 ) (0 .0 52 9) (0 .1 34 ) (0 .2 56 ) (0 .0 00 0) (0 .0 00 0) n ot es : a ll m od el s ar e es tim at ed u si ng th e a re lla no a nd b on d dy na m ic p an el g m m e st im at io ns ( st at a xt ab on d co m m an d) . t he v ar ia bl es a re d efi ne d as fo llo w s: l e v 2 = to ta l d eb t/t ot al a ss et s; n t d s = a nn ua l d ep re ci at io n e xp en se s/ to ta l a ss et s; t a n g = n et f ix ed a ss et s/ to ta l a ss et s; p r o fi t = e b it /t ot al a ss et s; r is k = e b it o ve r to ta l a ss et s; s iz e = n at ur al l og o f to ta l a ss et s; g r o w t h = m ar ke t v al ue o f e qu ity /b oo k v al ue o f e qu ity ; l iq u id it y = c ur re nt a ss et s ov er c ur re nt l ia bi lit ie s; s pp = f ir st d if fe re nc e of th e y ea r e nd s ha re p ri ce . f ig ur es in th e pa re nt he se s ar e tst at is tic s. ** *, * *, a nd * in di ca te s ig ni fic an ce a t t he 1 % , 5 % , a nd 1 0% le ve ls , r es pe ct iv el y. determinants of capital structure for malaysian shariah-compliant firms: 45-74 67 of the new screening methodology (from the year 2014 to 2017), there are only four sectors (sector 1 (consumer products), 2 (industrial products), 4 (trading services) and 7 (technology) that have significant results at the 1% and 5% significance level of the estimated coefficient for the target leverage 1 (total conventional debt to total assets). furthermore, the speed of adjustment is approximately 0.4 to 0.9, and would take roughly 1 to 2 years for firms to reach the target leverage. table 9 shows all sectors from the year 1999 to 2013 for the malaysian shariah-compliant firms to achieve the target leverage 2 (total debt to total assets). the speed of adjustment to reach the target leverage is approximately 0.3 to 0.7. moreover, it takes about 1 to 3 years for all sectors to achieve the target leverage level. from the year 2014 to 2017, the results for leverage 1 (total conventional debt to total assets) are significant for the sectors 1, 2, 4, and 7, whereby the results are at the 1% and 5% significance level for the speed of adjustment to target leverage 2 (total debt to total assets). the malaysian shariah-compliant firms would take around 1 to 2 years to achieve target leverage 2 (total debt to total assets) from the firms current leverage. this concludes that the malaysian shariah-compliant firms for all sectors close approximately 30% to 70% of the gap between current and target leverage of leverage 1 (total conventional debt to total assets) and leverage 2 (total debt to total assets) from the year 1999 to 2013 (before the revised screening methodology was introduced). however, from the year 2014 to 2017, only a few sectors reached the target leverage (after the revised screening methodology was introduced). this suggests that the speed of adjustment for malaysian shariah-compliant firms to achieve the target leverage from their current leverage was substantially affected when the securities commission malaysia (sc) revised the screening methodology in 2013. based on table 10, the findings imply that after the implementation of the revised screening methodology in november 2013, the speed of adjustment improved. the revised screening methodology led to faster adjustment to the target leverage level. as can be observed in the consumer products sector, the speed of adjustment improved from approximately 2 years (year 1999-2013) to 1 year (2014-2017). in addition, the level of conventional debt was found to be lower after the introduction of the revised screening methodology. 68 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 ta bl e 9. im pa ct o n l ev er ag e 2 (t ot al d eb t t o to ta l a ss et s) o f r ev is ed s cr ee ni ng m et ho do lo gy y ea r 1 99 9 2 01 3 v a r ia b l e s (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) c on su m er p ro du ct in du st ri al p ro du ct s c on st ru ct io n tr ad in g se rv ic es pr op er tie s pl an ta tio n te ch no lo gy in fr as tr uc tu re l e v 2 (1) 0. 67 5* ** 0. 57 1* ** 0. 51 5* ** 0. 32 3* ** 0. 48 5* ** 0. 66 7* ** 0. 60 8* ** 0. 29 4* (0 .0 48 0) (0 .0 26 2) (0 .0 56 9) (0 .0 39 9) (0 .0 56 2) (0 .0 39 1) (0 .0 79 2) (0 .1 55 ) n t d s 2. 29 0* ** -0 .2 51 -0 .3 17 -0 .5 91 -1 9. 11 ** * -0 .4 86 0. 67 1 -0 .1 61 (0 .3 74 ) (1 .0 01 ) (1 .8 43 ) (1 .8 37 ) (6 .1 73 ) (1 .9 34 ) (0 .4 44 ) (0 .3 78 ) ta n g 0. 10 9* ** 0. 02 05 0. 05 81 0. 90 5* ** 0. 12 8* * -0 .0 01 32 -0 .1 82 ** * 0. 20 5 (0 .0 40 6) (0 .0 62 7) (0 .0 65 3) (0 .0 91 0) (0 .0 60 9) (0 .0 45 3) (0 .0 68 5) (0 .1 35 ) pr o fi t -0 .4 27 ** * -0 .9 13 ** * -0 .1 81 ** * -0 .4 92 ** * -0 .5 07 ** * -0 .4 03 ** * -0 .2 60 ** * -0 .6 18 ** * (0 .0 66 3) (0 .0 73 3) (0 .0 56 2) (0 .0 59 2) (0 .1 33 ) (0 .0 79 5) (0 .0 77 4) (0 .1 70 ) r is k 0. 00 18 ** 0. 00 74 ** * -0 .0 00 4 -3 .1 6e -0 7 -0 .0 00 3 0. 00 03 -0 .0 01 3 -0 .0 10 2* (0 .0 00 9) (0 .0 01 3) (0 .0 00 5) (2 .0 4e -0 6) (0 .0 00 9) (0 .0 00 4) (0 .0 01 7) (0 .0 05 7) si z e 0. 00 96 ** * 0. 01 65 ** * 0. 01 12 ** * -0 .0 03 9 0. 01 25 ** * 0. 01 44 ** * 0. 00 79 ** * 0. 15 4* * (0 .0 01 2) (0 .0 02 0) (0 .0 01 4) (0 .0 03 2) (0 .0 02 3) (0 .0 02 2) (0 .0 02 3) (0 .0 69 2) g r o w t h 0. 00 30 0. 00 36 3 0. 03 66 -8 .2 2e -0 5 0. 00 99 ** * 0. 00 07 0. 01 21 0. 05 93 ** * (0 .0 27 0) (0 .0 26 3) (0 .0 63 5) (6 .1 0e -0 5) (0 .0 02 6) (0 .0 18 0) (0 .0 49 1) (0 .0 18 6) l iq u id it y -0 .0 08 3* ** -0 .0 24 8* ** -0 .0 06 0 0. 00 23 -0 .0 01 1 -0 .0 00 4 0. 00 61 0. 01 70 (0 .0 02 2) (0 .0 03 6) (0 .0 07 8) (0 .0 01 9) (0 .0 04 4) (0 .0 01 1) (0 .0 09 0) (0 .0 22 6) sp p -0 .0 05 2 0. 00 51 -0 .0 11 4* -0 .0 17 6* ** -0 .0 11 6 -0 .0 23 9* * -0 .0 20 8* * -0 .0 70 4* (0 .0 05 5) (0 .0 05 9) (0 .0 06 6) (0 .0 03 5) (0 .0 09 9) (0 .0 11 1) (0 .0 09 1) (0 .0 36 9) (c on tin ue d) determinants of capital structure for malaysian shariah-compliant firms: 45-74 69 y ea r 2 01 4 2 01 7 v a r ia b l e s (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) c on su m er p ro du ct in du st ri al p ro du ct s c on st ru ct io n tr ad in g se rv ic es pr op er tie s pl an ta tio n te ch no lo gy in fr as tr uc tu re l e v 2 (1) 0. 14 4* * 0. 54 5* ** 0. 95 3 0. 56 7* * -0 .5 45 0. 76 7 0. 06 89 ** * 0. 00 00 (0 .0 65 4) (0 .1 44 ) (2 .5 80 ) (0 .2 70 ) (0 .6 07 ) (0 .5 68 ) (0 .0 00 0) (0 .0 00 0) n t d s 4. 95 3* ** 1. 57 1 -2 .9 79 -4 .7 76 ** * 8. 90 4 7. 68 3 0. 00 00 34 .2 5 (1 .4 96 ) (1 .0 14 ) (1 1. 96 ) (1 .3 12 ) (9 .9 40 ) (5 .9 58 ) (0 .0 00 0) (0 .0 00 0) ta n g -0 .1 82 ** * 0. 09 15 -1 .6 48 0. 22 1* -0 .0 12 9 -0 .4 00 * -0 .9 12 ** * -1 .8 13 (0 .0 64 9) (0 .0 57 5) (1 .6 09 ) (0 .1 30 ) (0 .1 38 ) (0 .2 31 ) (0 .0 00 0) (0 .0 00 0) pr o fi t -1 .3 98 ** * -0 .1 27 -0 .1 86 -0 .0 51 1 0. 30 5 -1 .7 35 * -0 .3 53 ** * 12 .9 0 (0 .1 70 ) (0 .1 51 ) (1 .8 93 ) (0 .3 57 ) (0 .5 01 ) (0 .9 31 ) (0 .0 00 0) (0 .0 00 0) r is k -1 .6 6e -0 5 -0 .0 00 1 -0 .0 07 4 -0 .0 27 0* * 0. 01 36 -0 .0 10 9 0. 11 9* ** (9 .7 9e -0 5) (0 .0 00 5) (0 .0 50 4) (0 .0 13 1) (0 .0 11 3) (0 .0 12 3) (0 .0 00 0) si z e 0. 02 16 ** * 0. 00 63 ** * 0. 02 16 ** * 0. 01 03 ** * 0. 00 69 ** 0. 02 25 ** * 0. 03 27 ** * (0 .0 01 8) (0 .0 01 6) (0 .0 04 8) (0 .0 02 9) (0 .0 02 7) (0 .0 07 9) (0 .0 00 0) g r o w t h 0. 00 52 0. 02 64 0. 74 9 -0 .1 67 0. 18 0 1. 40 5 -0 .2 57 ** * (0 .0 05 5) (0 .0 67 7) (0 .9 14 ) (0 .1 33 ) (0 .1 49 ) (1 .0 26 ) (0 .0 00 0) l iq u id it y -0 .0 21 4* ** -0 .0 08 3* ** 0. 01 73 -0 .0 56 8* ** 0. 03 23 * -0 .0 02 5 -0 .0 43 7* ** (0 .0 02 8) (0 .0 02 3) (0 .0 57 8) (0 .0 16 2) (0 .0 18 3) (0 .0 01 7) (0 .0 00 0) sp p -0 .0 24 6 0. 01 56 ** -0 .0 54 0 0. 04 06 -0 .0 65 5 0. 22 5 -0 .0 51 5* ** 0. 00 00 (0 .0 31 9) (0 .0 07 6) (0 .1 71 ) (0 .0 52 9) (0 .0 51 0) (0 .2 56 ) (0 .0 00 0) (0 .0 00 0) n ot es : a ll m od el s ar e es tim at ed u si ng th e a re lla no a nd b on d dy na m ic p an el g m m e st im at io ns (s ta ta x ta bo nd c om m an d) . t he v ar ia bl es ar e de fin ed a s f ol lo w s: l e v 2 = to ta l d eb t/t ot al a ss et s; n t d s = a nn ua l d ep re ci at io n e xp en se s/ to ta l a ss et s; t a n g = n et f ix ed a ss et s/ to ta l a ss et s; p r o fi t = e b it /t ot al a ss et s; r is k = e b it o ve r to ta l a ss et s; s iz e = n at ur al l og o f to ta l a ss et s; g r o w t h = m ar ke t v al ue o f e qu ity /b oo k v al ue o f e qu ity ; l iq u id it y = c ur re nt a ss et s ov er c ur re nt l ia bi lit ie s; s pp = f ir st d if fe re nc e of th e y ea r e nd sh ar e pr ic e. f ig ur es in th e pa re nt he se s ar e tst at is tic s. * ** , * *, a nd * in di ca te s ig ni fic an ce a t t he 1 % , 5 % , a nd 1 0% le ve ls , r es pe ct iv el y. 70 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 ta bl e 10 . s um m ar y of t ar ge t c ap ita l s tr uc tu re m at ur ity in s pe ed a dj us tm en t i n im pl em en ta tio n of r ev is ed s cr ee ni ng m et ho do lo gy a t n ov 2 01 3 ty pe of le ve ra ge y ea r (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) c on su m er pr od uc t in du st ri al pr od uc ts c on st ru ct io n tr ad in g se rv ic es pr op er tie s pl an ta tio n te ch no lo gy in fr as tr uc tu re l e v 1 (1) sp ee d of a dj us tm en t 19 99 20 13 0. 64 3* ** 0. 57 1* ** 0. 50 5* ** 0. 33 1* ** 0. 47 5* ** 0. 68 0* ** 0. 60 8* ** 0. 57 5* ** 0. 35 7 0. 42 9 0. 49 5 0. 66 9 0. 52 5 0. 32 0. 39 2 0. 42 5 y ea r 2. 80 1 2. 33 1 2. 02 1. 49 5 1. 90 5 3. 12 5 2. 55 1 2. 35 3 sp ee d of a dj us tm en t 20 14 20 17 0. 14 4* * 0. 54 5* ** 0. 95 3 0. 56 7* * 1. 80 0 0. 76 7 0. 06 89 ** * 0 0. 85 6 0. 45 5 0. 43 3 0. 93 11 y ea r 1. 16 8 2. 19 8 2. 30 9 1. 07 4 l e v 2 (1) sp ee d of a dj us tm en t 19 99 20 13 0. 67 5* ** 0. 57 1* ** 0. 51 5* ** 0. 32 3* ** 0. 48 5* ** 0. 66 7* ** 0. 60 8* ** 0. 29 4* 0. 32 5 0. 42 9 0. 48 5 0. 67 7 0. 51 5 0. 33 3 0. 39 2 0. 70 6 y ea r 3. 07 7 2. 33 1 2. 06 2 1. 47 7 1. 94 2 3. 00 3 2. 55 1 1. 41 6 sp ee d of a dj us tm en t 20 14 20 17 0. 14 4* * 0. 54 5* ** 0. 95 3 0. 56 7* * -0 .5 45 0. 76 7 0. 06 89 ** * 0 0. 85 6 0. 45 5 0. 43 3 0. 93 11 y ea r 1. 16 8 2. 19 8 2. 30 9 1. 07 4 n ot es : * ** , * *, a nd * d en ot e si gn ifi ca nt a t 1 % , 5 % a nd 1 0% le ve ls , r es pe ct iv el y. t he c om pu ta tio n of s pe ed o f a dj us tm en t i s de ri ve d fr om o ne m in us th e co ef fic ie nt s on l e v 1 (1) a nd l e v 2 (1) . determinants of capital structure for malaysian shariah-compliant firms: 45-74 71 5. conclusion and discussion the main objectives for this study are to examine the determinants of capital structure for each sectors among malaysian shariah-compliant firms, and whether the inclusion of islamic debt (leverage 1 and leverage 2) has led to different results due to changes in the screening methodology. this study seeks to investigate the motive behind the reduced number of shariah-compliant firms in november 2013, for which these companies have had high levels of conventional debt that surpassed the benchmarks set by the securities commission malaysia. by identifying the factors that influence capital structure and capital structure decisions, firms will be able to maximize value and strengthen the company’s shariah-compliant financial stability, while remaining listed as shariahcompliant securities. the empirical analysis and results reported in this study asserts that firms have target capital structure, and there are specific determinants that would affect the capital structure of shariah-compliant firms in malaysia. this study concludes that the lagged dependent variables (lagged leverage 1 and leverage 2) have positive significance on capital structure with speed of adjustment at approximately 2 years. hence, shariah-compliant firms close approximately 30% to 70% of the gap between current and target capital structure within one and two years. these findings reaffirm that there exist target leverage for all sectors of the shariah-compliant firms in malaysia. moreover, the speed of adjustment from current leverage to the target leverage became faster when the securities commission malaysia (sc) introduced a revised screening methodology in november 2013. most of the malaysian shariah-compliant firms for each sector were considerably affected by the revised screening methodology, particularly with regards to the speed of adjustment towards achieving the target leverage. the findings have also shown certain implications to large firms. large firms tend to generate more income and profit, however they require more debt to support investment activities. with regards to profitability, this study identified a negative relationship between profitability and leverage for shariah-compliant firms for all sectors. shariah-compliant firms that have high profitability will use a lower leverage for financial activities. therefore, the results strongly support the pecking order theory. the findings on the target leverage level imply that after the introduction of the revised screening methodology in november 2013, the speed of adjustment improved. the revised screening methodology led to a faster adjustment towards the target leverage level. for example, the speed of adjustment for the consumer products sector improved from approximately 2 years (year 1999-2013) to 1 year (2014-2017). in addition, the level of conventional debt was found to be lower after the revised screening methodology was introduced. it is important for firms to achieve the target leverage when making financial decisions, which strengthens the firm’s shariah-compliant financial stability and sustainability, and subsequently enable firms to remain listed as shariah-compliant securities. future research on capital structure should be aimed towards the investigation of the effects of target leverage on sustainable growth rate. 72 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 acknowledgement the authors would like to acknowledge the research and innovation management centre (rimc) usim, nilai and ministry of higher education malaysia (mohe) for the financial support of this research. this research is supported by mohe under the fundamental research grant scheme (frgs) with reference code: frgs/1/2018/ss01/usim/01/2. references abdeljawad, i., mat-nor, f., ibrahim, i., & abdul-rahim, r. (2013). dynamic capital structure trade-off theory: evidence from malaysia. international review of business research papers, 9(6), 102-110. ali, m. m., ibrahim, m. k., mohammad, r., zain, m. m., & alwi, m. r. (2009). malaysia: value relevance of accounting numbers. in global practices of corporate social responsibility. berlin, heidelberg: springer berlin heidelberg, 201-231. al-najjar, b. (2011). empirical modelling of capital structure jordanian evidence. journal of emerging market finance, 10(1), 1-19. al-yahyaee, k. h., pham, t. m., & walter, t. s. (2013). capital structure and stock returns: evidence from an emerging market with unique financing arrangements. applied financial economics, 29(14). antoniou, a., guney, y., & paudyal, k. (2002). the determinants of corporate capital structure: evidence from european countries. centre for empirical research in finance, 23-26. arellano, m., & bond, s. (1991). some tests of specification for panel data: monte carlo evidence and an application to employment equations. the review of economic studies, 58(2), 277. arioglu, e., & tuan, k. (2014). speed of adjustment: evidence from borsa istanbul. borsa istanbul review. baxamusa, m., & jalal, a. (2014). does religion affect capital structure? research in international business and finance, 31, 112-131. bidyut jyoti bhattacharjee, & faculty of commerce, b.h college, howly, assam, i. (2010). capital structure decision: an empirical investigation in indian information technology sector. smart journal of business management studies, 6(2), 18-24. bonaimé, a. a., öztekin, o., & warr, r. s. (2014). capital structure, equity mispricing, and stock repurchases. journal of corporate finance, 26, 182-200. bouraoui, t., & li, t. (2014). the impact of adjustment in capital structure in mergers & acquisitions on us acquirers’ business performance. journal of applied business research, 30(1), 27-41. determinants of capital structure for malaysian shariah-compliant firms: 45-74 73 brendea, g. (2014). financing behavior of romanian listed firms in adjusting to the target capital structure. finance a uver – czech, journal of economics and finance, 64(4), 312-329. byoun, s. (2008). how and when do firms adjust their capital structures toward targets? journal of finance, 63(6), 3069-3096. chan, c. c., & chang, y. h. (2008). firm-specific stock return variation and capital structure decisions. applied economics letters, 15(4), 293-299. deangelo, h., deangelo, l., & whited, t. m. (2011). capital structure dynamics and transitory debt. journal of financial economics, 99(2), 235-261. drobetz, w., & wanzenried, g. (2006). what determines the speed of adjustment to the target capital structure? applied financial economics, 16, 941-958. drobetz, w., gounopoulos, d., merikas, a., & schröder, h. (2013). capital structure decisions of globally-listed shipping companies. transportation research part e: logistics and transportation review. getzmann, a., lang, s., & spremann, k. (2014). target capital structure and adjustment speed in asia. asia-pacific journal of financial studies, 43(1), 1-30. haron, r. (2014). key factors influencing target capital structure of property firms in malaysia. asian social science, 10(3), 62-69. haron, r. (2016). do indonesian firms practice target capital structure? a dynamic approach. journal of asia business studies, 10(3), 318-334. haron, r., & ibrahim, k. (2011). determinants of target capital structure: evidence on south east asia countries. journal of business and policy research, 6(3), 39-61. haron, r., & ibrahim, k. (2012). target capital structure and speed of adjustment: panel data evidence on malaysia shariah compliant securities, journal of economics management and accounting, 20(202), 87-107. haron, r., ibrahim, k., mat-nor, f., & ibrahim, i. (2013). factors affecting speed of adjustment to target leverage: malaysia evidence. global business review, 14(2), 243-262. hovakimian, a., & li, g. (2011). in search of conclusive evidence: how to test for adjustment to target capital structure. journal of corporate finance, 17(1), 33-44. johnson, r., & soenen, l. (2003). indicators of successful companies. european management journal, 21(3), 364–369. md. hashim, a., habib, f., isaacs, z., & gadhoum, m. (2007). isra – bloomberg shariah stock screening and income cleansing methodologies: a conceptual paper. isra international journal of islamic finance, 9(1), 27-42. muhammad, j. (2015). shariah-compliant stocks and stock market reaction in malaysia. universiti putra malaysia press. najib, n. h., hamid, i. r., nasarudin, s. n., & saiti, b. (2014). the comparison of shariah screening methodology for stocks between malaysia security 74 the international journal of banking and finance, vol. 14, 2018-2019: 45-74 commission criteria and dow jones method: a critical assessment. paper presented at unikl business school management and entrepreneurship conference, malaysia. suffian, m. t. m., sanusi, z. m., osman, a. h., & azhari, m. i. m. (2015). manipulation of earnings: the pressure of opportunistic behavior and monitoring mechanisms in malaysian shariah-compliant companies. procedia economics and finance, 31, 213–227. zainudin, n., miskam, s., & sulaiman, m. (2014). revised shariah screening methodology for shariah-compliant securities: new standard to meet global expectation. paper presented at the conference on management and muamalah, international islamic university college selangor. 67 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 how to cite this article: alfi, c. f. (2023). a meta-analysis of the relationship between religiosity and saving behaviour. international journal of banking and finance, 18(1), 67-94. https://doi. org/10.32890/ ijbf2023.18.1.4 a meta-analysis of the relationship between religiosity and saving behaviour coky fauzi alfi state polytechnic of sriwijaya, indonesia coky.fauzi.alfi@polsri.ac.id received: 2/2/2022 revised: 2/3/2022 accepted: 30/3/2022 published: 5/1/2023 abstract the purpose of this study was to synthesize the findings of previous studies on the relationship between religiosity and saving behaviour by using a meta-analysis approach. it also sought to determine the strength of the relationship, besides its direction. eleven studies which met the five criteria and four techniques used in the study were used as samples for the meta-analytic analysis. the size of the effect in each study was then determined by pearson’s product-moment correlations (r). to estimate the average distribution of relationship true effects, the fisher r-to-z transformation and random-effects methods were used. the empirical evidence showed that there was a positive correlation between religiosity and saving behaviour. however, according to guilford’s convention, the true effect size (r = 0.303) would mean that religiosity had a weak correlation with saving behaviour. it is recommended that authorities and financial institutions use the findings of this study to develop plans focused on advocating and facilitating saving behaviour among religious people. https://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 68 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 keywords: meta-analysis, religiosity, saving behaviour, fisher r-to-z transformation, random-effects method. jel classification: z120, g410. introduction it appears that the global community is becoming more religious. based on a pew research center (2015) survey, all major religions are estimated to show a rise in the number of followers by 2050. the survey found that 84 percent of the world population was religiously affiliated in 2010, with projections predicting that this share would rise to 87 percent by 2050. these findings and projections appear to contradict the views of several influential scholars, such as karl marx, emile durkheim, and max weber, who predicted that religion would be less important in various socioeconomic activities as industrialization progressed, economic markets expanded, and science, technology, and education advanced rapidly (basedau et al., 2018). furthermore, the findings of various studies in economics (e.g. azzi & ehrenberg, 1975; iannaccone, 1998; iyer, 2016), sociology (e.g. geertz, 1973; inglehart, 2018; lenski, 1961), and psychology (e.g. allport & ross, 1967; berry et al., 2002; pargament, 1999) have acknowledged the importance of religion in human society. for example, it plays an important role in, energising people to work for social change, promoting mental well-being, or acting as a social control agent. the investigation into the relationship between religion and economic growth has received considerable attention, ever since max weber (1905) recognized the significance of religious affiliations in economic performance. he argued that the values in protestant teachings would shape their adherents’ work ethic, resulting in professionalism and efficiency in economic activities. more than a century after weber’s thesis, a large body of literature has noticed a link between religion and macroeconomic prosperity. for instance, it has been discovered that religious beliefs, particularly beliefs in hell and heaven, have a positive effect on economic attitudes, leading to higher incomes and gross domestic product (gdp), and christianity is the religion with 69 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 the greatest impact on economic growth, with protestants being more capitalists than other christians (barro & mccleary, 2003; filipova, 2012; guiso et al., 2003; hayward & kemmelmeier, 2011). moreover, religion has long been associated with the teachings of thriftiness and customary living. the research findings however, show that there are differences in which religions adhere to the most frugal and conventional ways of life. according to keister (2003), guiso et al. (2003), and renneboog and spaenjers (2012), catholics appeared to value frugality and convenient living more than protestants, whereas arruñada (2010) and filipova (2012) discovered the opposite. although many studies found a correlation between religious belief and thriftiness, their findings are less convincing when used to explain a link between religious belief and saving decisions. this is due to the distinction between thriftiness and saving decisions. thriftiness is the trait to try and reduce spending, whereas saving decisions are initiated by residual income. therefore, research into how religion influences economic behaviour and financial decisions at the microeconomic level seems to remain limited (see klaubert, 2010; yayeh, 2014). in terms of empirical assessments that link individual saving attitudes to religious preferences or practices, it needs to be explored further. this investigation should be beneficial because it could help us to solve pressing issues in the national economy, for instance, pressing concerns such as wealth inequality (bilen, 2016; keister, 2003) and consumerism (tjahjono, 2014), or even the issues of conserving energy and natural resources (singh et al., 2021). since the investigation of religiosity on saving behaviour is an emerging research area, the present study is interested in knowing the ‘true’ effect size of the relationship between these variables. therefore, this study has performed a meta-analysis to gain a more objective, robust, and less biased understanding of the relationship between the variables by investigating the distribution of effect sizes. a meta-analysis is an approach for aggregating effect-size indices from multiple studies (borenstein et al., 2011). it contributes to answering the question of whether the observed variations in effect sizes across studies are due to a single population effect size (law, 1995). to date, there has been no other study examining religiosity and saving behaviour across samples, methodologies, and time. this study has utilized meta-analysis as a quantitative tool to synthesize the findings 70 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 of previous studies, and to determine the strength of the relationship between religiosity and saving behaviour, as well as the direction of that relationship. as a result, the strength of the correlation between previous findings and their direction, whether positive or negative, has been held to the same standards, as long as they meet the inclusion criteria set for the study sample. one of the inclusion criteria, for example, was that the study sample would include religiosity, religious belief, or religious faith as an independent variable. in addition, the other objective is to contribute to the growth of the literature in this area of study. this study can provide a retrospective summary of the existing literature and provide further empirical evidence of the true effect of religiosity on saving behaviour. it could help shape new research by describing what was already known and synthesizing the new body of evidence. after reviewing previous studies and establishing the inclusion criteria, eleven journal articles were identified as the study sample; all together these sources provided a total of 1,063 participants coming from various locations. more specifically, yayeh (2014) collected samples in ethiopia, while ababio and mawutor (2015) did so in ghana. satsios and hadjidakis (2017) gathered data in greece. in indonesia, questionnaires were administered by murdayanti et al. (2020); prastiwi (2021); priyo nugroho et al. (2017); wijaya et al. (2019). meanwhile, data was collected in malaysia by abdullah and abd. majid (2001); ismail et al. (2018); kassim et al. (2019); mei teh et al. (2019). as a result, this meta-analytic study was able to generate numerous plot functions, such as the forest plot, standardized residual plot, and cook’s distance plot, as well as measurements, such as the random-effect model, heterogeneity statistics, outliers, and influential case diagnostics. literature review the relationship between religiosity and saving behaviour is typically measured using one of two methods: methods which are either economically or psychologically oriented. in the economic approach, the goal is to create forecasts about behaviour as accurately as possible. however, this approach often neglects to explore the true underlying causes of why individuals behave the way they do (nyhus, 71 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 2017). researchers who employ this approach rely on secondary data surveys, such as the world values survey (wvs), the international social survey program (issp), the gallup millennium survey, the panel study of income dynamics (psid), or the konda araştrma ve danşmanlk, to determine an individual’s religious behaviour. from these data sources, researchers discovered that the ‘average’ person’s religiosity could be measured in the following five ways, namely participation in religious services, belief in heaven and hell, belief in the afterlife, faith in god, and self-identification as a religious person (barro & mccleary, 2003). these religious aspects are then examined in relation to the adherents’ amount of income or consumption using various econometric methodologies so as to understand the significance of religiosity in saving behaviour. klaubert (2010), for example, used the psid to investigate the link between individual saving decisions and religiosity, as measured by church attendance, in the united states. using the konda data survey, davutyan and öztürkkal (2016) investigated the effect of religious affiliation on financial behaviour in turkey. they discovered however, weak evidence that religious people have distinct preferences for saving decisions. this was due to there being no difference between religious and non-religious people when it comes to saving decisions. meanwhile, guiso et al. (2003) discovered a link between religious intensity and thriftiness in a crossnational study using the wvs sample statistics. on the other hand, the psychological viewpoint begins from a different place. this approach frequently concentrates on psychological factors, and examines individual differences rather than average human behaviour. therefore, various explanatory variables and methods have been employed in the analysis of the relationship between religiosity and saving behaviour, which makes it different from the economic approach. the psychological viewpoint often employs primary data sources and applies behavioural theories, for example the theory of planned behavior (ajzen, 1991) or social learning theory (bandura, 1977). the theory of planned behavior identifies specific factors, namely intentions and perceived behavioral control, that can be utilised to estimate and describe human behavior in various contexts. intentions are motivational variables that demonstrate how far individuals are willing to go and how much effort they intend to put in, whereas perceived behavioral control refers to the perception of how easy or difficult it is to control an interest (ajzen, 1991). meanwhile, 72 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 attitudes toward behavior, subjective norms, and perceived behavioral control can all have an impact on intentions. furthermore, social learning theory is often used as a base theory to describe the role of financial literacy in saving behavior. the theory hypothesizes that the cognitive abilities of individuals, i.e., knowledge and skills, impact on changing their behaviors. this cognitive ability can be learned by seeing, imitating, practicing, and processing information from the behaviour of others and its environments, including families, friends, neighbours, the workplace, or the media. the psychological viewpoint also employs religiosity measurement scales, such as the orthodoxy measurement (glock, 1962) or the religious orientations (allport & ross, 1967). the orthodoxy measurement uses the following five scales: belief, practice, knowledge, experience, and consequences, and these would inform the preferred faith. belief is an ideological dimension that a religious person will adhere to. prayer, fasting, involvement in special sacraments, worship, and other ritualistic activities are included in the practice. knowledge refers to the understanding of the fundamental tenets of a religious person’s faith and its sacred scriptures. experience gives a religious emotional experience, and consequences are all of the religious prescriptions for what a religious person should do. meanwhile, the measurement of religious orientation uses the following two dimensions: intrinsic and extrinsic, and they would inform us of the primary motive for life in religion. those who are intrinsically motivated find that their primary motive in religion is to live according to their religious convictions and prescriptions. however, extrinsically oriented people may find religion useful in a variety of ways, including stability and reassurance, social connection and diversionary tactics, status, and self-justification. few researchers have adopted this approach in their studies. for example, priyo nugroho et al. (2017) expanded the theory of planned behaviour by including two new variables: religiosity and self-efficacy. they then employed allport and ross’s (1967) scale for measuring religiosity to investigate the saving behaviour of islamic bank customers. in the meantime, kassim et al. (2019) who used the social learning theoretical framework and the religiosity measurement scale which had its root in glock’s (1962) work, discovered that whereas religiosity had no effect on saving behaviour, financial literacy did. 73 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 methodology criteria and search procedure the samples for the present meta-analytic study were selected because they had discussed the influence of religiosity on saving behaviour directly. to be included in the meta-analytic sample, the studies must fulfil five criteria. they are as follows: • the studies used religiosity, religious belief, or religious faith as an independent variable. • the studies used saving behaviour or saving habits, saving money, or saving decisions as a dependent variable. • the studies used a quantitative research approach. • the studies used primary data at a micro analytical level. • the studies presented the pearson’s r effect size clearly or could be processed using another statistical method. studies would be excluded if they had found a relationship between religiosity and saving behaviour indirectly. finding studies from various journals, such as journals on economics, business, management, finance, marketing, religion, culture, and social science, that fit the inclusion and exclusion criteria for a metaanalysis study was challenging. for example, to avoid the possibility that this might turn out to be a time-consuming process, an effective search strategy was used from start to finish. the present study has implemented four techniques to conduct a wide-ranging literature search. they were as follows: (1) deciding search terms and keywords, (2) searching for specific phrases, (3) using truncated and wildcard searches as well as boolean logic, and (4) using citation searching. firstly, these terms and keywords were applied in the search process: religiosity, religious belief, religious faith, saving behaviour, saving habits, saving money, and saving decisions. secondly, quotation marks were used for words which appear next to each other, e.g., “religious belief,” “religious faith,” “saving behaviour,” “saving habits,” “saving money,” “saving decisions.” thirdly, the search used combined truncation and wildcard searches with boolean logic, e.g., “religio*” and “saving behavio?r”. fourthly, articles that were cited in other publications were also included in the search. these techniques were then employed to search for studies in the various research search engines and databases, such as semantic scholar, 74 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 google scholar, research gate, ebscohost, proquest, jstor, and wiley online library. all potentially relevant titles and abstracts were then saved and managed systematically for the next stage, which was screening. the first stage of the screening was to import all the references into a reference management software package and de-duplicate them. in this case, a software called endnote was used. the next stage was to read and identify all the saved articles of study, and filter them out if they were irrelevant. the stages of screening resulted in 11 journal articles identified as the relevant data selected for the meta-analysis. data extraction once screening has been done and all relevant articles selected for the study have been identified, the next step is the data extraction (see teshome et al., 2018; zuckerman et al., 2013). in this process, the key aspects that will be used for the statistical meta-analysis will have to be extracted from the articles. some key aspects of the articles are set, namely the authors’ name and year of publication, sampling methods, measurement techniques, variables identification (independent and dependent), methods of statistical analysis, and a summary of the results (see table 1). the characteristics of each article that met the criteria for inclusion were also highlighted. for example, the eleven studies used various themes related to religiosity and saving behaviour as independent and dependent variables, respectively. these are described in the column on variables. in another column, such as the measurement technique, it is stated that all studies applied a questionnaire survey to ensure that primary data was used. the most useful information, however, is in the results column. it discusses the significance of the relationship between religiosity and saving behaviour, as well as various goodness of fit tests, e.g., chi-square, odds ratio, or t-statistic, that can be used to compute the effect size r. effect size computation following data extraction, the next task was to determine the size of the effect in each study and ensure that this effect size was expressed in the same way. the effect sizes are used to describe the strength of the relationship between the variables. there are two common types of effect size: the r type and the d type. the two most commonly 75 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 used of the r type are pearson’s product-moment correlations (r) and fisher’s r-to-z transformation (zr), whereas the three most commonly used d type are cohen’s d, hedges’s g, and glass’s d (rosenthal, 1995). in this study, pearson’s r was the preferred effect size. in this regard, it involved calculating the r value for each study carrying out the meta-analysis. there was no need to do anything if a study had used the r value. however, because some studies had no effect size value and only provided various fit test indicators (e.g., chi-square, odds ratio, t-test statistic), a conversion to pearson’s r was performed via an online calculator at www.psychometrica.de/effect_size.html (lenhard & lenhard, 2016). meanwhile, if the authors did not provide the indicators and could not compute a conversion to the r value, they were contacted via email to gain the relevant information. a reminder was sent if they did not reply. method of analysis the analysis is carried out using the fisher r-to-z transformed correlation coefficient as the outcome measure. the fisher’s r-to-z transformation is commonly used because samples from a metaanalysis contain a variety of effect sizes. it is also to achieve normality in the effect sizes (cheung et al., 2012). there are three steps in implementing this method (borenstein et al., 2011; field & gillett, 2010). to begin, use fisher’s r-to-z transformation to convert the effect size in each study into a standard normal metric. the fisher’s r-to-z transformation formula is given as where is the effect size in each study. after that, for each study, a weighted average of scores are computed by the formula where is the number of studies and is the sample size. finally, it should be converted back to using the formula in addition, the random-effects statistical model was applied to estimate the average distribution of true effects. the random-effects method was chosen because the effect size was extracted from a series of studies conducted by various authors in various populations at various times. the present analysis also reported on the estimate of the index, the h2 index, the i2 index, and the q-test (cochran, 1954) with a p-value as the heterogeneity statistics outcome. the q-test was used to assess the null hypothesis that all effect sizes a meta-analysis of the relationship between religiosity and saving behaviour method of analysis 𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒𝑒𝑒 � 1+𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 1−𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 �, 𝑧𝑧𝑧𝑧�̅�𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 , 𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖−1 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖+1 . τ2 a meta-analysis of the relationship between religiosity and saving behaviour method of analysis 𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒𝑒𝑒 � 1+𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 1−𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 �, 𝑧𝑧𝑧𝑧�̅�𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 , 𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖−1 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖+1 . τ2 a meta-analysis of the relationship between religiosity and saving behaviour method of analysis 𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒𝑒𝑒 � 1+𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 1−𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 �, 𝑧𝑧𝑧𝑧�̅�𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 , 𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖−1 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖+1 . τ2 a meta-analysis of the relationship between religiosity and saving behaviour method of analysis 𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒𝑒𝑒 � 1+𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 1−𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 �, 𝑧𝑧𝑧𝑧�̅�𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 , 𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖−1 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖+1 . τ2 a meta-analysis of the relationship between religiosity and saving behaviour method of analysis 𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒𝑒𝑒 � 1+𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 1−𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 �, 𝑧𝑧𝑧𝑧�̅�𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 , 𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖−1 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖+1 . τ2 a meta-analysis of the relationship between religiosity and saving behaviour method of analysis 𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒𝑒𝑒 � 1+𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 1−𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 �, 𝑧𝑧𝑧𝑧�̅�𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 , 𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖−1 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖+1 . τ2 three steps to implementing this method (borenstein et al., 2011; field & gillett, 2010). to begin, use fisher's r-to-z transformation to convert the effect size in each study into a standard normal metric. the fisher's r-to-z transformation formula is given by 𝑧𝑧𝑟𝑟𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒 ( 1+𝑟𝑟𝑖𝑖 1−𝑟𝑟𝑖𝑖 ), where 𝑟𝑟𝑖𝑖 is the effect size in each study. after that, for each study, a weighted average of 𝑧𝑧𝑟𝑟 scores are computed by �̅�𝑧𝑟𝑟𝑖𝑖 = ∑ 𝑛𝑛𝑖𝑖𝑧𝑧𝑟𝑟𝑖𝑖 𝑘𝑘 𝑖𝑖=1 ∑ 𝑛𝑛𝑖𝑖𝑘𝑘𝑖𝑖=1 , where 𝑘𝑘 is the number of studies and 𝑛𝑛𝑖𝑖 is the sample size. finally, it should be converted back to 𝑟𝑟𝑖𝑖 using the formula 𝑟𝑟𝑖𝑖 = 𝑒𝑒2�̅�𝑧𝑟𝑟𝑖𝑖−1 𝑒𝑒2�̅�𝑧𝑟𝑟𝑖𝑖+1 . in addition, the random-effects statistical model is applied to estimate the average distribution of true effects. the random-effects method was chosen because the effect size was extracted from a series of studies conducted by various authors in various populations at various times. the analysis also reports the estimate of the 2 index, the h2 index, the i2 index, and the q-test (cochran, 1954) with a p-value as the heterogeneity statistics outcome. the q-test is used to assess the null hypothesis that all effect sizes from all studies are homogenous (chen & peace, 2021). if the p-value is less than  (the typical significance level is 0.05), the null hypothesis should be rejected, indicating that the effect sizes from all studies are not homogenous. meanwhile, 2, h2, and i2 are used to determine the strength of the distribution of true effect sizes. the 2 index is estimated using the hedges’ estimator (hedges & olkin, 1985) to measure the variance of the true effect sizes, and the index should be greater than zero. the h2 index is quantified using higgins and thompson’s (2002) formula to inform the relative extent of heterogeneity in comparison to all studies, and the index should be greater than 1. the i2 index is also calculated using higgins and thompson's (2002) formula to determine the percentage of observed heterogeneity versus real heterogeneity. as a rule of thumb, the i2 index could be considered as having low heterogeneity (i2 = 25%), moderate heterogeneity (i2 = 50%), and high heterogeneity (i2 = 75%). to display the conclusions of meta-analyses, forest plots are generated. forest plots provide information about each study’s effect size and confidence interval, as well as the average distribution of true effects. the analysis also examines whether studies may be outliers and/or influential in the random-effect model. they can have a significant impact on the value of the estimated random-effect model coefficients, i.e., the intercept. if they had remained in the analysis, they could have changed the entire outcome. the standardized residuals are used to detect outliers, while the cook's distances (cook, 1977) and dffits (difference in fits) are applied to diagnose the influential studies. studies are considered as potential outliers if they have a standardized residual larger than 3 or smaller than -3 (rstudent > ± 3), while they are considered to be influential if the cook's distance value is more than 1 (cook.d > 1) and dffits is larger than 2 (dffits > 2) (gerbing, 2014). the meta-analysis was carried out with the help of open-source statistical software jamovi version 1.6.23 (the jamovi project, 2021). the major meta-analysis module library was used to compute r-to-z transformations, as well as to generate a random-effect model, heterogeneity statistics, a forest plot, and outlier and influential case diagnostics. results table 1 shows the results of data extraction processing. the eleven studies were published between 2001 and 2021, with abdullah and abd. majid (2001) is the longest and prastiwi (2021) is the most recent. the studies used two types of sampling methods: probability sampling and non-probability sampling. ababio and mawutor (2015); kassim et al. (2019); murdayanti et al. (2020); and yayeh (2014) applied the probability sampling method, whereas ismail et al. (2018); mei teh et al. (2019); priyo nugroho et al. three steps to implementing this method (borenstein et al., 2011; field & gillett, 2010). to begin, use fisher's r-to-z transformation to convert the effect size in each study into a standard normal metric. the fisher's r-to-z transformation formula is given by 𝑧𝑧𝑟𝑟𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒 ( 1+𝑟𝑟𝑖𝑖 1−𝑟𝑟𝑖𝑖 ), where 𝑟𝑟𝑖𝑖 is the effect size in each study. after that, for each study, a weighted average of 𝑧𝑧𝑟𝑟 scores are computed by �̅�𝑧𝑟𝑟𝑖𝑖 = ∑ 𝑛𝑛𝑖𝑖𝑧𝑧𝑟𝑟𝑖𝑖 𝑘𝑘 𝑖𝑖=1 ∑ 𝑛𝑛𝑖𝑖𝑘𝑘𝑖𝑖=1 , where 𝑘𝑘 is the number of studies and 𝑛𝑛𝑖𝑖 is the sample size. finally, it should be converted back to 𝑟𝑟𝑖𝑖 using the formula 𝑟𝑟𝑖𝑖 = 𝑒𝑒2�̅�𝑧𝑟𝑟𝑖𝑖−1 𝑒𝑒2�̅�𝑧𝑟𝑟𝑖𝑖+1 . in addition, the random-effects statistical model is applied to estimate the average distribution of true effects. the random-effects method was chosen because the effect size was extracted from a series of studies conducted by various authors in various populations at various times. the analysis also reports the estimate of the 2 index, the h2 index, the i2 index, and the q-test (cochran, 1954) with a p-value as the heterogeneity statistics outcome. the q-test is used to assess the null hypothesis that all effect sizes from all studies are homogenous (chen & peace, 2021). if the p-value is less than  (the typical significance level is 0.05), the null hypothesis should be rejected, indicating that the effect sizes from all studies are not homogenous. meanwhile, 2, h2, and i2 are used to determine the strength of the distribution of true effect sizes. the 2 index is estimated using the hedges’ estimator (hedges & olkin, 1985) to measure the variance of the true effect sizes, and the index should be greater than zero. the h2 index is quantified using higgins and thompson’s (2002) formula to inform the relative extent of heterogeneity in comparison to all studies, and the index should be greater than 1. the i2 index is also calculated using higgins and thompson's (2002) formula to determine the percentage of observed heterogeneity versus real heterogeneity. as a rule of thumb, the i2 index could be considered as having low heterogeneity (i2 = 25%), moderate heterogeneity (i2 = 50%), and high heterogeneity (i2 = 75%). to display the conclusions of meta-analyses, forest plots are generated. forest plots provide information about each study’s effect size and confidence interval, as well as the average distribution of true effects. the analysis also examines whether studies may be outliers and/or influential in the random-effect model. they can have a significant impact on the value of the estimated random-effect model coefficients, i.e., the intercept. if they had remained in the analysis, they could have changed the entire outcome. the standardized residuals are used to detect outliers, while the cook's distances (cook, 1977) and dffits (difference in fits) are applied to diagnose the influential studies. studies are considered as potential outliers if they have a standardized residual larger than 3 or smaller than -3 (rstudent > ± 3), while they are considered to be influential if the cook's distance value is more than 1 (cook.d > 1) and dffits is larger than 2 (dffits > 2) (gerbing, 2014). the meta-analysis was carried out with the help of open-source statistical software jamovi version 1.6.23 (the jamovi project, 2021). the major meta-analysis module library was used to compute r-to-z transformations, as well as to generate a random-effect model, heterogeneity statistics, a forest plot, and outlier and influential case diagnostics. results table 1 shows the results of data extraction processing. the eleven studies were published between 2001 and 2021, with abdullah and abd. majid (2001) is the longest and prastiwi (2021) is the most recent. the studies used two types of sampling methods: probability sampling and non-probability sampling. ababio and mawutor (2015); kassim et al. (2019); murdayanti et al. (2020); and yayeh (2014) applied the probability sampling method, whereas ismail et al. (2018); mei teh et al. (2019); priyo nugroho et al. three steps to implementing this method (borenstein et al., 2011; field & gillett, 2010). to begin, use fisher's r-to-z transformation to convert the effect size in each study into a standard normal metric. the fisher's r-to-z transformation formula is given by 𝑧𝑧𝑟𝑟𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒 ( 1+𝑟𝑟𝑖𝑖 1−𝑟𝑟𝑖𝑖 ), where 𝑟𝑟𝑖𝑖 is the effect size in each study. after that, for each study, a weighted average of 𝑧𝑧𝑟𝑟 scores are computed by �̅�𝑧𝑟𝑟𝑖𝑖 = ∑ 𝑛𝑛𝑖𝑖𝑧𝑧𝑟𝑟𝑖𝑖 𝑘𝑘 𝑖𝑖=1 ∑ 𝑛𝑛𝑖𝑖𝑘𝑘𝑖𝑖=1 , where 𝑘𝑘 is the number of studies and 𝑛𝑛𝑖𝑖 is the sample size. finally, it should be converted back to 𝑟𝑟𝑖𝑖 using the formula 𝑟𝑟𝑖𝑖 = 𝑒𝑒2�̅�𝑧𝑟𝑟𝑖𝑖−1 𝑒𝑒2�̅�𝑧𝑟𝑟𝑖𝑖+1 . in addition, the random-effects statistical model is applied to estimate the average distribution of true effects. the random-effects method was chosen because the effect size was extracted from a series of studies conducted by various authors in various populations at various times. the analysis also reports the estimate of the 2 index, the h2 index, the i2 index, and the q-test (cochran, 1954) with a p-value as the heterogeneity statistics outcome. the q-test is used to assess the null hypothesis that all effect sizes from all studies are homogenous (chen & peace, 2021). if the p-value is less than  (the typical significance level is 0.05), the null hypothesis should be rejected, indicating that the effect sizes from all studies are not homogenous. meanwhile, 2, h2, and i2 are used to determine the strength of the distribution of true effect sizes. the 2 index is estimated using the hedges’ estimator (hedges & olkin, 1985) to measure the variance of the true effect sizes, and the index should be greater than zero. the h2 index is quantified using higgins and thompson’s (2002) formula to inform the relative extent of heterogeneity in comparison to all studies, and the index should be greater than 1. the i2 index is also calculated using higgins and thompson's (2002) formula to determine the percentage of observed heterogeneity versus real heterogeneity. as a rule of thumb, the i2 index could be considered as having low heterogeneity (i2 = 25%), moderate heterogeneity (i2 = 50%), and high heterogeneity (i2 = 75%). to display the conclusions of meta-analyses, forest plots are generated. forest plots provide information about each study’s effect size and confidence interval, as well as the average distribution of true effects. the analysis also examines whether studies may be outliers and/or influential in the random-effect model. they can have a significant impact on the value of the estimated random-effect model coefficients, i.e., the intercept. if they had remained in the analysis, they could have changed the entire outcome. the standardized residuals are used to detect outliers, while the cook's distances (cook, 1977) and dffits (difference in fits) are applied to diagnose the influential studies. studies are considered as potential outliers if they have a standardized residual larger than 3 or smaller than -3 (rstudent > ± 3), while they are considered to be influential if the cook's distance value is more than 1 (cook.d > 1) and dffits is larger than 2 (dffits > 2) (gerbing, 2014). the meta-analysis was carried out with the help of open-source statistical software jamovi version 1.6.23 (the jamovi project, 2021). the major meta-analysis module library was used to compute r-to-z transformations, as well as to generate a random-effect model, heterogeneity statistics, a forest plot, and outlier and influential case diagnostics. results table 1 shows the results of data extraction processing. the eleven studies were published between 2001 and 2021, with abdullah and abd. majid (2001) is the longest and prastiwi (2021) is the most recent. the studies used two types of sampling methods: probability sampling and non-probability sampling. ababio and mawutor (2015); kassim et al. (2019); murdayanti et al. (2020); and yayeh (2014) applied the probability sampling method, whereas ismail et al. (2018); mei teh et al. (2019); priyo nugroho et al. 76 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 from all studies were homogenous (chen & peace, 2021). if the p-value was less than (the typical significance level is 0.05), the null hypothesis should be rejected, indicating that the effect sizes from all studies were not homogenous. meanwhile, , h2, and i2 were used to determine the strength of the distribution of true effect sizes. the index was estimated using the hedges’ estimator (hedges & olkin, 1985) to measure the variance of the true effect sizes, and the index should be greater than zero. the h2 index was quantified using the higgins and thompson’s (2002) formula to inform the relative extent of heterogeneity in comparison to all studies, and the index should be greater than 1. the i2 index was also calculated using the higgins and thompson’s (2002) formula to determine the percentage of observed heterogeneity versus real heterogeneity. as a rule of thumb, the i2 index could be considered as having low heterogeneity (i2 = 25%), moderate heterogeneity (i2 = 50%), and high heterogeneity (i2 = 75%). to display the conclusions of the meta-analyses, forest plots were generated. forest plots provided information about each study’s effect size and confidence interval, as well as the average distribution of true effects. the analysis also examined whether studies may be outliers and/or influential in the random-effect model. they can have a significant impact on the value of the estimated random-effect model coefficients, i.e., the intercept. if they had remained in the analysis, they could have changed the entire outcome. the standardized residuals are used to detect outliers, while the cook’s distances (cook, 1977) and dffits (difference in fits) are applied to diagnose the influential studies. studies are considered as potential outliers if they have a standardized residual larger than 3 or smaller than -3 (rstudent > ± 3), while they are considered to be influential if the cook’s distance value is more than 1 (cook.d > 1) and dffits is larger than 2 (dffits > 2) (gerbing, 2014). the meta-analysis was carried out with the help of an open-source statistical software the jamovi version 1.6.23 (the jamovi project, 2021). the major meta-analysis module library was used to compute r-to-z transformations, as well as to generate a random-effect model, heterogeneity statistics, a forest plot, and outlier and influential case diagnostics. three steps to implementing this method (borenstein et al., 2011; field & gillett, 2010). to begin, use fisher's r-to-z transformation to convert the effect size in each study into a standard normal metric. the fisher's r-to-z transformation formula is given by 𝑧𝑧𝑟𝑟𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒 ( 1+𝑟𝑟𝑖𝑖 1−𝑟𝑟𝑖𝑖 ), where 𝑟𝑟𝑖𝑖 is the effect size in each study. after that, for each study, a weighted average of 𝑧𝑧𝑟𝑟 scores are computed by �̅�𝑧𝑟𝑟𝑖𝑖 = ∑ 𝑛𝑛𝑖𝑖𝑧𝑧𝑟𝑟𝑖𝑖 𝑘𝑘 𝑖𝑖=1 ∑ 𝑛𝑛𝑖𝑖𝑘𝑘𝑖𝑖=1 , where 𝑘𝑘 is the number of studies and 𝑛𝑛𝑖𝑖 is the sample size. finally, it should be converted back to 𝑟𝑟𝑖𝑖 using the formula 𝑟𝑟𝑖𝑖 = 𝑒𝑒2�̅�𝑧𝑟𝑟𝑖𝑖−1 𝑒𝑒2�̅�𝑧𝑟𝑟𝑖𝑖+1 . in addition, the random-effects statistical model is applied to estimate the average distribution of true effects. the random-effects method was chosen because the effect size was extracted from a series of studies conducted by various authors in various populations at various times. the analysis also reports the estimate of the 2 index, the h2 index, the i2 index, and the q-test (cochran, 1954) with a p-value as the heterogeneity statistics outcome. the q-test is used to assess the null hypothesis that all effect sizes from all studies are homogenous (chen & peace, 2021). if the p-value is less than  (the typical significance level is 0.05), the null hypothesis should be rejected, indicating that the effect sizes from all studies are not homogenous. meanwhile, 2, h2, and i2 are used to determine the strength of the distribution of true effect sizes. the 2 index is estimated using the hedges’ estimator (hedges & olkin, 1985) to measure the variance of the true effect sizes, and the index should be greater than zero. the h2 index is quantified using higgins and thompson’s (2002) formula to inform the relative extent of heterogeneity in comparison to all studies, and the index should be greater than 1. the i2 index is also calculated using higgins and thompson's (2002) formula to determine the percentage of observed heterogeneity versus real heterogeneity. as a rule of thumb, the i2 index could be considered as having low heterogeneity (i2 = 25%), moderate heterogeneity (i2 = 50%), and high heterogeneity (i2 = 75%). to display the conclusions of meta-analyses, forest plots are generated. forest plots provide information about each study’s effect size and confidence interval, as well as the average distribution of true effects. the analysis also examines whether studies may be outliers and/or influential in the random-effect model. they can have a significant impact on the value of the estimated random-effect model coefficients, i.e., the intercept. if they had remained in the analysis, they could have changed the entire outcome. the standardized residuals are used to detect outliers, while the cook's distances (cook, 1977) and dffits (difference in fits) are applied to diagnose the influential studies. studies are considered as potential outliers if they have a standardized residual larger than 3 or smaller than -3 (rstudent > ± 3), while they are considered to be influential if the cook's distance value is more than 1 (cook.d > 1) and dffits is larger than 2 (dffits > 2) (gerbing, 2014). the meta-analysis was carried out with the help of open-source statistical software jamovi version 1.6.23 (the jamovi project, 2021). the major meta-analysis module library was used to compute r-to-z transformations, as well as to generate a random-effect model, heterogeneity statistics, a forest plot, and outlier and influential case diagnostics. results table 1 shows the results of data extraction processing. the eleven studies were published between 2001 and 2021, with abdullah and abd. majid (2001) is the longest and prastiwi (2021) is the most recent. the studies used two types of sampling methods: probability sampling and non-probability sampling. ababio and mawutor (2015); kassim et al. (2019); murdayanti et al. (2020); and yayeh (2014) applied the probability sampling method, whereas ismail et al. (2018); mei teh et al. (2019); priyo nugroho et al. a meta-analysis of the relationship between religiosity and saving behaviour method of analysis 𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒𝑒𝑒 � 1+𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 1−𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 �, 𝑧𝑧𝑧𝑧�̅�𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 , 𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖−1 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖+1 . τ2 a meta-analysis of the relationship between religiosity and saving behaviour method of analysis 𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 0.5𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑒𝑒𝑒𝑒 � 1+𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 1−𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 �, 𝑧𝑧𝑧𝑧�̅�𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖𝑧𝑧𝑧𝑧𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 ∑ 𝑛𝑛𝑛𝑛𝑖𝑖𝑖𝑖 𝑘𝑘𝑘𝑘 𝑖𝑖𝑖𝑖=1 , 𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖 = 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖−1 𝑒𝑒𝑒𝑒2𝑧𝑧𝑧𝑧�𝑟𝑟𝑟𝑟𝑖𝑖𝑖𝑖+1 . τ2 77 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 ta bl e 1 o ve rv ie w o f s tu di es in cl ud ed in m et aa na ly si s a ut ho rs (y ea r) sa m pl in g m et ho d m ea su re m en t te ch ni qu e v ar ia bl es st at is tic al a na ly si s r es ul t i nd ep en de nt d ep en de nt a bd ul la h an d a bd . m aj id (2 00 1) n ot m en tio ne d q ue st io nn ai re su rv ey r el ig io si ty in de x an d in co m e sa vi ng m ul tip le li ne ar re gr es si on “… th er e ex is t a c on cl us iv e re la tio ns hi p be tw ee n sa vi ng an d r el ig io si ty in de x … ” (t -s ta tis tic s = 1. 99 3, p < 0 .0 5) (p . 7 5) . y ay eh (2 01 4) m ul tis ta ge cl us te r s am pl in g an d pr ob ab ili ty pr op or tio na l t o si ze (p ps ) s am pl in g q ue st io nn ai re su rv ey re lig io n af fil ia tio n, re lig io us at te nd an ce , r el ig io n id en tit y, ho us eh ol d ne t i nc om e pe r m on th , g en de r, ho us eh ol d ac ce pt in g in te re st p ay m en t, le ve l o f e du ca tio n, fa m ily si ze , a ge , m ar ita l s ta tu s, w ea lth , a nd k no w le dg e ab ou t sa vi ng in te re st p ay m en t. sa vi ng l in ea r a nd p ro bi t re gr es si on “… th e m or e of te n pe op le go in g to c hu rc h/ m os qu e, i. e. th e m or e re lig io us th ey a re , th e lo w er is th ei r pr op en si ty to s av e m on ey .” (w al d ch isq ua re o f 6 2. 58 w ith p -v al ue of 0 .0 00 ) a ba bi o an d m aw ut or (2 01 5) si m pl e ra nd om sa m pl in g an d co nv en ie nc e sa m pl in g q ue st io nn ai re su rv ey re lig io si ty , u nc er ta in ty , liq ui di ty c on st ra in t, st ag e in lif e, a nd in te rg en er at io na l ef fe ct in co m e sa vi ng l og it m od el c hu rc h at te nd an ce v er y si gn ifi ca nt ly e xp la in s th at re lig io si ty e ff ec ts s av in g be ha vi or (o dd s ra tio = 0 .8 22 , p < 0. 05 ) ( p. 5 5) . (c on tin ue d) 78 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 a ut ho rs (y ea r) sa m pl in g m et ho d m ea su re m en t te ch ni qu e v ar ia bl es st at is tic al a na ly si s r es ul t i nd ep en de nt d ep en de nt pr iy o n ug ro ho et a l. (2 01 7) pu rp os iv e sa m pl in g q ue st io nn ai re su rv ey se lf -e ffi ca cy , r el ig io si ty , a tti tu de , a nd s ub je ct iv e no rm in te nt io n an d b eh av io r si m ul ta ne ou s e qu at io n m od el in g “… r el ig io si ty h as a p os iti ve an d si gn ifi ca nt in flu en ce o n cu st om er b eh av io r us in g pr od uc ts a nd s er vi ce s of is la m ic b an ks ” (i .e . b an k sa vi ng s or d ep os its ) ( p. 4 4) . sa ts io s an d h ad jid ak is (2 01 7) sn ow ba ll sa m pl in g q ue st io nn ai re su rv ey re lig io si ty a nd s el fm as te ry fiv e in te nt io ns to sa vi ng s ub sc al es : th ri ft , s av in g in vo lv em en t, sa vi ng ha bi ts , s ha m e of de bt a nd n o ne ed to s av e pe ar so n co rr el at io n “… , r el ig io si ty is s ig ni fic an tly po si tiv el y co rr el at ed w ith al l 5 in te nt io n su bs ca le s, ... ” (p . 2 0) . t he s ub sc al e of sa vi ng h ab its is s ig ni fic an tly po si tiv el y co rr el at ed w ith re lig io si ty (r (1 00 ) = 0 .3 32 , p < 0. 01 ). is m ai l e t a l. (2 01 8) pu rp os iv e sa m pl in g q ue st io nn ai re su rv ey se rv ic e qu al ity , r el ig io us be lie f, an d kn ow le dg e. sa vi ng b eh av io r m ul tip le li ne ar re gr es si on “… r el ig io us b el ie f i s si gn ifi ca nt ly r el at ed to s av in g be ha vi ou r (t = 4 .6 0, p = 0. 00 ).” (p . 1 07 6) k as si m e t a l. (2 01 9) d is pr op or tio na te st ra tifi ed s am pl in g q ue st io nn ai re su rv ey fa m ily b ac kg ro un d, r el ig io si ty , a tti tu de , l ite ra cy , h ou se ho ld in co m e, a ge , l ev el o f e du ca tio n, a nd l oc al ity . sa vi ng b eh av io r m ul tip le li ne ar re gr es si on “… th e re su lts d em on st ra te th at r el ig io si ty , … a re n ot si gn ifi ca nt ly r el at ed to s av in g be ha vi or .” (p . 2 48 ) ( tst at is tic s = 1. 41 8) (c on tin ue d) 79 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 a ut ho rs (y ea r) sa m pl in g m et ho d m ea su re m en t te ch ni qu e v ar ia bl es st at is tic al a na ly si s r es ul t i nd ep en de nt d ep en de nt m ei t eh e t a l. (2 01 9) c on ve ni en ce sa m pl in g q ue st io nn ai re su rv ey in di vi du al c ha ra ct er is tic , so ci al is at io n, c og ni tiv e ab ili ty , r el ig io n fa ith , a nd se lf -e ffi ca cy . pr iv at e sa vi ng l og is tic re gr es si on “a s fo r re lig io us fa ith , d iv in e gu id an ce (o dd s ra tio = 6 .5 1) si gn ifi ca nt ly p re di ct ed a n in di vi du al ’s lik el ih oo d to s av e m on ey .” (p . 1 0) w ija ya e t a l. (2 01 9) c on ve ni en ce sa m pl in g q ue st io nn ai re su rv ey r el ig io si ty le ve l sa vi ng d ec is io ns c hi -s qu ar e te st “… a c hi -s qu ar e te st b et w ee n re lig io si ty le ve l a nd s av in g de ci si on s cr ite ri a, w hi ch sh ow ed th er e is a s ig ni fic an t di ffe re nc e (p < 0 .0 1) . m or e th an 6 0 pe r ce nt o f t he re sp on de nt s de ci de d to s av e m on ey in b m ts b ec au se of th ei r pr od uc ts b ei ng in ac co rd an ce w ith s ha ri a. ” (p . 14 75 ) ( ch isq ua re = 6 .4 63 67 ) m ur da ya nt i e t al . ( 20 20 ) pr op or tio na te st ra tifi ed ra nd om sa m pl in g q ue st io nn ai re su rv ey fi na nc ia l k no w le dg e, s el fco nt ro l, an d re lig io us b el ie fs . sa vi ng b eh av io r pa rt ia l l ea st sq ua re “… r el ig io us b el ie fs h av e a si gn ifi ca nt p os iti ve e ffe ct o n sa vi ng s be ha vi or , … ” (p . 8 ) (t -s ta tis tic s = 6. 77 , p < 0 .0 01 ) pr as tiw i ( 20 21 ) n ot m en tio ne d q ue st io nn ai re su rv ey r el ig io si ty , e nv ir on m en t, an d r ep ut at io n. sa vi ng d ec is io n m ul tip le li ne ar re gr es si on “… r el ig io si ty , . .. ha ve a si gn ifi ca nt p os iti ve e ffe ct o n sa vi ng d ec is io ns .” (p . 2 22 ) (t -s ta tis tic s = 2. 16 1, p < 0 .0 5) (c on tin ue d) 80 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 ta bl e 2 o ve rv ie w o f d es cr ip tiv e st at is tic s an d e ffe ct s iz e a ut ho rs (y ea r) n a ge (y /o ) g en de r m ar ita l s ta tu s r el ig io n l oc at io n (c ou nt ry ) pe ar so n r a bd ul la h an d a bd . m aj id (2 00 1) 16 0 78 .1 3% 1 8 to 2 3 19 .3 8% 2 4 to 2 9 2. 50 % 3 0 to 3 5 34 .3 8% m al e 65 .6 3% f em al e 95 .6 3% s in gl e 4. 38 % m ar ri ed m us lim in te rn at io na l i sl am ic u ni ve rs ity m al ay si a (i iu m ) se la ng or (m al ay si a) 0. 15 76 y ay eh (2 01 4) 38 4 42 (m ea n) n ot m en tio ne d 74 % m ar ri ed 16 % w id ow ed 67 .5 % o rt ho do x c hr is tia n 30 .8 % m us lim 1. 7% p ro te st an t w es t a m ha ra n at io na l r eg io na l st at e (e th io pi a) 0. 40 37 a ba bi o an d m aw ut or (2 01 5) 20 0 n ot m en tio ne d n ot m en tio ne d n ot m en tio ne d c hr is tia n a cc ra m et ro po lit an (g ha na ) 0. 05 4 pr iy o n ug ro ho et a l. (2 01 7) 22 0 42 % le ss 3 1 45 % 3 1 to 4 0 13 % a bo ve 4 0 n ot m en tio ne d n ot m en tio ne d m us lim y og ya ka rt a (i nd on es ia ) 0. 56 6 sa ts io s an d h ad jid ak is (2 01 7) 10 0 n ot m en tio ne d n ot m en tio ne d n ot m en tio ne d m us lim x an th i, r od op i a nd e vr os t hr ac e (g re ec e) 0. 33 2 is m ai l e t a l. (2 01 8) 15 0 2. 7% le ss 2 0 54 % 2 0 to 3 0 29 .3 % 3 1 to 4 0 14 % a bo ve 4 0 42 .7 % m al e 57 .3 % f em al e n ot m en tio ne d 61 .3 % m us lim 15 .3 % b ud dh a 12 .7 % h in du 7. 3% c hr is tia n 3. 3% o th er s (m al ay si a) 0. 37 56 (c on tin ue d) 81 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 a ut ho rs (y ea r) n a ge (y /o ) g en de r m ar ita l s ta tu s r el ig io n l oc at io n (c ou nt ry ) pe ar so n r k as si m e t a l. (2 01 9) 53 1 n ot m en tio ne d n ot m en tio ne d 51 .2 % m ar ri ed 48 .8 % s in gl e m us lim se la ng or (m al ay si a) 0. 06 15 m ei t eh e t a l. (2 01 9) 22 4 16 to 6 0 n ot m en tio ne d n ot m en tio ne d m us lim (m al ay si a) 0. 45 88 w ija ya e t a l. (2 01 9) 52 3 2. 68 % 1 0 to 2 0 13 % 2 1 to 2 5 18 .1 6% 2 6 to 3 0 20 .6 5% 3 1 to 3 5 21 .0 3% 3 6 to 4 0 24 .4 7% a bo ve 40 50 .8 6% m al e 49 .1 4% f em al e n ot m en tio ne d m us lim su ra ka rt a an d su ko ha rj o (i nd on es ia ) 0. 11 12 m ur da ya nt i e t al . ( 20 20 ) 26 8 13 to 2 0 n ot m en tio ne d n ot m en tio ne d m us lim d ar un na ja h is la m ic b oa rd in g sc ho ol ja ka rt a (i nd on es ia ) 0. 41 35 pr as tiw i ( 20 21 ) 10 0 12 % le ss 1 8 27 % 1 8 to 2 5 36 % 2 5 to 3 0 25 % 3 0 to 4 0 n ot m en tio ne d 45 % m al e 55 % f em al e m us lim k sp ps b m t a m an ah u m m ah s ur ab ay a (i nd on es ia ) 0. 21 61 82 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 results table 1 shows the results of data extraction processing. the 11 studies were published between 2001 and 2021, with abdullah and abd. majid (2001) being the oldest and prastiwi (2021) is the most recent. the studies used two types of sampling methods: probability sampling and non-probability sampling. ababio and mawutor (2015); kassim et al. (2019); murdayanti et al. (2020); and yayeh (2014) applied the probability sampling method, whereas ismail et al. (2018); mei teh et al. (2019); priyo nugroho et al. (2017); satsios and hadjidakis (2017); and wijaya et al. (2019) employed the non-probability sampling method. meanwhile, abdullah and abd. majid (2001); and prastiwi (2021) there was no mention of the method used in their studies. to collect primary data, all studies developed a self-administered questionnaire. furthermore, various themes of religiosity and saving behaviour, such as religious attendance (yayeh, 2014), religious belief (ismail et al., 2018; murdayanti et al., 2020), religion faith (mei teh et al., 2019), saving habits (satsios & hadjidakis, 2017), and saving decisions (prastiwi, 2021; wijaya et al., 2019), were used to as the independent and dependent variables. various statistical analyses were also applied, namely pearson correlation (satsios & hadjidakis, 2017), chi-square test (wijaya et al., 2019), multiple linear regression (abdullah & abd. majid, 2001; ismail et al., 2018; kassim et al., 2019; prastiwi, 2021), logit regression (ababio & mawutor, 2015; mei teh et al., 2019), probit regression (yayeh, 2014), partial least square (murdayanti et al., 2020), and simultaneous equation modelling (priyo nugroho et al., 2017). on the other hand, various fit test indicators, such as chi-square (wijaya et al., 2019; yayeh, 2014), odds ratio (ababio & mawutor, 2015; mei teh et al., 2019), and t-statistic (abdullah & abd. majid, 2001; ismail et al., 2018; kassim et al., 2019; prastiwi, 2021), have been used to assess the significance of the relationship between religiosity and saving behaviour. these indicators had to be converted into pearson’s r before they are used in a meta-analysis. meanwhile, priyo nugroho et al. (2017) have provided another fit test indicator, namely the critical ratio (8.395) (personal communication). table 2 summarizes the descriptive statistics from each study, including the number of observations, age, gender, marital status, religion, and location, as well as the estimated effect sizes. as previously stated, 11 articles were used as samples for the meta 83 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 analytic study, and these gave a total of 1,063 observations. kassim et al. (2019) observed 531 people, making it the largest population sample, while prastiwi (2021); as well as satsios and hadjidakis (2017) had 100 observations, making it the smallest. meanwhile, the age range varied from 13 to more than 40 years. furthermore, only four studies provided gender information, i.e., abdullah and abd. majid (2001); ismail et al. (2018); prastiwi (2021); and wijaya et al. (2019), in which most of their respondents were women. similarly, marital status was only provided by three studies, i.e., abdullah and abd. majid (2001); kassim et al. (2019); and yayeh (2014), with the majority of their respondents being married. on the other hand, all studies provided religious information, with muslims representing the majority of their respondents. furthermore, the study locations reveal that respondents were from a variety of countries, including ethiopia, ghana, greece, malaysia, and indonesia. meanwhile, all studies were ready to use the same method to express effect sizes in pearson’s r. the study conducted by priyo nugroho et al. (2017) had the largest effect size (0.566), while the study performed by ababio and mawutor (2015) had the smallest one (0.054). table 3 random-effect model estimate se z p ci lower bound ci upper bound intercept 0.303 0.0587 5.17 < .001 0.188 0.418 table 4 heterogeneity statistics i² h² df q p 0.181 0.0328 (se= 0.0171) 89.08% 9.155 10 108.574 < .001 table 3 provides the random-effects model, and table 4 presents the heterogeneity statistics. based on the random-effects model, the estimated average fisher r-to-z transformed correlation coefficient was 0.303. (95% ci: 0.188 to 0.418) and was statistically significant (z = 5.17, p < 0.001). it showed that statistically there was a positive correlation between religiosity and saving behaviour, with a true effect table 3 random-effect model estimate se z p ci lower bound ci upper bound intercept 0.303. 0.0587 5.17 < .001 0.188 0.418 table 4 heterogeneity statistics  ² i² h² df q p 0.181 0.0328 (se= 0.0171) 89.08% 9.155 10 108.574 < .001 note: random-effects model (k = 11); ² estimator: hedges table 5 the source of heterogeneity analysis heterogeneity source coefficients t p 95% confidence interval lower upper publication year -0.131 -0.395 0.703 -0.895 0.634 sample size -0.334 -1.007 0.343 -1.098 0.431 muslim religion -0.535 -1.79 0.111 -1.22 0.153 figure 1 forest plot table 3 random-effect model estimate se z p ci lower bound ci upper bound intercept 0.303. 0.0587 5.17 < .001 0.188 0.418 table 4 heterogeneity statistics  ² i² h² df q p 0.181 0.0328 (se= 0.0171) 89.08% 9.155 10 108.574 < .001 note: random-effects model (k = 11); ² estimator: hedges table 5 the source of heterogeneity analysis heterogeneity source coefficients t p 95% confidence interval lower upper publication year -0.131 -0.395 0.703 -0.895 0.634 sample size -0.334 -1.007 0.343 -1.098 0.431 muslim religion -0.535 -1.79 0.111 -1.22 0.153 figure 1 forest plot 84 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 size of 0.303 and a significance value less than 0.001. meanwhile, according to heterogeneity statistics, the true effects appear to be nonhomogenous (q(10) = 108.574, p < 0.001). by using the hedges’ estimator, the index (0.0328) agreed with the q-test result, indicating that there was some between-study heterogeneity in the data, whereas = 0.181 indicated that the true effect sizes had an estimated standard deviation of sd = 0.181. in addition, the h2 (9.155) and i2 (89.08%) indices confirmed that true effect size differences account for more than half of the variation in the studies, which meant that the level of heterogeneity was high. table 5 the source of heterogeneity analysis heterogeneity source coefficients t p 95% confidence interval lower upper publication year -0.131 -0.395 0.703 -0.895 0.634 sample size -0.334 -1.007 0.343 -1.098 0.431 muslim religion -0.535 -1.79 0.111 -1.22 0.153 based on a random-effects model, figure 1 depicts correlation coefficients with corresponding 95 percent confidence intervals for each study graphically. the estimated r from each study ranged from 0.0541 (ababio & mawutor, 2015) to 0.6416 (priyo nugroho et al., 2017). meanwhile, the weights ranged from 7.99 percent to 9.92 percent, in prastiwi (2021), with satsios and hadjidakis (2017) having the lowest and kassim et al. (2019) having the highest. on the other hand, with kassim et al. (2019) having the shortest interval and prastiwi (2021) having the longest interval, the 95 percent confidence intervals ranged from (-0.02, 0.15) to (0.02, 0.42). furthermore, the majority of confidence intervals were completely positive of zero, indicating that the majority of studies had a statistically significant positive effect. however, in some studies, i.e., ababio and mawutor (2015) and kassim et al. (2019), the confidence intervals were not entirely positive of zero, indicating that these studies had a statistically insignificant positive effect (0.05 and 0.06). thus, all observed dispersions reflected genuine differences in effect size and p-value between studies. however, a meta-analysis method only uses the effect size from each study rather than the p-value (borenstein et al., 2011). 85 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 figures 2, 3, and 4 provide the outcomes of outlier and influential case diagnostics. figure 2 shows the standardized residual values, which are used to identify outliers, while figures 3 and 4 show the dffits values and cook’s distances, which are used to detect influential cases. a look at the studentized residuals revealed that none of the studies had a value greater than ±3, indicating that there were no outliers in the random-effect model. meanwhile, based on the dffits values of no more than 2 and cook’s distances of no more than 1, none of the studies could be considered overly influential. figure 1 forest plot figure 2 standardized residual table 3 random-effect model estimate se z p ci lower bound ci upper bound intercept 0.303. 0.0587 5.17 < .001 0.188 0.418 table 4 heterogeneity statistics  ² i² h² df q p 0.181 0.0328 (se= 0.0171) 89.08% 9.155 10 108.574 < .001 note: random-effects model (k = 11); ² estimator: hedges table 5 the source of heterogeneity analysis heterogeneity source coefficients t p 95% confidence interval lower upper publication year -0.131 -0.395 0.703 -0.895 0.634 sample size -0.334 -1.007 0.343 -1.098 0.431 muslim religion -0.535 -1.79 0.111 -1.22 0.153 figure 1 forest plot 14 figure 2 standardized residual figure 3 dffits values 86 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 figure 3 dffits values figure 4 cook’s distances the current study’s goal was to examine the existing empirical evidence on the relationship between religiosity and saving behaviour through a meta-analysis approach. the findings show that religiosity has a positive, but weak correlation with saving behavior with r = 0.303 (p < 0.001; 95% ci = (0.188; 0.418)). as a rule of thumb, the correlation strength is described as negligible (r < 0.2), low (0.2 ≤ r < 0.4), moderate (0.4 ≤ r < 0.7), high (0.7 ≤ r < 0.9), or very high (r ≥ 0.9) (guilford & fruchter, 1973). according to this rule of thumb, the correlation will be ignored when its strength is less than 0.2. this finding implies that changes in religiosity have little impact on changes in saving behaviour. moreover, the low r value was caused by the fact that the r values of the observed studies were mostly low (see figure 1), and even two of them were negligible, i.e., ababio and 14 figure 2 standardized residual figure 3 dffits values 15 figure 4 cook's distances the current study's goal was to examine the existing empirical evidence on the relationship between religiosity and saving behaviour through a meta-analysis approach. the findings show that religiosity has a positive, but weak correlation with saving behavior with r = 0.303 (p < 0.001; 95% ci = (0.188; 0.418)). as a rule of thumb, the correlation strength is described as negligible (r < 0.2), low (0.2 ≤ r < 0.4), moderate (0.4 ≤ r < 0.7), high (0.7 ≤ r < 0.9), or very high (r ≥ 0.9) (guilford & fruchter, 1973). according to this rule of thumb, the correlation will be ignored when its strength is less than 0.2. this finding implies that changes in religiosity have little impact on changes in saving behaviour. moreover, the low r value was caused by the fact that the r values of the observed studies were mostly low (see figure 1), and even two of them were negligible, i.e., ababio and mawutor (2015), and kassim et al. (2019). meanwhile, there were only four studies that provided empirical evidence that religiosity had a moderate impact on saving behaviour, namely mei teh et al. (2019), murdayanti et al. (2020), priyo nugroho et al. (2017), and yayeh (2014). on the other hand, ababio and mawutor (2015) obtained the lowest r value (0.05), showing that religious belief had no effect on household savings. they assumed that religious beliefs that promoted values like frugality, hard work, and honesty did not increase the savings habits of households in ghana. likewise, kassim et al. (2019) found that religiosity had no effect on saving behaviour (r = 0.06). they assumed that an individual's level of religiosity would lead to the preference to spend money in god's name rather than saving it. unfortunately, neither of them tried to dig a bit deeper into this finding, such as linking it to the lowest frequency value of the saving questions in their questionnaire. in fact, it could be a clue to crucial evidence showing the low impact of the relationship between religiosity and saving behaviour. the result obtained also supports the notion that religious people have varying understandings of saving behaviour. these understandings stem from two opposing perspectives on saving money taught by various religions, including christianity and islam. these religions consider saving to be either a positive or negative practice (yayeh, 2014). in terms of saving as a positive practice, christianity and islam, for example, teach saving when there is enough to go around in times of scarcity (genesis 37-50; sahih albukhari 5357). religious people who put into practice this belief will also use their saving behaviour to support their frugal lifestyle (agarwala et al., 2019). meanwhile, in terms of saving as a negative practice, religions such as christianity and islam, for example, warn against the dangers of hoarding wealth (luke 12:16-21; quran 9:34). religious people who engage in this practice are motivated by fears that if saving money becomes a passion, they will become trapped in a cycle of amassing wealth and being stingy one 87 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 mawutor (2015), and kassim et al. (2019). meanwhile, there were only four studies that provided empirical evidence that religiosity had a moderate impact on saving behaviour, namely mei teh et al. (2019), murdayanti et al. (2020), priyo nugroho et al. (2017), and yayeh (2014). on the other hand, ababio and mawutor (2015) obtained the lowest r value (0.05), showing that religious belief had no effect on household savings. they assumed that religious beliefs that promoted values like frugality, hard work, and honesty did not increase the savings habits of households in ghana. likewise, kassim et al. (2019) found that religiosity had no effect on saving behaviour (r = 0.06). they assumed that an individual’s level of religiosity would lead to the preference to spend money in god’s name rather than saving it. unfortunately, neither of them tried to dig a bit deeper into this finding, such as linking it to the lowest frequency value of the saving questions in their questionnaire. in fact, it could be a clue to crucial evidence showing the low impact of the relationship between religiosity and saving behaviour. the result obtained also supports the notion that religious people have varying understandings of saving behaviour. these understandings stem from two opposing perspectives on saving money taught by various religions, including christianity and islam. these religions consider saving to be either a positive or negative practice (yayeh, 2014). in terms of saving as a positive practice, christianity and islam, for example, teach saving when there is enough to go around in times of scarcity (genesis 37-50; sahih al-bukhari 5357). religious people who put into practice this belief will also use their saving behaviour to support their frugal lifestyle (agarwala et al., 2019). meanwhile, in terms of saving as a negative practice, religions such as christianity and islam, for example, warn against the dangers of hoarding wealth (luke 12:16-21; quran 9:34). religious people who engage in this practice are motivated by fears that if saving money becomes a passion, they will become trapped in a cycle of amassing wealth and being stingy one day (sawyer, 1954). as a result, in studies where there is empirical evidence that religiosity has a moderate impact on saving behaviour, their respondents are likely to view saving as a positive practice. in contrast, respondents in studies with empirical evidence that religiosity has a low impact on saving behaviour may be less open to the idea of saving as a positive practice. in addition, the current study discovered heterogeneity at a high level (q(10) = 108.574, p < 0.001; i2 = 89.08%). heterogeneity may exist, 88 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 because of differences in study quality (e.g., values for effect size and significance), methodology, sample size, demographic factors, and respondent characteristics. from a statistical perspective, quantifying heterogeneity can be used to determine whether or not population effect sizes are likely to be consistent or varying (borenstein et al., 2011; hedges & olkin, 1985). meanwhile, the summary information shown in tables 1 and 2 demonstrates that the characteristics of studies were not all the same. there were differences even within the studies themselves, like variation in religious factors (see table 2). for instance, ismail et al. (2018); and yayeh (2014) collected data from a variety of religious adherents, whereas other studies only collected data from a single religious adherent. as a result, the present study was motivated to investigate the cause of heterogeneity using a technique known as meta-regression analysis. in meta-regression, the effect size of each study as the independent variable is regressed on the study characteristics as the dependent variable (chen & peace, 2021). furthermore, the year of publication, sample size, and muslim religion were investigated to determine the source of heterogeneity. they acted as independent variables in the meta-regression analysis because they were relatively complete data. the results of metaregression analysis then revealed that publication year, sample size, and muslim religion were not statistically significant for the presence of heterogeneity (see table 5). conclusion and future studies the current study used a meta-analysis to synthesize the findings of previous studies to determine the effect of religiosity on saving behaviour. the meta-analysis of 11 journal articles and 1,063 respondents revealed that religiosity has a low impact on saving behaviour. the current study also confirmed the notion that religious people have two different perspectives on saving behaviour, holding the divergent view that saving can be either a negative or positive practice. because the study’s findings indicated that religiosity has little influence on saving behaviour, it is possible that people with a high level of religiosity have a less rigid perspective on saving as a positive behaviour. the findings have important implications for the development of theories in the field of saving behaviour. in researching religiosity and 89 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 saving behaviour, this study has pioneered a new analytic approach known as meta-analysis. it has provided a retrospective summary of the existing literature on the relationship between religiosity and saving behaviour. it could help researchers make a more informed decision about which variables to use. furthermore, this study has added to the body of knowledge about the true effect of the relationship between religiosity and saving behaviour. it could be used as a model for future research as well as a tool for additional analysis. the findings also have practical implications for increasing public awareness. this study has provided insights for authorities and financial institutions interested in encouraging religious people to save. more understanding of the findings can aid them in the improvement of plans focused on savings advocacy and savings facilitation. savings advocacy would help religious households and individuals understand the importance of saving and resource management. it would be a critical step in developing their saving habits. meanwhile, financial institutions can create a type of savings account that corresponds to the religious predisposition. it would promote the notion that saving money is a good deed. the plans are beneficial in encouraging religious people to engage in responsible financial behaviour. collective savings will have a significant impact on the economy if these individuals have a strong savings predisposition. however, a number of important limitations of the study have to be considered. first, the current meta-analytic method has synthesized only a few studies. although the limited sample used in the present study was acceptable, the findings would be more reliable if a larger number of studies could be synthesized. second, of the 11 previous studies that were synthesized, islam was the religion which had the highest proportion of adherents, with only a small proportion of adherents from other religions. although that was the nature of the meta-analytic samples, the current pattern of results could be more generalizable if the proportion of adherents was more balanced. third, the current study synthesized 11 studies had used primary data and measured variables with a questionnaire. this technique opens the possibility of a person’s proclivity to respond to a questionnaire with a positive self-image, also known as a socially desirable response (van de mortel, 2008). although bias has been reduced by selecting studies from reputable journals, it is still possible to be biased when answering sensitive questions, such as religious ones. 90 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 furthermore, more research is needed to examine the relationship between religiosity and saving behaviour, more specifically, the importance of using longitudinal studies at multiple points in time. this type of research could shed light on the minor impact of the relationship and contribute to a critical explanation for this phenomenon. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. references ababio, a. g., & mawutor, g. (2015). does religion matter for savings habit of households in ghana? singaporean journal of business economics and management studies, 4(8). https:// doi.org/10.12816/0019680 abdullah, n., & abd. majid, m. s. (2001). saving behaviour in islamic framework: the case of international islamic university malaysia. jurnal syariah, 9(2), 61-84. https://mjs.um.edu.my/ index.php/js/article/view/22905. agarwala, r., mishra, p., & singh, r. (2019). religiosity and consumer behavior: a summarizing review. journal of management, spirituality & religion, 16(1), 32-54. https://doi.org/10.1080/ 14766086.2018.1495098 ajzen, i. (1991). the theory of planned behavior. organizational behavior and human decision processes, 50(2), 179-211. https://doi.org/10.1016/0749-5978(91)90020-t allport, g. w., & ross, j. m. (1967). personal religious orientation and prejudice. journal of personality and social psychology, 5(4), 432-443. https://doi.org/10.1037/h0021212 arruñada, b. (2010). protestants and catholics: similar work ethic, different social ethic. the economic journal, 120(547), 890918. https://doi.org/10.1111/j.1468-0297.2009.02325.x azzi, c., & ehrenberg, r. (1975). household allocation of time and church attendance. journal of political economy, 83(1), 27-56. https://doi.org/10.1086/260305 bandura, a. (1977). social learning theory. prentice-hall. 91 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 barro, r. j., & mccleary, r. m. (2003). religion and economic growth across countries. american sociological review, 68(5), 760. https://doi.org/10.2307/1519761 basedau, m., gobien, s., & prediger, s. (2018). the multidimensional effects of religion on socioeconomic development: a review of the empirical literature. journal of economic surveys, 32(4), 1106-1133. https://doi.org/10.1111/joes.12250 berry, j. w., poortinga, y. h., segall, m. h., & dasen, p. r. (2002). cross-cultural psychology: research and applications: cambridge university press. bilen, m. (2016). global wealth inequality: islamic economic perspective for market or state oriented solution. turkish journal of islamic economics, 3(1), 17-35. https://doi. org/10.15238/tujise.2014.3.1.17-35 borenstein, m., hedges, l. v., higgins, j. p. t., rothstein, h. r., & higgins, d. j. p. t. (2011). introduction to meta-analysis. john wiley & sons. chen, d.-g. d., & peace, k. e. (2021). applied meta-analysis with r and stata. crc press. cheung, m. w.-l., ho, r. c. m., lim, y., & mak, a. (2012). conducting a meta-analysis: basics and good practices. international journal of rheumatic diseases, 15(2), 129-135. https://doi.org/10.1111/j.1756-185x.2012.01712.x cochran, w. g. (1954). some methods for strengthening the common χ2 tests. biometrics, 10(4), 417-451. https://doi. org/10.2307/3001616 cook, r. d. (1977). detection of influential observation in linear regression. technometrics, 19(1), 15-18. https://doi. org/10.2307/1268249 davutyan, n., & öztürkkal, b. (2016). determinants of savingborrowing decisions and financial inclusion in a high middle income country: the turkish case. emerging markets finance and trade, 52(11), 2512-2529. https://doi.org/10.1080/154049 6x.2016.1187596 field, a. p., & gillett, r. (2010). how to do a meta-analysis. british journal of mathematical and statistical psychology, 63(3), 665-694. https://doi.org/10.1348/000711010x502733 filipova, l. (2012). are attitudes conducive to economic growth stronger in protestants than in others? https://nupi.brage. unit.no/nupi-xmlui/bitstream/handle/11250/277240/wp--804lenka-r.pdf?sequence=3 92 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 geertz, c. (1973). the interpretation of cultures: selected essays (vol. 5019): basic books. gerbing, d. (2014). influence of data points on the estimated model. school of business administration portland state university. http://web.pdx.edu/~gerbing/515/resources/influence.pdf glock, c. y. (1962). on the study of religious commitment. religious education, 57(sup4), 98-110. https://doi. org/10.1080/003440862057s407 guilford, j. p., & fruchter, b. (1973). fundamental statistics in psychology and education (5th ed.). mcgraw-hill. guiso, l., sapienza, p., & zingales, l. (2003). people’s opium? religion and economic attitudes. journal of monetary economics, 50(1), 225-282. https://doi.org/10.1016/s03043932(02)00202-7 hayward, r. d., & kemmelmeier, m. (2011). weber revisited: a cross-national analysis of religiosity, religious culture, and economic attitudes. journal of cross-cultural psychology, 42(8), 1406-1420. https://doi.org/10.1177/0022022111412527 hedges, l., & olkin, i. (1985). statistical methods for meta-analysis. academic press. higgins, j. p., & thompson, s. g. (2002). quantifying heterogeneity in a meta-analysis. statistics in medicine, 21(11), 1539-1558. https://doi.org/10.1002/sim.1186 iannaccone, l. r. (1998). introduction to the economics of religion. journal of economic literature, 36(3), 1465-1495. inglehart, r. (2018). culture shift in advanced industrial society. princeton university press. ismail, s., khairuddin, n. s., alias, n. e., loon-koe, w., & othman, r. (2018). an empirical analysis of saving behavior among malaysian employees. international journal of academic research in business and social sciences, 8(10), 1070-1080. https://doi.org/10.6007/ijarbss/v8-i10/4822 iyer, s. (2016). the new economics of religion. journal of economic literature, 54(2), 395-441. https://doi.org/10.1257/ jel.54.2.395 kassim, a. a. m., mohamed, j., jahari, n. a., nasyikin, n., & zain, m. (2019). personal saving behavior among muslim employees. international journal of academic research in business and social sciences, 9(11). https://doi.org/10.6007/ ijarbss/v9-i11/6545 keister, l. a. (2003). religion and wealth: the role of religious affiliation and participation in early adult asset accumulation. 93 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 social forces, 82(1), 175-207. https://doi.org/10.1353/ sof.2003.0094 klaubert, a. (2010). striving for savings: religion and individual economic behavior working paper series in economics. lüneburg: leuphana universität lüneburg, institut für volkswirtschaftslehre. law, k. s. (1995). the use of fisher’s z in schmidt-hunter. type meta-analyses. journal of educational and behavioral statistics, 20(3), 287-306. https://doi.org/10.2307/1165404 lenhard, w., & lenhard, a. (2016). calculation of effect sizes. https://www.psychometrica.de/effect_size.html lenski, g. (1961). the religious factor: a sociological study of religion’s impact on politics, economics, and family life. garden city. mei teh, g., palil, m. r., mohd zain, n. h., & mamat, n. h. (2019). predicting private saving within malay community. asian academy of management journal, 24(supp.1), 1-17. https:// doi.org/10.21315/aamj2019.24.s1.1 murdayanti, y., ulupui, i., gurendrawati, e., & purwohedi, u. (2020). an assessment of saving behavior in the religious education institution. journal of southwest jiaotong university, 55(4). https://doi.org/10.35741/issn.0258-2724.55.4.27 nyhus, e. k. (2017). saving behaviour: economic and psychological approaches. in r. ranyard (ed.), economic psychology (pp. 206). john wiley & sons. pargament, k. i. (1999). the psychology of religion and spirituality? yes and no. the international journal for the psychology of religion, 9(1), 3-16. https://doi.org/10.1207/ s15327582ijpr0901_2 pew research center. (2015). the future of world religions: population growth projections, 2010-2050. https://assets.pewresearch. org/wpcontent/uploads/sites/11/2015/03/pf_15.04.02_ projectionsfullreport.pdf prastiwi, i. e. (2021). analysis of the influence of religiosity, environment, and reputation factors on customer decisions on saving at kspps bmt amanah ummah. international journal of economics, business and accounting research (ijebar), 5(1). https://doi.org/10.29040/ijebar.v5i1.2237 priyo nugroho, a., hidayat, a., & kusuma, h. (2017). the influence of religiosity and self-efficacy on the saving behavior of the іslamic banks. banks and bank systems, 12(3), 35-47. https:// doi.org/10.21511/bbs.12(3).2017.03 94 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 67–94 renneboog, l., & spaenjers, c. (2012). religion, economic attitudes, and household finance. oxford economic papers, 64(1), 103– 127. https://doi.org/10.1093/oep/gpr025 rosenthal, r. (1995). writing meta-analytic reviews. psychological b u l l e t i n , 11 8 ( 2 ) , 1 8 3 . h t t p s : / / d o i . o r g / 1 0 . 1 0 3 7 / 0 0 3 3 2909.118.2.183 satsios, n., & hadjidakis, s. (2017). the effect of religiosity and self mastery on saving behaviour of pomak households in thrace. journal sustaibable development, culture, traditions, 17-27. https://doi.org/10.26341/issn.2241-4002-2017-1b-2 sawyer, w. (1954). christ’s teachings on the use of money. https:// scholarworks.harding.edu/wyatt-sawyersermons/281 singh, g., sharma, s., sharma, r., & dwivedi, y. k. (2021). investigating environmental sustainability in small familyowned businesses: integration of religiosity, ethical judgment, and theory of planned behavior. technological forecasting and social change, 173, 121094. https://doi.org/10.1016/j. techfore.2021.121094 the jamovi project. (2021). jamovi (version 1.6) [computer software]. https://www.jamovi.org teshome, h. m., ayalew, g. d., shiferaw, f. w., leshargie, c. t., & boneya, d. j. (2018). the prevalence of depression among diabetic patients in ethiopia: a systematic review and metaanalysis. depression research and treatment, 6135460. https:// doi.org/10.1155/2018/6135460 tjahjono, g. (2014). religiosity and the intention to buy luxury goods among young indonesian muslims in jakarta (unpublished master’s thesis). university of western sydney. van de mortel, t. f. (2008). faking it: social desirability response bias in self-report research. australian journal of advanced nursing, 25(4), 40-48. weber, m. (1905). the protestant ethic and the spirit of capitalism. charles scribner’s sons. wijaya, i. f., hakim, a. r., saputro, n., & mulyadi, m. (2019). religiosity level and saving decisions in baitul maal wat tamwil: the case of indonesia. journal of islamic marketing, 11(6), 1465-1483. https://doi.org/10.1108/jima-09-2018-0160 yayeh, f. a. (2014). the impact of religion on household saving behavior. microeconomics: intertemporal consumer choice & savings ejournal. https://doi.org/10.2139/ssrn.2518129 zuckerman, m., silberman, j., & hall, j. a. (2013). the relation between intelligence and religiosity: a meta-analysis and some proposed explanations. personality and social psychology review, 17(4), 325-354. https://doi.org/10.1177/1088868313497266 115 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 how to cite this article: alananga, s. s. (2022). sector lending concentration and credit risk: an evaluation of lender perceptions in tanzania. international journal of banking and finance, 17(2), 115-151. https://doi. org/10.32890/ ijbf2022.17.2.5 sector lending concentration and credit risk: an evaluation of lender perceptions in tanzania samwel sanga alananga department of business studies ardhi university, tanzania corresponding author: salanangasanga@gmail.com received: 29/5/2021 revised: 9/9/2021 accepted: 19/9/2021 published:27/6/2022 abstract this study contributes to the debate on the effect of lending concentration (lc) on credit risk. it is based on data from an online survey of 151 employees from 37 lending institutions, employees with diverse experience in the different sections of their institutions. following a successful three-factor solution on lc based on an exploratory factor analysis, binary logistic regression models were implemented to determine the perceived lending risk (plr) based on three types of lc, namely social status lending concentration (sslc), private sector lending concentration (pslc) and public employee lending concentration (pelc). it was noted that overconcentration based on social status provided an explanation for the increase in non-performing loans (npl) risk among both large and small lenders. since lending concentration reverses the effect of macroeconomic variables, such as credit risk management practices (crmps), credit processing considerations (cpcs), as http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 116 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 well as collateral types, assessing the degree at which the lender is concentrated across sectors is imperative, prior to any credit risk management initiative. although lc directly affects lending risks perceptions alongside the traditional and corporate finance theories, the indirect lc effect via crmps, cpcs, bank size and originality, as well as the various collateral typologies seems to provide new insights into this area of research. keywords: lending concentration, non-performing loan, credit risk, credit risk management, bank lending and borrowing. jel classification: g21. introduction banks and other financial intermediaries exist partly because borrowers often have more information on their respective abilities to finance the loans than lenders (mol-gómez-vázquez, et al., 2018). borrowers’ advantage over information of projects requires monitoring by lenders to ensure that borrowers abstain from acting in their own selfinterest (abu hussain & al-ajmi, 2012). however, lenders may fail to accurately predict the behavior of borrowers, leading to the huge accumulation of non-performing loans (npl) ugoani, 2016; urt, 2019; francisa et al., 2018; le & diep, 2020; berger et al., 2017). as monitoring is also costly, financial intermediaries must rely on credit risk management practices (crmps) to allow credit information production and monitoring (purda, 2008). according to the bank for international settlements (2006), two forms of lc can be identified; the first is “name concentration”, which relates to the imperfect diversification of idiosyncratic risk in the portfolio, either because of its small size or because of large exposures to specific individual obligators (beck & de jonghe, 2013); and the second is “sector concentration”, which is synonymous to imperfect diversification across systematic components of risk, namely sectoral factors (simpasa & pla, 2016). along these lines, diversification may be associated with relatively higher lending risk, especially when venturing into new and competitive markets (le & diep, 2020; adzobu et al., 2017). diversification may also induce a higher npl risk via the reduced incentives to monitor in response 117 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 to agency conflict between equity holders and creditors of the lender (adzobu et al., 2017; paravisini et al., 2015; beck & de jonghe, 2013; acharya et al., 2006). beck and de jonghe (2013) however, suggests that lc causes higher lending risks among banks. to date there is no consensus despite the acknowledgement that lc increases information at the cost of higher portfolio risk, thus may not necessarily yield overall lower perceived lending risk (plr) among lenders (berger et al., 2017). at least two theoretical perspectives have been used to explain lc risks facing financial institutions. the first pertains to the standpoint of traditional portfolio theory (tpt), which suggests that diversification, whether across lending sectors or name/products, and services, provide an opportunity for lenders to mitigate plr. from the perspective of the tpt, lc is susceptible to the economic volatility of the sectors of lc than when they are well diversified (adzobu et al., 2017). the tpt’s suggestions are in favor of diversification, principally drawing from markowitz (1959) and supported by diamond (1984). the second theoretical perspective pertains to a tenet of corporate finance theory (cft) which claims that most of the lc effects accrue from the built-in nurturing of expertise (francisa et al., 2018). according to the cft, lenders should specialize by concentrating their activities on specific sectors, or lines of business in order to enjoy the comparative advantage of developing expertise in the areas of their focus (jensen, 1986; denis, denis, & sarin, 1997; adzobu et al., 2017). the cft has suggested that lc can also result in better screening of potential borrowers and loan applications, and there is more efficiency in monitoring, hence leading to a lower npl risk (beck & de jonghe, 2013). lc gives room for experienced expertise to develop, thus facilitating detection of deteriorating businesses early and taking the necessary appropriate measures (adzobu et al., 2017; böve et al., 2010). similarly, it might be possible to prevent risk shifting by borrowers (beck & de jonghe, 2013). this study intends to contribute to the debate on the effect of lc on npl risks by evaluating lender's perceptions on lending risks and the associated degree of lc. despite the abundant literature on lc and diversification in the developed world, little is known with regard to developing countries, mainly because of the problem of limited data (abu hussain & al-ajmi, 2012; adzobu et al., 2017). for example, 118 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 tanzania experienced a credit crunch and lenders suffered from huge losses as a result of the high npl rates (richard et al., 2008). similarly, lending has been concentrated more on the secondary rather than primary sectors (kira, 2013), whereby the incorporated companies have more chances of getting loans than the unincorporated ones. with regard to the crmps, the bank of tanzania (bot) requires each lending institution to prepare a comprehensive risk management programme (rmp) designed to its needs and circumstances under which it operates (bot, 2010). despite the treatment provided through the crmps, the bot (2019) has reported that the npls ratio declined to 9.8 percent from 10.4 percent, which however, was still above the desirable level. since the tanzania’s lending market has been highly concentrated (imf, 2018), it is possible that lc could be responsible for the persistently higher npl rates. efforts to curb the npl through the crmps and credit considerations in tanzania seemed to be massive, but the npl’s responses seemed to be very marginal. this study is carried out under the assumption that the npl does respond to certain specific lending considerations and the crmps, but not to all of them despite the fact that all are intended to lower the npl risk. literature review sector lending concentration sector specific observations in vietnam and lithuania have suggested that lc in sectors such as agriculture, manufacturing, transport and communication, construction and real estate and other communitywide sectors (electricity, gas and water) has been associated with higher perceived lending risk (plr), compared to service sectors such as housing, hotels and restaurants (le & diep, 2020; pham & lensink, 2008; skridulytė & freitakas, 2012). simpasa & pla (2016) provided evidence that for sectors where commercial lenders had high lc, i.e., private sector lending concentration (pslc), the corresponding npl risk tended to be lower to reflect monopoly rents accruing from superior information and monitoring efficiency or expertise in a given sector. these observations though contradictory, have pointed to the hypothesis that plr might be low in the pslc than in other forms of lc. 119 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 social status lending concentration (sslc) describes lc in terms of individual characteristics (hunter & nixon, 1999). theoretically, the existence of at least one major customer/borrower who accounts for 10 percent or more of the supplier’s induces a high plr among lenders (liu et al., 2019). beck and de jonghe (2013) observed that individual lender differentiation would contribute to more systemwide risks in the financial system. under the sslc, borrowers might transfer a part of their businesses’ achieved results to reduce the financial expenses associated with loan repayment (laib, 2013). similarly, in some western countries an individual’s share of the npl would be determined by external factors, i.e., institutional quality or lack of adequate crmps by lenders, which could contribute further to a higher npl risk (hunter & nixon, 1999; skridulytė & freitakas, 2012; liu et al., 2019). pham and lensink (2007) has suggested that in developing countries, an individual’s share of the npl might be linked to informality, whereby formal employment borrowers were more likely to default than their informal counterparts. these notable findings seem to suggest support for the hypothesis that the sslc could be associated with a high or low plr depending on the context. the public employee lending concentration (pelc) on the other hand, may be considered the safest given the certainty in income flow (le & nguyen, 2018; pham & lensink, 2008). liu et al. (2019) and yang (2017) however observes that, lc is associated with higher interest rate among bondholders to reflect the need to compensate for a higher plr. instead of reducing loans, lenders are more likely to increase monitoring through a shorter-term debt when facing a higher pelc (yang, 2017). borrowers’ occupation in lending (lenders) also reduces default risk (pham & lensink, 2008). as such, it is hypothesized that the pelc is associated with a relatively lower plr, compared to other sectors. the macroeconomic environment in addition to lc, macroeconomic conditions may contribute to increasing the npl risk (beck & de jonghe, 2013; trautmann & vlahu, 2013). the channel through which lc can lead to a higher plr can be via loan losses and difficulties in managing interest rate risk (paravisini, rappoport, & schnabl, 2015; beck & de jonghe, 2013; le & diep, 2020). lassoued (2017) notes that while gdp growth generally reflects higher income flows for small borrowers and an 120 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 increase in the profitability of firms, and therefore, better solvency, higher inflation rates constrain households in their consumption freedom. similarly, both the gdp and inflation experience affect the npl because of either lost lending relationship in times of low growth (trautmann & vlahu, 2013), or weak institutions or repeated crises where the contagious aggravation of crises are through wordof-mouth (trautmann & vlahu, 2013; iyer & puri, 2012). in terms of perceptions, psychological studies in the so called ‘pollyanna principle’ (dember & penwell, 1980), have well established that even with limited information, better fortunes are often perceived better (overstated) while misfortunes are perceived worse (understated) (de meza & southey, 1996). therefore, a booming economy might signal a higher repayment ability, thus attracting more borrowing even where actual repayment is not guaranteed. lender specific characteristics kira (2013), al-kayed (2020) and abu hussain & al-ajmi (2012) have suggested that lender typologies (islamic and non-islamic) determine the type and level of credit risk facing a financial institution. furthermore, relatively larger lenders tend to experience a higher npl risk from loan loss provisions (adzobu et al., 2017). however, if a large lender is diversified rather than concentrated, the ability to grab economies of scales is enhanced, leading to a relatively lower plr than that achieved by smaller lenders (beck & de jonghe, 2013). in contrast, simpasa and pla (2016) have suggested that there would be a higher npl risk in response to lc, regardless of a lender’s size. in addition, in another study by lassoued (2017), it was suggested that lender size had an insignificant effect on the plr. the effect of bank size based on these studies could be either positive or negative, it depended on other factors. similarly, the experience of the lender might affect risk perceptions. from the viewpoint of institutional memory hypothesis, whereby the less experienced key decision makers are, the higher the likelihood of lending risks underestimation (burakov, 2014; berger & udell, 2004). credit processing considerations a number of credit processing considerations may be pivotal in reducing the plr. the first is existing debt (cdebt), which however, is rarely considered important to microfinance institutions (mfis) and 121 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 small and medium enterprises (smes) who rely on soft information (abdulsaleh & worthington, 2016; dezoort, wilkins, & justice, 2017; lassoued, 2017; harif, hoe, & zali, 2011; berger, minnis, & sutherland, 2017). the second is income, capacity and savings, all of which captures the ability of the borrower to meet the periodic repayment of the interest and principle (adzobu et al., 2017; laib, 2013; abdulsaleh & worthington, 2016; koomson et al., 2016). there were cases however, where these capacity-related considerations had not been significant (lassoued, 2017). with regard to income (cincome) the plr tended to be relatively lower among high-income borrowers than low-income ones, thus providing an explanation for credit refusal among the poor (le & nguyen, 2018; pham & lensink, 2008; koomson et al., 2016); and credit history (credh) (abdulsaleh & worthington, 2016). additional considerations included concerns with collaterals (ccolla) and business plan (cbplan) (abdulsaleh & worthington, 2016). both of these reflected the applicant’s creditworthiness and borrower’s character, especially in the context of lending to highly risky sectors, such as the mfis and smes (abdulsaleh & worthington, 2016; laib, 2013; pham & lensink, 2007). other concerns being demographics, such as age whereby, the older are considered less risky (koomson et al., 2016; kira, 2013); the number of dependants, i.e., the more the number of dependants, the higher the plr (koomson et al., 2016; pham & lensink, 2007); and education, thereby those with a vocational or university degree would be associated with a low plr (kira, 2013). further considerations include the purpose of the loan (loanpur) and borrower’s employment status (cemploy). studies by le and nguyen (2018), klyuev (2008) and harrison et al. (2004) have suggested that the plr depended on the stability of income and borrowers’ business experience. it was found that those with higher business experience were associated with a low plr (kira, 2013), though abdulsaleh and worthington (2016) noted that if the loan was for a startup, experience was irrelevant. credit risk management practices lenders mitigate credit risk by using several practices, including the following: risk-based pricing (rbp) to diversify risk when lenders charge a high interest rate on risky borrowers (wood & kellman, 122 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 2010); impose different bank covenants (dbc) to borrowers to reflect borrowers’ creditworthiness (choppari & rajeshwar, 2015; liu et., 2019; campello and gao, 2017); use of credit insurance for loan granted (cilg) to transfer risk from the lender to the seller (insurer) in exchange for payment (prasad, 2016); the use of credit reference bureau (crb) to assess credit history of borrowers (ondabu, 2019); credit review (crev) of borrower income to assess current creditworthiness; bank guarantee to cover debt (gcd) in case of default; unlock loan of potential borrowers upon request (lpb) (seyram, 2013); transfer of loan to another bank; offer private banking to specific borrowers; and differential consideration of individual borrowers based on their employment status. the last three have however not been investigated with regard to its plr effect. different collateral types provision of collateral for many lenders is a policy issue, without it the loan application is most likely to be rejected (abdulsaleh & worthington, 2016). different types of collateral are accepted by lenders including the following: real estate, machinery, equipment and vehicles, products and goods (inventory), personal guarantees and accounts receivable (abdulsaleh & worthington, 2016). banks require a collateral in order to address any adverse selection (chan & thakor, 1987; stiglitz & weiss, 1981), and the moral hazard problem. it is a means to monitor borrowers’ behavior and ensure the success of the project for which the loan was granted (aghion & bolton, 1992). in rare cases, the collateral may indicate the purpose of the loan, thus increasing the safety margin of the lender (abdulsaleh & worthington, 2016). the npl risk reduction effect of these different types of collateral is however, a matter of empirical observations. theoretical determinants of perceived lending risks the conceptual model for this study is as provided in figure 1. figure 1 shows that enhancing the intensity of the use of the crmps have a positive and significant impact on the plr, while reducing such intensity increases the lender’s plr. lc also shapes the plr, however, its effect is not direct as lc is moderated by multiple exogenous variables. thus, enhanced lender’s lc in a particular sector leads to significant plr reduction, only if such lc is also associated with favorable perceptions on the macroeconomic environment (me), borrower differentiation (bd) via credit processing considerations and bank/lender characteristics (bc). 123 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 figure 1 conceptual model of determinants of lenders’ perceived lending risk (plr) methodology to evaluate the relationships portrayed in figure 1, an online survey was distributed to lenders via google forms from august 2020 to march 2021. the contact details of all respondents were obtained based on online social media, specifically linkedin and the personal contact details of the authors. the study also utilized a limited secondary data, specifically macroeconomic data from the world bank development indicators to compute the gdp and inflation experience. data description by 2015, tanzania had 63 banks and other financial institutions, of which 36 were commercial banks, 12 were regional and cooperative banks, three (3) were financial institutions, three (3) were 7 figure 1 conceptual model of determinants of lenders’ perceived lending risk (plr) methodology to evaluate the relationships portrayed in figure 1, an online survey was distributed to lenders via google forms from august 2020 to march 2021. the contact details of all respondents were obtained based on online social media, specifically linkedin and the personal contact details of the authors. the study also utilized a limited secondary data, specifically macroeconomic data from the world bank development indicators to compute the gdp and inflation experience. data description by 2015, tanzania had 63 banks and other financial institutions, of which 36 were commercial banks, 12 were regional and co-operative banks, three (3) were financial institutions, three (3) were microfinance banks, two (2) were development financial institutions, one (1) was the tanzania mortgage refinancing company, three (3) were lender characteristics macroeconomic environment (me) borrowers’ differentiation (bd) credit risk management practices (crmp) credit risk management practices (crmps) perceived lending risk (plr) customer lending concentration (lc) credit processing considerations (cpcs) + 124 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 microfinance banks, two (2) were development financial institutions, one (1) was the tanzania mortgage refinancing company, three (3) were finance leasing, one (1) was a representative bank and two (2) were credit reference bureau11. major commercial banks, such as the nmb bank, crdb bank ltd and akiba commercial bank (acb) have established considerable presence in rural productive sectors such as in agriculture. the tpb bank has a countrywide network of post offices, which could facilitate savings mobilization and money transfers. in addition to banks and financial institutions, tanzania also has several mfis, which are registered by different authorities and legislations22. the total number of employees in these banks, financial intermediaries and micro-finance institutions (mfis) is however, not determined for the purpose of this study. hence a convenient sampling strategy was adopted. similarly, since the focus of this study was on the experience of the lenders’ operatives through employees, it was necessary to capture the experience of the employees, given their current and previous employments. in sum, the experience of the respondents in the lending industry was of prime importance in this study. the questionnaire instrument was divided into four major parts; the first intended to capture basic information about the respondents and lenders, such as their work experience and in which types of employment departments. the second part captured the crmps in detail, using a five point likert scale to gather the associated reasons for practice variation across lenders. the third part paid some specific attention to practices such as the crb and loan insurance33. the fourth part captured considerations in granting loans, risk mitigation strategies, as well as lc across sectors. the last part captured the collateral types relevant to credit processing. to maximize the probability of getting new insights into the dynamics of the lending market in tanzania, this study has shown a preference for more than one respondent per lender, with varying experience, although several non-bank financial institutions had provided only a single response. when the survey was ended, a total of 151 respondents from 36 lending institutions had returned the duly completed questionnaire. out of the 36 lending institutions, 33 were banks while four were non-bank financial institution. the detailed classification of the sample of financial institutions and the number of 125 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 respondents is provided in appendix 2, while a summary of the data used for this study is provided in appendix 1. data analysis the framework in figure 1 was operationalized through a four-stage analysis: i) the preliminary lc and plr indices were computed based on the mean transformed scores of the likert scale indicators, and based on equation 1. [1] where is the actual factor score is the maximum possible score for indicatoor n is the total number of respondents in the sample (max 151) ii) in the second stage, exploratory factor analysis (efa) was used to identify three perceived lending concentration (lc) sectors – i.e., sslc, pslc and pelc based on responses on a total of 15, five point likert scale questions on the level for which the loans market was sought to be concentrated along pre-identified areas of economic activities (see figure 2 & figure 3). iii) in the third stage the significant determinants of the lc were established and dummified (scores above average =1, 0 otherwise) from factors loadings in (ii) based on the three-factor principle components (pcs) scores as dependent variables and the independent variables summarized in figure 1. using binary logistic regression, four equations have been estimated for this purpose. the binary logistic model for this purpose has been specified as; [2] where, refers to the probability of lc facing a bank as perceived by lender i; is the dummy for perceived lc score based on factor loading greater than average for lender i in sectors. 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = � 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑁𝑁𝑁𝑁𝑖𝑖𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑁𝑁𝑁𝑁 𝑖𝑖𝑖𝑖=1⁄ 𝐼𝐼𝐼𝐼 𝑖𝑖𝑖𝑖=1 126 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 refers to the jth measurement of a macroeconomic variable. in this case inflation and growth experience. these two measures were the average for the period under which the respondent has been in the lending industry. for example, the growth experience is computed as is the gdp growth rate during the respondents’ years of experience from the year of entry to the current year being the kth measurement of the borrowers’ differentiation variables as perceived by lender i; and being the lth measurement of borrower characteristics, as perceived by lender i. iv) in the final stage, the sector’s prediction of lc in equation 2, for each respondent who has worked at lender were then combined in another set of binary logistic regressions for low (up to 20% npl) and high risk (over 20% npl) perceptions (pcrs) as dependent variables. other levels of the plr have been included for clarity. results general observations at the core of this study is the overall effect of lc on the npl risk, as presented in figure 2 and figure 3. the findings presented thus far, have found support from the study by pham and lensink (2008), where the slc which was in favor of individual establishments had a lower plr perception, an indicator that the sslc could be justified as a low-risk lending avenue in tanzania. the fact that the sslc accrued a low plr seemed to suggest that the lending could be along social networks (pham & lensink, 2007), which if not trustworthy might plunge the banking sector into serious exposure and could provide the reason for the ever-increasing npl rates. this observation is however, contrary to western studies which have associated individual lending to a relatively higher plr than corporate borrowers (beck & de jonghe, 2013; liu et al., 2019; laib, 2013; hunter & nixon, 1999).figure 3 provides the scores of the various plr per sector. by comparing it with figure 2, it is evident that high lc and high plr sectors include smes, cbd businesses and non-partner individual borrowers. the low lc and low pcr sectors were dsm private sector organizations, interbank lending, as well as state corporations. 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 1 sector lending concentration and credit risk: an evaluation of lenders’ perceptions in tanzania 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖⁄ ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1 ∑ 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖𝑁𝑁𝑖𝑖=1⁄ 𝐼𝐼 𝑖𝑖=1 …………………………………………………………………...... where 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝐴𝐴𝐴𝐴𝐴𝐴𝑓𝑓𝑓𝑓 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 𝑖𝑖 𝑀𝑀𝑀𝑀𝐴𝐴𝐴𝐴𝑖𝑖 is the 𝑀𝑀𝐴𝐴𝐼𝐼𝑖𝑖𝑀𝑀𝐴𝐴𝑀𝑀 𝑝𝑝𝑓𝑓𝑠𝑠𝑠𝑠𝑖𝑖𝑝𝑝𝐴𝐴𝐼𝐼 𝑠𝑠𝐴𝐴𝑓𝑓𝑓𝑓𝐼𝐼 for indicator 𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝐴𝐴𝐼𝐼 ( 𝜋𝜋𝑖𝑖𝑖𝑖 1−𝜋𝜋𝑖𝑖𝑖𝑖 ) = 𝛽𝛽0 + 𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗 + 𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 + 𝛽𝛽3𝐵𝐵𝐿𝐿𝑖𝑖𝑖𝑖 …………………………………………… where, 𝜋𝜋𝑖𝑖𝑖𝑖 = 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 1 + 𝐼𝐼 𝛽𝛽0+𝛽𝛽1𝑀𝑀𝑀𝑀𝑗𝑗+𝛽𝛽2𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖+𝛽𝛽3𝐵𝐵𝐵𝐵𝑖𝑖𝑖𝑖 𝜋𝜋𝑖𝑖𝑖𝑖 𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀j 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 𝑀𝑀𝑀𝑀𝑖𝑖𝑖𝑖 = 1 𝐺𝐺 ∑ 𝐺𝐺𝑖𝑖𝑖𝑖𝐺𝐺𝑖𝑖=1 . 𝐺𝐺𝑖𝑖𝑖𝑖 𝑔𝑔 = 1 to the current year 𝑔𝑔 = 𝐺𝐺. 𝐵𝐵𝐵𝐵𝑖𝑖k 𝐵𝐵𝐿𝐿𝑖𝑖l i (plris1 ), 127 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 figure 2 perceived customer lending concentration across sectors furthermore, the observations have suggested that sectors where the plr was relatively high, had included the commercialized sector where expectations based on simpasa and pla (2016) were swayed in favor of the low risk. the observations to the contrary have suggested the possibility of a sslc induced plr, whereby low risk sectors might be perceived risky if lending concentration exceeded a certain threshold. this study however, did not venture more into the threshold effect of lc. thus, this research avenue is reserved for future studies. the pslc is also a better option than the pelc, contrary to the longenshrined belief that those with employment and permanent income were perceived to be relatively less risky (le & nguyen, 2018; pham & lensink, 2008; koomson et al., 2016). in this study however, income was not observed to be an important consideration during loan processing, while employment considerations have induced a relatively high rather than low plr, contrary to the findings in le and nguyen (2018); klyuev (2008) and harrison et al. (2004). this was perhaps due to the fact that the government employee lending market could be oversaturated, or that lenders would consider the employment of the borrower only when they believed or faced a high npl risk. 10 at the core of this study is the overall effect of lc on the npl risk, as presented in figure 2 and figure 3. the findings presented thus far, have found support from the study by pham & lensink (2008), where the slc which was in favor of individual establishments had a lower plr perception, an indicator that the sslc could be justified as a low-risk lending avenue in tanzania. the fact that the sslc accrued a low plr seemed to suggest that the lending could be along social networks (pham & lensink, 2007), which if not trustworthy might plunge the banking sector into serious exposure and could provide the reason for the everincreasing npl rates. this observation is however, contrary to western studies which have associated individual lending to a relatively higher plr than corporate borrowers (beck & de jonghe, 2013; liu et al., 2019; laib, 2013; hunter & nixon, 1999). figure 2 perceived customer lending concentration across sectors figure 3 provides the scores of the various plr per sector. by comparing it with figure 2, it is evident that high lc and high plr sectors include smes, cbd businesses and non-partner individual borrowers. the low lc and low pcr sectors were dsm private sector organizations, interbank lending, as well as state corporations. furthermore, the observations have suggested that sectors where the plr was relatively high, had included the commercialized sector where expectations based on simpasa and pla (2016) were swayed in favor of the low risk. the observations to the contrary have suggested the possibility of a sslc induced plr, whereby low risk sectors might be perceived risky if lending 128 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 figure 3 perceived credit default risk across sectors table 1 descriptive statistics for variables in the final models s variable name n min max mean std. dev. skewness kurtosis a perceived non-performing loans risk a1 below 10% 150 .12 .94 .58 .20 -.11 -.96 a2 between 10 20% 133 .00 1.00 .45 .31 .29 -1.18 a3 below 20% 150 .13 .98 .64 .24 -.27 -1.06 a4 20 40% 128 .00 1.00 .63 .37 -.55 -1.31 a5 over 40% 142 .00 1.00 .52 .39 -.16 -1.61 a6 over 20% 151 .00 1.00 .47 .50 .12 -2.01 b lender’s/respondents’ characteristics b0 size large 151 0 1 .47 b1 local_large 151 0 1 .36 b2 org. local 134 0 1 .54 b3 giant_2 151 0 1 .15 b4 giant_3 151 0 1 .05 11 concentration exceeded a certain threshold. this study however, did not venture more into the threshold effect of lc. thus, this research avenue is reserved for future studies. figure 3 perceived credit default risk across sectors the pslc is also a better option than the pelc, contrary to the long-enshrined belief that those with employment and permanent income were perceived to be relatively less risky (le & nguyen, 2018; pham & lensink, 2008; koomson et al., 2016). in this study however, income was not observed to be an important consideration during loan processing, while employment considerations have induced a relatively high rather than low plr, contrary to the findings in le and nguyen (2018); klyuev (2008) and harrison et al. (2004). this was perhaps due to the fact that the government employee lending market could be oversaturated, or that lenders would consider the employment of the borrower only when they believed or faced a high npl risk. table 1 descriptive statistics for variables in the final models s variable name n min max mean std. dev. skewness kurtosis a perceived non-performing loans risk (continued) 129 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 s variable name n min max mean std. dev. skewness kurtosis b5 giant_1 151 0 1 .13 b6 oib 151 0 1 .17 b7 olb 151 0 1 .13 b8 back 151 0 1 .74 b9 manage 151 0 1 .22 b10 credit 151 0 1 .25 b11 administration 151 0 1 .16 b12 riskf 151 0 1 .27 b13 operations 151 0 1 .56 b14 exper 143 1.00 25.00 6.97 4.23 1.37 2.78 c macroeconomic environment (me) c1 grow_exper 143 3.45 1.46 5.64 1.68 .65 -.23 c2 infl_exper 143 5.62 6.41 6.14 .22 -1.42 .87 d lenders application of credit risk management practices (crmps) d1 cilg 151 .24 1.21 1.00 .27 -1.20 .56 d2 dcb 151 .00 1.31 1.00 .29 -.85 .40 d3 gcd 151 .30 1.51 1.00 .38 -.27 -.96 d4 crev 151 .00 1.19 1.00 .28 -1.48 1.65 d5 lpb 151 .00 1.44 1.00 .36 -.60 -.07 d6 pbb 151 .00 1.27 1.00 .30 -.94 .08 d7 rbp 151 .00 1.32 1.00 .31 -.91 .51 e considerations in credit processing e1 age 150 .00 1.38 1.01 .40 -.96 .10 e2 busexp 150 .00 1.30 1.00 .36 -1.43 1.52 e3 ccap 150 .00 1.26 1.01 .34 -1.68 2.35 e4 cdebt 150 .00 1.21 1.01 .31 -2.00 3.76 e5 cdep 150 .00 1.91 1.01 .55 .06 -.81 e6 charct 150 .00 1.26 1.01 .37 -1.55 1.39 e7 collat 150 .00 1.22 1.01 .33 -1.84 2.83 e8 employ 150 .00 1.32 1.01 .37 -1.27 1.06 e9 loanpur 150 .00 1.25 1.00 .35 -1.57 1.73 e10 credh 150 .00 1.18 1.01 .31 -2.13 3.93 f lender’s use of different collateral types f1 land 150 .00 1.47 1.01 .43 -.73 -.30 f2 pcp 150 .00 1.57 1.01 .39 -.65 .41 f3 cp 150 .00 1.80 1.01 .57 .01 -1.28 f4 bsa 150 .00 1.84 1.01 .60 .03 -1.35 f5 invent 150 .00 1.61 1.01 .42 -.67 .28 f6 lien 150 .00 1.61 1.01 .42 -.55 .01 f7 invoice 150 .00 1.57 1.01 .41 -.64 .12 g customer lending concentrations (lc) g1 sslc 150 .01 .99 .57 .32 -.39 -1.21 g2 pslc 142 .00 .99 .56 .29 -.31 -.92 g4 pelc 142 .01 .99 .56 .26 -.27 -.84 130 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 descriptive statistics to analyze the relationship between the plr and lc, it was necessary to first establish a quantitative measure of lc based on the 15 items evaluated in the questionnaire. factor analysis suggested a threefactor solution, in which their respective principal components were considered as slc indicators. prior to further analyses, descriptive statistics have been considered in table 1. it can be observed that some variables were skewed with a kurtosis value being far away from zero, thus calling for methods other than linear regression in determining the predictors. the dispersion was however not very alarming, as it seemed to suggest some stability. table 2 determinants of the different levels of plr (low) models summary model fit model classification no yes overall npl <= 10 percent -2 log likelihood 149.98 no 30 20 cox & snell 0.17 yes 23 55 nagelkerke 0.23 percent 56.6 73.33 66.41 npl = 20 40 percent -2 log likelihood 120.09 no 59 15 cox & snell 0.36 yes 10 44 nagelkerke 0.48 percent 85.51 74.58 80.47 npl > 40 percent -2 log likelihood 131.8 no 23 9 cox & snell 0.23 yes 21 75 nagelkerke 0.31 percent 52.27 89.29 76.56 determinants of the plr table 2 and table 3 provide the summary information for “low” and “high” plr models respectively. from table 2, it can be seen that the pseudo r2s were adequately high and provided an acceptable classification of cases, all being beyond 75 percent for two (2) models. however, the first one was below acceptable standards. this provides an indication of good model fit for the last two (2) models and a justification for further interpretation of the results. on the high npl, risk perception, the pseudo r2; i.e., correlation and the classification 131 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 table are provided in table 3. all the models are generally fit and provide at least 85 percent classification accuracy. table 3 determinants of the different levels of plr (high) models summary model fit model classification no yes overall npl <= 10 percent -2 log likelihood 74.71 no 40 5 cox & snell 0.52 yes 7 76 nagelkerke 0.71 percent 85.11 93.83 90.63 npl = 20 40 percent -2 log likelihood 85.22 no 52 10 cox & snell 0.51 yes 9 57 nagelkerke 0.68 percent 85.25 85.07 85.16 npl > 40 percent -2 log likelihood 79.94 no 30 6 cox & snell 0.47 yes 11 81 nagelkerke 0.65 percent 73.17 93.1 86.72 the effect of lender and employee characteristics with regard to employee characteristics, this study would like to suggest that those in management and those with a relatively broader experience in the banking industry have a higher plr, an observation which is in line with the experience hypothesis, and which has also been suggested in dember and penwell (1980) and de meza and southey (1996). a study in urt (2019) has suggested that tanzania is facing a higher npl risk and that employees in the credit department are more aware of this. unexpectedly however, it was further noted that employees in the credit department have suggested a significantly lower npl risk. it is evident here that risk is overstated by those in management, as they view it as bad to their organization (de meza & southey, 1996), while those in operations (credit) understate risk as they prefer being optimistic in order to encourage borrowing. further analysis among those in the credit department has further suggested that they also understated risk. this was because on average, they had less experience across lcs, which is an expected form of behavior (burakov, 2014; berger & udell, 2004). 132 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 table 4 binary logistic model results from determinants of “low” plr less than 10% npl between 10 20 % npl below 20% npl exp(b) sig. exp(b) sig. exp(b) sig. constant 0.53 0.13 0.45 (0.70) (1.32) (0.77) a lender/respondent characteristics origlocal(1) 4.03 * (0.81) sizelarge(1) 14.89 *** (0.93) local_large(1) 0.03 *** (1.27) giant_3(1) 19.17 ** (1.40) operations(1) 2.10 * (0.42) riskf(1) 3.36 ** 5.26 *** 4.66 *** (0.50) (0.63) (0.56) credit(1) 3.64 ** 2.60 * (0.57) (0.56) c lender application of credit risk management practices (crmps) rpai_cilg 0.05 *** (1.15) rpai_lpb 6.19 ** (0.88) rpai_rbp 31.15 ** (1.45) rpai_dbc 0.01 *** (1.47) d considerations in credit processing rcc_loanpur 11.74 *** (0.88) (continued) 133 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 less than 10% npl between 10 20 % npl below 20% npl exp(b) sig. exp(b) sig. exp(b) sig. rcc_ccap 9.47 *** (0.70) rcc_charct 4.30 ** (0.59) e lender use of different collateral types rai_bsa 0.30 *** 0.21 *** 0.13 *** (0.37) (0.50) (0.47) rai_invent 0.07 *** (0.93) rai_lien 14.70 *** (0.89) f customer lending concentrations (lc) sslc 7.57 ** 3.65 * (0.84) (0.72) as reflected in the data displayed in table 4 and figure 4, it can be seen that this study points to at least three main observations with regard to the employees of the three lending giants in tanzania. the first relates to employees of giant_3, where the plr was generally low and this has been enhanced by having most of its loans alongside the sslc, which seemed to provide an additional advantage in terms of soft information (pham & lensink, 2008). it was found to have a relatively high plr with regard to the pelc and the pslc. as a result, the ability of giant_3 to generate useful credit information from these sectors was lower, in line with the tpt paradigm (diamond, 1984; markowitz, 1959). thus, both public employees and the private sector are not for giant_1 an avenue for the slc. the lender would be better-off diversifying across sectors if lending on this market. like giant_3’s employees, the ones in giant_1 will also be able to achieve a lower plr if the sslc dominates and when it is against the pelc. it appears that the sslc can provide important perceived shock absorbers to address the plr among employees of this lender. however, unlike giant_3’s employees giant_1’s employees seemed to associate the pslc with high risk, an observation that was contrary to the previous observation that lc to commercialized sectors, such agriculture, manufacturing, transport and communication, 134 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 construction and real estate would have a relatively lower plr (le & diep, 2020; pham & lensink, 2008; skridulytė & freitakas, 2012). this observation could be linked to the relatively lower ability of giant_1 to channel monopoly rents from superior information and monitoring efficiency expertise into the private sector (simpasa & pla, 2016). figure 4 lending giants npl risk in relation to lending concentration lastly, there is ample evidence in this study that giant_2’s employees are unlikely to perceive a high npl risk regardless of the slc, an indication that giant_2 has proven to be a lower/stable npl risk avenue. employees from this lending giant have suggested that a declining plr in response to the sslc is an indicator of the strong reliance on social relationship in their operatives, just like the other banks while it is an increasing function of the pelc as an indicator of slc saturation. 16 less than 10% npl between 10 20 % npl below 20% npl exp(b) sig. exp(b) sig. exp(b) sig. (0.84) (0.72) as reflected in the data displayed in table 4 and figure 4, it can be seen that this study points to at least three main observations with regard to the employees of the three lending giants in tanzania. the first relates to employees of giant_3, where the plr was generally low and this has been enhanced by having most of its loans alongside the sslc, which seemed to provide an additional advantage in terms of soft information (pham & lensink, 2008). it was found to have a relatively high plr with regard to the pelc and the pslc. as a result, the ability of giant_3 to generate useful credit information from these sectors was lower, in line with the tpt paradigm (diamond, 1984; markowitz, 1959). thus, both public employees and the private sector are not for giant_1 an avenue for the slc. the lender would be better-off diversifying across sectors if lending on this market. figure 4 lending giants npl risk in relation to lending concentration (a) giant_1’s plr in relation to sslc (b) giant_1’s plr in relation to pslc (c) giant_1’s plr in relation to pelc (d) giant_2’s plr in relation to sslc (e) giant_2’s plr in relation to pslc (f) giant_2’s plr in relation to pelc 16 less than 10% npl between 10 20 % npl below 20% npl exp(b) sig. exp(b) sig. exp(b) sig. (0.84) (0.72) as reflected in the data displayed in table 4 and figure 4, it can be seen that this study points to at least three main observations with regard to the employees of the three lending giants in tanzania. the first relates to employees of giant_3, where the plr was generally low and this has been enhanced by having most of its loans alongside the sslc, which seemed to provide an additional advantage in terms of soft information (pham & lensink, 2008). it was found to have a relatively high plr with regard to the pelc and the pslc. as a result, the ability of giant_3 to generate useful credit information from these sectors was lower, in line with the tpt paradigm (diamond, 1984; markowitz, 1959). thus, both public employees and the private sector are not for giant_1 an avenue for the slc. the lender would be better-off diversifying across sectors if lending on this market. figure 4 lending giants npl risk in relation to lending concentration (a) giant_1’s plr in relation to sslc (b) giant_1’s plr in relation to pslc (c) giant_1’s plr in relation to pelc (d) giant_2’s plr in relation to sslc (e) giant_2’s plr in relation to pslc (f) giant_2’s plr in relation to pelc (a) giant_1's plr in relation to sslc (b) giant_1's plr in relation to pslc (c) giant_1's plr in relation to pelc (a) giant_2's plr in relation to sslc (b) giant_2's plr in relation to pslc (c) giant_2's plr in relation to pelc 135 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 figure 5 npl risk in relation to lending concentration in relation to originality and size of lender this study further notes a relatively lower plr in favor of lenders with local origin than those originating elsewhere, suggesting some typological effect in the plr (abu hussain & al-ajmi, 2012; kira, 2013). however, slc in favor of social status or public employees modifies this effect by suggesting that if local lenders’ slc is in favor of social status or government employees, it will end-up with a relatively high plr among its employees. it is evident that the effect of origin of bank is stable only when a lender is focused on the private sector. the panels (a) – (d) in figure 5 provide additional information on the moderating effect of lc on the effect of lender size on the plr. there is evidence to the effect that local large banks are unlikely to fall prey to the higher plr among its employees, to reflect a higher ability to grab economies of scale (beck & de jonghe, 2013), but high levels of the plr among international large lenders who may be unable to benefit from the same, if most such economies emanates from social networks. 17 like giant_3’s employees, the ones in giant_1 will also be able to achieve a lower plr if the sslc dominates and when it is against the pelc. it appears that the sslc can provide important perceived shock absorbers to address the plr among employees of this lender. however, unlike giant_3’s employees giant_1’s employees seemed to associate the pslc with high risk, an observation that was contrary to the previous observation that lc to commercialized sectors, such agriculture, manufacturing, transport and communication, construction and real estate would have a relatively lower plr (le & diep, 2020; pham & lensink, 2008; skridulytė & freitakas, 2012). this observation could be linked to the relatively lower ability of giant_1 to channel monopoly rents from superior information and monitoring efficiency expertise into the private sector (simpasa & pla, 2016). lastly, there is ample evidence in this study that giant_2’s employees are unlikely to perceive a high npl risk regardless of the slc, an indication that giant_2 has proven to be a lower/stable npl risk avenue. employees from this lending giant have suggested that a declining plr in response to the sslc is an indicator of the strong reliance on social relationship in their operatives, just like the other banks while it is an increasing function of the pelc as an indicator of slc saturation. figure 5 npl risk in relation to lending concentration in relation to originality and size of lender (a) plr across lender’s origin in relation to sslc (b) plr across lender’s origin in relation to pslc (c) plr across lender’s origin in relation to pelc 17 like giant_3’s employees, the ones in giant_1 will also be able to achieve a lower plr if the sslc dominates and when it is against the pelc. it appears that the sslc can provide important perceived shock absorbers to address the plr among employees of this lender. however, unlike giant_3’s employees giant_1’s employees seemed to associate the pslc with high risk, an observation that was contrary to the previous observation that lc to commercialized sectors, such agriculture, manufacturing, transport and communication, construction and real estate would have a relatively lower plr (le & diep, 2020; pham & lensink, 2008; skridulytė & freitakas, 2012). this observation could be linked to the relatively lower ability of giant_1 to channel monopoly rents from superior information and monitoring efficiency expertise into the private sector (simpasa & pla, 2016). lastly, there is ample evidence in this study that giant_2’s employees are unlikely to perceive a high npl risk regardless of the slc, an indication that giant_2 has proven to be a lower/stable npl risk avenue. employees from this lending giant have suggested that a declining plr in response to the sslc is an indicator of the strong reliance on social relationship in their operatives, just like the other banks while it is an increasing function of the pelc as an indicator of slc saturation. figure 5 npl risk in relation to lending concentration in relation to originality and size of lender (a) plr across lender’s origin in relation to sslc (b) plr across lender’s origin in relation to pslc (c) plr across lender’s origin in relation to pelc (a) plr across lender's origin in relation to sslc (b) plr across lender's origin in relation to pslc (c) plr across lender's origin in relation to pelc (d) plr across lender's size in relation to sslc (e) plr across lender's size in relation to pslc (f) plr across lender's size in relation to pelc 136 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 effect of the macroeconomic environment this study also examined the macroeconomic environment (me) in relation to the plr and it became evident that both long lived inflationary and growth experiences were unlikely to increase the probability of a relatively higher plr. evaluations of the moderations of lc panels (a) – (c) of figure 6 seems to suggest that lc in favor of both the sslc and the pslc does not modify the effect of inflation experience. high inflation experience is associated with lower plr across lc levels only if such lc has been in favor of social status, or private sector firms. this observation suggests that a significant lc based on social status and private sectors may cushion the lender from the negative effect of unprecedented inflationary experience on the npl. although the inflationary environment could restraint the household and spending behavior firms (trautmann & vlahu, 2013), both the private sector and individuals were relatively more flexible to re-adjust their earnings and expenditure to reflect changes in prices. however, with lc in favor of government employees, the lender is likely to suffer a higher npl risk, since the earnings of borrowers (employees’ salary) are hardly adjustable in the short-run. this study also analyzed the effect of the gdp growth as experienced by lenders throughout their lending career. the observations suggest that although the high gdp growth experience is generally unlikely to be associated with a high plr, the higher slc in favor of social status induces a higher plr in response to the higher gdp growth. as noted in le and diep (2020), the channel of effect could be loan loss provisions and difficulties in managing interest rate risk. lenders operating during boom are not expected to fall prey to a high plr, as repayment is guaranteed by the certainty of income flow (lassoued, 2017). however, if the lender is under the sslc, see figure 6 (d), then a reversal of lc effect may be anticipated. this is despite the generalization that lc has a lower plr, a booming economy might cast a negative shadow on the sslc, leading to a high plr. the effect of credit risk management based on the results displayed in table 4, three crmps appeared in this study to provide intuitive explanations with regard to the plr. the application of the rbp to mitigate credit risk is unquestionable in as long as lc is in favor of private sector firms, as well as government employees (wood & kellman, 2010). however, this study did not 137 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 find any significant effect when lc favors social status. for a lender who has not been concentrated in any sector, unlocking the loans of a potential borrower (lbp) would provide an effective mechanism to lower the plr, as was suggested by seyram (2013). thus, the effectiveness of the, lbp as a tool to mitigate credit risk is higher when lenders are against lc in its entirety. figure 6 npl risk in relation to lending concentration, originality and size of lender with moderate sslc and with either very low or very high pslc, the lender guarantee to cover debt (gcd) works to lower the plr. a very low lc into the private sector seemed to suggest that more lending was in favor of other criteria, which could entail a strong use of the gcd to avoid the associated plr, especially from government employees, as well as individuals. when lc is within the private sector, the gcd is also very relevant; this is because concentration may be associated with a high degree of uncertainty and thus, the lender may require a cushion against the npl risk in the form of the gcd. 19 based on the results displayed in table 4, three crmps appeared in this study to provide intuitive explanations with regard to the plr. the application of the rbp to mitigate credit risk is unquestionable in as long as lc is in favor of private sector firms, as well as government employees (wood & kellman, 2010). however, this study did not find any significant effect when lc favors social status. for a lender who has not been concentrated in any sector, unlocking the loans of a potential borrower (lbp) would provide an effective mechanism to lower the plr, as was suggested by seyram, (2013). thus, the effectiveness of the, lbp as a tool to mitigate credit risk is higher when lenders are against lc in its entirety. figure 6 npl risk in relation to lending concentration, originality and size of lender (a) plr across lender’s origin in relation to sslc (b) plr across lender’s origin in relation to pslc (c) plr across lender’s origin in relation to pelc (d) plr across lender’s size in relation to sslc (e) plr across lender’s size in relation to pslc (f) plr across lender’s size in relation to pelc with moderate sslc and with either very low or very high pslc, the lender guarantee to cover debt (gcd) works to lower the plr. a very low lc into the private sector seemed to suggest that more lending was in favor of other criteria, which could entail a strong use of the gcd to avoid the associated plr, especially from government employees, as well as individuals. when lc is within the private sector, the gcd is also very relevant; this is because concentration may be associated with a high degree of uncertainty and thus, the lender may require a cushion against the npl risk in the form of the gcd. (a) plr across lender's origin in relation to sslc (b) plr across lender's origin in relation to pslc (c) plr across lender's origin in relation to pelc (d) plr across lender's size in relation to sslc (e) plr across lender's size in relation to pslc (f) plr across lender's size in relation to pelc 19 based on the results displayed in table 4, three crmps appeared in this study to provide intuitive explanations with regard to the plr. the application of the rbp to mitigate credit risk is unquestionable in as long as lc is in favor of private sector firms, as well as government employees (wood & kellman, 2010). however, this study did not find any significant effect when lc favors social status. for a lender who has not been concentrated in any sector, unlocking the loans of a potential borrower (lbp) would provide an effective mechanism to lower the plr, as was suggested by seyram, (2013). thus, the effectiveness of the, lbp as a tool to mitigate credit risk is higher when lenders are against lc in its entirety. figure 6 npl risk in relation to lending concentration, originality and size of lender (a) plr across lender’s origin in relation to sslc (b) plr across lender’s origin in relation to pslc (c) plr across lender’s origin in relation to pelc (d) plr across lender’s size in relation to sslc (e) plr across lender’s size in relation to pslc (f) plr across lender’s size in relation to pelc with moderate sslc and with either very low or very high pslc, the lender guarantee to cover debt (gcd) works to lower the plr. a very low lc into the private sector seemed to suggest that more lending was in favor of other criteria, which could entail a strong use of the gcd to avoid the associated plr, especially from government employees, as well as individuals. when lc is within the private sector, the gcd is also very relevant; this is because concentration may be associated with a high degree of uncertainty and thus, the lender may require a cushion against the npl risk in the form of the gcd. 138 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 the effect of considerations in credit processing in terms of considerations in credit processing, the data in table 4 seem to suggest that assessment of character emerged as the unequivocal consideration, an observation well supported in the literature (abdulsaleh & worthington, 2016; laib, 2013). regardless of lc, lenders ought to consider borrowers’ character, as such a consideration yields relatively lower plr. these observations are however, contrary to the findings in pham and lensink (2008), who observed a limited correlation between character and the plr. though the consideration of loan purpose is seen as relevant, this study has noted that such consideration makes sense in reducing the plr only if the different types of lc are either very low or very high. on the lowest extreme of the slc, lenders do consider the purpose of the loan as an important criterion in granting loans; this is because borrowers are from diverse sectors for which the lender have limited knowledge (beck & de jonghe, 2013; francisaet al., 2018). therefore, detailed information on the purpose of the loan would facilitate the lender in getting some insights on the kind of business and the probability of repayment of the loan from the general behavior of the different purposes of loans (adzobuet et al., 2017; böve et al., 2010). at the highest end, the lender is concentrated and faces the risk of putting “all eggs in one basket” alongside the tpt postulates (le & diep, 2020). although the lenders are now more knowledgeable about borrower behavior, they still require information in the appropriate basket – the one which lending should be concentrated. as such, details on the loan purpose will provide a mechanism for the lender to identify the best lending option among the many that exists. the observations from this study also seem to suggest that the effect of capital consideration is dependent on both the type and level of lc. to achieve lower plr in response to capital consideration, as suggested in the literature (richard et al., 2008; adzobu et al., 2017; koomson et al., 2016), both the lower and very high sslc and pslc can be targeted. the slc in favor of social status, as well as lc for private firms both would require capital to achieve a lower plr, because such borrowers would often have capital to declare. at low levels of such slc, the lender has limited knowledge of the borrower and capital provision to the lender should provide a guarantee to the lender on the ability of individual borrowers, as well as firms to recover the 139 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 advanced loan from its assets in case of default. at high slc, the lender is knowledgeable of the borrower, but capital adequacy is still needed to address moral hazard problems and motivate the borrower to honor its obligation. in terms of slc to government employees, capital adequacy is unnecessary, as has been suggested by lassoued (2017), but when availed it can provide mechanisms to address the plr by ironing out the moral hazards in the borrower’s mind (aghion & bolton, 1992). the effect of collateral types lastly, this study points to the effect of two collateral types on the plr. the first is the use of the business savings account (bsa), which was seen as inducing a high plr perception, contrary to expectations (koomson et al., 2016). however, a high pslc reverses such an effect. the use of the bsa can be a useful tool to mitigate the plr only if the lender lc is in favor of the private sector. in this market, business accounts are more prevalent and having one induces a comfort zone to lenders. similarly, inventories as collateral can reduce the plr when lc is either very low or when it is very high for all lc typologies. in either case inventories operate in the same manner as capital (richard et al., 2008; adzobu et al., 2017): at the lowest level of the slc, inventories address the information gap (chan & thakor, 1987; stiglitz & weiss, 1981), while at the highest the slc provide a cushion against moral hazard behavior among lenders (aghion & bolton, 1992). conclusion the findings of this study, based on the contradictory observations in tanzania that intensive credit crunch via the crmps seemed to provide a marginal outcome in terms of lowering the npl risk, could be explained by the fact that there had been lending overconcentration based on social status across large and small lenders. furthermore, it is evident that lending concentration reverses the effect of the macroeconomic variables, credit risk management practices (crmps), credit processing considerations (cpcs), as well as collateral types. it is evident that only the intensification of “character”, “capital” and “inventory” scrutiny during loan processing has had a direct npl risk-reducing effect, while all other 140 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 considerations or crmps were dependent on the degree of lending concentration. therefore, assessing the degree of lender lc across customers or sectors is imperative prior to any credit risk management initiative. furthermore, it is notable that although lc has directly affected the plr, as predicted by the traditional and corporate finance theories, the indirect lc effect via crmps, cpcs, bank size and originality, as well as the various collateral typologies has provided some new insights into this area of research. in terms of policy implications, it would be advisable that the central bank (bot), instead of requiring lenders to have in place a holistic credit management program, should demand for the adoption of flexible lending policies to reflect the macroeconomic environment, as well as the respective magnitude and direction of lc. restrictions should be placed on lenders with a sslc, i.e., lending based on the characteristics of borrowers to the effect that they refrain from excessively expanding loans during high economic growth, and unless they re-direct such loans in favor of other criteria, especially employment and productivity of enterprises. such restrictions may also extend to specific crmps. for example, the findings of this study have suggested that the rbp is an effective strategy to militate against the npl risk, only if the lending favors the pslc or the pelc, while the lbp is effective for lenders who are balanced in their lending behavior. similarly, loan guarantees are effective only when the sslc and the pslc are low, while a business savings account scrutiny during loan processing works to reduce the plr under high pslc. a policy to restrict or control the use of the rbp should be advocated only in an environment where the lending concentration is not well guided. it should apply only to lenders whose lending is biased in favor of the sslc, or the use of the lbp by lenders who are heavily concentrated, or the use of a loan guarantee to a lender who faces the pelc. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. 141 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 endnotes 1 for the updated list of financial institutions in tanzania visit https:// www.bot.go.tz/banksupervision/institutions 2 such legislations include business registration and licensing agency; the national payment systems act, 2015, cooperative societies act, 2013; companies act 2002; societies act cap 337 of 1954; ngo act 2002; trustees’ incorporation act 2002 (cap 318). 3 this data was however, not the subject of the current study. references abdulsaleh, a. m., & worthington, a. (2016). bankers’ perceptions of successful smes loan applications: a case study from libya. the mena journal of business case studies, doi: 10.5171/2016.449044. abu hussain, h., & al-ajmi, j. (2012). risk management practices of conventional and islamic banks in bahrain. the journal of risk finance, 13(3), 215-239. acharya, v., hasan, i., & saunders, a. (2006). should banks be diversified? evidence from individual bank portfolios. journal of business, 79(3), 1355-1412. adzobu, l. d., agbloyor, e. k., & aboagye, a. (2017). the effect of loan portfolio diversification on banks’ risks and return evidence from an emerging market. managerial finance, 43(11), 1274-1291. aghion, p., & bolton, p. (1992). an incomplete contracts approach to financial contracting. the review of economic studies, 59(3), 473-494. al-kayed, l. t. (2020). effects of focus versus diversification on bank risk and return: evidence from islamic banks’ loan portfolios. journal of islamic accounting and business research, 11(10), 2155-2168. doi 10.1108/jiabr-10-2019-0192. bank for international settlements. (2006). an overview of the issues and a synopsis of the results from the research task force project. studies on credit risk concentration, p. working paper no. 15. beck, t., & de jonghe, o. (2013). lending concentration, bank performance and systemic risk: exploring cross-country variation. world bank policy research working paper 6604. 142 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 berger, a., & udell, g. (2004). the institutional memory hypothesis and the procyclicality of bank lending behavior. journal of financial intermediation, 13, 459-495. doi: 10.1016/j. jfi.2004.06.006. berger, p. g., minnis, m., & sutherland, a. (2017). commercial lending concentration and bank expertise: evidence from borrower financial statements. journal of accounting and economics. doi: 10.1016/j.jacceco.2017.06.005. bot. (2010). risk management guidelines for banks and financial institutions. https://www.bot.go.tz/publications/acts,%20 regulations,%20circulars,%20guidelines/guidelines/ en/2020091513165478.pdf bot. (2019). tanzania financial stability report. tanzania: directorate of financial sector supervision, bank of tanzania. böve, r., düllmann, k., & pfingsten, a. (2010). do specialization benefits outweigh concentration risks in credit portfolios of german banks. discussion paper no. 10/2010,. burakov, d. v. (2014). does framing affect risk attitude? experimental evidence from credit market. american journal of applied sciences, 11(3), 391-395. doi:10.3844/ajassp.2014.391.395. campello, m., & gao, j. (2017). customer concentration and loan contract terms. journal of financial economics, 123, 108–136. ttp://dx.doi.org/10.1016/j.jfineco.2016.03.010. chan, y. s., & thakor, a. (1987). collateral and competitive equilibria with moral hazard and private information. the journal of finance, 42(2), 345363. choppari, n., & rajeshwar, r. b. (2015). credit risk management practices of indian commercial banks. international journal of management and social science, 3(1), 89 94. de meza, d., & southey, c. (1996). the borrower’s curse: optimism, finance and entrepreneurship. the economic journal, 106(435), 375-386. dember, w. n., & penwell, l. (1980). happiness, depression, and the pollyanna principle. bulletin of the psychonomic society, 15(5), 321-323. denis, d., denis, d., & sarin, a. (1997). agency problems, equity ownership and corporate diversification. the journal of finance, 52(1), 135-160. dezoort, f. t., wilkins, a., & justice, s. e. (2017). the effect of sme reporting framework and credit risk on lenders’ judgments and decisions. journal of accounting and public policy, 36(4), 302-315. https://doi.org/10.1016/j.jaccpubpol.2017.05.003. 143 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 diamond, d. w. (1984). financial intermediation and delegated monitoring. the review of economic studies, 51(3), 393-414. https://doi.org/10.2307/2297430. feder, g., & just, r. e. (1980). a model for analysing lenders’ perceived risk. applied economics, 12(2), 125-144. http:// dx.doi.org/10.1080/00036848000000020. francisa, b. b., hasan, i., küllüc, a. m., & zhou, m. (2018). should banks diversify or focus? know thyself: the role of abilities. economic systems, 42, 106–118. harif, m., hoe, c., & zali, s. (2011). business financing for malaysian smes: what are the banks’ determining factors. world review of business research, 1(3), 78-101. harrison, d. m., noordewier, t. g., & yavas, a. (2004). do riskier borrowers borrow more? real estate economics, 32(3), 385– 411. hayden, e., porath, d., & von westernhagen, n. (2006). does diversification improve the performance of german banks? evidence from individual bank portfolios. discussion paper no. 110. hunter, c., & nixon, j. (1999). the discourse of housing debt the social construction of landlords, lenders, borrowers and tenants. housing, theory and society, 16, 165–178. imf. (2018). united republic of tanzania financial system stability assessment. bank of tanzania, capital markets department. dar es salaam: bot. iyer, r., & puri, m. (2012). understanding bank runs: the importance of depositor–bank relationships and networks. american economic review, 102, 1414–1445. jensen, m. (1986). agency costs of free cash flow, corporate finance, and takeovers. the american economic review, 76(2), 323329. kira, a. r. (2013). the evaluation of the factors influence the access to debt financing by tanzanian smes. european journal of business and management, 5(7), https://www.iiste.org/ journals/index.php/ejbm/article/view/4701/5076. klyuev, v. (2008). show me the money: access to finance for small borrowers in canada. imf working paper western hemisphere department, wp/08/22. koomson, i., annim, s. k., peprah, & atta, j. (2016). loan refusal, household income and savings in ghana: a dominance analysis approach. african journal of economic and sustainable development, 5(2), 172 – 191. 144 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 laib, y. (2013). determinants of bank financing for small and medium enterprise. association de recherches et publications en management, gestion 200-2013/3, 30, pp. 29 -47. lassoued, n. (2017). what drives credit risk of microfinance institutions? international evidence international. journal of managerial finance, 13(5), 541-559. doi 10.1108/ijmf-032017-0042. le, c. h., & nguyen, h. l. (2018). collateral quality and loan default risk: the case of vietnam. comparative economic studies. doi. org/10.1057/s41294-018-0072-6. le, t. t., & diep, t. t. (2020). the effect of lending structure concentration on credit risk: the evidence of vietnamese commercial banks. journal of asian finance, economics and business, 7(7), 59–72 59. doi:10.13106/jafeb.2020.vol7. no7.059. liu, c., xiao, z., & xie, h. (2019). customer concentration, institutions, and corporate bond contracts. international journal of finance and economics, 25(1), 90-119. https://doi. org/10.1002/ijfe.1731. markowitz, h. m. (1959). portfolio selection efficient diversification of investments. new haven, connecticut: yale university press. mol-gómez-vázquez, a., hernández-cánovas, g., & koëter-kan, j. (2018). bank market power and the intensity of borrower discouragement: analysis of smes across developed and developing european countries. small business economics, https://doi.org/10.1007/s11187-018-0056-y. ondabu, j. b. (2019). role of credit reference bureau on financial intermediation: evidence from the commercial banks in kenya. paravisini, d., rappoport, v., & schnabl, p. (2015). specialization in bank lending: evidence from exporting firms. nber working paper series, p. working paper 21800. pham, t. t., & lensink, r. (2007). lending policies of informal, formal and semiformal lenders evidence from vietnam. economics of transition, 15(2), 181–209. pham, t. t., & lensink, r. (2008). household borrowing in vietnam: a comparative study of default risks of formal, informal and semi-formal credit. journal of emerging market finance, 7(3), 237–261. doi: 10.1177/097265270800700302. prasad, s. (2016). the efficacy of credit insurancea study of the quantitative impact on trade receivables and receivables turnover ratio. yrkeshögskolan arcada. arcada. https:// core.ac.uk/download/pdf/80992729.pdf 145 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 purda, l. d. (2008). risk perception and the financial system. journal of international business studies, 39, 1178–1196 https://doi. org/10.1057/palgrave.jibs.8400411. richard, e., chijoriga, m., kaijage, e., peterson, c., & bohman, h. (2008). credit risk management system of a commercial bank in tanzania. international journal of emerging markets, 3(3), 323-332. doi 10.1108/17468800810883729. seyram, p. k. (2013). risk management practices among commercial banks in ghana. europian journal of business and management. simpasa, a., & pla, l. (2016). sectoral credit concentration and bank performance in zambia. african development bank group wrrking paper series, p. working paper no. 245. skridulytė, r., & freitakas, e. (2012). the measurement of concentration risk in loan portfolios. economics & sociology, 5(1), 51-61. stiglitz, j., & weiss, a. (1981). credit rationing in markets with imperfect information. the american economic review, 71(3), 393-410. trautmann, s. t., & vlahu, r. (2013). strategic loan defaults and coordination: an experimental analysis. journal of banking & finance, 37 , 747–760. doi.org/10.1016/j.jbankfin.2012.10.019. ugoani, j. n. (2016). non-performing loans portfolio and its effect on bank profitability in nigeria. independent journal of management & production (ijm&p). urt. (2019). bank of tanzania ( bot) annual report. dar es salaam: urt. wood, a., & kellman, a. (2010). risk management practices by barbadian banks. international journal of business and social research, 3(5), 22-33, doi: https://doi.org/10.18533/ijbsr. v3i5.3. yang, z. (2017). customer concentration, relationship, and debt contracting. journal of applied accounting research, 18(2), 185-207. doi:10.1108/jaar-04-2016-004. 146 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 appendices appendix 1: description of the variables s/n variables name abbrev. description and measurement a perceived nonperforming loans risk plr a dummy constructed from the probability of experiencing each of the npl rates/ range and converted into dummies for each range of npl rate i.e. below 10%, 10 -20%, 20 – 40% and above 40% b lender’s/respondents’ characteristics b0 lender size is large sizelarge a dummy for large lender taking a value of 1 if true and 0 otherwise b1 lender is local and large local_large a dummy for large local lender taking a value of 1 if true and 0 otherwise b2 lender’s origin is local orglocal a dummy for local lender taking a value of 1 if true and 0 otherwise b3 works at giant_2 bank giant_2 a dummy for giant_2 bank taking a value of 1 if true and 0 otherwise b4 works at giant_3 bank giant_3 a dummy for giant_3 bank taking a value of 1 if true and 0 otherwise b5 works at giant_1 bank giant_1 a dummy for giant_1 bank taking a value of 1 if true and 0 otherwise b6 works at other international lender oib a dummy for an international lender other than major banks taking a value of 1 if true and 0 otherwise b7 works at other local bank olb a dummy for an local lender other than major banks taking a value of 1 if true and 0 otherwise b8 work at back office operation back a dummy for work in back-office operations taking a value of 1 if true and 0 otherwise b9 work as a manager manage a dummy for work as manager taking a value of 1 if true and 0 otherwise b10 works in the credit department credit a dummy for employed in the credit department taking a value of 1 if true and 0 otherwise b11 works in the general administration department administration a dummy for employed in the general administration department taking a value of 1 if true and 0 otherwise b12 risk and/ or finance departments riskf a dummy for employed in the finance and/or risk department taking a value of 1 if true and 0 otherwise b13 work in the operations department operations a dummy for work as supervisor taking a value of 1 if true and 0 otherwise (continued) 147 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 s/n variables name abbrev. description and measurement b14 work experience exper experience in lending business in years c macroeconomic environment (me) c1 experienced growth grow_exper average growth rate over the lending experience years based on gdp growth from wdi c2 experienced inflation infl_exper average inflation rate over the lending experience years based on cpi from wdi d lenders application of credit risk management practices (crmps) d1 credit insurance to loans granted cilg an index of the level at which lenders apply credit insurance to loan granted as specified borrowers as computed based on eq. 1 d2 apply apply different bank covenants to loans granted dcb an index of the level at which lenders apply apply different bank covenants to loans granted as computed based on eq. 1 d3 apply bank guarantee to loans granted gcd an index of the level at which lenders apply guarantee to loans granted as computed based on eq. 1 d4 apply credit review before granting loans crev an index of the level at which lenders apply credit review before granting loans as computed based on eq. 1 d5 unlock loans of potential borrowers lpb an index of the level at which lenders unlock loans of potential borrowers as computed based on eq. 1 d6 apply premium lending to specified borrowers pbb an index of the level at which lenders apply premium lending to specified borrowers as computed based on eq. 1 d7 apply risk based pricing to loans granted rbp an index of the level at which lenders apply risk based pricing to loans granted as computed based on eq. 1 e considerations in credit processing e1 age age an index of the level at which lenders consider age of the borrower when processing loans applications as computed based on eq. 1 e2 business experience busexp an index of the level at which lenders consider business experience of the borrower when processing loans applications as computed based on eq. 1 (continued) 148 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 s/n variables name abbrev. description and measurement e3 current capital ccap an index of the level at which lenders consider current capita of the borrower when processing loans applications as computed based on eq. 1 e4 existing debt cdebt an index of the level at which lenders consider current debt of the borrower when processing loans applications as computed based on eq. 1 e5 current number of dependants cdep an index of the level at which lenders consider current dependants of the borrower when processing loans applications as computed based on eq. 1 e6 character charct an index of the level at which lenders consider character of the borrower when processing loans applications as computed based on eq. 1 e7 collateral collat an index of the level at which lenders consider collateral of the borrower when processing loans applications as computed based on eq. 1 e8 employment employ an index of the level at which lenders consider employment of the borrower when processing loans applications as computed based on eq. 1 e9 income income an index of the level at which lenders consider income of the borrower when processing loans applications as computed based on eq. 1 e10 loan purpose loanpur an index of the level at which lenders consider borrower’s loan purchase when processing loans applications as computed based on eq. 1 credit history credh an index of the level at which lenders consider borrower’s credit history when processing loans applications as computed based on eq. 1 f lender’s use of different collateral types f1 land plot land an index of the level at which lenders use land plots as collateral as computed based on eq. 1 (continued) 149 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 s/n variables name abbrev. description and measurement f2 partially completed property pcp an index of the level at which lenders use incomplete property as collateral as computed based on eq. 1 f3 completed property cp an index of the level at which lenders use completed property as collateral as computed based on eq. 1 f4 business saving account bsa an index of the level at which lenders use borrower saving account as collateral as computed based on eq. 1 f5 inventory invent an index of the level at which lenders use inventory as collateral as computed based on eq. 1 f6 lien lien an index of the level at which lenders use land plots as collateral as computed based on eq. 1 f7 invoice invoice an index of the level at which lenders use invoice as collateral as computed based on eq. 1 g customer lending concentrations (lc) g1 social status lending concentration sslc probability of lending to individuals from the household sector as computed based on eq. 2 g2 private sector lending concentration pslc probability of lending to incorporated private sector borrowers as computed based on eq. 2 g4 public employee lending concentration pelc probability of lending to government employee borrowers as computed based on eq. 2 150 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 appendix 2: number of respondents from different lender institutions bank origin and size total local & large local & small international & large international & small bank financial institutions (lenders) crdb bank plc 23 0 0 0 23 nmb bank plc 20 0 0 0 20 national bank of commerce ltd 7 0 0 0 7 tib development bank 2 0 0 0 2 tanzania agricultural development bank 2 0 0 0 2 bank of tanzania (bot) 1 0 0 0 1 tpb bank plc 0 6 0 0 6 azania bank ltd 0 2 0 0 2 maendeleo bank plc 0 2 0 0 2 bpz bank 0 1 0 0 1 canara bank tanzania ltd 0 1 0 0 1 dcb 0 1 0 0 1 i & m bank 0 1 0 0 1 mwalimu commercial bank 0 1 0 0 1 absa bank tanzania 0 0 8 0 8 standard chartered bank ltd 0 0 4 0 4 stanbic bank tanzania ltd 0 0 2 0 2 banc abc bank 0 0 1 0 1 citibank 0 0 1 0 1 (continued) 151 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 115–151 bank origin and size total local & large local & small international & large international & small equity bank tanzania ltd 0 0 0 9 9 access bank tanzania ltd 0 0 0 7 7 bank of africa (tanzania) ltd 0 0 0 5 5 exim bank tanzania ltd 0 0 0 5 5 united bank for africa (tanzania) ltd 0 0 0 4 4 dtb 0 0 0 3 3 ncba bank tanzania limited 0 0 0 3 3 china bank 0 0 0 1 1 ecobank tanzania 0 0 0 1 1 fbl 0 0 0 1 1 fnb 0 0 0 1 1 guaranty trust bank tanzania 0 0 0 1 1 icb 0 0 0 1 1 finca microfinance bank 0 0 0 2 2 non-bank financial institutions (lenders) vision fund tanzania 0 0 0 1 1 bankable financial services intermediary 0 1 0 0 1 ambroce france shirima 0 1 0 0 1 bumaco insurance 0 1 0 0 1 not mentioned 0 0 0 17 17 54 19 17 61 151 1 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: dagogo, d.w., & ajadi, s. k. (2021). private cost of capital and incremental business value of mid-market firms. international journal of banking and finance, 16(1), 1-20. private cost of capital and incremental business value of mid-market firms 1daibi w. dagogo & 2saheed k. ajadi 1department of banking & finance rivers state university, nigeria 2zartech ltd. nigeria 1corresponding author: dagogo.daibi@ust.edu.ng received: 2/5/2020 revised: 18/8/2020 accepted: 8/9/2020 published: 30/1/2021 abstract this study examined the implications of private cost of capital on the incremental business value (ibv) of middle market firms in nigeria. specifically, three costs were identified as follows: private cost of debt (pcd), private cost of equity (pce), and overall private cost of capital (pcoc). the purpose was to investigate the extent to which private cost of capital, which is calculated differently from weighted average cost of capital for large enterprises, could contribute to incremental business value of middle market (mid-market) firms. two panel data regression models were specified with one dependent variable (incremental business value). the first model has private cost of equity and private cost of debt as independent variables, while the second has private cost of capital as the independent variable. the panel comprised 10 middle market enterprises registered as members of the nigerian association of stock dealers (nasd). https://doi.org/10.32890/ijbf2021.16.1.1 2 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 middle market enterprises are operators in the private sector whose total assets (excluding land and building) are above one hundred and fifty thousand usd but not more than one million five hundred thousand usd. the study adopted the fixed effect model as the best linear estimator after a model validation with the aid of the hausman test. we found that private cost of debt, private cost of equity, and overall private cost of capital have negative and significant effects on the incremental business value of middle market firms. we concluded that incremental business value is more elastic to changes in private cost of equity than private cost of debt, and that this is as a result of two phenomena: firstly, higher explicit private cost of equity than debt, and secondly, greater proportion of private equity than private debt in the capital structure of middle market firms in nigeria. keywords: private cost of capital, private cost of equity, private cost of debt, incremental business value, capital market, middle markets, financial dualism, and capital access point. jel classification: g30 introduction a very disturbing phenomenon in the financial sector of less developed countries is the concept of financial dualism. myint (1985) broadly described it as the coexistence of the formal and informal players in a nation’s financial markets. this phenomenon has policy implications for both macroeconomic management and private enterprise development. for the former, it calls for normative consideration in monetary and fiscal policies. for the latter, it touches on financial decisions of enterprises relating to investing, financing, liquidity, dividend and valuation. incidentally, the same less developed nations with dualistic money, capital and foreign exchange markets also show very weak institutional structures toward doing away with financial dualism. in nigeria, formal and informal financial markets are patronised, respectively, by the bifurcated private sector operators: large and small. we have carefully excluded micro enterprises because this work 3 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 focuses on enterprises with capacity for substantial value creation and transcendental growth. in spite of the lofty argument for financial sector liberalisation and the avoidance of market distortions, there is sufficient literature to support enterprise development interventions by government and multilateral agencies through the provision of soft loans, common facilities, technical supports, etc. (iprc-oecd, 2011). indeed, these interventions are neither easily nor frequently accessible as desired. (dagogo, 2006; dagogo & ohaka, 2015). when supplied through the government media, it is further muddled with other economic phenomena like adverse selection and hoarding. generally, the rates of interest and foreign exchange in the formal markets are less than the rates in the informal markets. besides, transactions in the formal markets are far less risky than those in the informal market place. large enterprises are active players in the formal markets where prices are low. they can also drag prices even lower because of their sheer size and associated economies of scale. again, in the event of government intervention, large enterprises have the social and political networks needed to access such windows at rates lower than the market offer. in the process, they become instruments of “crowding out” the intervention incentives meant for private enterprise development to the disadvantage of small enterprises. meanwhile, small enterprises have difficulties competing with the large enterprises in the market for capital where the highest bidder takes all. when they turn to the government for intervention, they meet insurmountable bottlenecks which they are unable to confront. they may then be left with two options: first is private capital, if their business models are quite attractive for those classes of investors. second is informal financial market where funds are available at usurious rates. very frequently, the second option is the devil’s alternative and the bitter pill small enterprises have to swallow. while it is somewhat easy to articulate the financial decisions of large and small firms along the path of established theories and implicit behaviours in a bifurcated sense (myers & majluf, 1984; leach & melicher, 2012), it does not appear too easy to establish or contemplate precisely what these decisions might be for the middle player of a trifurcated private enterprise sector. it will seem at first that medium enterprises will share in the features of both large and 4 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 small enterprises discussed above. however, this paper assumes ab initio that the contextual circumstances for enterprises (also known as middle market enterprises) will be different, and this will influence their financial decisions differently. the distinctiveness of middle market enterprises has led to a wide spread acceptance of middle market finance theory, a calculated departure from the mainstream finance theory that takes account of the peculiarities of the middle market enterprises on the one hand, and away from the drudgery of the micro and small-sized firms on the other hand. the need for an isolated treatment of middle market firms arises not only from their transient position but also from the expeditious nature of their financing, investing, dividend, liquidity and valuation decisions. thus, they constitute an indispensable linkage in the enterprise development continuum, with an implicit perception of failure where this transition does not occur within a reasonable period. (dagogo & ohaka, 2015). in addition, they differ from small-sized firms in terms of relative contribution to gdp, capacity for bootstrapping, management efficiency, corporate governance, tangibility of assets, informational asymmetry, agency costs, etc. with reference to financing decisions, these differences explain the reason for an adjusted cost of capital model applicable to mid-market firms. (slee, 2011). key questions arising from the background and justifications above are presented here to give the required thought directions: a. is there any significant effect of private cost of capital on the ibv of middle market enterprises? b. is there any significant difference in the effects of private cost of debt and private cost of equity on the ibv of middle market enterprises? literature review this section reviews existing literature in the following sequence: a review of cost of capital, a justification of the use of incremental business value, and an empirical relevance of mid-market finance theory. 5 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 cost of capital the concept of cost of capital is one source of difference between accounting and economic profits. for instance, accounting profit only recognises cost of debt as interest charges without recourse to the associated implicit costs. generally, an investor’s cost of capital is the minimum required rate of return on his investment (vernimmen, quiry, dallocchio, lefur, & salvi, 2014; pouraghajan et al, 2012). put succinctly, an investee’s cost of capital is an investor’s return. it therefore implies that investees would seek to minimise their cost of capital by carefully selecting the most efficient mix of capital structure (referring to an optimal combination of debt and equity), whereas investors would carefully choose a portfolio of asset classes (bond, public equity, private debt, private equity) that minimises their regrets or risks while maximising their returns, always ensuring that their risks are adequately priced as reasonable return. as dagogo and imegi (2017) described it, every mention of capital structure has a corresponding mention of its implications for overall cost of capital or the weighted average cost of capital (wacc). albeit, there are competing views about the reality of creating value only by altering the ratio of debt-to-equity in a manner that reduces wacc. barring any dissenting views and letting the traditional view to prevail, an investee’s ability to create incremental business value is enhanced by reducing wacc (casseli & negri, 2018). capitalisation of private firm of the mid-market range may attract mezzanine, venture capital, securitisation, bank lending, leveraged buyout, management buyout (or buy-in), and other private debt or equity investments. markets for these alternative assets are less efficient than markets for public equity or bond investments because there are fewer organised markets for investors, and therefore deals are usually on an ‘ad hoc’ basis, and fundraising is transacted on a deal-by-deal basis. (dagogo & imegi, 2017; slee 2011) eventually, the financial costs attracted by mid-market firms reflect the perceived riskiness of their investment projects measured by the project’s propensity to generate expected cash flow, and perceived financial risk measured by the degree of financial leverage or their sensitivity to credit and liquidity risks. therefore, to determine the private cost of capital, we consider a risk-free rate plus a risk premium, which does not only cover operating and financial risks but also includes 6 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 a premium for inefficient market, asset intangibility, asymmetric information, and the likes. in other words, risks are priced by capital type, capital access point, and investee type, so that the models of cost of capital based on public equities and bonds such as the capital asset pricing model (capm) and dividend growth model are unsuitable for determining the cost of capital in mid-market firms. (slee, 2011). for example, after-tax cost of debt is generally given as: (1) where rf equals risk-free rate. this measure is helpful in understanding the overall rate paid by a firm for a given type of debt capital. the measure can also give investors an idea of the firm’s risk level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine 5 level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: 𝐾𝐾𝑒𝑒 = 𝑟𝑟𝑟𝑟 + 𝛽𝛽(𝑚𝑚 − 𝑟𝑟𝑟𝑟) (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine the rate of return required by private capital investors. it is of the following form: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = ∑ [(𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 + 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖𝑁𝑁𝑖𝑖=1 ) × 𝑀𝑀𝑀𝑀𝑖𝑖 ∑ 𝑀𝑀𝑀𝑀𝑖𝑖𝑁𝑁𝑖𝑖=1 ] (3) where: n equals number of sources of capital; 𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals median expected return for capital type 𝑖𝑖; 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals specific cap risk adjustment for capital type 𝑖𝑖; while 𝑀𝑀𝑀𝑀𝑖𝑖 equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt(𝑘𝑘𝑑𝑑), preference (𝑘𝑘𝑝𝑝), ordinary equity (𝑘𝑘𝑒𝑒 ) and retained earnings(𝑘𝑘𝑟𝑟 ), pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value 4 cost of capital the concept of cost of capital is one source of difference between accounting and economic profits. for instance, accounting profit only recognises cost of debt as interest charges without recourse to the associated implicit costs. generally, an investor’s cost of capital is the minimum required rate of return on his investment (vernimmen, quiry, dallocchio, lefur, & salvi, 2014; pouraghajan et al, 2012). put succinctly, an investee’s cost of capital is an investor’s return. it therefore implies that investees would seek to minimise their cost of capital by carefully selecting the most efficient mix of capital structure (referring to an optimal combination of debt and equity), whereas investors would carefully choose a portfolio of asset classes (bond, public equity, private debt, private equity) that minimises their regrets or risks while maximising their returns, always ensuring that their risks are adequately priced as reasonable return. as dagogo and imegi (2017) described it, every mention of capital structure has a corresponding mention of its implications for overall cost of capital or the weighted average cost of capital (wacc). albeit, there are competing views about the reality of creating value only by altering the ratio of debt-to-equity in a manner that reduces wacc. barring any dissenting views and letting the traditional view to prevail, an investee’s ability to create incremental business value is enhanced by reducing wacc (casseli & negri, 2018). capitalisation of private firm of the mid-market range may attract mezzanine, venture capital, securitisation, bank lending, leveraged buyout, management buyout (or buy-in), and other private debt or equity investments. markets for these alternative assets are less efficient than markets for public equity or bond investments because there are fewer organised markets for investors, and therefore deals are usually on an ‘ad hoc’ basis, and fundraising is transacted on a deal-by-deal basis. (dagogo & imegi, 2017; slee 2011) eventually, the financial costs attracted by mid-market firms reflect the perceived riskiness of their investment projects measured by the project’s propensity to generate expected cash flow, and perceived financial risk measured by the degree of financial leverage or their sensitivity to credit and liquidity risks. therefore, to determine the private cost of capital, we consider a risk-free rate plus a risk premium, which does not only cover operating and financial risks but also includes a premium for inefficient market, asset intangibility, asymmetric information, and the likes. in other words, risks are priced by capital type, capital access point, and investee type, so that the models of cost of capital based on public equities and bonds such as the capital asset pricing model (capm) and dividend growth model are unsuitable for determining the cost of capital in mid-market firms. (slee, 2011). for example, after-tax cost of debt is generally given as: (𝑟𝑟𝑟𝑟 + 𝑐𝑐𝑟𝑟𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑠𝑠𝑠𝑠𝑟𝑟𝑐𝑐𝑠𝑠𝑐𝑐)𝑥𝑥(1 − 𝑐𝑐𝑠𝑠𝑥𝑥 𝑟𝑟𝑠𝑠𝑐𝑐𝑐𝑐) (1) where rf equals risk-free rate. this measure is helpful in understanding the overall rate paid by a firm for a given type of debt capital. the measure can also give investors an idea of the firm’s risk 7 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 the rate of return required by private capital investors. it is of the following form: (3) where: n equals number of sources of capital; equals median expected return for capital type equals specific cap risk adjustment for capital type while equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt preference, ordinary equity and retained earnings pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value incremental business value (ibv) is the result of generating a return in excess of the corresponding cost of capital. recently, ibv has become the most preferred value metric because it applies economic definition of value rather than accounting book value. secondly, it is dynamic, applicable for continuous value determination processes, and is multipurposed such as for appraising whole entity, strategic business units, product lines, specific projects or employees. it is also useful for capital allocation decisions and in measuring the efficiency of capital employed. for private firms, the expected cash flow is discounted by a rate derived upon the use of private cost of capital model given in equation 3. it is given thus: (4) where ibv is incremental business value; and ebitda is earnings before interest, tax depreciation and amortisation, which is adjusted for owners’ discretionary expenses and one-time enterprise expenses. 5 level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: 𝐾𝐾𝑒𝑒 = 𝑟𝑟𝑟𝑟 + 𝛽𝛽(𝑚𝑚 − 𝑟𝑟𝑟𝑟) (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine the rate of return required by private capital investors. it is of the following form: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = ∑ [(𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 + 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖𝑁𝑁𝑖𝑖=1 ) × 𝑀𝑀𝑀𝑀𝑖𝑖 ∑ 𝑀𝑀𝑀𝑀𝑖𝑖𝑁𝑁𝑖𝑖=1 ] (3) where: n equals number of sources of capital; 𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals median expected return for capital type 𝑖𝑖; 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals specific cap risk adjustment for capital type 𝑖𝑖; while 𝑀𝑀𝑀𝑀𝑖𝑖 equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt(𝑘𝑘𝑑𝑑), preference (𝑘𝑘𝑝𝑝), ordinary equity (𝑘𝑘𝑒𝑒 ) and retained earnings(𝑘𝑘𝑟𝑟 ), pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value 5 level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: 𝐾𝐾𝑒𝑒 = 𝑟𝑟𝑟𝑟 + 𝛽𝛽(𝑚𝑚 − 𝑟𝑟𝑟𝑟) (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine the rate of return required by private capital investors. it is of the following form: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = ∑ [(𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 + 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖𝑁𝑁𝑖𝑖=1 ) × 𝑀𝑀𝑀𝑀𝑖𝑖 ∑ 𝑀𝑀𝑀𝑀𝑖𝑖𝑁𝑁𝑖𝑖=1 ] (3) where: n equals number of sources of capital; 𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals median expected return for capital type 𝑖𝑖; 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals specific cap risk adjustment for capital type 𝑖𝑖; while 𝑀𝑀𝑀𝑀𝑖𝑖 equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt(𝑘𝑘𝑑𝑑), preference (𝑘𝑘𝑝𝑝), ordinary equity (𝑘𝑘𝑒𝑒 ) and retained earnings(𝑘𝑘𝑟𝑟 ), pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value 5 level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: 𝐾𝐾𝑒𝑒 = 𝑟𝑟𝑟𝑟 + 𝛽𝛽(𝑚𝑚 − 𝑟𝑟𝑟𝑟) (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine the rate of return required by private capital investors. it is of the following form: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = ∑ [(𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 + 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖𝑁𝑁𝑖𝑖=1 ) × 𝑀𝑀𝑀𝑀𝑖𝑖 ∑ 𝑀𝑀𝑀𝑀𝑖𝑖𝑁𝑁𝑖𝑖=1 ] (3) where: n equals number of sources of capital; 𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals median expected return for capital type 𝑖𝑖; 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals specific cap risk adjustment for capital type 𝑖𝑖; while 𝑀𝑀𝑀𝑀𝑖𝑖 equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt(𝑘𝑘𝑑𝑑), preference (𝑘𝑘𝑝𝑝), ordinary equity (𝑘𝑘𝑒𝑒 ) and retained earnings(𝑘𝑘𝑟𝑟 ), pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value 5 level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: 𝐾𝐾𝑒𝑒 = 𝑟𝑟𝑟𝑟 + 𝛽𝛽(𝑚𝑚 − 𝑟𝑟𝑟𝑟) (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine the rate of return required by private capital investors. it is of the following form: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = ∑ [(𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 + 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖𝑁𝑁𝑖𝑖=1 ) × 𝑀𝑀𝑀𝑀𝑖𝑖 ∑ 𝑀𝑀𝑀𝑀𝑖𝑖𝑁𝑁𝑖𝑖=1 ] (3) where: n equals number of sources of capital; 𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals median expected return for capital type 𝑖𝑖; 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals specific cap risk adjustment for capital type 𝑖𝑖; while 𝑀𝑀𝑀𝑀𝑖𝑖 equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt(𝑘𝑘𝑑𝑑), preference (𝑘𝑘𝑝𝑝), ordinary equity (𝑘𝑘𝑒𝑒 ) and retained earnings(𝑘𝑘𝑟𝑟 ), pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value 5 level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: 𝐾𝐾𝑒𝑒 = 𝑟𝑟𝑟𝑟 + 𝛽𝛽(𝑚𝑚 − 𝑟𝑟𝑟𝑟) (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine the rate of return required by private capital investors. it is of the following form: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = ∑ [(𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 + 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖𝑁𝑁𝑖𝑖=1 ) × 𝑀𝑀𝑀𝑀𝑖𝑖 ∑ 𝑀𝑀𝑀𝑀𝑖𝑖𝑁𝑁𝑖𝑖=1 ] (3) where: n equals number of sources of capital; 𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals median expected return for capital type 𝑖𝑖; 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals specific cap risk adjustment for capital type 𝑖𝑖; while 𝑀𝑀𝑀𝑀𝑖𝑖 equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt(𝑘𝑘𝑑𝑑), preference (𝑘𝑘𝑝𝑝), ordinary equity (𝑘𝑘𝑒𝑒 ) and retained earnings(𝑘𝑘𝑟𝑟 ), pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value 5 level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: 𝐾𝐾𝑒𝑒 = 𝑟𝑟𝑟𝑟 + 𝛽𝛽(𝑚𝑚 − 𝑟𝑟𝑟𝑟) (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine the rate of return required by private capital investors. it is of the following form: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = ∑ [(𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 + 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖𝑁𝑁𝑖𝑖=1 ) × 𝑀𝑀𝑀𝑀𝑖𝑖 ∑ 𝑀𝑀𝑀𝑀𝑖𝑖𝑁𝑁𝑖𝑖=1 ] (3) where: n equals number of sources of capital; 𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals median expected return for capital type 𝑖𝑖; 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals specific cap risk adjustment for capital type 𝑖𝑖; while 𝑀𝑀𝑀𝑀𝑖𝑖 equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt(𝑘𝑘𝑑𝑑), preference (𝑘𝑘𝑝𝑝), ordinary equity (𝑘𝑘𝑒𝑒 ) and retained earnings(𝑘𝑘𝑟𝑟 ), pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value 5 level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: 𝐾𝐾𝑒𝑒 = 𝑟𝑟𝑟𝑟 + 𝛽𝛽(𝑚𝑚 − 𝑟𝑟𝑟𝑟) (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine the rate of return required by private capital investors. it is of the following form: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = ∑ [(𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 + 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖𝑁𝑁𝑖𝑖=1 ) × 𝑀𝑀𝑀𝑀𝑖𝑖 ∑ 𝑀𝑀𝑀𝑀𝑖𝑖𝑁𝑁𝑖𝑖=1 ] (3) where: n equals number of sources of capital; 𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals median expected return for capital type 𝑖𝑖; 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals specific cap risk adjustment for capital type 𝑖𝑖; while 𝑀𝑀𝑀𝑀𝑖𝑖 equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt(𝑘𝑘𝑑𝑑), preference (𝑘𝑘𝑝𝑝), ordinary equity (𝑘𝑘𝑒𝑒 ) and retained earnings(𝑘𝑘𝑟𝑟 ), pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value 5 level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: 𝐾𝐾𝑒𝑒 = 𝑟𝑟𝑟𝑟 + 𝛽𝛽(𝑚𝑚 − 𝑟𝑟𝑟𝑟) (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine the rate of return required by private capital investors. it is of the following form: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = ∑ [(𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 + 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖𝑁𝑁𝑖𝑖=1 ) × 𝑀𝑀𝑀𝑀𝑖𝑖 ∑ 𝑀𝑀𝑀𝑀𝑖𝑖𝑁𝑁𝑖𝑖=1 ] (3) where: n equals number of sources of capital; 𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals median expected return for capital type 𝑖𝑖; 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals specific cap risk adjustment for capital type 𝑖𝑖; while 𝑀𝑀𝑀𝑀𝑖𝑖 equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt(𝑘𝑘𝑑𝑑), preference (𝑘𝑘𝑝𝑝), ordinary equity (𝑘𝑘𝑒𝑒 ) and retained earnings(𝑘𝑘𝑟𝑟 ), pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value 5 level compared to others because riskier firms generally have higher cost of debt (van binsbergen, graham & yang, 2010). meanwhile, cost of equity represents a firm’s hurdle rate, marginal rate of efficiency of capital invested or the internal rate of return. it is often used as the capital budgeting threshold for required return. a firm’s cost of equity represents the compensation the market demands in exchange for owning the asset and bearing the risk of ownership. capm seems to be more suitable for a wider spectrum of firms (including non-dividend paying firms) than the dividend growth model. that said, the theory behind capm is more complex, as it is based on an equity stock’s volatility and the level of individual stock’s risk exposure relative to the market risk. it is given as: 𝐾𝐾𝑒𝑒 = 𝑟𝑟𝑟𝑟 + 𝛽𝛽(𝑚𝑚 − 𝑟𝑟𝑟𝑟) (2) where ke equals cost of equity capital, rf equals risk-free rate or the rate of return paid on risk-free investments such as treasury bills, and β equals the beta or measure of risk calculated as a regression on the firms’ stock price. the higher the volatility, the higher the beta and relative risk. and m equals market rate of return (mclaney, 2009; vernimmen et al., 2014). since the central theme of this study is on external financing, retained earnings are assumed away and not discussed in this literature. besides, the effect of retained earnings on lowering the overall cost of capital is not in doubt. unfortunately, the resulting overall cost of capital (ko) formulated with large enterprises return data cannot be applied to derive the private cost of capital. slee (2011) recommends a private discount rate model to determine the rate of return required by private capital investors. it is of the following form: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = ∑ [(𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 + 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖𝑁𝑁𝑖𝑖=1 ) × 𝑀𝑀𝑀𝑀𝑖𝑖 ∑ 𝑀𝑀𝑀𝑀𝑖𝑖𝑁𝑁𝑖𝑖=1 ] (3) where: n equals number of sources of capital; 𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals median expected return for capital type 𝑖𝑖; 𝑆𝑆𝑃𝑃𝐶𝐶𝑃𝑃𝑖𝑖 equals specific cap risk adjustment for capital type 𝑖𝑖; while 𝑀𝑀𝑀𝑀𝑖𝑖 equals market value of outstanding securities. the process of determining pcoc involves four steps. first is to determine the appropriate capital types; second is to determine the market value of each capital type; third is to apply a specific capital type (scap) risk adjustment to the selected median capital type; and fourth is to calculate the percentage of capital structure for each capital access point and add the individual percentages to derive pcoc. accordingly, the difference between wacc and pcoc lies in the manner of risk weighting. while the overall cost of capital for public firms is riskweighted in the known broad classifications such as debt(𝑘𝑘𝑑𝑑), preference (𝑘𝑘𝑝𝑝), ordinary equity (𝑘𝑘𝑒𝑒 ) and retained earnings(𝑘𝑘𝑟𝑟 ), pcoc accounts for risk properties of each specific capital type within the debt and equity classifications. incremental business value 6 incremental business value (ibv) is the result of generating a return in excess of the corresponding cost of capital. recently, ibv has become the most preferred value metric because it applies economic definition of value rather than accounting book value. secondly, it is dynamic, applicable for continuous value determination processes, and is multi-purposed such as for appraising whole entity, strategic business units, product lines, specific projects or employees. it is also useful for capital allocation decisions and in measuring the efficiency of capital employed. for private firms, the expected cash flow is discounted by a rate derived upon the use of private cost of capital model given in equation 3. it is given thus: 𝐼𝐼𝐼𝐼𝐼𝐼 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑎𝑎𝑎𝑎𝑎𝑎 𝐸𝐸𝐼𝐼𝐼𝐼𝐸𝐸𝐸𝐸𝐸𝐸 − (𝐼𝐼𝐼𝐼𝐼𝐼𝑅𝑅𝑎𝑎𝑎𝑎𝐼𝐼𝑅𝑅𝐼𝐼𝑎𝑎 × 𝑝𝑝𝑝𝑝𝑝𝑝𝐼𝐼𝑎𝑎𝑎𝑎𝑅𝑅 𝑅𝑅𝑐𝑐𝑎𝑎𝑎𝑎 𝑐𝑐𝑜𝑜 𝑅𝑅𝑎𝑎𝑝𝑝𝑝𝑝𝑎𝑎𝑎𝑎𝑐𝑐) (4) where ibv is incremental business value; and ebitda is earnings before interest, tax depreciation and amortisation, which is adjusted for owners’ discretionary expenses and one-time enterprise expenses. for investment, the relevant figure is the greater of the total amount of capital investment or the financial market value. a positive ibv reflects shareholders’ value accretion and a negative ibv indicates depletion. while drucker (1998) alluded that ibv is the genuine economic profit that arises from generating revenue beyond the corresponding economic cost, miller and modigliani (1958; 1963) found it as the key indicator of the required rate of market return, which is sufficient to compensate for risk and economic income. it was this understanding that laid the foundation for discounted cash flow methodology, and the use of net present value (npv). another metric that became more acceptable in the industry than npv is economic value added (eva), patented by stern stewart & co. the difference between ibv and eva lies in the nature of expected cash flow to be discounted. in eva the expected cash flow is the same as operating profit, whereas in ibv it is ebitda (slee, 2011; copeland, koller, & murrin, 1996; and kanabali & kashinath, 2015). garcio and aguilera (2014) made a novel departure in the process of identifying and measuring value by redefining the taxonomy of value creation and appropriation. they proposed incremental value creation and appropriation, which in their view, attended to the difficulty in estimating customers and suppliers value-added. they assumed that the sum of value created should not always be equal to the sum of value appropriated, leading to a zero-sum game at all times. in other words, their position is that value creation and appropriation could be positive, negative or zero sum. also, their research notes asserted that value is not necessarily created by fund providers alone but by multiple stakeholders. to that extent, they classified previously used accounting measures of value as too absolute. while we acknowledge the wider scope and application of coverage of incremental value, as noted above, this work is concerned only with private capital elasticities of incremental business value. middle market finance theory 8 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 for investment, the relevant figure is the greater of the total amount of capital investment or the financial market value. a positive ibv reflects shareholders’ value accretion and a negative ibv indicates depletion. while drucker (1998) alluded that ibv is the genuine economic profit that arises from generating revenue beyond the corresponding economic cost, miller and modigliani (1958; 1963) found it as the key indicator of the required rate of market return, which is sufficient to compensate for risk and economic income. it was this understanding that laid the foundation for discounted cash flow methodology, and the use of net present value (npv). another metric that became more acceptable in the industry than npv is economic value added (eva), patented by stern stewart & co. the difference between ibv and eva lies in the nature of expected cash flow to be discounted. in eva the expected cash flow is the same as operating profit, whereas in ibv it is ebitda (slee, 2011; copeland et al., 1996; and kanabali & kashinath, 2015). garcio and aguilera (2014) made a novel departure in the process of identifying and measuring value by redefining the taxonomy of value creation and appropriation. they proposed incremental value creation and appropriation, which in their view, attended to the difficulty in estimating customers and suppliers value-added. they assumed that the sum of value created should not always be equal to the sum of value appropriated, leading to a zero-sum game at all times. in other words, their position is that value creation and appropriation could be positive, negative or zero sum. also, their research notes asserted that value is not necessarily created by fund providers alone but by multiple stakeholders. to that extent, they classified previously used accounting measures of value as too absolute. while we acknowledge the wider scope and application of coverage of incremental value, as noted above, this work is concerned only with private capital elasticities of incremental business value. middle market finance theory middle market finance theory finds a dividing line between small and medium enterprises finance theories on the one hand, and between medium and large enterprises on the other hand. in doing so, the midmarket finance theory identifies a triadic integration of valuation, capitalisation and transfer of ownership interests that explains capital market decisions of medium-sized firms. slee (2011) argues that 9 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 valuation forms the base of the interconnection, and maintains a balance with the other two components, without which much of it will be done in isolation from the market and will amount to vanity. absence of this balance will be a source of disequilibrium in the case of private securities that do not have access to the active trading market. they must rely on point-in-time appraisal or discrete rather than continuous pricing and value determination. next, capitalisation rests on valuation to enable private capital markets price and allocate capital, given their risk-return sensitivity. although every round of capitalization requires preand post-money valuation, the inefficiency of the private capital markets, arising from informational asymmetry, distorts the valuation results. (leleux, swaay & megally, 2015). finally, transfer of ownership interest relies on capitalisation and valuation. it takes different forms and involves parties from within and without the firm. the shortcomings of the earlier pillars inhibit the ease of transfers of ownership of private firms. lack of liquidity hinders diversification, which increases the riskiness of private transfer markets. it is the coherent importance of all three concepts at any given time in exhibiting the same weakness of mid-market firms that inspired the thinking of constituting the midmarket finance theory. methodology ex-post facto research design was adopted since the research relied on historical data generated from annual reports of mid-market enterprises. besides, because of our limited ability to subject the study environment to an acceptable degree of control and our normative processes in the inclusion of independent variables, the study fits the quasi-experimental research design. the population consists of all enterprises in nigeria whose total assets (excluding land and building) are above one hundred and fifty thousand usd but not exceeding one million five hundred thousand usd and with a total workforce above fifty employees but not exceeding one hundred and ninety-nine employees (smedan, 2013). a sample of ten such enterprises listed under the nigerian association of stock dealers (nasd), an overthe-counter securities market, was selected not without the following considerations: maintenance of balanced panel, ease of data collection, reliability of data, and completeness of requisite data. (cscs, 2018). 10 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 ten years data were collected covering 2009 to 2018 from a cross section of the sample. it may seem at first that the time series data for ten years were insufficient, nevertheless, the study remains robust on account of the cross sectional data, which sufficiently compensated for the limited time series, leaving us with a panel dataset of 100 data points. we employed a two-stage analytical process. the first stage examined the panel regression models, showing two alternate models: fixed effect and random effect models. the second stage tested for serial autocorrelation to show if there is a serial relationship between a variable in time t and the same variable in time t l. two models were formulated with ibv as dependent variables in each case: cost of debt (represented by the one most important source of debt capital in the firms’ capital structure) and cost of equity (represented by one most important source of equity in the firms’ capital structure) as independent variables for one model; and pcoc, as an independent variable for the other model. it is the proxy for overall cost of capital for private firms. the first model examined the individual effects of each source of capital for the mid-market firm while the second looked at the global effect. both models were isolated to avoid multicollinearity. the model follows the classical linear regression equation of the following form: (4) (5) where ibv equals incremental business value, calculated for each enterprise as defined in equation 3; pcd equals private cost of debt; pce equals private cost of equity; pcoc equals private cost of capital; b0 equals constant term; b1 b2 equal coefficients of predictors; and equals error term or stochastic variable representing the uncontrolled country-specific factors such as demand volatility, business cycle, labour market, etc. (pourghajan et al., 2012). two panel regression models were specified in line with fixed and random effects estimation procedures, respectively (see results presented in table 2). 8 serial autocorrelation to show if there is a serial relationship between a variable in time t and the same variable in time t l. two models were formulated with ibv as dependent variables in each case: cost of debt (represented by the one most important source of debt capital in the firms’ capital structure) and cost of equity (represented by one most important source of equity in the firms’ capital structure) as independent variables for one model; and pcoc, as an independent variable for the other model. it is the proxy for overall cost of capital for private firms. the first model examined the individual effects of each source of capital for the mid-market firm while the second looked at the global effect. both models were isolated to avoid multicollinearity. the model follows the classical linear regression equation of the following form: 𝐼𝐼𝐼𝐼𝐼𝐼 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃 + 2𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑖𝑖 (4) 𝐼𝐼𝐼𝐼𝐼𝐼 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑖𝑖 (5) where ibv equals incremental business value, calculated for each enterprise as defined in equation 3; pcd equals private cost of debt; pce equals private cost of equity; pcoc equals private cost of capital; 0 equals constant term; 1 2 equal coefficients of predictors; and  equals error term or stochastic variable representing the uncontrolled country-specific factors such as demand volatility, business cycle, labour market, etc. (pourghajan et al., 2012). two panel regression models were specified in line with fixed and random effects estimation procedures, respectively (see results presented in table 2). a. fixed effects: this focuses on whether there are differences by using a fixed intercept for each of the different cross-sectional structures, as the difference may be due to special features of each mid-market firm. it is given thus: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 1 + 2 d2𝑖𝑖 + 3 d3𝑖𝑖 + 4 d4𝑖𝑖 + 5 d5𝑖𝑖 + ⋯ 10 d10𝑖𝑖 + 11𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 12𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝑖𝑖𝑖𝑖 (6) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝑖𝑖𝑖𝑖 (7) because the fixed effects account for both cross-sectional and time-series data, the increased covariance caused by individual-firm differences is eliminated, thereby increasing its estimation-result efficiency. in this case, 1 equals intercept value of firm 1; 2 – 10 represent the differences of the other firms from firm 1, measured by their differences of their coefficients; d2𝑖𝑖 − d10𝑖𝑖 represent 9 dummy variables representing 9 firms; 11 12 equal coefficients of predictors; and  equals error term or stochastic variable representing other 8 serial autocorrelation to show if there is a serial relationship between a variable in time t and the same variable in time t l. two models were formulated with ibv as dependent variables in each case: cost of debt (represented by the one most important source of debt capital in the firms’ capital structure) and cost of equity (represented by one most important source of equity in the firms’ capital structure) as independent variables for one model; and pcoc, as an independent variable for the other model. it is the proxy for overall cost of capital for private firms. the first model examined the individual effects of each source of capital for the mid-market firm while the second looked at the global effect. both models were isolated to avoid multicollinearity. the model follows the classical linear regression equation of the following form: 𝐼𝐼𝐼𝐼𝐼𝐼 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃 + 2𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑖𝑖 (4) 𝐼𝐼𝐼𝐼𝐼𝐼 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑖𝑖 (5) where ibv equals incremental business value, calculated for each enterprise as defined in equation 3; pcd equals private cost of debt; pce equals private cost of equity; pcoc equals private cost of capital; 0 equals constant term; 1 2 equal coefficients of predictors; and  equals error term or stochastic variable representing the uncontrolled country-specific factors such as demand volatility, business cycle, labour market, etc. (pourghajan et al., 2012). two panel regression models were specified in line with fixed and random effects estimation procedures, respectively (see results presented in table 2). a. fixed effects: this focuses on whether there are differences by using a fixed intercept for each of the different cross-sectional structures, as the difference may be due to special features of each mid-market firm. it is given thus: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 1 + 2 d2𝑖𝑖 + 3 d3𝑖𝑖 + 4 d4𝑖𝑖 + 5 d5𝑖𝑖 + ⋯ 10 d10𝑖𝑖 + 11𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 12𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝑖𝑖𝑖𝑖 (6) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝑖𝑖𝑖𝑖 (7) because the fixed effects account for both cross-sectional and time-series data, the increased covariance caused by individual-firm differences is eliminated, thereby increasing its estimation-result efficiency. in this case, 1 equals intercept value of firm 1; 2 – 10 represent the differences of the other firms from firm 1, measured by their differences of their coefficients; d2𝑖𝑖 − d10𝑖𝑖 represent 9 dummy variables representing 9 firms; 11 12 equal coefficients of predictors; and  equals error term or stochastic variable representing other 8 serial autocorrelation to show if there is a serial relationship between a variable in time t and the same variable in time t l. two models were formulated with ibv as dependent variables in each case: cost of debt (represented by the one most important source of debt capital in the firms’ capital structure) and cost of equity (represented by one most important source of equity in the firms’ capital structure) as independent variables for one model; and pcoc, as an independent variable for the other model. it is the proxy for overall cost of capital for private firms. the first model examined the individual effects of each source of capital for the mid-market firm while the second looked at the global effect. both models were isolated to avoid multicollinearity. the model follows the classical linear regression equation of the following form: 𝐼𝐼𝐼𝐼𝐼𝐼 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃 + 2𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑖𝑖 (4) 𝐼𝐼𝐼𝐼𝐼𝐼 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑖𝑖 (5) where ibv equals incremental business value, calculated for each enterprise as defined in equation 3; pcd equals private cost of debt; pce equals private cost of equity; pcoc equals private cost of capital; 0 equals constant term; 1 2 equal coefficients of predictors; and  equals error term or stochastic variable representing the uncontrolled country-specific factors such as demand volatility, business cycle, labour market, etc. (pourghajan et al., 2012). two panel regression models were specified in line with fixed and random effects estimation procedures, respectively (see results presented in table 2). a. fixed effects: this focuses on whether there are differences by using a fixed intercept for each of the different cross-sectional structures, as the difference may be due to special features of each mid-market firm. it is given thus: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 1 + 2 d2𝑖𝑖 + 3 d3𝑖𝑖 + 4 d4𝑖𝑖 + 5 d5𝑖𝑖 + ⋯ 10 d10𝑖𝑖 + 11𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 12𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝑖𝑖𝑖𝑖 (6) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝑖𝑖𝑖𝑖 (7) because the fixed effects account for both cross-sectional and time-series data, the increased covariance caused by individual-firm differences is eliminated, thereby increasing its estimation-result efficiency. in this case, 1 equals intercept value of firm 1; 2 – 10 represent the differences of the other firms from firm 1, measured by their differences of their coefficients; d2𝑖𝑖 − d10𝑖𝑖 represent 9 dummy variables representing 9 firms; 11 12 equal coefficients of predictors; and  equals error term or stochastic variable representing other 11 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 fixed effect this focuses on whether there are differences by using a fixed intercept for each of the different cross-sectional structures, as the difference may be due to special features of each mid-market firm. it is given thus: (6) (7) because the fixed effects account for both cross-sectional and timeseries data, the increased covariance caused by individual-firm differences is eliminated, thereby increasing its estimation-result efficiency. in this case, b1 equals intercept value of firm 1; b2 b10 represent the differences of the other firms from firm 1, measured by their differences of their coefficients; d2i d10i represent 9 dummy variables representing 9 firms; b11 b12 equal coefficients of predictors; and equals error term or stochastic variable representing other variables that are merely explained away. i = 1, 2, ….10 mid-market firms; and t = 1, 2, ......10 years. random effects this focuses on the relationship with the study sample as a whole, thus the samples are randomly selected, as opposed to using the entire population. the total sample regression function of the random effect can be expressed as: (8) (9) if this is represented with random variables, then + i, which indicates that the difference occurs randomly, and the expected value of and that is, the error term is a sum of two parts: is the combined time series and cross-section 8 serial autocorrelation to show if there is a serial relationship between a variable in time t and the same variable in time t l. two models were formulated with ibv as dependent variables in each case: cost of debt (represented by the one most important source of debt capital in the firms’ capital structure) and cost of equity (represented by one most important source of equity in the firms’ capital structure) as independent variables for one model; and pcoc, as an independent variable for the other model. it is the proxy for overall cost of capital for private firms. the first model examined the individual effects of each source of capital for the mid-market firm while the second looked at the global effect. both models were isolated to avoid multicollinearity. the model follows the classical linear regression equation of the following form: 𝐼𝐼𝐼𝐼𝐼𝐼 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃 + 2𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑖𝑖 (4) 𝐼𝐼𝐼𝐼𝐼𝐼 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑖𝑖 (5) where ibv equals incremental business value, calculated for each enterprise as defined in equation 3; pcd equals private cost of debt; pce equals private cost of equity; pcoc equals private cost of capital; 0 equals constant term; 1 2 equal coefficients of predictors; and  equals error term or stochastic variable representing the uncontrolled country-specific factors such as demand volatility, business cycle, labour market, etc. (pourghajan et al., 2012). two panel regression models were specified in line with fixed and random effects estimation procedures, respectively (see results presented in table 2). a. fixed effects: this focuses on whether there are differences by using a fixed intercept for each of the different cross-sectional structures, as the difference may be due to special features of each mid-market firm. it is given thus: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 1 + 2 d2𝑖𝑖 + 3 d3𝑖𝑖 + 4 d4𝑖𝑖 + 5 d5𝑖𝑖 + ⋯ 10 d10𝑖𝑖 + 11𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 12𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝑖𝑖𝑖𝑖 (6) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝑖𝑖𝑖𝑖 (7) because the fixed effects account for both cross-sectional and time-series data, the increased covariance caused by individual-firm differences is eliminated, thereby increasing its estimation-result efficiency. in this case, 1 equals intercept value of firm 1; 2 – 10 represent the differences of the other firms from firm 1, measured by their differences of their coefficients; d2𝑖𝑖 − d10𝑖𝑖 represent 9 dummy variables representing 9 firms; 11 12 equal coefficients of predictors; and  equals error term or stochastic variable representing other 9 variables that are merely explained away. i = 1, 2, ….10 mid-market firms; and t = 1, 2, ......10 years. b. random effects: this focuses on the relationship with the study sample as a whole, thus the samples are randomly selected, as opposed to using the entire population. the total sample regression function of the random effect can be expressed as: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 2𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (8) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (9) if this is represented with random variables, then 0𝑖𝑖 =  + i, which indicates that the difference occurs randomly, and the expected value of 0𝑖𝑖 is 0, and w𝑖𝑖𝑖𝑖 = ε𝑖𝑖𝑖𝑖 + μ𝑖𝑖𝑖𝑖 , that is, the error term w𝑖𝑖𝑖𝑖 is a sum of two parts: ε𝑖𝑖𝑖𝑖 is the combined time series and cross-section error component, and μ𝑖𝑖𝑖𝑖 is the cross-section or firm-specific error component. c. hausman test: this is the most commonly used method for evaluating fixed and random effects. it shows whether sectional unit differences are fixed or random. it has an asymptotic chi-square distribution with degrees of freedom equal to the number of independent variables. if the variables are significantly correlated, then the fixed effects estimation is consistent and efficient, and should be adopted. on the other hand, if the variables are not significantly correlated, then the random effects estimation is consistent and efficient, and should be adopted. (gujurati, porter, & gunasakar, 2013). serial autocorrelation was also conducted to test the existence of serial relationship between one variable in time t and the same variable in time t 1. the result shows no serial correlation amongst the variables. table 1 breush-godfrey serial correlation lm test model 1 f-statistic 3.181682 prob. f(2,105) 0.0041 obs*r-squared 0.379355 prob. chi-square (2) 0.8272 model 2 f-statistic 0.196212 prob. f(2,106) 0.8221 obs*r-squared 0.405730 prob. chi-square(2) 0.8164 9 variables that are merely explained away. i = 1, 2, ….10 mid-market firms; and t = 1, 2, ......10 years. b. random effects: this focuses on the relationship with the study sample as a whole, thus the samples are randomly selected, as opposed to using the entire population. the total sample regression function of the random effect can be expressed as: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 2𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (8) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (9) if this is represented with random variables, then 0𝑖𝑖 =  + i, which indicates that the difference occurs randomly, and the expected value of 0𝑖𝑖 is 0, and w𝑖𝑖𝑖𝑖 = ε𝑖𝑖𝑖𝑖 + μ𝑖𝑖𝑖𝑖 , that is, the error term w𝑖𝑖𝑖𝑖 is a sum of two parts: ε𝑖𝑖𝑖𝑖 is the combined time series and cross-section error component, and μ𝑖𝑖𝑖𝑖 is the cross-section or firm-specific error component. c. hausman test: this is the most commonly used method for evaluating fixed and random effects. it shows whether sectional unit differences are fixed or random. it has an asymptotic chi-square distribution with degrees of freedom equal to the number of independent variables. if the variables are significantly correlated, then the fixed effects estimation is consistent and efficient, and should be adopted. on the other hand, if the variables are not significantly correlated, then the random effects estimation is consistent and efficient, and should be adopted. (gujurati, porter, & gunasakar, 2013). serial autocorrelation was also conducted to test the existence of serial relationship between one variable in time t and the same variable in time t 1. the result shows no serial correlation amongst the variables. table 1 breush-godfrey serial correlation lm test model 1 f-statistic 3.181682 prob. f(2,105) 0.0041 obs*r-squared 0.379355 prob. chi-square (2) 0.8272 model 2 f-statistic 0.196212 prob. f(2,106) 0.8221 obs*r-squared 0.405730 prob. chi-square(2) 0.8164 9 variables that are merely explained away. i = 1, 2, ….10 mid-market firms; and t = 1, 2, ......10 years. b. random effects: this focuses on the relationship with the study sample as a whole, thus the samples are randomly selected, as opposed to using the entire population. the total sample regression function of the random effect can be expressed as: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 2𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (8) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (9) if this is represented with random variables, then 0𝑖𝑖 =  + i, which indicates that the difference occurs randomly, and the expected value of 0𝑖𝑖 is 0, and w𝑖𝑖𝑖𝑖 = ε𝑖𝑖𝑖𝑖 + μ𝑖𝑖𝑖𝑖 , that is, the error term w𝑖𝑖𝑖𝑖 is a sum of two parts: ε𝑖𝑖𝑖𝑖 is the combined time series and cross-section error component, and μ𝑖𝑖𝑖𝑖 is the cross-section or firm-specific error component. c. hausman test: this is the most commonly used method for evaluating fixed and random effects. it shows whether sectional unit differences are fixed or random. it has an asymptotic chi-square distribution with degrees of freedom equal to the number of independent variables. if the variables are significantly correlated, then the fixed effects estimation is consistent and efficient, and should be adopted. on the other hand, if the variables are not significantly correlated, then the random effects estimation is consistent and efficient, and should be adopted. (gujurati, porter, & gunasakar, 2013). serial autocorrelation was also conducted to test the existence of serial relationship between one variable in time t and the same variable in time t 1. the result shows no serial correlation amongst the variables. table 1 breush-godfrey serial correlation lm test model 1 f-statistic 3.181682 prob. f(2,105) 0.0041 obs*r-squared 0.379355 prob. chi-square (2) 0.8272 model 2 f-statistic 0.196212 prob. f(2,106) 0.8221 obs*r-squared 0.405730 prob. chi-square(2) 0.8164 9 variables that are merely explained away. i = 1, 2, ….10 mid-market firms; and t = 1, 2, ......10 years. b. random effects: this focuses on the relationship with the study sample as a whole, thus the samples are randomly selected, as opposed to using the entire population. the total sample regression function of the random effect can be expressed as: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 2𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (8) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (9) if this is represented with random variables, then 0𝑖𝑖 =  + i, which indicates that the difference occurs randomly, and the expected value of 0𝑖𝑖 is 0, and w𝑖𝑖𝑖𝑖 = ε𝑖𝑖𝑖𝑖 + μ𝑖𝑖𝑖𝑖 , that is, the error term w𝑖𝑖𝑖𝑖 is a sum of two parts: ε𝑖𝑖𝑖𝑖 is the combined time series and cross-section error component, and μ𝑖𝑖𝑖𝑖 is the cross-section or firm-specific error component. c. hausman test: this is the most commonly used method for evaluating fixed and random effects. it shows whether sectional unit differences are fixed or random. it has an asymptotic chi-square distribution with degrees of freedom equal to the number of independent variables. if the variables are significantly correlated, then the fixed effects estimation is consistent and efficient, and should be adopted. on the other hand, if the variables are not significantly correlated, then the random effects estimation is consistent and efficient, and should be adopted. (gujurati, porter, & gunasakar, 2013). serial autocorrelation was also conducted to test the existence of serial relationship between one variable in time t and the same variable in time t 1. the result shows no serial correlation amongst the variables. table 1 breush-godfrey serial correlation lm test model 1 f-statistic 3.181682 prob. f(2,105) 0.0041 obs*r-squared 0.379355 prob. chi-square (2) 0.8272 model 2 f-statistic 0.196212 prob. f(2,106) 0.8221 obs*r-squared 0.405730 prob. chi-square(2) 0.8164 9 variables that are merely explained away. i = 1, 2, ….10 mid-market firms; and t = 1, 2, ......10 years. b. random effects: this focuses on the relationship with the study sample as a whole, thus the samples are randomly selected, as opposed to using the entire population. the total sample regression function of the random effect can be expressed as: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 2𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (8) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (9) if this is represented with random variables, then 0𝑖𝑖 =  + i, which indicates that the difference occurs randomly, and the expected value of 0𝑖𝑖 is 0, and w𝑖𝑖𝑖𝑖 = ε𝑖𝑖𝑖𝑖 + μ𝑖𝑖𝑖𝑖 , that is, the error term w𝑖𝑖𝑖𝑖 is a sum of two parts: ε𝑖𝑖𝑖𝑖 is the combined time series and cross-section error component, and μ𝑖𝑖𝑖𝑖 is the cross-section or firm-specific error component. c. hausman test: this is the most commonly used method for evaluating fixed and random effects. it shows whether sectional unit differences are fixed or random. it has an asymptotic chi-square distribution with degrees of freedom equal to the number of independent variables. if the variables are significantly correlated, then the fixed effects estimation is consistent and efficient, and should be adopted. on the other hand, if the variables are not significantly correlated, then the random effects estimation is consistent and efficient, and should be adopted. (gujurati, porter, & gunasakar, 2013). serial autocorrelation was also conducted to test the existence of serial relationship between one variable in time t and the same variable in time t 1. the result shows no serial correlation amongst the variables. table 1 breush-godfrey serial correlation lm test model 1 f-statistic 3.181682 prob. f(2,105) 0.0041 obs*r-squared 0.379355 prob. chi-square (2) 0.8272 model 2 f-statistic 0.196212 prob. f(2,106) 0.8221 obs*r-squared 0.405730 prob. chi-square(2) 0.8164 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = β1 + β2 d2𝑖𝑖𝑖𝑖 + β3 d3𝑖𝑖𝑖𝑖 + β4 d4𝑖𝑖𝑖𝑖 + β5 d5𝑖𝑖𝑖𝑖 + ⋯β10 d10𝑖𝑖𝑖𝑖 + β11𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 + µ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 8 serial autocorrelation to show if there is a serial relationship between a variable in time t and the same variable in time t l. two models were formulated with ibv as dependent variables in each case: cost of debt (represented by the one most important source of debt capital in the firms’ capital structure) and cost of equity (represented by one most important source of equity in the firms’ capital structure) as independent variables for one model; and pcoc, as an independent variable for the other model. it is the proxy for overall cost of capital for private firms. the first model examined the individual effects of each source of capital for the mid-market firm while the second looked at the global effect. both models were isolated to avoid multicollinearity. the model follows the classical linear regression equation of the following form: 𝐼𝐼𝐼𝐼𝐼𝐼 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃 + 2𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑖𝑖 (4) 𝐼𝐼𝐼𝐼𝐼𝐼 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑖𝑖 (5) where ibv equals incremental business value, calculated for each enterprise as defined in equation 3; pcd equals private cost of debt; pce equals private cost of equity; pcoc equals private cost of capital; 0 equals constant term; 1 2 equal coefficients of predictors; and  equals error term or stochastic variable representing the uncontrolled country-specific factors such as demand volatility, business cycle, labour market, etc. (pourghajan et al., 2012). two panel regression models were specified in line with fixed and random effects estimation procedures, respectively (see results presented in table 2). a. fixed effects: this focuses on whether there are differences by using a fixed intercept for each of the different cross-sectional structures, as the difference may be due to special features of each mid-market firm. it is given thus: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 1 + 2 d2𝑖𝑖 + 3 d3𝑖𝑖 + 4 d4𝑖𝑖 + 5 d5𝑖𝑖 + ⋯ 10 d10𝑖𝑖 + 11𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 12𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝑖𝑖𝑖𝑖 (6) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝑖𝑖𝑖𝑖 (7) because the fixed effects account for both cross-sectional and time-series data, the increased covariance caused by individual-firm differences is eliminated, thereby increasing its estimation-result efficiency. in this case, 1 equals intercept value of firm 1; 2 – 10 represent the differences of the other firms from firm 1, measured by their differences of their coefficients; d2𝑖𝑖 − d10𝑖𝑖 represent 9 dummy variables representing 9 firms; 11 12 equal coefficients of predictors; and  equals error term or stochastic variable representing other 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = β1 + β2 d2𝑖𝑖𝑖𝑖 + β3 d3𝑖𝑖𝑖𝑖 + β4 d4𝑖𝑖𝑖𝑖 + β5 d5𝑖𝑖𝑖𝑖 + ⋯β10 d10𝑖𝑖𝑖𝑖 + β11𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 + µ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 9 variables that are merely explained away. i = 1, 2, ….10 mid-market firms; and t = 1, 2, ......10 years. b. random effects: this focuses on the relationship with the study sample as a whole, thus the samples are randomly selected, as opposed to using the entire population. the total sample regression function of the random effect can be expressed as: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 2𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (8) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (9) if this is represented with random variables, then 0𝑖𝑖 =  + i, which indicates that the difference occurs randomly, and the expected value of 0𝑖𝑖 is 0, and w𝑖𝑖𝑖𝑖 = ε𝑖𝑖𝑖𝑖 + μ𝑖𝑖𝑖𝑖 , that is, the error term w𝑖𝑖𝑖𝑖 is a sum of two parts: ε𝑖𝑖𝑖𝑖 is the combined time series and cross-section error component, and μ𝑖𝑖𝑖𝑖 is the cross-section or firm-specific error component. c. hausman test: this is the most commonly used method for evaluating fixed and random effects. it shows whether sectional unit differences are fixed or random. it has an asymptotic chi-square distribution with degrees of freedom equal to the number of independent variables. if the variables are significantly correlated, then the fixed effects estimation is consistent and efficient, and should be adopted. on the other hand, if the variables are not significantly correlated, then the random effects estimation is consistent and efficient, and should be adopted. (gujurati, porter, & gunasakar, 2013). serial autocorrelation was also conducted to test the existence of serial relationship between one variable in time t and the same variable in time t 1. the result shows no serial correlation amongst the variables. table 1 breush-godfrey serial correlation lm test model 1 f-statistic 3.181682 prob. f(2,105) 0.0041 obs*r-squared 0.379355 prob. chi-square (2) 0.8272 model 2 f-statistic 0.196212 prob. f(2,106) 0.8221 obs*r-squared 0.405730 prob. chi-square(2) 0.8164 9 variables that are merely explained away. i = 1, 2, ….10 mid-market firms; and t = 1, 2, ......10 years. b. random effects: this focuses on the relationship with the study sample as a whole, thus the samples are randomly selected, as opposed to using the entire population. the total sample regression function of the random effect can be expressed as: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 2𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (8) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (9) if this is represented with random variables, then 0𝑖𝑖 =  + i, which indicates that the difference occurs randomly, and the expected value of 0𝑖𝑖 is 0, and w𝑖𝑖𝑖𝑖 = ε𝑖𝑖𝑖𝑖 + μ𝑖𝑖𝑖𝑖 , that is, the error term w𝑖𝑖𝑖𝑖 is a sum of two parts: ε𝑖𝑖𝑖𝑖 is the combined time series and cross-section error component, and μ𝑖𝑖𝑖𝑖 is the cross-section or firm-specific error component. c. hausman test: this is the most commonly used method for evaluating fixed and random effects. it shows whether sectional unit differences are fixed or random. it has an asymptotic chi-square distribution with degrees of freedom equal to the number of independent variables. if the variables are significantly correlated, then the fixed effects estimation is consistent and efficient, and should be adopted. on the other hand, if the variables are not significantly correlated, then the random effects estimation is consistent and efficient, and should be adopted. (gujurati, porter, & gunasakar, 2013). serial autocorrelation was also conducted to test the existence of serial relationship between one variable in time t and the same variable in time t 1. the result shows no serial correlation amongst the variables. table 1 breush-godfrey serial correlation lm test model 1 f-statistic 3.181682 prob. f(2,105) 0.0041 obs*r-squared 0.379355 prob. chi-square (2) 0.8272 model 2 f-statistic 0.196212 prob. f(2,106) 0.8221 obs*r-squared 0.405730 prob. chi-square(2) 0.8164 12 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 error component, and is the cross-section or firm-specific error component. hausman test this is the most commonly used method for evaluating fixed and random effects. it shows whether sectional unit differences are fixed or random. it has an asymptotic chi-square distribution with degrees of freedom equal to the number of independent variables. if the variables are significantly correlated, then the fixed effects estimation is consistent and efficient, and should be adopted. on the other hand, if the variables are not significantly correlated, then the random effects estimation is consistent and efficient, and should be adopted. (gujurati, porter & gunasakar, 2013). serial autocorrelation was also conducted to test the existence of serial relationship between one variable in time t and the same variable in time t 1. the result shows no serial correlation amongst the variables. table 1 breush-godfrey serial correlation lm test model 1 f-statistic 3.181682 prob. f(2,105) 0.0041 obs*r-squared 0.379355 prob. chi-square (2) 0.8272 model 2 f-statistic 0.196212 prob. f(2,106) 0.8221 obs*r-squared 0.405730 prob. chi-square(2) 0.8164 data analysis and results the following tables explain the dynamic relationship between the dependent and independent variables. first is the analysis of hausman test. the null hypothesis states that the fixed effect does not have a better model fit than random effect. here, null hypothesis was rejected, implying that the fixed effect model does have a better model fit than the random effect model. the above result now leaves us with the analysis of fixed effect model alone. 9 variables that are merely explained away. i = 1, 2, ….10 mid-market firms; and t = 1, 2, ......10 years. b. random effects: this focuses on the relationship with the study sample as a whole, thus the samples are randomly selected, as opposed to using the entire population. the total sample regression function of the random effect can be expressed as: 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 2𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (8) 𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖 = 0 + 1𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + w𝑖𝑖𝑖𝑖 (9) if this is represented with random variables, then 0𝑖𝑖 =  + i, which indicates that the difference occurs randomly, and the expected value of 0𝑖𝑖 is 0, and w𝑖𝑖𝑖𝑖 = ε𝑖𝑖𝑖𝑖 + μ𝑖𝑖𝑖𝑖 , that is, the error term w𝑖𝑖𝑖𝑖 is a sum of two parts: ε𝑖𝑖𝑖𝑖 is the combined time series and cross-section error component, and μ𝑖𝑖𝑖𝑖 is the cross-section or firm-specific error component. c. hausman test: this is the most commonly used method for evaluating fixed and random effects. it shows whether sectional unit differences are fixed or random. it has an asymptotic chi-square distribution with degrees of freedom equal to the number of independent variables. if the variables are significantly correlated, then the fixed effects estimation is consistent and efficient, and should be adopted. on the other hand, if the variables are not significantly correlated, then the random effects estimation is consistent and efficient, and should be adopted. (gujurati, porter, & gunasakar, 2013). serial autocorrelation was also conducted to test the existence of serial relationship between one variable in time t and the same variable in time t 1. the result shows no serial correlation amongst the variables. table 1 breush-godfrey serial correlation lm test model 1 f-statistic 3.181682 prob. f(2,105) 0.0041 obs*r-squared 0.379355 prob. chi-square (2) 0.8272 model 2 f-statistic 0.196212 prob. f(2,106) 0.8221 obs*r-squared 0.405730 prob. chi-square(2) 0.8164 13 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 table 2 panel regression results model 1 fixed effects random effects variables coeff. prob. coeff. prob. c -82.7790 0.4058 34.4733 0.0000 pcd -2.4083 0.0071 -0.0047 0.9874 pce -10.7421 0.0120 0.0011 0.9785 r2 0.7705 0.0000 f-test 4.6576 0.0006 dw 2.2809 1.9583 model ii c 33.4716 0.0000 32.6833 0.0000 pcoc -3.0168 0.0354 0.1837 0.7158 r2 0.4165 0.0012 f-test 8.0923 0.1333 dw 2.1202 2.0830 table 3 summary statistics of hausman test test summary for model 1 chi-sq statistic chi-sq. d.f. prob. cross-section random 458.9261 2 0.0000 period random 0.0000 2 1.0000 cross-section and period random 3.9923 2 0.1359 test summary for model ii cross-section random 0.8326 1 0.0015 period random 3.8778 1 0.0489 cross-section and period random 0.0593 1 0.8077 * period test variance is invalid. hausman statistic set to zero. 14 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 second, the analysis above shows goodness-of-fit (r2) of 0.77, meaning that 77 percent of the changes in the dependent variables were caused by the changes in the independent variables. the f-test of 4.7 with a p-value of zero shows that the model is statistically significant. third, the analysis shows a statistically significant but negative relationship between private cost of equity and ibv. fourth, there is a statistically significant but negative relationship between private cost of debt and ibv. a one unit increase in private cost of debt causes 2.4 units decrease in ibv. finally, there is again a statistically significant but negative relationship between pcoc and ibv, as a unit increase in pcoc causes 3.01 units decrease in ibv. in all, the best fit regression line takes the following form: (10) (11) another look at equation 11 above shows that pce with a coefficient of -10.7 exerts a higher degree of elasticity on ibv than pcd with a coefficient of -2.4. put another way, a unit change in private cost of equity will affect ibv much more than a unit change in the private cost of debt. this may be interpreted in two ways: upside reward arising from a drop in pce and downside risk arising from an increase in the same. finally, a comparison between the coefficient of pcoc (-3.01) and the previous two shows that the degree of elasticity of pcoc falls between the coefficients of pce and pcd, thus affirming that pcoc exerts an equilibrating mechanism in the capital structure of mid-market firms as does wacc in the capital structure of large firms. discussions the existence of relationship between various private costs of capital and incremental business value might not have been in doubt ab initio. it was the difference in their spatio-temporal degrees of elasticity that incited the authors. there are two issues here: the first is the relative response of ibv to changes in pcoc. for lack of existing studies in this area, we compared this result with the study of dagogo and imegi (2017) which examined the implications of various costs of capital on 11 first is the analysis of hausman test. the null hypothesis states that the fixed effect does not have a better model fit than random effect. here, null hypothesis was rejected, implying that the fixed effect model does have a better model fit than the random effect model. the above result now leaves us with the analysis of fixed effect model alone. second, the analysis above shows goodness-of-fit (r2) of 0.77, meaning that 77 percent of the changes in the dependent variables were caused by the changes in the independent variables. the f-test of 4.7 with a p-value of zero shows that the model is statistically significant. third, the analysis shows a statistically significant but negative relationship between private cost of equity and ibv. fourth, there is a statistically significant but negative relationship between private cost of debt and ibv. a one unit increase in private cost of debt causes 2.4 units decrease in ibv. finally, there is again a statistically significant but negative relationship between pcoc and ibv, as a unit increase in pcoc causes 3.01 units decrease in ibv. in all, the best fit regression line takes the following form: 𝐼𝐼𝐼𝐼𝐼𝐼 = −82.77 − 10.7𝑃𝑃𝑃𝑃𝑃𝑃 − 2.4𝑃𝑃𝑃𝑃𝑃𝑃 (10) 𝐼𝐼𝐼𝐼𝐼𝐼 = 33.47 − 3.01𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 (11) another look at equation 11 above shows that pce with a coefficient of -10.7 exerts a higher degree of elasticity on ibv than pcd with a coefficient of -2.4. put another way, a unit change in private cost of equity will affect ibv much more than a unit change in the private cost of debt. this may be interpreted in two ways: upside reward arising from a drop in pce and downside risk arising from an increase in the same. finally, a comparison between the coefficient of pcoc (-3.01) and the previous two shows that the degree of elasticity of pcoc falls between the coefficients of pce and pcd, thus affirming that pcoc exerts an equilibrating mechanism in the capital structure of mid-market firms as does wacc (k𝑜𝑜) in the capital structure of large firms. discussions the existence of relationship between various private costs of capital and incremental business value might not have been in doubt ab initio. it was the difference in their spatio-temporal degrees of elasticity that incited the authors. there are two issues here: the first is the relative response of ibv to changes in pcoc. for lack of existing studies in this area, we compared this result with the study of dagogo and imegi (2017) which examined the implications of various costs of capital on the market value and profit of entrepreneurial firms. it was concluded that values of capitalised funds depend more on the vagaries of market forces than on profitability prospects of the firms’ assets. both studies reported similar results in spite of the differences in the formula between pcoc and wacc adopted in dagogo and imegi, (2017). it affirms that with the traditional capital structure theory, a company can improve its incremental business value or market value as it drives 11 first is the analysis of hausman test. the null hypothesis states that the fixed effect does not have a better model fit than random effect. here, null hypothesis was rejected, implying that the fixed effect model does have a better model fit than the random effect model. the above result now leaves us with the analysis of fixed effect model alone. second, the analysis above shows goodness-of-fit (r2) of 0.77, meaning that 77 percent of the changes in the dependent variables were caused by the changes in the independent variables. the f-test of 4.7 with a p-value of zero shows that the model is statistically significant. third, the analysis shows a statistically significant but negative relationship between private cost of equity and ibv. fourth, there is a statistically significant but negative relationship between private cost of debt and ibv. a one unit increase in private cost of debt causes 2.4 units decrease in ibv. finally, there is again a statistically significant but negative relationship between pcoc and ibv, as a unit increase in pcoc causes 3.01 units decrease in ibv. in all, the best fit regression line takes the following form: 𝐼𝐼𝐼𝐼𝐼𝐼 = −82.77 − 10.7𝑃𝑃𝑃𝑃𝑃𝑃 − 2.4𝑃𝑃𝑃𝑃𝑃𝑃 (10) 𝐼𝐼𝐼𝐼𝐼𝐼 = 33.47 − 3.01𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 (11) another look at equation 11 above shows that pce with a coefficient of -10.7 exerts a higher degree of elasticity on ibv than pcd with a coefficient of -2.4. put another way, a unit change in private cost of equity will affect ibv much more than a unit change in the private cost of debt. this may be interpreted in two ways: upside reward arising from a drop in pce and downside risk arising from an increase in the same. finally, a comparison between the coefficient of pcoc (-3.01) and the previous two shows that the degree of elasticity of pcoc falls between the coefficients of pce and pcd, thus affirming that pcoc exerts an equilibrating mechanism in the capital structure of mid-market firms as does wacc (k𝑜𝑜) in the capital structure of large firms. discussions the existence of relationship between various private costs of capital and incremental business value might not have been in doubt ab initio. it was the difference in their spatio-temporal degrees of elasticity that incited the authors. there are two issues here: the first is the relative response of ibv to changes in pcoc. for lack of existing studies in this area, we compared this result with the study of dagogo and imegi (2017) which examined the implications of various costs of capital on the market value and profit of entrepreneurial firms. it was concluded that values of capitalised funds depend more on the vagaries of market forces than on profitability prospects of the firms’ assets. both studies reported similar results in spite of the differences in the formula between pcoc and wacc adopted in dagogo and imegi, (2017). it affirms that with the traditional capital structure theory, a company can improve its incremental business value or market value as it drives 11 first is the analysis of hausman test. the null hypothesis states that the fixed effect does not have a better model fit than random effect. here, null hypothesis was rejected, implying that the fixed effect model does have a better model fit than the random effect model. the above result now leaves us with the analysis of fixed effect model alone. second, the analysis above shows goodness-of-fit (r2) of 0.77, meaning that 77 percent of the changes in the dependent variables were caused by the changes in the independent variables. the f-test of 4.7 with a p-value of zero shows that the model is statistically significant. third, the analysis shows a statistically significant but negative relationship between private cost of equity and ibv. fourth, there is a statistically significant but negative relationship between private cost of debt and ibv. a one unit increase in private cost of debt causes 2.4 units decrease in ibv. finally, there is again a statistically significant but negative relationship between pcoc and ibv, as a unit increase in pcoc causes 3.01 units decrease in ibv. in all, the best fit regression line takes the following form: 𝐼𝐼𝐼𝐼𝐼𝐼 = −82.77 − 10.7𝑃𝑃𝑃𝑃𝑃𝑃 − 2.4𝑃𝑃𝑃𝑃𝑃𝑃 (10) 𝐼𝐼𝐼𝐼𝐼𝐼 = 33.47 − 3.01𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 (11) another look at equation 11 above shows that pce with a coefficient of -10.7 exerts a higher degree of elasticity on ibv than pcd with a coefficient of -2.4. put another way, a unit change in private cost of equity will affect ibv much more than a unit change in the private cost of debt. this may be interpreted in two ways: upside reward arising from a drop in pce and downside risk arising from an increase in the same. finally, a comparison between the coefficient of pcoc (-3.01) and the previous two shows that the degree of elasticity of pcoc falls between the coefficients of pce and pcd, thus affirming that pcoc exerts an equilibrating mechanism in the capital structure of mid-market firms as does wacc (k𝑜𝑜) in the capital structure of large firms. discussions the existence of relationship between various private costs of capital and incremental business value might not have been in doubt ab initio. it was the difference in their spatio-temporal degrees of elasticity that incited the authors. there are two issues here: the first is the relative response of ibv to changes in pcoc. for lack of existing studies in this area, we compared this result with the study of dagogo and imegi (2017) which examined the implications of various costs of capital on the market value and profit of entrepreneurial firms. it was concluded that values of capitalised funds depend more on the vagaries of market forces than on profitability prospects of the firms’ assets. both studies reported similar results in spite of the differences in the formula between pcoc and wacc adopted in dagogo and imegi, (2017). it affirms that with the traditional capital structure theory, a company can improve its incremental business value or market value as it drives 15 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 the market value and profit of entrepreneurial firms. it was concluded that values of capitalised funds depend more on the vagaries of market forces than on profitability prospects of the firms’ assets. both studies reported similar results in spite of the differences in the formula between pcoc and wacc adopted in dagogo and imegi, (2017). it affirms that with the traditional capital structure theory, a company can improve its incremental business value or market value as it drives down its overall cost of capital (ko), and asserted that the same implication holds at every stage of the enterprise. the second issue is to evaluate the fact that a unit change in private cost of equity will affect ibv much more than a unit change in the private cost of debt would. this may be interpreted in two ways: upside reward arises in an increasing proportion to marginal drop in pce and downside risk arises in an increasing proportion to marginal increase in same. this conveys a signal for risk-return axiom. it illustrates that private equity capital is more expensive than private debt capital and must therefore attract higher returns, and that the opportunity cost of private equity capital must be sufficiently priced to compensate the risk of bankruptcy of mid-market firms. this draws us back to appreciate the pecking order theory of myer and majluf (1984), which is in tandem with the reasons for private equity financing of start-up and earlystage growth firms, and to escape from fixed interest charges and financial risks associated with early-stage debt financing. however, mid-market firms cannot be so declassified as to lack the capacity to leverage on less costly fixed interest securities. higher impact of the cost of private equity capital alludes first to a greater proportion of the same in mid-market firms’ capital structure in spite of its higher cost, and second to the existence of market inefficiency caused by asymmetric information, oligopolistic market structure, higher transaction cost, and low liquidity. the lesser impact of the cost of private debt simply alludes to lesser utilisation of mezzanine, factoring or leveraged buyout in preference for bank, government or multilateral organisations’ funding. unfortunately, this study exposes the fact that nigerian mid-market enterprises still exhibit the nuances of small enterprise financing, 16 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 and remains a far cry from the standards of mid-market enterprises in developed economies but basking under the universality of the acronym, sme, without seeking means of being strengthened for advancement deserving of their class. the above discussions underscore the logic in support of financial leverage and value creation. simply put, an entrepreneur creates more value with debt financing on two accounts: first, debt finance is traditionally cheaper than equity finance. second, it offers opportunity to increase operating profit by expanding business operations. the only precaution is the risk of insolvency, which serves as a check to excessive utilisation of debt finance. put together, we expected a result that will reflect a greater or balanced weight of debt-to-equity. that was not the case as shown above. conclusion firstly, reduction in private cost of capital (debt or equity) causes accretion in ibv. secondly, both pcoc (a pragmatic description of the overall cost of capital for emerging or mid-market firms) and wacc have similar effects on any value world under study. thirdly, there is greater weight of private equity than private debt in the capital structure of mid-market firms in nigeria. this is the case in spite of lower cost of debt capital from banks, multinational institutions and government agencies. it suggests the following: difficulty in accessing low cost capital from debt sources, oligopolistic private capital securities market, high transaction cost, crowding out of private sector capital allocation by large enterprises, and asymmetric information. fourthly, mid-market firms have not distinguished themselves as belonging to an entirely separate investment asset class but still seek funds in the same category as small enterprises in the ‘sme’ construct. this no doubt has also rubbed on their ability to create incremental business value, which explains the reason some mid-market firms have remained in this category for unacceptably longer periods. in addition, we acknowledged that this work is not without some limitations. firstly, we restricted ourselves to incremental value of fund providers in the midst of several other claimants of the firm’s value 17 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 created. secondly, we considered that a comparative analysis of the effects of public and private costs of capital on incremental business value could reveal further insights about the alternative models, and demonstrate greater reliance of the mid-market finance theory. however, this was not our objective. finally, dynamic econometric models could be used to account for the occurrence of lag in both dependent and independent variables. each of these limitations is a valid reason for advancing further research in this area. acknowledgment this research received no specific grant from any funding agency. references caselli, s., & negri, g. (2018). private equity and venture capital in europe: market, techniques, and deals. academic press. central securities clearing system (cscs). (2018). listed firms on the nasd. https://www.cscsnigeriaplc.com/home/dailypricelist copeland, k., koller, t., & murrin, a. (1996). valuation: measuring and managing the value of companies. john wiley & sons. dagogo, d. w. (2006). venture capital financing and enterprise development in nigeria (unpublished doctoral dissertation). rivers state university of science & technology, port harcourt. dagogo, d. w., & ohaka, j., (2015). critical evaluation of small business financial strategies in nigeria: a panel study. international journal of finance and accounting, 6(2) 37–45. dagogo, d. w., & imegi, j. c. (2017). market value maximisation, after-tax profit and the cost of capital of entrepreneurial firms. asian journal of economics, business and accounting 3(2), 1– 9. drucker, p. (1998). peter drucker takes the long view. fortune magazine. (sept ed.). garcia-castro, r., & aguilera, r. v. (2014). incremental value creation and appropriation in a world with multiple stakeholders. 18 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 http://www.wileyonlinelibrary.com john wiley & sons. doi: 10.1002/smj.2241 gujurati, d., porter, c., & gunasakar, s. (2013). basic econometrics. tata mcgraw-hill education. iprc-oecd (2011). enterprise development and economic transformation: creating the enabling environment. chinedac study group. addis ababa. kanaballi, b., & kashinath, r. b. (2015). assessment of economic value creation of select indian public sector firms: a comparative study. clear international journal of research in commerce & management, 6(1), 54–61. leach, j. c., & melicher, r. w. (2012). entrepreneurial finance. south-western cengage learning leleux, m., swaay, h., & megally, e. (2015). private equity 4.0: reinventing value creation. chichester: john wiley & sons. mclaney, e. (2009). business finance: theory & practice. pearson education. modigliani, f., & miller, m. h. (1958). the cost of capital, corporation finance and the theory of investment. american economic review, 48(3), 261–297. modigliani, f., & miller, m. h. (1963). corporation income taxes and the cost of capital: a correction. american economic reviews, 53, 433–443. myers, s. c., & majluf, n. s., (1984). corporate financing and investment decisions when firms have information that investors do not have. journal of financial economics, 13(2), 187–221. myint, h. (1985). organisational dualism and economic development. asian development review 3(1) 24–42. pouraghajan, a., tabari, n. a., ramezani, a., mansourinia, e., enamgholipour, m., & majd, p. (2012). relationship between cost of capital and accounting criteria of corporate performance evaluation: evidence from tehran stock exchange. world applied sciences journal, 20(5), 666–673. small and medium enterprises development agency of nigeria (smedan). (2012). survey report on msme in nigeria. http://www.smedan.gov.ng/images/pdf/2013-msme-surveysummary-report.pdf slee, r. t. (2011). private capital markets valuation, capitalisation, and transfer of private business interests. john wiley & sons. 19 the international journal of banking and finance, vol. 16, number 1 (january) 2021, pp: 1–20 van binsbergen, j. h., graham, j. r., & yang, j., (2010). the cost of debt. the journal of finance, 65(6), 2089–2136. vernimmen, p., quiry, p., dallocchio, m., lefur, y., & salvi, a. (2014). corporate finance: theory & practice. john wiley & sons. appendix figure 1. middle market expected returns. 15 appendix 40% 41.80% 35% 37.50% 30% 30% 25% 20% 21% 15% 10% 8% 5% 5% 0% bank abl & s mezzanine pe vc &ba factoring lbo mbo/bi key: abl = asset-bank loans & securitisation; lbo = leveraged buyout; pe = private equity; mbo/bi = management buyout/buy-in vc & ba = venture capital and business angels source: pepperdine private capital market line. slee (2011) figure 1: middle market expected returns. 95 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 how to cite this article: shittu, i., & abdulkadir, h. (2022). the moderating effect of cost per loan asset ratio on the relationship between credit risk and financial performance of listed deposit money banks in nigeria. international journal of banking and finance, 18(1), 95115. https://doi. org/10.32890/ ijbf2023.18.1.5 the moderating effect of cost per loan asset ratio on the relationship between credit risk and financial performance of listed deposit money banks in nigeria ¹isah shittu & ²hannafi abdulkadir 1department of accounting ahmadu bello university, zaria, nigeria 2jaiz bank nigeria plc (business development) 2corresponding author: abdulkadir.hannafi@jaizbankplc.com received: 22/10/2021 revised: 31/3/2022 accepted: 7/4/2022 published: 5/1/2023 abstract in recent years, banks in nigeria have experienced a significant increase in delinquent loan portfolios, which has contributed immensely to the financial difficulties in this sector. due to the trust of depositors, banks should be responsible for the efficient utilization of resources to achieve cost efficiency, which in turn contributes to raising income. this paper seeks to investigate the moderating role of the cost per loan asset ratio (clar) on the relationship between credit risk and return on asset (roa) of nigerian deposit money banks (dmbs). this study employs panel data analysis followed by the use of gls regression models to examine the relationship in question. the population consists of all fifteen (15) listed dmbs in the nigerian https://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 96 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 stock market as at december 31st, 2018, while the adjusted population was eleven (11). the results revealed a significant positive moderating relationship between the non-performing loan ratio (nplr) and capital adequacy ratio (car), while the loan loss provision ratio (llpr) and asset quality ratio (aqr) were negative, but statistically significant. moreover, the cost per loan asset ratio was found to have an inverse moderating effect on the relationship between the loan and advance ratio (ladr) and the bank’s probability, even though it was not statistically significant. based on the research findings, the study recommends that policymakers focus on capital regulation as measured by the capital adequacy ratio, risk level, liquidity, and operational cost efficiency. in addition, banks should have effective and efficient strategies to manage credit risks, which might help to enhance their performance. keywords: credit risk, return on assets, loans and advances, nonperforming loans, loan loss provision, cost of loan asset quality, asset ratio. jel classification: g21. introduction the constant rise in lending rates and other borrowing costs in nigeria have consistently raised concerns among investors, policymakers, and other stakeholders in the economy. this is the result of the rigidity in the monetary and structural policies in the economic system, which in turn has led to an increase in costs associated with funds sourcing by nigerian deposit money banks (dmbs). these two factors have contributed to high lending rates. in addition, increases in overhead costs, contributions of sinking funds by banks to the asset management corporation of nigeria (amcon) and premium payments to the nigerian deposit insurance corporation (ndic) have exerted extra demands on the cost of funds and earnings for banks. furthermore, owners and bank managers are seriously concerned with the cost of funds since they can earn a higher return if these costs are minimized. in addition, customers are also concerned with costs since they could end up paying a higher amount for the use of the bank’s products and services. similarly, governments and other bank regulators are 97 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 concerned about costs because if banks are unable to reduce their costs, it can have an impact on their profits and liquidity (ware & wambua, 2014). banking sector profitability is an important indicator of a stable financial sector. the high cost of funds can affect the level of bank profit in unique ways. bank financial profitability has received significant attention from scholars, bank management, supervisors, financial markets, and other stakeholders in different sectors of the economy. the primary objective of every entity in the economy is to make profits and minimize costs at a given level of risk. good financial performance shows management effort in making use of the bank’s resources. there is ample evidence that a profitable bank can withstand negative shocks, as well as contribute to stability in the sector. in addition, the earnings of dmbs in nigeria are affected by various factors, which include internal and external forces (alhadab & alsahawneh, 2016). however, the lack of proper monitoring and evaluation of loan portfolios can lead to the accumulation of npls, which have become a major threat to the nigerian banking sector. npls are caused by the lack of liquidity and a bank’s loan expansion. moreover, the npl could be seen as one of the most serious causes of banking failure in the nigerian banking industry (sheefeni, 2015). many steps have been taken to curtail the negative consequences of such risks by local and international organizations. these organizations have developed several procedures and policies that should be followed to mitigate the above mentioned risks. one of the provisions of the policies is the loan loss provision. furthermore, loan loss provision (llp) plays an important role in strengthening the financial position of banks (olabamiji & michael, 2018). in addition, banks play a vital role in the channeling of funds from surplus units (investors) and extending them to the deficit units (borrowers) within the economy. loans and advances represent the major income of the banking industry, and it can lead to bank failure if the loan portfolio is not properly managed and utilized. many bank failures that occurred in nigeria were the result of poor loans and advances, which had led to the accumulation of npls. to protect the bank from future uncertainty, the cbn (central bank of nigeria) and other regulators require all banks to maintain a certain 98 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 percentage of capital adequacy ratio. in addition, a good capital adequacy ratio may protect banks from any unforeseen circumstances that may pose a threat to their existence and survival (balango & rao, 2017). asset quality is regarded as an important indicator of the banking industry’s financial dependability and health. many scholars have considered it one of the key causes of severe banking distress. loans form an important part of a bank’s assets, as delays in repaying it can cause serious problems for the bank, investors (depositors), and the economy in general (iheanyi & sotonye, 2017). the cbn released its financial stability report concerning commercial banks in the country, which revealed that three banks failed a stress test conducted on financial institutions in the country. the cbn reports that loans and advances given by the bank are not performing and cannot be recovered. this has led to an accumulation of bad debts in customer accounts. according to the komolafe (2017), as of the first half of 2017, 14 banks had recorded a total of n177.3 billion in delinquent facilities. furthermore, investors and other stakeholders have started to lose confidence due to the failure of some banks to manage their investments within a given level of risk. the continuous diminishing of a bank’s asset quality over the years was due to the negligence of some bank officials in managing and handling their portfolios, which reduced the confidence of depositors, shareholders, and foreign investors in the nigerian banking sector (olabamiji & michael, 2018). the most recent bank failure in nigeria was that of diamond bank plc. this was due to the economic meltdown, huge exposure to the oil sector, breach of corporate governance by the bank management, and board disharmony, among others. this has led to the accumulation of huge non-performing loans (npls), which has forced the bank to sell its foreign subsidiaries. according to its financial statement, the bank’s management acted contrary to its prudential guidelines. as a result, the bank has a 12.6 percent npl rate, compared to the industry average of 5 percent, and high impairment charges. similarly, these problems have caused the bank to lose value in its share price, falling from n7 to n1.37. despite some policies and measures undertaken by regulators like the cbn and the ndic to curtail the implications of default risk in the sector, there is still a rise in the number of npls. therefore, there is a need for additional research in the area to re-examine and identify 99 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 the causes of such default risks. there have been several studies conducted on credit risk and the performance of banks, but there are no studies conducted on the moderating aspect in either developing or least developed countries. therefore, the aim of the present study is to contribute to the existing literature by examining the moderating effect of cost per loan asset ratio on the association between default risk and the roa of listed dmbs in nigeria, as well as to identify the relationship between npls and the roa. the study spanned the period from 2010 to 2018, and looked at loan loss provisions (llps) on the roa, loans and advances (lad) and the roa, capital adequacy ratio (car) and the roa, and the impact of asset quality (aq) on the roa of nigerian listed dmbs. the rest of the paper is organized as follows: section 2 reviews the literature and provides testable hypotheses. section 3 introduces the research methodology. section 4 presents the empirical results and reports on the robustness of testing, and section 5 concludes the study with recommendations literature review and hypotheses development in this section, the nexus between credit risk and profitability is explored by reviewing the available body of literature. more specifically, the focus was on previous studies that covered the credit risk of banks and the profitability of both developed and developing countries. lastly, the mechanisms and hypotheses of development will be discussed. literature on non-performing loans and bank performance bhusare et al. (2018) assessed the effect of default risk on the profitability of commercial banks in india for a period of eight years, from 2009 to 2016.theirstudy adopted a multiple regression model in analyzing the data. the results showed that npls measured by gross nplas to gross loans and advances hada positive relationship with the roi, but were statistically insignificant. there was the possibility of high sampling error in the cluster sampling used in this research. similarly, kishori and sheeba (2017) examined the effect of credit risks on the return of capital employed by indian banks over a period of 20 years, starting from 1997 to 2016. a regression model was 100 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 used in the analysis of the data. findings showed that npls had a significant impact onthe roce. the roce used in this research did not account for the depreciation and amortization of the capital employed. regression analysis was used in another study conducted by singh and sharma (2018) on the relationship between credit risk and the roa of public sector banks in india over a six-year period (2011–2016).the result indicated that the nplr and roa had a significant negative relationship. the random sampling technique used in this study might be subjected to a misrepresentation or inaccurate reflection of the population. based on the literature and theory, the following hypotheses were developed in the null form: h01 : cost per loan asset ratio (clar) has no significant moderating effect on the association between non-performing loans and the roa of deposit money banks in nigeria. literature on loan loss provision and bank performance kutum (2017) investigated the effect of credit risk on the roa of palestine banks listed on the stock exchange for six years (2010– 2015). the findings showed that llps had a positive association with the roa, but it was not statistically significant. the technique used in this study only models relationship that is linear and is sensitive to outliers. in their study, mwaurah et al. (2017) examined the implication of default risk on the stock returns of listed nairobi commercial banks from 2006 to 2015. the study was analyzed using the gls regression model. regression output showed that loan loss provisions (llps) had significant positive relationships with the stock returns. in contrast, hamza (2017) analyzed the impact of credit risk on the returns of pakistan’s commercial banks for a six-year period (2006–2011). a pooled ols was used in the analysis. the findings indicated that the loan loss provision ratio (llpr) had a negative impact on bank returns, which was measured using the return on equity, but it was statistically significant. however, the risk of using pooled ols in this study is that there will be some form of endogeneity. based on the above studies, the following hypotheses were developed in the null form: 101 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 h02 : cost per loan asset ratio (clar) has no significant moderating effect on the relationship between loan loss provision and the roa of deposit money banks in nigeria. literature on loan advances and bank performance kayode et al. (2015) evaluated the relationship between default risk and the roa in nigerian commercial banks for the duration of fourteen (14) years from 2000 to 2013. the findings revealed that the lad significantly affected the profitability of banks in nigeria. the period covered in this study had been taken over by events, and the study now is not considered reliable because so many things had happened in the banking industry in nigeria from 2013 to date. for instance, the cbn 2016 new prudential guideline pointed out that the methodology used in the study was flawed due to design and data collection problems, distortions of measurement errors, and the method was unpredictable. in another study, muriithi et al. (2016) investigated the effect of loan risk on the roa of banks in kenya for ten (10) years, from 2005 to 2014. a generalized method of moments (gmm) was employed as a technique for data analysis. results indicated that the ladr had significant implications on the banks’ roa in kenya. this indicates that as the ladr is increased, profitability is decreasing. samuel (2015) investigated the effects of bank profitability on the roa in nigeria over a five-year period (2008–2015) and found that loans and advances had no significant effect on the profitability of banks in nigeria. the technique used in the above study was sensitive to outliers. similarly, test statistics might be unreliable if the data were not normally distributed. based on the above findings; the following hypotheses were developed in the null form: h03 : cost per loan asset ratio (clar) has no significant moderating effect on the association between loans and advances and the roa of deposit money banks in nigeria. literature on capital adequacy ratio and bank performance isanzu (2017) conducted research on the relationship between credit risk and the roa of chinese banks for seven years from 2008 to 2014. the results were analyzed using a balanced panel data regression model. they found that the car significantly affected the profitability of banks in china. there is distortion of measurement 102 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 errors and selectivity problems in the method used in their study. mendoza and rivera (2017) evaluated the consequences of default risks and the car on returns of philippine rural banks covering the period of five years from 2009 to 2013. the data was subjected to arellano-bond regression. dynamic results revealed that the car had significant implications for profitability. while bhattarai (2017) measured the impact of credit default risk on the roa of nepalese banks for the duration of 8 years from 2009 to 2016.the data were analyzed using the pooled ols method. according to the findings of the analysis, the car had a significant negative correlation with bank profitability. the study used a convenience sampling method that considers samples to represent the entire populations. this may lead to over or under representation of some groups in the samples. based on the reviewed literature, the following hypotheses were developed in the null form: h04 : cost per loan asset ratio (clar) has no significant moderating effect on the relationship between capital adequacy ratio and the roa of deposit money banks in nigeria. asset quality literature and bank performance cheruiyot (2016) examined the impact of asset quality on the roa of banks in kenya for the period of five years (2011–2015). the study adopted a multiple regression model and concluded that asset quality had a positive effect on the roa of banks in kenya. according to the research findings, as the ratio of default rates was lower, asset quality would also be lower, and vice versa. the capital asset pricing model used in this research is a market-based theory, and it is applicable in the market. ogbulu and eze (2016) used a multiple regression model to examine the relationship between asset quality management and the performance of nigerian dmbs over a 24-yearperiod (1990– 2013). they discovered that asset quality had a significant positive correlation with performance. furthermore, the study did not provide a theory to underpin the variables in the study. in contrast, iheanyi et al. (2017) examined the implications of good credit quality on the profitability of nigerian banks. this research adopted the camel model, and it spanned a period of 19 years, from 1996 to 2014. the researchers found that 103 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 credit quality in nigerian banks had negative implications for their profit. the technique used in this study was very sensitive to outliers. similarly, the test statistics of the study might be unreliable because the data was not normally distributed. based on the results and analysis of the above studies, the following hypotheses were developed in the null form: h05 : cost per loan asset ratio (clar) has no significant moderating effect on the association between asset quality and the roa of deposit money banks in nigeria. overall, in their analyses, the above studies did not consider the moderating role of cost per asset ratio. the present study has contributed to the existing literature by examining the moderating role of cost per loan asset ratio on the association between credit risk and the financial performance of deposit money banks in nigeria. it has applied the gls regression to correct potential endogeneity, heteroscedasticity, cross-section dependence, and autocorrelation problems. in addition, the literature reviewed had used asset quality as a credit risk management yardstick, whereas the present research used it as a variable to determine credit risk level. hence, this study has also contributed to the existing literature by examining the role of asset quality in determining the credit risk level on profitability and modern portfolio theory, which is rarely the focus in the literature. conceptual framework the conceptual framework has been developed from the review of literature and the underpinning theories are as presented in the above diagram (see figure 1). the diagram represents the various variables used in the study; it shows the interrelationships between the independent variables, such as the non-performing loans, loan loss provision, loan and advances, capital adequacy, asset quality, and the dependent variable roa. similarly, the roa was used as a dependent variable as it could measure the ability of the bank’s management to generate income by utilizing company assets at their disposal. the five independent variables directly affected the financial performance of the banks. furthermore, cost per loan asset was used as a moderating variable, which was expected to influence the relationship between the two variables. 104 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 figure 1 framework of the study research methodology in this section, the research design, sources and methods of data collection, population and sample size, techniques of analysis, model specifications, and variables of measurement used in carrying out the present study will be discussed. moreover, the research employed a correlation research design due to the effectiveness of the design in establishing a relationship among variables. the population for this research comprised 15 (fifteen) dmbs quoted on the nigerian stock exchange, as at december 31st, 2018. a filter was introduced in the study to include banks according to the following criteria: any banks with available annual reports and accounts over the research period; banks that were not acquired during the period of the study and have branches in all the state capitals; and lastly, any banks that are quoted on the nigerian stock exchange (nse). based on the above criteria, eleven (11) banks were selected from the fifteen (15) banks quoted on the nigerian stock exchange market during the nine-year period starting from 2010to 2018. over the period covered by the research, secondary sources of data from published annual reports of eleven (11) listed dmbs under the nse were employed. 5 cheruiyot (2016) examined the impact of asset quality on the roa of banks in kenya for the period of five years (2011–2015). the study adopted a multiple regression model and concluded that asset quality had a positive effect on the roa of banks in kenya. according to the research findings, as the ratio of default rates was lower, asset quality would also be lower, and vice versa. the capital asset pricing model used in this research is a market-based theory, and it is applicable in the market. ogbulu and eze (2016) used a multiple regression model to examine the relationship between asset quality management and the performance of nigerian dmbs over a 24-yearperiod (1990–2013). they discovered that asset quality had a significant positive correlation with performance. furthermore, the study did not provide a theory to underpin the variables in the study. in contrast, iheanyi et al. (2017) examined the implications of good credit quality on the profitability of nigerian banks. this research adopted the camel model, and it spanned a period of 19 years, from 1996 to 2014. the researchers found that credit quality in nigerian banks had negative implications for their profit. the technique used in this study was very sensitive to outliers. similarly, the test statistics of the study might be unreliable because the data was not normally distributed. based on the results and analysis of the above studies, the following hypotheses were developed in the null form: h05: cost per loan asset ratio (clar) has no significant moderating effect on the association between asset quality and the roa of deposit money banks in nigeria. overall, in their analyses, the above studies did not consider the moderating role of cost per asset ratio. the present study has contributed to the existing literature by examining the moderating role of cost per loan asset ratio on the association between credit risk and the financial performance of deposit money banks in nigeria. it has applied the gls regression to correct potential endogeneity, heteroscedasticity, cross-section dependence, and autocorrelation problems. in addition, the literature reviewed had used asset quality as a credit risk management yardstick, whereas the present research used it as a variable to determine credit risk level. hence, this study has also contributed to the existing literature by examining the role of asset quality in determining the credit risk level on profitability and modern portfolio theory, which is rarely the focus in the literature. moderating variable independent variable dependent variable non-performing loans (npls) loan loss provisions (llps) loan and advances ratio (lad) capital adequacy ratio (car) asset quality ratio (aqr) return on assets (roa) cost per loan asset ratio (cla) 105 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 table 1 presents the variables used in this study. the measurement for each variable is also provided. in this study, profitability (profit after tax divided by total assets) was applied as a dependent variable. independent variables consisted of non-performing loans, loan loss provision, loans and advances, capital adequacy, and asset quality. the moderating variables were cost per loan asset ratio and are in line with previous studies (poudel, 2012; bhattarai, 2016). table 1 variables and their measurement variables types of variables variable measurement sources roa dependent profit after tax to total assets kargi (2011) non-performing loans independent npls to loan and advances alshatti (2015) loan loss provisions independent llp to loan & advances alhadabetal (2016) loan and advances independent lad to total deposit bizuayehu (2015) capital adequacy independent shareholders fund to total assets abiola (2 014) assets quality independent net nonperforming assets to total assets abata etal. (2014) cost per loan assets moderator total operating cost to total loan and advance poudel (2012) note. based on the conceptual and empirical literature reviewed this study has employed the panel data method to examine the moderating effect of cost per loan asset ratio on the association between credit risk and the roa. this method allows the elimination of unobservable heterogeneity for each observation in the sample of a study. panel data analysis was carried out using the generalized least square (gls) model. the studyapplied gls regression to correct for potential endogeneity, heteroscedasticity, cross-section dependence, and autocorrelation problems. the regression models for the panel data must comply with some assumptions; either they must be linear, unbiased, have a lag structure, or contain important results prior to 106 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 testing the model. to examine the moderating effect of cost per loan asset ratio, this research used the following model: roait denotes the profitability ratio of bank i at time t, calculated as net profit after tax divided by total assets. nplrit represents the non-performing loan ratio of bank i at time t and is calculated as nonperforming loan to loan and advances. llprit is the loan loss provision ratio of bank i at time t and is calculated as loan loss provisions to loan and advances. ladritis the loan and advances of a bank at time t, which is measured as loan and advances to total deposits. carit represents the capital adequacy ratio of a bank at time t, measured as shareholders’ funds to total assets. aqri, t is the asset quality ratio of bank i at time t, which is measured as net non-performing assets to total assets, and clarit is the cost associated with loans and advances of bank i at time t, measured as total operating cost to total loans and advances. furthermore, the main advantage of using this model is that it allows for the relationship between non-performing loans, loan loss provisions, loans and advances, capital adequacy, asset quality, and financial performance. this study uses the gls estimator proposed by previous research studies to investigate the effects of non-performing loans, loan loss provisions, loan and advances, capital adequacy, asset quality, and cost per asset ratio on financial performance. results and discussion results from the various tests conducted for the sample of banks from the period 2010 to 2018 were presented, analyzed, and interpreted. the overall aim is to examine whether cost per loan asset ratio (clar) has a significant moderating effect on the relationship between nonperforming loans, loan loss provision, loan and advances, capital adequacy ratio, asset quality and return on assets of deposit money banks in nigeria. the section also presented analyses of descriptive statistics of the selected variables, the correlation matrix, robustness test, hausman specification test, generalized least squares (gls) regression results, test of hypotheses, and discussion of findings. 7 capital adequacy, and asset quality. the moderating variables were cost per loan asset ratio and are in line with previous studies (poudel, 2012; bhattarai, 2016). table 1 variables and their measurement variables types of variables variable measurement sources roa dependent profit after tax to total assets kargi (2011) non-performing loans independent npls to loan and advances alshatti (2015) loan loss provisions independent llp to loan& advances alhadabetal (2016) loan and advances independent lad to total deposit bizuayehu (2015) capital adequacy independent shareholders fund to total assets abiola (2 014) assets quality independent net nonperforming assets to total assets abata etal. (2014) cost per loan assets moderator total operating cost to total loan and advance poudel (2012) note. based on the conceptual and empirical literature reviewed this study has employed the panel data method to examine the moderating effect of cost per loan asset ratio on the association between credit risk and the roa. this method allows the elimination of unobservable heterogeneity for each observation in the sample of a study. panel data analysis was carried out using the generalized least square (gls) model. the studyapplied gls regression to correct for potential endogeneity, heteroscedasticity, cross-section dependence, and autocorrelation problems. the regression models for the panel data must comply with some assumptions; either they must be linear, unbiased, have a lag structure, or contain important results prior to testing the model. to examine the moderating effect of cost per loan asset ratio, this research used the following model: roait=β0+β1nplrit+β2llprit+β3ladrit+β4carit +β5aqrit+β6clarit+clar (β7nplrit+β8llprit+β9ladrit+β10carit+β11aqrit+ℇit) roait denotes the profitability ratio of bank i at time t, calculated as net profit after tax divided by total assets. nplrit represents the non-performing loan ratio of bank i at time t and is calculated as non-performing loan to loan and advances. llprit is the loan loss provision ratio of bank i at time t and is calculated as loan loss provisions to loan and advances. ladritis the loan and advances of a bank at time t, which is measured as loan and advances to total deposits. carit represents the capital adequacy ratio of a bank at time t, measured as shareholders' funds to total assets. aqri, t is the asset quality ratio of bank i at time t, which is measured as net non-performing assets to total assets, and clarit is the cost associated with loans and advances of bank i at time t, measured as total operating cost to total loans and advances. furthermore, the main advantage of using this model is that it allows for the relationship between non-performing loans, loan loss provisions, loans and advances, capital adequacy, asset quality, and financial performance. this study uses the gls estimator proposed by previous research studies to investigate the effects of non-performing loans, loan loss provisions, loan and advances, capital adequacy, asset quality, and cost per asset ratio on financial performance. results and discussion 107 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 descriptive statistics the descriptive statistics below show the nature of the data used. table 2 presents the descriptive statistics such as the mean, standard deviation, and variance, minimum and maximum for each variable, which are return on assets, non-performing loans, loan loss provisions, loan and advances, capital adequacy ratio, asset quality, and cost per loan asset ratio of each dependent and independent variable. table 2 descriptive statistics mean std. deviation minimum maximum observations roa 0.021 0.053 -0.099 0.482 99 nplr 0.111 0.169 0.004 1.044 99 llpr 0.067 0.089 0.009 0.539 99 ladr 0.654 0.196 0.036 1.064 99 car 0.118 0.213 -1.547 0.281 99 aqr 0.045 0.064 0.001 0.381 99 clar 0.141 0.275 0.035 99 note. the table describes the nature of the variables in terms of the vv mean, std deviation, minimum values, maximum values and number of observations. the results of the descriptive statistics from table 2 show the mean value of the roa, which is 2.1 percent. this means that in the period between2010 and 2018, the total assets of sample dmbs generated a 2.1 percent return on average. while it deviated 5.3percent, this indicated variation in theroa among the sample banks. table 2 also shows that the npls have a mean of 11.1percent, while their deviation stood at 16.9 percent.the npls had a minimum value of 0.4 percent while their maximum was104.4 percent. this indicates thehuge npls among the sample of banks during the period under investigation. the spread between the maximum and minimum value was 104 percent. it can be seen from table 2 that the llps on average were 6.7 percent, while their minimum and maximum stood at 0.9 percent and 53.9 percent, respectively; this implies that the dmbs were slightly above the required 8 percent provision by the basel accord. the results showed that the lad has a 65.4 percent average and it deviated by 19.6 percent. in addition, it had a maximum value of 106.4 percent, while the minimum value was 3.6 percent. this means that the dmbs 108 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 concentrated more attention on giving loan facilities that have more risk. table 2 also shows that the capital adequacy ratio had an average of 11.8 percent, with a standard deviation of 21.3 percent. similarly, the minimum stood at -154.7 percent, while the maximum value was 28.1 percent. the spread between the maximum and minimum value was -126.6 percent. however, the maximum value of the car was more than the 15 percent that was regulated by the cbn. this also means that the dmbs’ investments were mainly derived from long-term debts or liabilities. asset quality was also maintained at an average of 4.5 percent, with a minimum value of 0.1 percent, while the maximum value stood at 38.1 percent. this shows that the dmbs concentrated more on financing businesses which was riskier than other investments. regardingthe la, the average mean was 14.1 percent, while the minimum value was 3.5 percent and the maximum value was 273.1percent. this indicates a huge cost among thesample of dmbs. correlation matrix for dependent and independent variables table 3 represents the correlation matrix, which shows the relationship between each two pairs of thevariables in the model. the correlation matrix is a preliminary test to check for the possibility of multicollinearity. however, in this study, a further test of multicollinearity was conducted using the variance inflation factor (vif) and tolerance value (tv). table 3 correlation matrix roa npls llp lad car aq cla roa 1.0000 npl 0.056 1.0000 llp -0.167 0.155 1.0000 lad 0.157 -0.237 -0.151 1.0000 car 0.182 -0.217 0.056 0.454 1.0000 aq 0.126 0.877 0.154 -0.026 0.074 1.0000 cla -0.261 0.283 0.087 -0.431 -0.644 -0.014 1.0000 note. table 3 shows the correlation relationship between the variables signifying absence of multicolinearity, because all correlation between all the variables is below 70%. 109 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 table 3 shows the correlation coefficients between variables. the pearson correlation coefficient indicates a positive correlation between the roa and the npl. moreover, the npl has a correlation coefficient value of 0.056. the correlation coefficient of 0.056 (npl/ la) indicates an increase in the roa with every rise in the npl. this is unusual because, according to the theory, the npl and the roa have an indirect relationship. moreover, the positive effect of npls on returns indicated that npls were increasing proportionately to profitability, even though there was credit default. the correlation matrix above shows that the coefficient of return on assets (roa) and loan loss provisions (llps) is -0.167. this implies that loan provisions (llps) have a negative correlation with financial performance as measured by the roa. the correlation coefficient of -0.167 (llps/ la) indicated a decrease in the roa with every rise in npls. the correlation output from the table above shows a positive relationship between loans and advances (lad) as credit risk indicators and profitability. the correlation coefficient of 0.157 indicated a rise in profitability with every increase in the risk factors of loans and advances. there was a direct correlation between the car and the return on assets. the car ratio had a correlation coefficient of 0.182, which shows a positive correlation between the car and the roa. this implies that as the capital adequacy ratio is increased, roa will increase. the roa had a positive correlation with asset quality in table 3. it had a correlation coefficient of 0.126. the correlation output shows a positive relationship between good asset quality and the roa, with the coefficient of correlation being 0.126. in terms of the clar and the roa, they were seen as inversely related. the correlation coefficient of -0.261 means that as the clar rises, the roa will decrease. from table 4, it can be observed that the r2 is 0.22, which means that 22 percent of the variation in the financial performance of deposit money banks in nigeria is explained jointly by the independent variables captured in the model of the study. the f-statistic is 114.04, which is greater than the critical value at the one percent level of significance. therefore, the variables (credit risk components) are jointly significant in explaining the variations in return on assets in the random effects specification. the wild-performance, which is significant at 1percent, indicates that the model is fit. similarly, a positive moderating effect was found on the association between nonperforming loans and the roa, with a coefficient of 2.068 percent 110 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 and a statistical significance of 1 percent. according to the analysis, one naira rise in the npl could increase profitability by 2.068 percent. this states that banks with a high npl can obtain a high profit. this is in line with the predictions of previous theory, which stated that the higher the risk, the higher the return. the positive sign also indicates that riskier banks are more cost-efficient and profitable. this result does not contradict the preliminary hypothesis (h01), and it is supported by bhusare et al. (2018), kishori and sheeba (2017), but it is refuted by singh and sharma (2018) and bhattarai (2016). table 4 gls regression results varibles coefficient z-value p>|z| npl -0.147 -1.74 0.085 llp 0.054 1.15 0.253 lad 0.068 0.92 0.358 car -0.201 -3.14 0.002** aq 0.866 2.00 0.049 cost per loan asset 0.231 0.67 0.502 moderation of npls 2.068 3.81 0.000* moderation of llps -0.876 -2.60 0.011* moderation of lad -0.419 -0.77 0.444 moderation of car 0.735 2.96 0.004** moderation of aq -8.373 -2.66 0.009** r-squared: 0.22 adjusted r-squared: 0.17 f-statistic: 114.04 prob.: 0.0000 note. *** significant at 1%, ** significant at 5%, * significant at 10% in addition, a negative moderating effect was found inthe relationship between loan loss provisions and the roa, with a coefficient of -0.876 (87.6%). this means that a one naira increase in loan loss provisions will reduce profitability by 87.6percent. the negative relationship is in line with the prior expectations of this study. furthermore, the llpr has a p-value of 0.01, which is at a 10 percent significant level. this outcome offers unswerving support to the preliminary hypothesis (h02) and is further corroborated in the studies by hamza (2017), alhadab and alsahawneh (2016). however, this was the case in the findings of kutum (2017) and mwaurah et al. (2017). also, table 4 shows 111 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 a negative moderating role of the cla on the association between the lad/td and the roa, with a coefficient of -0.419 (41.9%). this shows that as loans and advances increased by one, profitability would decline by 41.9%, even though the increase is not significant. the coefficient of tl/td was negative, which indicated that more aggressive banks were involved in lending activities as compared to deposits and, therefore, were less profitable. in the case of developed countries, the scenario may be different (bhattarai, 2016). however, in the case of developing countries like nigeria, where people have a tendency to default on loans, a higher tl/td may hinder efficiency because it increases the cost related to cash collection. on the basis of this, h03 would not be rejected and is supported by kayode et al. (2015) and bizuayehu. (2015), but opposed by muriithi et al. (2016) and samuel, (2015). from table 4, it can be seen that the clar has a negative moderating effect on the association between capital adequacy ratio and the roa, with a coefficient of 0.735 (73.5%), and it was statistically significant at the 5 percent level of significance. this implies that capitalized banks are in a better position to withstand any negative shock. in a developing country like nigeria, the cost of capital is higher, and therefore, it is expensive to maintain more capital. hence, banks with high capital have higher capital efficiency. this finding is inconsistent with the hypothesis h04 and contrasts with the findings in isanzu, (2017) and bhattarai (2016). it could also be seen from the results that the clar moderates the relationship between the aq and the roa negatively, with a coefficient of -8.373 percent, but is significant at 5 percent. this indicates that a one naira increase in the aq will reduce profitability by 8.373 percent. the negative relationship contradicts the prior expectations of this study. this contradicts the initial hypothesis (h05), but is supported by cheruiyot (2016), and ogbulu and eze (2016), who reported insignificant relationships. conclusion and recomendations the findings of the study point to the conclusion that the cost per loan asset ratio (clar) moderates the relationship between npls and the financial performance of listed dmbs positively. secondly, it can also be concluded that the cost per loan asset ratio (clar) moderates the association negatively between llps and the roa of deposit money banks. since llps normally serve as a financial backup for banks, 112 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 they can absorb losses. moreover, a one-naira increase in the ratio of llps to loans and advances (lad) will decrease bank performance. thirdly, the study has revealed that as measured by the roa, and even though it was not statistically significant, the cost per loan asset ratio (clar) moderates negatively the relationship between loans and advances and profitability. the study also showed that loans and advances (lad) played an important role in explaining changes in the roa. thus, the research has concluded that loans and advances are a determinant of the roa of nigerian listed dmbs. fourthly, the results also confirmed that the clar moderates the relationship between the car and return on assets positively. this finding is in line with prior expectations. the significant result of the car means that the car may influence the bank’s roa. lastly, results show that the clar moderates the relationship between the roa and asset quality of listed dmbs in nigeria and is statistically significant. there is clear evidence that asset quality is one of the credit risk attributes that may influence the roa of banks. this means banks with high asset quality and lower npls earn more profits than others. in summary, it is clear that the clar has a moderate effect of asset quality on the roa of dmbs. in light of this finding, the study recommends that the management of dmbs in nigeria should take comprehensive measures, such as the use of camel indicators to curtail the negative implications of non-performing loans on financial performance. the study also recommends that bank management should place a greater emphasis on deposit mobilization because it can affect their returns. the study also recommends that the capital adequacy ratio should be managed because adequate liquidity can reduce the banks’ sensitivity to credit risk. furthermore, this could affect the financial performance positively. finally, the study recommends that regulators develop effective prudential guidelines and policies to strengthen the management of credit risk. this may include maintaining an adequate level of provisions for nonperforming and bad loans, transferring default risk to third parties, extending maturity, restructuring the duration of non-performing loans, and decreasing interest rates for loans. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. 113 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 refrences abata, d., & adeolu, m. (2014). asset quality and bank performance: a study of commercial banks in nigeria. research journal of finance and accounting, 5, 39–44. https://iiste.org/journals/ index.php/rjfa/article/view/16209/16592. abbas, a., zaidi, s. a. h., ahmad, w., & ashraf, r. u. (2014). credit risk exposure and performance of banking sector of pakistan. journal of basic applied science research, 4(january,2014),240-245.https://www.researchgate.net/ p u b l i c a t i o n / 2 7 1 3 2 7 7 2 1 _ c r e d i t _ r i s k _ e x p o s u r e _ a n d _ performance_of_banking_sector_of_pakistan abiola, i., & olausi, a. s. (2014). impact of credit risk management on the commercial banks performance in nigeria. international journal of management and sustainability, 3, 295–306. https:// doi.org/10.18488/journal.11/2014.3.5/11.5.295.306 alhadab, m., & alsahawneh, s. (2016). loan loss provision and the profitability of commercial banks: evidence from jordan. international journal of business and management, 11(12), 242. https://doi.org/10.5539/ijbm.v11n12p242 alshatti, a. s. (2015). effect of credit risk management on the financial performance of the jordanian commercial banks. journal of investment management and financial innovations, 12, 338345. https://doi.org/10.5897/ajbm2013.7171 asare, o. (2015). impact of credit risk on profitability of some selected banks in ghana. school of business knust, kenya, ghana. http://ir.knust.edu.gh/xmlui/handle/123456789/8653 balango, t. k., & rao, m. k. (2017). effect of npl on profitability of banks with reference to commercial bank of ethiopia. business and management research journal, 7(june), 45–50. https:// www.ijsrp.org/research-paper-0517/ijsrp-p6566.pdf bhattarai, y. r. (2016). effect of credit risk on the performance of commercial banks in nigeria. nrb economic review, 4(1), 42–64. https://doi.org/10.1007/s13398-014-0173-7.2 bhattarai, y. r. (2017). credit risk and commercial banks ’ profitability in nepal: a panel approach. journal for studies in management and planning, 3, 1–15. https://www.pen2print.org/2017/10/ credit-risk-and-commercial-banks.html bhusare, p. s., shukla, s., & indi, r. (2018). the effect of credit risk on returns of select public and private scheduled commercial banks in india. iosr journal of business and management, 2(june), 30–7. www.iosrjournals.org 114 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 bizuayehu, m. (2015). impact of credit risk on profitability performance of commercial banks in ethiopia (unpublished master’s thesis). deparment of accounting & finance, college of business and economics addis ababa, ethopia. https://doi. org/10.5897/ajbm2013.7171 cbn. (2016). cbn financial stability report-december, 2016. global economic and financial system stability outlook. https://www.cbn.gov.ng/out/2017/fprd/fsr%20december%20 2016%20(2). cheruiyot, r. k. (2016). effect of asset quality on profitability of commercial banks in kenya (unpublished master’s thesis). university of nairobi, kenya. hamza, s. m. (2017). impact of credit risk management on banks performance: a case study in pakistan banks. european journal of business and management, 9, 57–64. https://www.iiste.org/ journals/index.php/ejbm/article/view/34898. iheanyi, i. h., & sotonye, i. (2017). assessing the performance of nigeria’s bank through camel model. journal of accounting and finance management, 3, 14-22. https://www.iiardjournals. org/get/jafm/vol.%203%20no.%201202017/assessing%20the. isanzu, j. s. (2017). impact of credit risk on the financial performance of chinese banks. journal of international business research and marketing, 2(march), 14–17. https://doi.org/10.18775/ jibrm.1849-8558.2015.23.3002 kargi, h. s. (2011). credit risk and the performance of nigeria banks. european journal of finance management, 1(july, 2011), 1-24. https://www.ijsrp.org/research-paper-0517/ijsrp-p6566. kayode, o. f., obamuyi, t. m., owoputi, j. a., & adeyefa, f. a. (2015). credit risk and bank performance in nigeria. iopsr journal of economics and finance, 6(march-april), 21-28. https://doi.org/10.9790/5933-06222128 kishori, b., & sheeba, j. (2017). impact of credit risk on the profitability of state bank of india. international journal of science research and technology, 3, 2587-2594. https://doi. org/10.21917/ijms.2017.0073 komolafe, b. (2017, january 1). economic recession: the vanguard. nigeria edition. kutum, i. (2017).the impact of credit risk on the profitability of banks listed on the palestine exchange. research journal of finance and accounting, 8(november), 136–41. https://www.iiste.org/ journals/index.php/rjfa/article/view/36781 115 the international journal of banking and finance, vol. 18, number 1 (january) 2023, pp: 95–115 mendoza, r., & rivera, j. p. r. (2017). effect of credit risk and capital adequacy on the profitability of rural banks in the philippines. scientific annals of economics and business, 64, 83–96. https:// doi.org/10.1515/saeb-2017-0006 muriithi, j., waweru, k., & muturi, w. m.(2016). effect of credit risk on financial performance of commercial banks kenya. journal of economics and finance,7(4), 72–83. www.iosrjournals.org mwaurah, i., muturi, w., & waititu, a. (2017). the influence of credit risk on stock returns. international journal of scientific research publications, 7(5), 575–84. https://www.ijsrp.org/ research-paper-0517/ijsrp-p6566. mwongela, m. c. (2015). the relationship between interest rates and profitability of commercial banks in kenya (unpublished master’s thesis). department of business administration, university of nairobi, kenya. ogbulu, o. m., & eze, g. p. (2016). credit risk management and the performance of deposit money banks in nigeria: an error correction analysis. applied economics and finance,4, 97– 109. https://doi.org/10.11114/aef.v3i2.1356 olabamiji, o., & michael, o. (2018). impact of credit management practices and bank performance: evidence from first bank. south asian journal of social studies and economics, 1(19 april), 1–10. 10.9734/sajsse/2018/v1i125772 poudel, p. r. s. (2012). impact of credit risk management on financial performance of commercial banks in nepal. international journal of arts and commerce, 1(5), 9–15. https://doi. org/10.5897/ajbm2013.7171 samuel, o. l., & iwoye, a. (2015). effect of credit risk on the performance of commercial banks in nigeria. african journal of accounting, auditing and finance, 2, 118. https://doi. org/10.1007/s13398-014-0173-7.2 sheefeni, j. p. s. (2015). impact of macro-economic determinants on non-performing loans in namibia. international review of research in emerging markets and the global economy, 1, 612–632. https://doi.org/10.5901/ajis.2015.v4n3s1p534 singh, s., & sharma, k. (2018). impact of credit risk on profitability of indian public sector banks. international journal of research economics and social sciences (ijeress), 8(2), 492–8. http://euroasiapub.org.https://www.vanguardngr.com/2017/01/ economic-recession-constrained-banking-2016/ were, m., & wambua, j. (2014). what factors drive interest rate spread of commercial banks? emprical evidence from kenya. journal advance research, (december), 1-11. http://dx.doi. org/10.1016/j.rdf.2014.05.005 111 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: song, s. i., yazi, e., morni, f., & janang, j. t. (2021). covid-19 and stock returns: evidence from malaysia. international journal of banking and finance, 16(2), 111-140. https://doi.org/10.32890/ ijbf2021.16.2.5 covid-19 and stock returns: evidence from malaysia 1saw imm song, 2erimalida yazi, 3fareiny morni & 4jennifer tunga janang 1universiti teknologi mara, cawangan pulau pinang 2, 3 & 4universiti teknologi mara, cawangan sarawak 1corresponding author:songsi@uitm.edu.my received: 15/1/2021 revised: 30/3/2021 accepted: 30/3/2021 published: 15/6/2021 abstract the severe acute respiratory syndrome (sars) coronavirus or covid-19 has affected the world unprecedentedly. malaysia is not exempted from its impact. the malaysian government announced a nationwide lockdown in the middle of march 2020. the magnitude of the outbreak had caused panic to the public and financial panic in the stock market. this study examined the impact of covid-19 cases and the action taken by the government through movement control orders (mcos) and economic stimulus packages in the stock market. event study methodology was used to assess the impact of covid-19 on stock returns in bursa malaysia. consistent with the efficient market hypothesis, the study found that during the early stages of the mcos, the cumulative average abnormal returns (caar) reflected significant negative returns. however, it showed positive returns after 112 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 mco 3 and mco 4. the results implied that the market perceived that the pandemic was under control. the study also revealed a significant relationship between caar and the number of cases announced, supporting the notion that in a less to a moderately free country such as malaysia, investors showed a certain lack of trust in the number of cases reported by the authorities, and thus overreacted to the number of reported cases. the stimulus packages that were expected to stabilise the economy and society were found to be positively significant during the early stages of the mcos. keywords: covid-19, pandemic, malaysia, event study, efficient market hypothesis. jel classification: g010, g140. introduction the severe acute respiratory syndrome coronavirus 2 or covid-19 is an infectious disease that causes respiratory infections in humans. before december 2019, the virus was relatively unknown until an outbreak of pneumonia emerged in wuhan, china. subsequently, the virus spread across the world in mid-january 2020 when many chinese nationals flew home or abroad for the chinese lunar year celebrations. covid-19 was classified by the world health organization (who) as a pandemic on 12 march 2020. according to who, covid-19 has an incubation period of 14 days. the virus can be transmitted within that period (who, 2020), and the transmission can quickly occur through respiratory droplets. people, especially those with pre-existing medical condition and weak immune systems, are more vulnerable to the virus. generally, covid-19 cases in malaysia can be categorised by waves (shah et al., 2020) and clusters. during the early stage, the cases were confined to just a few imported cases (zack, 2020) involving foreigners who arrived in malaysia on 24 january 2020. the 2nd wave occurred on 4 march 2020 due to a sudden surge in the number of positive cases caused by a malaysian who returned from china. the situation became worse following a religious event held from 28 february until 2 march 2020 involving 12,500 attendees at masjid 113 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 seri petaling, kuala lumpur (roslan, 2020). this cluster had led to huge spikes in the number of local cases (ministry of health, 2020). malaysia recorded 2,000 active cases by the end of march 2020, and more than 5000 confirmed infections by 15 april 2020 (kaos, 2020). it was the largest cumulative number of confirmed covid-19 infections in southeast asia during the period. the highest number of covid-19 cases in malaysia in the second wave was reported on 5 june with 277 new cases. by 1 september 2020, the covid-19 outbreak had affected most countries globally, with 25,637,635 confirmed positive cases and 854,772 deaths. about 99 percent of patients experienced mild infections and recovered, but 1 percent (61,145) were in serious or critical condition (worldometer, 2020). since the outbreak of the disease in china which spread globally in early 2020, its impact has been especially severe on the tourism and airline industries due to lockdowns or restricted movements and travel bans imposed by various countries. following the steps taken by other countries affected by the pandemic, the malaysian government announced on 16 march 2020, that a 14day mco be imposed nationwide to control the spread of the disease. this led to the shutting down of all government and private premises except those which were categorised under essential services. it also restricted individuals from travelling to other states that had been declared as red zones. all citizens were prohibited from leaving the county, and the country’s borders were closed to foreigners. all sports, religious, social and cultural activities were also prohibited (bunyan, 2020). the duration of the first mco was from 18 march until 31 march 2020 (bunyan, 2020). however, it was further extended to 14 april 2020 (anis, 25 march 2020) and after that to 28 april 2020 (rodzi, 2020), and subsequently to 12 may 2020 (povera & harun, 2020) in response to the rising number of cases in the country. on 1 may 2020, a “conditional movement control order” (cmco) was announced by the prime minister. the deliberate easing of restrictions permitted most businesses to restart on 4 may under rigorous standard operating procedures (sop) until 9 june 2020. it was later replaced by the recovery movement control order (rmco) until 31 august (loo, 2020). 114 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 the stock market plummeted immediately (bursa malaysia, 2020) in response to the mco announcement on 16 march 2020. the hardest-hit industries were the airlines, tourism, and retail businesses, which led to an increasing number of people losing their jobs and a reduction in trade volume. small and medium-sized enterprises were also severely affected due to the decline in social spending and fixed expenses on rents, wages and interests (duan et al., 2020). the severity of the outbreak reported in the number of daily new cases was also expected to affect the economy and stock market. the main objectives of this paper are to assess the extent of the impact of the outbreak and how the measures taken by the government through the implementation of the mcos and the handing out of stimulus packages affected the stock market. the mco announcements which impacted stock prices can be explained based on market efficiency hypothesis (fama, 1970). we hypothesized that the control measures taken by the government to mitigate the spread of covid-19 helped restrain the panic and thus was expected to have had a positive effect on stock prices in line with investor sentiment driven hypothesis (baker & wurgler, 2007; chen et al., 2020; yu & yuan, 2011). similarly, we also expected the government’s announcements of the stimulus packages to exert a positive impact on stock prices. as the pandemic has been unprecedented, there is still no empirical study on the impact of covid-19 on the malaysian stock market. findings from this paper are crucial to policymakers, the investment community, and the corporate sector to make informed decisions based on the government’s measures and the response of market participants during this pandemic period. it also contributes to the literature which has investigated stock market reactions to various disasters and crises. we employed the event study methodology with a sample of 760 companies listed on bursa malaysia to achieve the study’s objectives. the sample period taken for the study was 144 days (2 january to 31 july 2020). various event windows were examined to assess the reaction of the market towards the mcos. the event day (to) was the official announcement day of the mcos. the results indicated that the market reacted negatively to the mco announcement during the early stages, but after mco 3 and mco 4, the market’s reaction turned positive. the results also showed a significant relationship between caar and the number of cases announced, which supported 115 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 the notion that in a less to moderately free country such as malaysia, investors could overreact to these announcements. following erdem (2020), we defined less free countries as those with relatively low political and civil liberties rights indicators. the stimulus packages which were expected to stabilise the economy and the society were only positively significant during the early stages of the mcos. the outline of the paper is as follows. in section 2, we review previous studies and discuss our paper’s contribution. this is followed by section 3, where we describe our data and the methodology used, specifically the event study method and regression analysis. in section 4, we examine the results of the event study and the regression results. finally, our concluding remarks are presented in section 5. literature review since the outbreak of covid-19 in china, the pandemic has posed unprecedented uncertainty to global markets. it is believed that this virus and sars which appeared in 2002 are quite similar, but the severity of the current infection is more dreadful than the earlier one. similar to a terrorist attack, burch et al. (2016), and chen et al. (2007) found that the sars epidemic in 2002 caused shock and panic selling in the stock market. on the other hand, nippani and washer (2004) focused on stock indices of eight countries which were seriously affected by sars and found that sars had no negative impact on the affected countries’ stock markets except for those based in china and vietnam. it may be because the most severe negative impacts of sars were seen on the local consumption side, especially in tourism and air travel-related services in those countries (siu & wong, 2004). lee and mckibbin (2004) evaluated the global economic impact of sars. they found that the sars epidemic spread rapidly through countries via cross border travel and affected the economies of many countries due to financial integration and globalisation. coupled with the widespread media coverage, marinč (2016) found that geographical proximity and information disseminated on an outbreak had also influenced the affected countries’ asset prices. the results indicated that the impact of the epidemic on stock prices was generally negative. the effect was prominent in smaller and more 116 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 volatile stocks and less established industries. using the infectious disease equity market volatility (emv) tracker, baker et al. (2020) showed that the early-phase impact of covid-19 looked similar to the impact of other infectious disease outbreaks in the past. however, as the threat became increasingly apparent and overwhelming, the dow jones industrial average, which was high in february 2020, crashed in february up to march 2020. baker et al. (2020) also attributed the adverse stock market reactions to the severity and ease of spread of the covid-19 virus and the high death rate among those infected with the virus. the high stock market volatility was also due to the rapid diffusion of information compared to the previous pandemics. the mcos had brought a severe plunge in demand for various services. it significantly reduced the flow of labour to businesses. the economist (2020) noted that “one of the ways virus damages the economy is to interfere with the supply of labour, goods and services. but more serious is its spill-over effect.” this spill-over effect on the capital market was expected to be negatively consistent with marinč’s (2016) study. however, as the pandemic has been unprecedented, it is vital to study the impact of this pandemic on the stock market to assess the market responses to this health crisis. effects of mco on the equity market as of 1 september 2020, malaysia reported 9340 infected cases, 127 death, and 9054 recoveries. malaysia was one of the earliest countries that implemented the mcos (shah et al., 2020) as the threat of the virus became increasingly apparent. many businesses were affected because of the ‘social distancing’ and ‘stay-at-home’ orders. tourism, airlines, event management, and sports industries were the most badly hit sectors. all events which drew crowds were cancelled. the suspension of tourism, business travel, and work visas severely impacted malaysia’s industries, which relied very much on its economic growth. it is estimated that there was a loss of over rm45 billion in the malaysian tourism industry due to the covid-19 pandemic for the first half of 2020 (bernama, 2020). the mcos that restricted and reduced people and factory activities led to reduced demand for energy products such as aviation fuel, coal and others, which subsequently caused a fall in oil prices. 117 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 coupled with the overproduction of oil by saudi arabia and russia in early march 2020, the us stock markets declined over 30 percent from february 2020, a record high, to a crash in early march 2020 indicating that volatilities of equities and oil prices had spiked to crisis levels (li, 2020). the spill-over effect on the world financial markets witnessed a plunge in their stock prices. the spill-over effect of oil crisis on stock prices was further supported by le and chang (2015) that changes in oil prices had a significant impact on stock market returns in malaysia. their study period included the asian financial crisis in 1998. le and chang (2015) established that the immediate response of the malaysian stock market was found to be positive, which afterwards turned negative in the four to nine months following the oil price shocks. ozili and arun (2020) highlighted that firm profits were likely to be lower due to fear and uncertainty. the global stock markets erased about us$6 trillion in wealth in one week from 24 to 28 february 2020. however, they found that the impact of the social distancing policies on the country’s stock market indices and economic activities were positive. they contended that the increasing number of confirmed coronavirus cases did not significantly affect the level of economic activities. alam et al. (2020) revealed that the market responded favourably with significantly positive average abnormal returns (aar) during the lockdown period in their study of 31 companies listed on the bombay stock exchange (bse). on the contrary, negative aar was found in the pre-lockdown period. they contended that this was because the government’s action to impose a lockdown was seen as an effective way to stop the virus from spreading, and the reaction was reflected in the buoyant stock market returns. on the other hand, the negative aar in the pre-lockdown period was due to global panic at the beginning of the pandemic. consistent with alam et al. (2020) results, narayan et al. (2020) showed that all policies, on aggregate, had positive effects on the g7 countries’ stock market excess returns. however, nationwide lockdown influenced returns the most in the majority of the countries researched, followed by stimulus packages and travel bans. in malaysia’s context, where the mcos were successfully enforced, had led to the flattening 118 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 of the covid-19 curve (abdul rashid, 2020). thus, we hypothesise that the stock market should have negative returns at the beginning of the mcos h1. however, it is expected to have positive returns once the pandemic is under control h2. the effect of covid-19 detection on the malaysian stock market bursa malaysia tumbled during the initial mco as investors sold securities due to the fear of expected economic impact caused by the virus (aruna, 2020). it has been found that most of the covid-19 research currently focuses more on specific countries (ramelli & wagner, 2020; 2020; al-awadhi et al., 2020). except for a study by ashraf (2020), who used daily confirmed new and death cases from 64 countries, and found that stock markets responded negatively to the growth in new confirmed cases. his findings showed that the stock markets reacted more on new confirmed cases rather than on the rise in the number of deaths. generally, adverse market reaction was more robust during the early stages of the pandemic, while the stock markets responded at varying speeds to the covid-19 news depending on the stage of the outbreak. ashraf’s (2020) findings were consistent with erdem (2020), where it was found that the pandemic had significant adverse effects on the stock market. they found that the growth in the number of new cases on stock returns was about three times more than that of growth in the number of deaths. using the freedom index as their independent variable, they found that the impact on the stock market returns in freer countries was less than those in less-free countries for every increase in the number of cases per million. in addition, lei and wisniewski (2018) concurred that expropriation was more likely to happen in autocratic states (the opposite of democratic or free state) amid economic challenges. there was a likelihood of increased mismanagement of firms which can lower firm value during the uncertainty period. malaysia is categorised as having a “flawed democracy”, relatively lesser in terms of full democracy and freedom practices as assessed by the democracy index of the economist intelligence unit (the economist, 2020). with a record of 52/100 freedom index (freedom house, 2020), malaysia is considered as partly free or relatively low to moderate in terms of political rights and civil liberties freedom measurements. thus, we hypothesise that 119 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 the numbers of new and death cases are expected to have a negative impact on stock returns h3. stimulus package and stock market performance to curb the effects of the covid-19 pandemic, the malaysian government has launched a total of rm295 billion worth of stimulus packages. the “caring package” or prihatin economic stimulus package of rm250 billion (povera, 2020) was announced on 27 march 2020, followed by drastic measures of a six-month loan moratorium offered by bank negara malaysia (bnm) to reduce the financial impact of the lockdown orders. of these, rm128 billion was distributed to protect the people’s welfare, and rm100 billion was channelled to support businesses, including small and medium enterprises (smes). on 6 april, a special stimulus plan worth rm10 billion was announced, which was aimed at helping small and medium-sized enterprises (smes) in mitigating the impact of mcos (daim, 2020). an additional short-term economic recovery plan (penjana) with 40 initiatives worth rm35 billion was also announced on 5 june 2020 to alleviate the effects of the pandemic (loheswar, 2020). various measures such as prohibiting short selling were also taken by the malaysian government to mitigate the capital market’s challenges due to the mco. the announcement was well-received by the public, and a positive reaction in the stock market was evident (zainul, 2020). the market reaction was also influenced by the situation in other economies as the process of globalisation has made economies worldwide become more interdependent. it increases synchronisation in the global stock markets, and are especially affected by the stock markets’ movement in the united states. in accordance with the efficient market hypothesis (emh), when investors perceive that news positively impacts the market then the investors will behave more optimistically. conversely, if the investors perceive the news as bad, then sentiments become relatively pessimistic, and the market will indicate a downward trend. based on this hypothesis, the stock prices reflect all available information, and stocks always trade at their fair value on exchanges. on the other hand, according to semi-strong emh, it contends that security prices have factored in publicly available market information but not material non-public information. superior gain is possible if material non-public information is available. 120 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 after the first mco announcement in march 2020, the stock market’s negative impact signalled a significant decline in earnings and a global economic recession (liew & surendran, 2020). however, with the stimulus packages’ announcement, it was expected that the malaysian market volume and participation would be stabilised. in the prihatin stimulus packages, the loan repayment moratorium, which was estimated to be worth rm100 billion, resulted in around rm15 billion excess cash a month in the market. the liquidity in the market partly contributed to the rally in the stock market which was a surprise to many. nevertheless, teng and surendran (2020) attributed the rally to the government’s low-interest rates policy. a low-interest-rate environment would motivate, especially institutional investors, to take greater risks to increase their returns. the low-interest-rate regime also encouraged personal and business spending, improved revenues and corporate earnings and thus better-expected stock returns. it was consistent with ozili and arun (2020) that a higher level of fiscal policy spending positively impacted economic activities. thus, we hypothesise that the stimulus packages provided by the malaysian government will have a positive impact on stock returns h4. methodology data this study investigated the effects of covid-19 cases and the actions taken by the government, on the stock returns in bursa malaysia. the required data was extracted from thomson reuters datastream database. the initial sample consisted of 826 public listed companies. due to incomplete information for some companies, the final sample consisted of 760 companies listed on bursa malaysia. daily stock prices, the market index and all financial data were obtained from the same database. the data was collected from 2 january 2020 to 31 july 2020 (144 observation days). the selected period would allow the analysis of the implication of the specific events between the announcement of the first covid-19 cases in malaysia and the announcement of the rmco on the market returns in malaysia. event study was employed to assess the short-term impact of the pandemic 121 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 on the stock returns. the announcement dates of the mcos (these include mcos, cmco, rmco announcements) were selected as the event day. to investigate the impact of the announcements on stock prices, we set up three event windows consisting of 15 observation days: (-7, 0), (0, +7) and (-7, +7). the dates for each event (stimulus packages, mcos and number of cases) were collected from various internet sources. event study methodology the event study methodology is an established method for measuring the abnormal stock returns when an event is announced (song et al., 2011; yazi, et al., 2015). the events include, for instance, announcements of the mcos, announcements of the government stimulus packages, announcements of the economic sectors to resume operations or announcements of covid-19 new or death cases. in assessing the abnormal returns of the stocks (the actual returns minus the required rate of returns), capital asset pricing model (capm) was employed to estimate the required rate of returns of the stocks: (1) where, r(rit) is the required rate of return of stock i on day t. rf denotes the daily risk-free rate derived from the average 3-months treasury bills for one year divided by 365 and adjusted for the inflation rate. βi is the beta measurement for stock i estimated from the historical betas’ average for two years prior to the event date and were extracted from datastream database. rmt is the return of the benchmark index, kuala lumpur composite index (klci) on day t. after establishing the required rate of return, the abnormal or unexpected return (arit) for each stock i on event day t is measured as follows: (2) where, arit is the daily abnormal returns of stock i on day t rit is the actual return of stock i on day t r(rit) is the required rate of return for stock i on day t 7 thus, we hypothesise that the stimulus packages provided by the malaysian government will have a positive impact on stock returns (h4). methodology data this study investigated the effects of covid-19 cases and the actions taken by the government, on the stock returns in bursa malaysia. the required data was extracted from thomson reuters datastream database. the initial sample consisted of 826 public listed companies. due to incomplete information for some companies, the final sample consisted of 760 companies listed on bursa malaysia. daily stock prices, the market index and all financial data were obtained from the same database. the data was collected from 2 january 2020 to 31 july 2020 (144 observation days). the selected period would allow the analysis of the implication of the specific events between the announcement of the first covid-19 cases in malaysia and the announcement of the rmco on the market returns in malaysia. event study was employed to assess the short-term impact of the pandemic on the stock returns. the announcement dates of the mcos (these include mcos, cmco, rmco announcements) were selected as the event day. to investigate the impact of the announcements on stock prices, we set up three event windows consisting of 15 observation days: (7, 0), (0, +7) and (-7, +7). the dates for each event (stimulus packages, mcos and number of cases) were collected from various internet sources. event study methodology the event study methodology is an established method for measuring the abnormal stock returns when an event is announced (song et al., 2011; yazi, et al., 2015). the events include, for instance, announcements of the mcos, announcements of the government stimulus packages, announcements of the economic sectors to resume operations or announcements of covid-19 new or death cases. in assessing the abnormal returns of the stocks (the actual returns minus the required rate of returns), capital asset pricing model (capm) was employed to estimate the required rate of returns of the stocks: 𝑅𝑅(𝑅𝑅𝑖𝑖𝑖𝑖) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖𝑖𝑖(𝑅𝑅𝑚𝑚𝑖𝑖 − 𝑅𝑅𝑓𝑓) (1) where, r(rit) is the required rate of return of stock i on day t. rf denotes the daily risk-free rate derived from the average 3-months treasury bills for one year divided by 365 and adjusted for the inflation rate. βi is the beta measurement for stock i estimated from the historical betas’ average for two years prior to the event date and were extracted from datastream database. rmt is the return of the benchmark index, kuala lumpur composite index (klci) on day t. after establishing the required rate of return, the abnormal or unexpected return (arit) for each stock i on event day t is measured as follows: 𝐴𝐴𝑅𝑅𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑖𝑖𝑖𝑖 − 𝑅𝑅(𝑅𝑅𝑖𝑖𝑖𝑖) (2) where, arit is the daily abnormal returns of stock i on day t rit is the actual return of stock i on day t r(rit) is the required rate of return for stock i on day t 7 thus, we hypothesise that the stimulus packages provided by the malaysian government will have a positive impact on stock returns (h4). methodology data this study investigated the effects of covid-19 cases and the actions taken by the government, on the stock returns in bursa malaysia. the required data was extracted from thomson reuters datastream database. the initial sample consisted of 826 public listed companies. due to incomplete information for some companies, the final sample consisted of 760 companies listed on bursa malaysia. daily stock prices, the market index and all financial data were obtained from the same database. the data was collected from 2 january 2020 to 31 july 2020 (144 observation days). the selected period would allow the analysis of the implication of the specific events between the announcement of the first covid-19 cases in malaysia and the announcement of the rmco on the market returns in malaysia. event study was employed to assess the short-term impact of the pandemic on the stock returns. the announcement dates of the mcos (these include mcos, cmco, rmco announcements) were selected as the event day. to investigate the impact of the announcements on stock prices, we set up three event windows consisting of 15 observation days: (7, 0), (0, +7) and (-7, +7). the dates for each event (stimulus packages, mcos and number of cases) were collected from various internet sources. event study methodology the event study methodology is an established method for measuring the abnormal stock returns when an event is announced (song et al., 2011; yazi, et al., 2015). the events include, for instance, announcements of the mcos, announcements of the government stimulus packages, announcements of the economic sectors to resume operations or announcements of covid-19 new or death cases. in assessing the abnormal returns of the stocks (the actual returns minus the required rate of returns), capital asset pricing model (capm) was employed to estimate the required rate of returns of the stocks: 𝑅𝑅(𝑅𝑅𝑖𝑖𝑖𝑖) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖𝑖𝑖(𝑅𝑅𝑚𝑚𝑖𝑖 − 𝑅𝑅𝑓𝑓) (1) where, r(rit) is the required rate of return of stock i on day t. rf denotes the daily risk-free rate derived from the average 3-months treasury bills for one year divided by 365 and adjusted for the inflation rate. βi is the beta measurement for stock i estimated from the historical betas’ average for two years prior to the event date and were extracted from datastream database. rmt is the return of the benchmark index, kuala lumpur composite index (klci) on day t. after establishing the required rate of return, the abnormal or unexpected return (arit) for each stock i on event day t is measured as follows: 𝐴𝐴𝑅𝑅𝑖𝑖𝑖𝑖 = 𝑅𝑅𝑖𝑖𝑖𝑖 − 𝑅𝑅(𝑅𝑅𝑖𝑖𝑖𝑖) (2) where, arit is the daily abnormal returns of stock i on day t rit is the actual return of stock i on day t r(rit) is the required rate of return for stock i on day t 122 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 then the average abnormal returns (aar) for a sample of n stock for each day are calculated as follows: (3) where, n is the number of firms with abnormal returns on day t. the cumulative average abnormal returns (caar) represents the average total effect of the event across all firms, thus: (4) to determine the significant value of the caar, the one sample t-test is conducted. regression analysis multiple regression analysis was conducted to test the hypotheses by assessing the impact of covid-19 cases based on the measures taken by the government (e.g. mcos and stimulus packages) on stock market performance. caart = α + β1ln(klci)t + β2ln(dji)t-1 + β3ln(oil)t-1 +β4newt +β5deatht + β6mcost + β7stimulust + ɛt (5) where caart = cumulative average abnormal returns ln(klci)t = natural log of klci index return; ln(dji)t = natural log dow jones index return; ln(oil)t = natural log wti crude oil return; newt = number of new cases; deatht = number death cases; mcost = dummy variable equal 1 on mco announcement day + subsequent 4 trading days, and zero for otherwise; stimulust = dummy variable equal 1 on stimulus package announcement day + 4 subsequent days, and zero for otherwise. 8 then the average abnormal returns (aar) for a sample of n stock for each day are calculated as follows: 𝐴𝐴𝐴𝐴𝐴𝐴𝑡𝑡 = 1 𝑁𝑁 ∑ 𝐴𝐴𝐴𝐴𝑖𝑖,𝑡𝑡𝑁𝑁𝑖𝑖=1 (3) where, n is the number of firms with abnormal returns on day t. the cumulative average abnormal returns (caar) represents the average total effect of the event across all firms, thus: 𝐶𝐶𝐴𝐴𝐴𝐴𝐴𝐴𝑡𝑡 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖,𝑡𝑡𝑛𝑛𝑖𝑖=1 (4) to determine the significant value of the caar, the one sample t-test is conducted. regression analysis multiple regression analysis was conducted to test the hypotheses by assessing the impact of covid-19 cases based on the measures taken by the government (e.g. mcos and stimulus packages) on stock market performance. caart = α + β1ln(klci)t + β2ln(dji)t-1 + β3ln(oil)t-1 +β4newt +β5deatht + β6mcost + β7stimulust + ɛt (5) where caart = cumulative average abnormal returns ln(klci)t = natural log of klci index return; ln(dji)t = natural log dow jones index return; ln(oil)t = natural log wti crude oil return; newt = number of new cases; deatht = number death cases; mcost = dummy variable equal 1 on mco announcement day + subsequent 4 trading days, and zero for otherwise; stimulust = dummy variable equal 1 on stimulus package announcement day + 4 subsequent days, and zero for otherwise. the klci, dji, and oil prices are used as the control variables as they might affect the caar (lee & mckibbin, 2004) for model with dependent variable caar. while klci is used as the barometer for movements of the malaysian market, dji is used as the leadership of price in the world market that influences international investors to respond to news from the united states due to their greater liquidity, capitalization and cheaper transaction costs (becker et al., 1995). the use of oil prices is in line with studies which have established the existence of a relationship between changes in oil prices and stock prices of oil exporting countries (le & chang, 2015; salisu & isah, 2017; nusair & al-khasawneh, 2018). le and chang (2015) established that changes in oil prices have a significant impact on stock market returns in malaysia. on the other hand, salisu and isah (2017) confirmed the existence of a nonlinear relationship between oil prices and stock prices where the relationship is found to be stronger among oil exporting countries and that rising oil prices increase stock returns only when stock markets are bullish while falling oil prices lower stock returns only when stock markets are bearish (nusair & al-khasawneh, 2018). 8 then the average abnormal returns (aar) for a sample of n stock for each day are calculated as follows: 𝐴𝐴𝐴𝐴𝐴𝐴𝑡𝑡 = 1 𝑁𝑁 ∑ 𝐴𝐴𝐴𝐴𝑖𝑖,𝑡𝑡𝑁𝑁𝑖𝑖=1 (3) where, n is the number of firms with abnormal returns on day t. the cumulative average abnormal returns (caar) represents the average total effect of the event across all firms, thus: 𝐶𝐶𝐴𝐴𝐴𝐴𝐴𝐴𝑡𝑡 = ∑ 𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖,𝑡𝑡𝑛𝑛𝑖𝑖=1 (4) to determine the significant value of the caar, the one sample t-test is conducted. regression analysis multiple regression analysis was conducted to test the hypotheses by assessing the impact of covid-19 cases based on the measures taken by the government (e.g. mcos and stimulus packages) on stock market performance. caart = α + β1ln(klci)t + β2ln(dji)t-1 + β3ln(oil)t-1 +β4newt +β5deatht + β6mcost + β7stimulust + ɛt (5) where caart = cumulative average abnormal returns ln(klci)t = natural log of klci index return; ln(dji)t = natural log dow jones index return; ln(oil)t = natural log wti crude oil return; newt = number of new cases; deatht = number death cases; mcost = dummy variable equal 1 on mco announcement day + subsequent 4 trading days, and zero for otherwise; stimulust = dummy variable equal 1 on stimulus package announcement day + 4 subsequent days, and zero for otherwise. the klci, dji, and oil prices are used as the control variables as they might affect the caar (lee & mckibbin, 2004) for model with dependent variable caar. while klci is used as the barometer for movements of the malaysian market, dji is used as the leadership of price in the world market that influences international investors to respond to news from the united states due to their greater liquidity, capitalization and cheaper transaction costs (becker et al., 1995). the use of oil prices is in line with studies which have established the existence of a relationship between changes in oil prices and stock prices of oil exporting countries (le & chang, 2015; salisu & isah, 2017; nusair & al-khasawneh, 2018). le and chang (2015) established that changes in oil prices have a significant impact on stock market returns in malaysia. on the other hand, salisu and isah (2017) confirmed the existence of a nonlinear relationship between oil prices and stock prices where the relationship is found to be stronger among oil exporting countries and that rising oil prices increase stock returns only when stock markets are bullish while falling oil prices lower stock returns only when stock markets are bearish (nusair & al-khasawneh, 2018). 123 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 the klci, dji, and oil prices are used as the control variables as they might affect the caar (lee & mckibbin, 2004) for model with dependent variable caar. while klci is used as the barometer for movements of the malaysian market, dji is used as the leadership of price in the world market that influences international investors to respond to news from the united states due to their greater liquidity, capitalization and cheaper transaction costs (becker et al., 1995). the use of oil prices is in line with studies which have established the existence of a relationship between changes in oil prices and stock prices of oil exporting countries (le & chang, 2015; salisu & isah, 2017; nusair & al-khasawneh, 2018). le and chang (2015) established that changes in oil prices have a significant impact on stock market returns in malaysia. on the other hand, salisu and isah (2017) confirmed the existence of a nonlinear relationship between oil prices and stock prices where the relationship is found to be stronger among oil exporting countries and that rising oil prices increase stock returns only when stock markets are bullish while falling oil prices lower stock returns only when stock markets are bearish (nusair & al-khasawneh, 2018). the dummy variable equals to 1 was assigned to the mco announcement dates and their subsequent four trading days (total of five trading days of observation). it was found that bursa malaysia did not reach its full efficiency level in a semi-strong form market. mohammed et al. (2010) found that the caar still exhibited clear upward or downward trends even a few days after announcement dates. thus, the observation of five trading days was to capture the short-term market reactions in a near efficient semi-strong form market. findings and discussion this section is divided into two parts; (1) an event study methodology measuring the reaction of stock prices (caar) following covid-19 events in malaysia; and (2) regression analysis to determine the relationship of stock returns with covid-19 events. caar analysis firstly, to gauge the impact of covid-19 events on the malaysian stock market, we chose specific events such as the announcement of 124 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 the first, 100th, 1000th, 5000th covid-19 cases as well as the highest cases in one day in malaysia. besides that, relevant announcements made by the malaysian government that could have had an economic implication such as mco announcements, loan moratorium announcement, and announcements of various stimulus packages to boost the different economic sectors affected by covid-19 were taken into consideration. the results of the caar and t-statistics for each of the chosen event windows are presented in table 1. table 1 selected covid-19 events and the impact on caar of malaysian stocks date obs. day event caar (%) t-statistic 3 – 25 jan 2020 17 announcement of first covid-19 cases in malaysia 3.37 3.2445*** 3 jan – 9 mar 2020 46 malaysia reached the 100th case mark -0.98 9.3871*** 15 – 31 mar 2020 12 pm tan sri muhyiddin yassin announces the first 14-day mco -6.69 -12.2577*** 18 – 31 mar 2020 10 first mco (18 – 31 march 2020) -3.44 -20.7979*** 3 jan – 20 mar 2020 55 malaysia reached the 1,000th case mark -10.47 2.8037*** 3 jan – 24 mar 2020 57 klse suspends short selling -9.43 1.7215** 3 jan – 25 mar 2020 58 announcement of mco extension until 14 april and moratorium -11.05 1.2631 3 jan – 27 mar 2020 60 announcement of prihatin stimulus package -9.63 0.5937 1 – 14 apr 2020 10 second mco (1 – 14 april 2020) 10.43 0.0437 (continued) 125 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 date obs. day event caar (%) t-statistic 10 – 28 apr 2020 13 announcement of mco extension until 28 april 6.18 6.1997*** 3 jan 16 apr 2020 74 malaysia reached the 5,000th case mark 3.19 0.2525 15 – 28 apr 2020 10 third mco (15 – 28 april 2020) 6.05 6.4990*** 23 apr – 12 may 2020 13 announcement of mco extension until 12 may 10.61 6.1997*** 29 apr – 12 may 2020 7 fourth mco (29 april – 12 may 2020) 6.70 11.3876*** 3 jan – 1 may 20201 85 announcement of economic sectors to resume operations 4 may 2020: economic sectors resume operations 10.77 1.9211** 3 jan – 5 june 2020 105 announcement of highest case in one day, 277 cases 8.44 5.1688*** 11 may – 9 june 2020 18 announcement of conditional mco until 9 june -6.31 19.1415*** 13 may – 9 june 2020 17 conditional mco (13 may – 9 june) -7.24 19.3970*** 7 june – 31 july 2020 39 announcement of recovery mco until 31 august 11.66 34.3054*** 10 june – 31 july 2020 39 recovery mco (10 june – 31 aug 2020) 11.95 36.5359*** note: significant at different levels: 1%, 5% and 10% denoted by *, **, ***, respectively. based on the results in table 1, it can be seen that the announcement of the detection of the first covid-19 cases in malaysia did not have a negative impact on caar. similarly, the announcement of the highest case in one day, also did not have a negative impact on caar. on the contrary, caar was found to be significant and positive on both dates. 126 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 however, the announcement of related covid-19 events such as the 100th and 1000th case mark, were found to have had a significant and negative caar. the negative caar could be attributed to several events happening simultaneously during the time: (1) the rapid increase in positive covid-19 cases i.e., new cases were in triple digits; (2) the announcement of the 100th and 1000th case mark were between a timespan of only two weeks; (3) the announcement of the first mco; and (4) the announcement of the first two covid-19 related deaths. these events were interpreted as negative news and absorbed by the malaysian stock market. we also calculated the impact of the announcement of the 5000th case mark on caar but it was found to be positive and insignificant. we attributed the results to stock market investors being affected by the announcement in a dissimilar manner to the first 100th and 1000th case mark. it could also be attributed to the investors being reassured by the government of malaysia of its actions in containing the spread of the virus; hence, an increase in covid-19 cases no longer had the same effect on the stock market as before. besides that, we also calculated the impact of mco announcements on caar. it was found that the nnouncement of the first and second mco had a significant and negative impact on caar. it reflected the absorption of information perceived by the market on the mco’s negative effects on malaysia’s various economic sectors. the mco required all non-essential public and private entities to cease operations with employees being arequired to work from home. moreover, in some industries, such as transportation, hotel and tourism, the mco caused their revenue to deteriorate significantly. however, subsequent announcements of the 3rd and 4th mco were found to be significantly positive which indicated that the economy had readjusted to the new normal brought by the mco and it could also be attributed to the various stimulus packages announced by the prime minister within the same period. it should also be noted that the announcement of the conditional mco (cmco) was found to have had a significantly negative effect on caar, indicating the stock returns’ negative reaction towards the news. on the other hand, the announcement of the recovery mco (rmco) was found to have had a significantly positive effect on caar. figure 1 depicts the caar from january 2020 to july 2020 for three events: (1) the announcement of the first covid-19 before the start of 127 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 mco 1; (2) beginning of mco 1 until the end of mco 4; and (3) the overall study period. besides that, the movement of klci was also added as a comparison. figure 1 klci and caar from january to july 2020. it can be seen from figure 1 that the movement of caar for malaysian stocks and klci as the main indicator of the stock market (bursa malaysia) are almost in tandem with each other, with some considerable differences. from 3 january to 3 march 2020 malaysian stocks were on an incline while klci was declining. however, from the beginning of the first mco announcement, both malaysian stocks and klci declined sharply. as mentioned earlier, this could be due to several events that could have connoted negative market performance such as the imposition of mco, the increase of covid-19 cases to triple digits, and the first deaths due to covid-19. however, caar rallied from the end of march, and since 7 april 2020, it could be seen that malaysian stocks were experiencing positive caar. there was a sudden decrease in caar from the middle of may 2020 which could be attributed to the introduction of conditional mco, which allowed most business sectors to resume operations yet movements were restricted such as curfews and social distancing. next, we examine the impact of the first mco to the fourth mco on short-term caar. the results are shown in table 2. 11 we also calculated the impact of the announcement of the 5000th case mark on caar but it was found to be positive and insignificant. we attributed the results to stock market investors being affected by the announcement in a dissimilar manner to the first 100th and 1000th case mark. it could also be attributed to the investors being reassured by the government of malaysia of its actions in containing the spread of the virus; hence, an increase in covid-19 cases no longer had the same effect on the stock market as before. besides that, we also calculated the impact of mco announcements on caar. it was found that the announcement of the first and second mco had a significant and negative impact on caar. it reflected the absorption of information perceived by the market on the mco’s negative effects on malaysia’s various economic sectors. the mco required all non-essential public and private entities to cease operations with employees being required to work from home. moreover, in some industries, such as transportation, hotel and tourism, the mco caused their revenue to deteriorate significantly. however, subsequent announcements of the 3rd and 4th mco were found to be significantly positive which indicated that the economy had readjusted to the new normal brought by the mco and it could also be attributed to the various stimulus packages announced by the prime minister within the same period. it should also be noted that the announcement of the conditional mco (cmco) was found to have had a significantly negative effect on caar, indicating the stock returns’ negative reaction towards the news. on the other hand, the announcement of the recovery mco (rmco) was found to have had a significantly positive effect on caar. figure 1 depicts the caar from january 2020 to july 2020 for three events: (1) the announcement of the first covid-19 before the start of mco 1; (2) beginning of mco 1 until the end of mco 4; and (3) the overall study period. besides that, the movement of klci was also added as a comparison. figure 1 klci and caar from january to july 2020. 1200 1250 1300 1350 1400 1450 1500 1550 1600 1650 -15 -10 -5 0 5 10 15 20 25 03/01/2020 03/02/2020 03/03/2020 03/04/2020 03/05/2020 03/06/2020 03/07/2020 overall study period ca ar 1st covid-19 case end of mco 4 caar mco 1 mco 4 caar klci 128 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 table 2 short-term impact of mco events on caar of malaysian stocks date obs. days / event window event caar (%) t-statistic 3 jan – 31 july 2020 144 overall data observations 20.11 9.9765*** 3 jan – 12 may 2020 73 first covid-19 case until end of mco 4 12.41 2.4862*** 18 mar – 12 may 2020 39 mco 1 – mco 4 16.49 0.7020 10 – 18 mar 2020 (-7, t0) mco 1 -6.76 -1.3656 18 – 26 mar 2020 (to, +7) -9.94 -8.0590*** 10 – 26 mar 2020 (-7, +7) -12.69 -3.9538*** 24 mar – 1 apr 2020 (-7, to) mco 2 -9.94 -8.0590*** 1 – 9 apr 2020 (to, +7) 6.55 -6.3426*** 24 mar – 9 apr 2020 (-7, +7) -1.78 -9.7708*** 7 – 15 apr 2020 (-7, to) mco 3 6.75 -0.4744 15 – 23 apr 2020 (to, +7) 1.71 7.8244*** 7 – 23 apr 2020 (-7, +7) 12.01 1.5694* 21 – 29 apr 2020 (-7, to) mco 4 4.33 7.9999*** 29 apr – 12 may 2020 (to, +7) 7.28 15.0091*** 21 apr – 12 may 2020 (-7, +7) 10.48 7.6132*** note: t0 is the announcement of the mco /extension of the mco *, **, *** denote statistical significance at 1%, 5% and 10%, respectively. the results in table 2 show the three types of analysis: (1) the overall period of this study; (2) from the detection of the first covid-19 case in malaysia; and (3) throughout the duration of all four mcos. malaysian stocks have experienced a significant and positive caar (figure 2) after mid-april which coincided with the flattening of the covid-19 curve. the caar were positive before and after the announcements of mco 3 and mco 4. this indicated that investors do not view news of mco extensions as negative. however, in the short-term, caar appeared to be negative throughout the period before and after the announcements of mco 1 and mco 2. thus, 129 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 the findings supported hypotheses 1 and 2 that at the early stages of the mcos, the reactions from the market were negative due to uncertainties but when the spread of the virus was reportedly under control, the returns were positive. figure 2 caar chart from mco 1 to mco 4 event windows descriptive statistics analysis table 3 descriptive statistics caar klci wti crude oil price no. of new cases no. of death cases mcos stimulus package dji mean 6.2162 1481.65 37.60 41.72 0.5764 0.2083 0.1597 25720.73 median 4.8201 1505.25 39.43 10.5 0 0 0 25868.12 std. dev. 7.4770 99.06 14.71 61.01 1.2932 0.4075 0.3676 2596.86 min. -11.05 1219.72 -37.63 0 0 0 0 18591.93 max. 20.10 1611.42 63.29 277 8 1 1 29551.42 n 144 144 144 144 144 144 144 144 13 coincided with the flattening of the covid-19 curve. the caar were positive before and after the announcements of mco 3 and mco 4. this indicated that investors do not view news of mco extensions as negative. however, in the short-term, caar appeared to be negative throughout the period before and after the announcements of mco 1 and mco 2. thus, the findings supported hypotheses 1 and 2 that at the early stages of the mcos, the reactions from the market were negative due to uncertainties but when the spread of the virus was reportedly under control, the returns were positive. figure 2 caar chart from mco 1 to mco 4 event windows descriptive statistics analysis table 3 presents the descriptive statistics of the variables. the observations were 144 days, from 3 january to 31 july 2020. the caar within this study period was between -11.05 percent and 20.10 percent with an average of 6.21. for klci, the lowest klci was 1219.72, and the highest was 1611.42, which was at the beginning of this study period. for crude oil prices, it was between -37.63 and 63.29 with an average of 36.60. meanwhile, for dji, the lowest was 18,591.93, and the highest was 29,551.42 with an average of 25,720.73. table 3 descriptive statistics caar klci wti crude oil price no. of new cases no. of death cases mcos stimulus package dji mean 6.2162 1481.65 37.60 41.72 0.5764 0.2083 0.1597 25720.73 -15 -10 -5 0 5 10 15 20 1 0 / 0 3 / 2 0 2 0 1 9 / 0 3 / 2 0 2 0 3 0 / 0 3 / 2 0 2 0 0 8 / 0 4 / 2 0 2 0 1 7 / 0 4 / 2 0 2 0 2 8 / 0 4 / 2 0 2 0 1 2 / 0 5 / 2 0 2 0 mco 1 mco 2 mco 3 mco 4 130 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 table 3 presents the descriptive statistics of the variables. the observations were 144 days, from 3 january to 31 july 2020. the caar within this study period was between -11.05 percent and 20.10 percent with an average of 6.21. for klci, the lowest klci was 1219.72, and the highest was 1611.42, which was at the beginning of this study period. for crude oil prices, it was between -37.63 and 63.29 with an average of 36.60. meanwhile, for dji, the lowest was 18,591.93, and the highest was 29,551.42 with an average of 25,720.73. regression results multiple regression analysis was conducted to identify variables which were most likely to affect both caar and klci. the regression analysis followed a base model consisting of only the control variables and an extended model to include the covid-19 variables (equation 5). in model 1 and model 2, caar was selected as the dependent variable while klci, the benchmark for market performance, crude oil price and the dow jones industrial average (dji) were selected as control variables. the number of new cases, number of death cases, dummy variable for mco and dummy variable for the injection of stimulus packages were selected as covid-19 specific variables. model 3 and model 4 used klci as the dependent variable. the purpose was to compare the effects of the covid-19 specific variables on caar as well as the benchmark index. we performed the regression analysis firstly with the control variables and later added all the covid-19 variables. from the first regression (table 4), model 1 indicated that all our control variables were significant in affecting caar where both the dji and crude oil price were significant and negative in affecting caar while klci was found to be significantly positive in affecting caar. in model 2, except for the mco dummy variable, all the other covid-19 variables had a significant and negative effect on caar. in model 3, we found that the control variables, dji and crude oil prices were significantly positive in affecting klci. in model 4, when covid-19 variables were added, it was found that only the announcements of new covid-19 cases were significant and negatively affected klci. all other covid-19 variables were insignificant in affecting klci. 131 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 ta bl e 4 m ul tip le r eg re ss io n: t he im pa ct o f c ov id -1 9 on c a a r a nd k lc i d ep en de nt v ar ia bl e: c a a r d ep en de nt v ar ia bl e: k l c i [1 ] [2 ] [3 ] [4 ] c oe f. tst at . c oe f. tst at . c oe f. tst at . c oe f. tst at . c o n st a n t -6 67 .9 47 [8 8. 18 0] -7 .5 75 ** * -4 61 .0 96 [9 5. 73 3] -4 .8 16 ** * -5 17 8. 12 4 [5 89 .5 25 ] -8 .7 84 ** * -4 35 5. 03 6 [6 76 .2 50 ] -6 .4 40 ** * l n k l c i 12 7. 98 7 [1 6. 29 6] 7. 85 4* ** 11 7. 81 0 [1 5. 38 6] 7. 65 7* ** l n d ji -2 2. 21 1 [1 1. 04 1] -2 .0 12 ** -3 4. 94 2 [1 0. 47 9] -3 .3 34 ** * 63 3. 89 6 [6 2. 58 7] 10 .1 28 ** * 55 8. 95 3 [6 9. 89 8] 7. 99 7* ** l n o il -9 .6 73 [2 .2 22 ] -4 .3 52 ** * -1 0. 07 0 [2 .1 70 ] -4 .6 40 ** * 63 .4 37 [1 6. 09 4] 3. 94 2* ** 49 .2 48 [1 7. 36 6] 2. 83 6* ** n e w c a se s -.0 23 [0 .0 12 ] -1 .8 47 * -.1 89 [.0 99 ] -1 .9 05 * d e a t h c a se s -1 .2 26 [0 .5 22 ] -2 .3 49 ** -1 .9 29 [4 .2 78 ] -.4 51 m c o s 1. 08 8 [1 .3 74 ] .7 92 -1 7. 99 4 [1 1. 13 8] -1 .6 15 st im u l u s pa c k a g e -3 .2 06 [1 .4 59 ] -2 .1 97 ** 4. 61 0 [1 1. 95 6] .3 86 r 2 0. 38 4 0. 49 9 0. 79 5 0. 80 6 a d j. r 2 0. 37 0 0. 47 3 0. 79 2 0. 79 7 f st a t. 28 .6 47 ** * 19 .1 02 ** * 27 1. 47 8* ** 93 .9 02 ** * n 14 4 14 4 14 4 14 4 n ot e: s ta nd ar d er ro rs a re in p ar en th es es . s ig ni fic an t a t d iff er en t l ev el s: 1 0% , 5 % , a nd 1 % d en ot ed b y *, * *, * ** , r es pe ct iv el y. 132 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 ta bl e 5 sp lit -w in do w p er io d: t he im pa ct o f c ov id -1 9 on c a a r d ep en de nt v ar ia bl e: c a a r 3 ja nu ar y– 6 a pr il 20 20 d ep en de nt v ar ia bl e: c a a r 7 a pr il– 31 j ul y 20 20 [5 ] [6 ] [7 ] [8 ] c oe f. tst at . c oe f. tst at . c oe f. tst at . c oe f. tst at . c o n st a n t -5 .5 59 [6 4. 86 0] -.0 86 11 2. 00 3 [8 8. 63 5] 1. 26 4 -1 81 .1 75 [1 33 .4 38 ] -1 .3 58 71 .8 47 [9 7. 72 8] .7 35 ln k l c i -4 5. 39 2 [1 3. 27 3] -3 .4 20 ** * -6 6. 13 4 [1 6. 52 5] -4 .0 02 ** * 3. 90 1 [2 1. 53 4] .1 81 -1 4. 93 8 [1 4. 84 1] -1 .0 07 ln d ji 29 .9 04 [6 .9 89 ] 4. 27 9* ** 32 .3 26 [6 .5 08 ] 4. 96 7* ** 14 .2 36 [2 3. 13 0] .6 15 2. 49 5 [1 5. 98 8] .1 56 ln o il 8. 72 4 [2 .0 42 ] 4. 27 2* ** 11 .1 73 [3 .0 05 ] 3. 71 8* ** 5. 85 9 [1 .9 79 ] 2. 96 1* ** 7. 50 2 [1 .5 22 ] 4. 93 0* ** n e w c a se s -.0 17 [.0 10 ] -1 .7 30 * -.0 31 [.0 07 ] -4 .5 61 ** * d e a t h c a se s .2 94 [.2 91 ] 1. 01 3 -.6 55 [.4 05 ] -1 .6 17 m c o s -.3 64 [1 .2 61 ] -.2 89 .7 83 [.8 44 ] .9 28 st im u l u s pa c k a g e 2. 52 7 [.7 86 ] 3. 21 5* ** -5 .8 22 [1 .0 07 ] -5 .7 82 ** * r 2 0. 79 7 0. 84 2 0. 41 2 0. 74 1 a d j. r 2 0. 78 7 0. 82 3 0. 38 8 0. 71 5 f st a t. 81 .1 79 ** * 44 .1 14 ** * 17 .0 41 ** * 28 .2 48 ** * n 66 66 78 78 n ot e: s ta nd ar d er ro rs a re in p ar en th es es . s ig ni fic an t a t d iff er en t l ev el s: 1 0% , 5 % , a nd 1 % d en ot ed b y *, * *, * ** , r es pe ct iv el y. 133 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 due to the difference between the results of caar and klci, we divided the period into two windows, from 3 january to 6 april 2020 as the first window period and from 7 april to 31 july 2020 as the second window period. the date 6 april was chosen as the end of the first window period because this was the turning point where caar was zero and from 7 april onwards, caar was positive (figure 1 and figure 2). the results are shown in table 5. when the period was split, it was found that all control variables were significant in affecting the caar in the first window period. however, in the second window period, it was found that only crude oil prices were significant in affecting caar. both the klci and dji were found to be insignificant in affecting caar. for covid-19 variables, it was found that in the first window period only the stimulus package was positively significant in affecting caar. the other variables were found to be insignificant. however, for the second window period, it was found that both new covid-19 cases and the stimulus package were negatively significant in affecting caar while the other covid-19 variables were found to be insignificant. overall, oil prices had a significant positive effect in both the window periods, indicating that oil prices were an important factor in determining market sentiment. the number of new cases reported had negatively affected the caar in both the window periods and was significant throughout all the models. it showed that market participants were very cautious about the uncertainty posed by the development of the pandemic. since malaysia’s political and civil liberties rights index showed a partly free with relatively low to moderate score (52/100), our results showed support of hypothesis 3 where the number of new and death cases are expected to impact stock returns due to overreaction and lack of confidence in the number of cases reported (erdem, 2020). the unprecedented political turmoil, which saw a sudden change in the government during the onset of the covid-19 outbreak, could possibly further justify the negative impact of the stock returns (bbc, 2020). for hypothesis 4, that the stimulus packages should positively impact caar was only supported in model 6, which was only during the early stage of the mco periods. for the subsequent period, the indication was negative. the positive impact during the early stage was consistent with the emh that the market perceived that the 134 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 stimulus would positively impact companies and the economy. at the later stage, all the information was already reflected in the stock market prices, the impact on the subsequent incremental assistance might not be obvious then, and thus the results could be distorted by other factors. conclusion this study examined the effects of covid-19 cases and the malaysian government’s measures on stock market returns. the study used market data extracted from thomson reuters datastream database to measure the abnormal returns. the final sample consisted of 760 companies listed on bursa malaysia. the event study method and capital asset pricing model were used to assess the pandemic’s shortterm impact on stock returns. the findings were consistent with alam et al. (2020) that the market reacted positively with significantly positive caar during the lockdown period. we found significant positive caar after mco 3 and mco 4 as the government has allowed almost all sectors to resume their business (anis, 1 may 2020). the relaxation of the mcos was in line with expectations that people were eager to go back to work as livelihoods were affected due to the two months’ lockdown. besides, the huge amount of stimulus packages has provided large liquidity in the market, and a low-interest rates regime may also lead to positive abnormal returns. the negative caar during the early stage of mco reflected the panic and uncertainty posed by the pandemic. as malaysia is relatively low in the freedom index, it was expected that people were inclined to comply with the regulations and government orders, and therefore the mco was successfully enforced. however, as contended by erdem (2020), less free countries are associated with autocratic states, leading to mismanagement during uncertain times and lower firm values and negatively affecting stock market performance. thus investors in less-free countries such as malaysia could lack trust; that the number of cases was being underreported, and therefore they overreacted to the announcements as compared to the freer countries. 135 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 consistent with the emh, the stimulus package was impactful at the beginning of the mco periods. it had a significant positive impact on the caar. the implication on emh can be linked to the level of freedom of the country in terms of information transparency as stock prices absorb information in relation to how the government manages the spread of covid-19 and how investors perceive the effect will be on business operations and performance. at the later stage, the effects on the subsequent incremental assistance may not be that impactful. the results could be distorted by other factors, such as the long period of mcos, which was extended to 31 december 2020 including political instability. the study was limited in its analysis to july 2020 when the mco was still enforced. future investigations should extend the research to cover the whole mco period and to include macroeconomic variables such as interest rates, liquidity, and the performance of specific sectors during the pandemic. acknowledgment this research received no specific grant from any funding agency in the public. references abdul rashid, h. r. (2020, april 24). m’sia managed to flatten covid-19 curve: noor hisham. malaysiakini. https://www. malaysiakini.com/news/522397 alam, m. n., alam, m. s., & chavali, k. (2020). stock market response during covid-19 lockdown period in india: an event study. journal of asian finance, economics and business, 7(7), 131–137. al-awadhi, a. m., alsaifi, k., al-awadhi, a., & alhammadi, s. (2020). death and contagious infectious diseases: impact of the covid-19 virus on stock market returns. journal of behavioral and experimental finance, 27, 100326. anis, m. n. (2020, march 25). mco period extended to april 14. the star. https://www.thestar.com.my/news/nation/2020/03/25/ mco-extended-to-april-14 136 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 anis, m. n. (2020, may 1). pm: almost all economic sectors, business activities can resume ops under conditional mco starting may 4. the star. https://www.thestar.com.my/news/ nation/2020/05/01/pm-almost-all-economic-sectors-businessactivities-can-resume-ops-under-conditional-mco-startingmay-4 aruna, p. (2020, january 29). coronavirus fears hit bursa malaysia. the star. https://www.thestar.com.my/business/businessnews/2020/01/29/coronavirus-fears-hit-bursa-malaysia ashraf, b. n. (2020). stock markets’ reaction to covid-19: cases or fatalities? research in international business and finance, 54: 101249. doi 10.1016/jribaf.2020.101249. baker, m., & wurgler, j. (2007). investor sentiment in the stock market. journal of economic perspectives, 21(2), 129–152. baker, s. r., bloom, n., davis steven j., j., k. k., sammon, m. c., & viratyosin, t. (2020). the unprecedented stock market impact of covid-19. nber working paper series, 26945, 1689– 1699. https://www.nber.org/papers/w26945 becker, k. g., finnerty, j. e., & friedman, j. (1995). economic news and equity market linkages between the u.s. and u.k. journal of banking and finance, 19, 1191–1210. bernama. (2020, june 27). covid-19: malaysia’s tourism industry hit with rm45 billion in losses. new straits times. https://www. nst.com.my/news/nation/2020/06/604012/covid-19-malaysiastourism-industry-hit-rm45-billion-losses bunyan, j. (2020, march 2020, merch 516). pm: malaysia under movement control order from wed until march 31, all shops closed except for essential services. the malay mail. https:// w w w. m a l a y m a i l . c o m / n e w s / m a l a y s i a / 2 0 2 0 / 0 3 / 1 6 / p m malaysia-in-lockdown-from-wed-until-march-31-all-shopsclosed-except-for/1847204 burch, t., emery, d. r., & fuerst, r. (2016). who moves markets in a sudden marketwide crisis? evidence from 9/11. journal of financial and quantitative analysis 51(2): 463–487. bursa malaysia. (2020). indices information. https://www. bursamalaysia.com/trade/trading_resources/listing_directory/ indices-profile?stock_code=0200i chen, c., liu, l., & zhao, n. (2020). fear sentiment, uncertainty, and bitcoin price dynamics: the case of covid-19. emerging markets finance and trade. 56(10), 2298-2309. 137 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 chen, m. h., jang, s. s., & kim, w. g. (2007). the impact of the sars outbreak on taiwanese hotel stock performance: an event-study approach. int. j. hosp. manag., 26(1), 200–212. daim, n. (2020, april 6). additional rm10bil for smes under prihatin economic stimulus package. new straits times. https:// www.nst.com.my/news/nation/2020/04/581812/additionalrm10bil-smes-under-prihatin-economic-stimulus-package duan, h., wang, s., & yang, c. (2020). coronavirus-limit the shortterm economic damage. nature, 578, 515. erdem, o. (2020). freedom and stock market performance during covid-19 outbreak. finance research letters, 36, 1–6. fama, e. f. (1970). efficient capital markets: a review of theory and empirical work. the journal of finance, 2, 383–417. freedom house. (2020, september 15). freedom in the world 2020. https://freedomhouse.org/country/malaysia/freedomworld/2020 head, j. (2020, march 16). how malaysia’s government collapsed in two years. british broadcasting corporation. https://www.bbc. com/news/world-asia-51716474 kaos, j., jr. (2020, april 15). covid-19: cases breach 5,000-mark, but fewest new cases since march 14. the star. https://www.thestar. com.my/news/nation/2020/04/15/covid-19-cases-breach-5000mark-but-fewest-new-cases-since-march-14 le, t. h., & chang, y. (2015). effects of oil price shocks on the stock market performance: do nature of shocks and economies matter? energy economics, 51, 261–274. lee, j.-w., & mckibbin, w. j. (2004). globalization and disease: the case of sars. asian economic papers, 3(1), 113–131. lei, x., & wisniewski, t. p. (2018). democracy and stock market returns. https://papers.ssrn.com/sol3/papers.cfm?abstract_ id=3198561 li, y. (2020, march 23). this was the fastest 30% sell-off ever, exceeding the pace of declines during the great depression. cnbc. https:// www.cnbc.com/2020/03/23/this-was-the-fastest-30percentstock-market-decline-ever.html liew j. t., & surendran, s. (2020, august 24). the edge. cover story: when will the stock market party end? the edge malaysia weekly. http://www.theedgemarkets.com/article/cover-stpry-when-willstock-market-party-end. loheswar, r. (2020, june 5). putrajaya announces rm35b penjana plan for economic regeneration. the malay mail. 138 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 https://www.malaymail.com/news/malaysia/2020/06/05/ putrajaya-announced-rm35b-penjana-plan-for-economicregeneration/1872767 loo, c. (2020, june 7). cmco ends june 9, recovery mco from june 10 to aug 31 (updated). the sun daily. https://www. thesundaily.my/home/cmco-ends-june-9-recovery-mco-fromjune-10-to-aug-31-updated-em2538754 marinč, r. i. m. (2016). geographic proximity of information to financial markets and impact on stock prices: evidence from the ebola outbreak. university of business and technology in kosono. https://doi.org/10.33107/ubt-ic.2016.19 ministry of health, malaysia. (2020). covid-19 malaysia. kementerian kesihatan malaysia. http://covid-19.moh.gov.my on august 21, 2020 mohammed, a. a., hussin, b. m., & ying, t. c. (2010). semi-strong form efficiency: market reaction to dividend and earnings announcements in malaysian stock exchange. research gate. https://www.researchgate.net/publication/228122538_semistrong_form_efficiency_market_reaction_to_dividend_ a n d _ e a r n i n g s _ a n n o u n c e m e n t s _ i n _ m a l a y s i a n _ s t o c k _ exchange narayan, p. k., phan, d. h. b., & liu, g. (2020). covid-19 lockdowns, stimulus packages, travel bans, and stock returns. finance research letters, 38, 101732. nippani, s., & washer, k. m. (2004). sars: a non-event for affected countries’ stock markets? applied financial economics 14(15), 1105–1110. nusair, s. a., & al-khasawneh, j. a. (2018). oil price shocks and stock market returns of the gcc countries: empirical evidence from quantile regression analysis. economic change and restructuring, 51(4), 339–372. ozili, p. k., & arun, t. (2020). spillover of covid-19: impact on the global economy, ssrn electronic journal. https://doi. org/10.2139/ssrn.3562570 povera, a. (2020, march 27). rm250 billion stimulus package announced. new straits times. https://www.nst.com.my/news/ nation/2020/03/578692/rm250-billion-stimulus-packageannounced 139 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 povera, a., & harun, h. n. (2020, april 23). mco phase 4 from april 29 to may 12. new straits times. https://www.nst.com.my/ news/nation/2020/04/586998/mco-phase-4-april-29-may-12 ramelli, s., & wagner, a. f. (2020). feverish stock price reactions to covid-19, swiss finance institute research paper no. 20 12. https://doi.org/10.2139/ssrn.3550274 rodzi, n. h. (2020, april 10). coronavirus: malaysia extends movement curbs by another two weeks to april 28. the straits times. https://www.straitstimes.com/asia/se-asia/coronavirusmalaysia-extends-movement-curbs-by-another-two-weeks-toapril-28 roslan, s. (2020, mac 22) hanya 12,500 bukannya 16,000 – jemaah tabligh. harian metro. https://www.hmetro.com.my/ mutakhir/2020/03/557146/hanya-12500-bukannya-16000jemaah-tabligh salisu, a. a., & isah, k. o. (2017). revisiting the oil price and stock market nexus: a nonlinear panel ardl approach. economic modelling, 66, 258–271. shah, a. u. m., safri, s. n. a., thevadas, r., noordin, n. k., rahman, a. a., sekawi, z., ideris, a., & sultan, m. t. h. (2020). covid-19 outbreak in malaysia: actions taken by the malaysian government. international journal of infectious diseases, 97, 108–116. siu, a., & wong, y. c. r. (2004). economic impact of sars: the case of hong kong. asian econ. pap. 3, 62–83. song, s. i., kueh, c.c., abdul rahman, r., & chu, e.y. (2011). wealth effects and financial performance of cross-border mergers and acquisitions in five east asian countries. american journal of finance and accounting, 2(3), 219–240. the economist. (2020). democracy index 2020: in sickness and in health? https://www.eiu.com/n/campaigns/democracy-index-2020/ the economist. (2020, march 7). sneezy money. https://espresso. economist.com/b0b9da81cf357c8884a06de8ef72bea0 world health organization. (2020). coronavirus disease 2019 (covid-19), situation report – 73. https://www.who.int/docs/ default-source/coronaviruse/situation-reports/20200402-sitrep73-covid-19.pdf?sfvrsn=5ae25bc7_6 worldometer. (2020). covid-19 coronavirus pandemic. https://www. worldometers.info/coronavirus/ 140 the international journal of banking and finance, vol. 16, number 2 (july) 2021, pp: 111–140 yazi, e., morni, f., & song, s. i. (2015). the effects of shariah compliance announcement towards share price changes in malaysia. journal of economics, business and management. 3(11), 1019–1023. yu, j., & yuan, y. (2011). investor sentiment and the mean-variance relation. journal of financial economics, 100(2), 367–381. zack, j. (2020, january 25). three chinese nationals test positive for coronavirus in m’sia, says health minister (updated). the star. https://www.thestar.com.my/news/nation/2020/01/25/threechinese-nationals-test-positive-for-coronavirus-in-m039siasays-health-minister zainul, i. f. (2020, march 28). stimulus boost for local equity market. the star. https://www.thestar.com.my/business/ business-news/2020/03/28/stimulus-boost-for-local-equitymarket how to cite this article: ho, c. m., & lin, w. y. (2020). comparison of the pass-through speed models of different markets: an empirical study of the markets of mainland china and taiwan. international journal of banking and finance, 15(1), 73-88. https://doi.org/10.32890/ijbf2020.15.1.9932 comparison of the pass-through speed models of different markets: an empirical study of the markets of mainland china and taiwan 1chi ming ho 2wu yih lin department of finance, southern taiwan university of science and technology 1corresponding author: z0q@stust.edu.tw; hcmccu@gmail.com 2ma380204@stust.edu.tw a r t i c l e i n f o article history: received 15 october 2019 revised 9 january 2020 accepted 15 january 2020 published 31 march 2020 keywords: bank competition, boone indicator, spread, pass-through speed model, error correction model. jel: g21, g28 a b s t r a c t this paper adopted the boone indicator, developed by boone et al. (2008) and van leuvensteijn et al. (2011; 2013), to investigate the influence of different pass-through spread models in the competition among banks in emerging markets. with the market share of banks as a dependent variable and marginal cost as an independent variable, this paper probed into the competition among banks regarding the loan market to determine whether competition on the loan interest rates of banks affected the pass-through of monetary policy-related interest rates. after analyzing approximately 5,657 entries of records of the banking industries in taiwan and mainland china, this paper reached three significant conclusions: 1) the boone indicator model pointed out that, competition in the banking market of mainland china the international journal of banking and finance, vol. 15. number 1, 2020: 73-88 73 74 the international journal of banking and finance, vol. 15, no 1, 2020 : 73-88 was more intense than that of taiwan; 2) empirical research based on the interest rate spread model indicated that the spread of mainland china was lower than that of taiwan; 3) the passthrough speed model implied that, the interest rate sensitivity of the market of mainland china was higher than that of the taiwan market. the above results indicate that the influence of monetary policy pass-through on the interest rate of the market in mainland china is faster than in taiwan. 1. introduction while many channels of monetary policies can shift policy effects to the economic system, the speed of the pass-through method plays the most vital role. the liberalization of the interest rate market will have a huge impact on the supervisory structure of commercial banks and social financing cost. some evidence (van leuvensteijn, bikker, van rixtel, & sorensen, 2011; van leuvensteijn, sorensen, bikker, & van rixtel, 2013) suggests that, the costs of financial intermediaries are extremely high (klein, 1971; monti, 1972). as the banking system does not have strong competitiveness, all such factors weaken the monetary pass-through method. in fact, competition is the key to the success of monetary policies. in a competitive market, the changes in policy-related interest rates can be quickly transferred to the banking system, such as the provision of new interest rates to customers or interactive changes in the interest rates of banks and markets. the implementation of monetary policies requires a stable market pass-through speed mechanism to achieve the goal of economic stability of a government. pass-through speed refers to the process of a government to adjust policy-based interest rates to change the interest rates of the monetary market. banks transfer the cost of such changes to their retail interest rates. most financial markets have forward-looking expectations, and even if the policyrelated interest rates of the government have not been changed, the retail interest rates of banks may have already reflected such changes. van leuvensteijn et al. (2013) asserted that, despite new financial reforms conducted by emerging countries, competition in the banking industry have declined over the past decade. as stipulated by the interest rate system, traditional indicators fail to correctly assess competition in the banking market. comparison of the pass-through speed models of different markets: an empirical 75 study of the markets of mainland china and taiwan: 73-88 comparatively speaking, if the boone indicator model was used to assess the banking business of mainland china, it would indicate that competition had increased (van leuvensteijn et al., 2013). this paper explores the influence of the loan interest rates of banks on the competition of the loan market, and utilizes the pass-through mechanism, through which bank competition drives monetary policies. the model of van leuvensteijn et al. (2013), and the error correction model (ecm) are applied to investigate the changes in interest rates of the two emerging economies of mainland china and taiwan. both china and taiwan markets are regional financial centers in asia, but there are still many comparable differences between the two in terms of financial supervision and economic policy. the chinese market adopts a fixed exchange rate system. as it is a planned economy, policies on domestic economic stabilization are prioritized. as a result, capital transfer is slow and regulated and interest rate market is subject to limited uncertainties. however, the taiwan market adopts a floating exchange rate. therefore, fiscal and monetary policies would change rapidly according to international situations. as a result, capital transfer is fast. however, there are gaps in policy implementation, along with increase in uncertainties. under the existing operations of two such emerging markets, how would the ultimate interest rate transmission change national income? there is no discussion on such a topic in the previous literature. in particular, the economic growth of china is critical to the prosperity of the global stock market, and therefore, it is a topic worth comparison and analysis. this model not only provides a consistent framework to compare the conclusions of the eu market, it also observes that, due to competition, market interest rates and bank interest rates are reflected in the monetary policies of central banks. 2. literature review the earliest literature on bank competition is from the 1970s. the monti-klein model of klein (1971) and monti (1972) assumed that the loan interest rates of banks had a certain degree of pricing power; lower demand elasticity led to higher intermediary profits (lower deposit interest rates led to higher loan interest rates) and weaker price competitiveness. maudos and fernandez de guevara (2004) believed that the growth in the power of the banking market (i.e. the decrease in competitive pressure) resulted in the increase in net interest margins. in addition, corvoisier and gropp (2002) identified the difference between the retail interest rates of banks and interest rates of the monetary market. the retail loan interest rates of specific bank products rose sharply in concentrated markets. leuvensteijn et al. (2013) analyzed the impact of loan interest rates of eight state-owned banks in the eurozone on the competition of the loan market from 1994 to 2004. they employed their innovative approach, meaning the boone indicator, to assess the extent of the competition. their empirical findings showed that, a competitive 76 the international journal of banking and finance, vol. 15, no 1, 2020 : 73-88 market hid significantly low interest rates. policy-related interest rates would change bank interest rates and formed pass-through speed. there were also findings in terms of the adjustment of loan interest rates via competition. cottarelli and kourelis (1994) and borio and fritz (1995) carried out cross-country analysis, and found that the monetary policy passthrough mechanism had significant effects in restraining competition. thus, in an environment with weaker competition, banks often found loan interest rates fairly tricky, which was due to the barriers to entry. mojon (2001) employed the index of deregulation, as established by gual (1999), to test the influence of the competition of the banking industry on the pass-through of loan interest rates of banks in the eurozone. mojon found that, when the interest rates of the monetary market declined, higher competition tended to exert pressure on banks, in order to adjust loan interest rates faster. scholnick (1996), heinemann and schu¨ler (2002), sander and kleimeier (2002), and kuan-min and thanh-binh nguyen (2010) found similar asymmetric pass-through effect. additionally, de bondt (2005) argued that the strong competition of other banks and the capital market could facilitate banks in the eurozone to accelerate the adjustment of interest rates. many country-specific studies found the same results; when market competition was weak, the entry of market interest rates into the pass-through mechanism of banks would also turn weak (heffernan,1997). de graeve et al. (2007) estimated the decisive factors of the pass-through speed behavior of belgian banks, and discovered that banks with stronger market power adopted less competitive pricing policies. lago-gonza´lez and salas-fuma´s (2005) conducted microeconomic analysis of spanish banks, and their evidence suggested that the combination of the price cost adjustment market power of banks led to price rigidity and asymmetric pass-through. bredin and o’reilly (2004) also showed similar results. kok sørensen and werner (2006) deemed that, pass-through and bank competition in different countries in the eurozone were different. moreover, gropp, kok sørensen, and lichtenberger (2007) focused on the eurozone, and held that the effect of pass-through provided positive impact on the level of competition in the banking sector. van leuvensteijn et al. (2013) employed new competition indicators, which were unlike previous measurement methods (boone, 2008; bikker & van leuvensteijn, 2008; van leuvensteijn et al., 2011). they argued that, in a highly competitive market, more efficient companies were likely to have greater market share. the basic idea of the boone indicator is market power hypothesis, that is, efficiency hypothesis. in a highly competitive market, more efficient companies are likely to have a greater market share. in other words, it assumes that market share, through marginal cost, determines if a bank is competitive in a competitive market. therefore, this paper further examines how an efficient market changes over time, how the changes in competition affect the loan costs of families and enterprises, and how policy-related interest rates are comparison of the pass-through speed models of different markets: an empirical 77 study of the markets of mainland china and taiwan: 73-88 passed through to the loan interest rates of banks. this paper intends to prove the following three hypotheses: 1) more competitive markets have smaller boone indicators; 2) in a highly competitive market, the spread is low; and 3) in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. 3. data and model 3.1 data sources and description the variable data used in the model of this paper were from the unbalance panel data of the taiwan economic journal (tej). the samples consisted of quarterly data from 2005 to 2016, and contained a total of 1,553 entries of records from 33 public banks. in addition to the tej, the data of mainland china came from the csmar solution journal. these samples were also from 2005 to 2016, and contained a total of 3,840 entries of the observed values of 96 banks, excluding missing values. the boone indictor was used to assess market competition, as well as the correlation between market share and marginal cost. the market share in this paper was the result of the sum of the quarterly loans of each bank for each year divided by the sum of the quarterly loans of an individual bank each year. marginal cost refers to the average cost and income ratio, as proposed by van leuvensteijn et al. (2011). the data of bank interest rates were the monthly average loan interest rates of the banks of taiwan and mainland china. in order to obtain sensitivity, market interest rates were the monthly average weighted overnight lending interest rates. a total of 264 entries of quarterly data from 2005 and 2016 in the markets of mainland china and taiwan were observed. 3.2 model development and research hypotheses 3.2.1 boone indicator model (3.1a) (3.1b) i: country. t: time. ms: market share of each bank. mc: marginal cost of each bank. ε: residuals. β: boone indicator. (α, β, γ, f, g): parameters. d: virtual variable. (-βt) means the negative parameter of βt, indicating efficient market share. hence, eq. (3.1a) is rewritten as eq. (3.1b) to describe the boone indicator of βt. theoretically, the taiwanese market is more competitive than that of mainland china, as the loan interest rates of taiwan during the research period 4 how policy-related interest rates are passed through to the loan interest rates of banks. this paper intends to prove the following three hypotheses: 1) more competitive markets have a smaller boone indicator; 2) in a highly competitive market, spread is low; and 3) in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. 3. data and model 3.1 data sources and description the variable data used in the model of this paper are from the unbalance panel data of the taiwan economic journal (tej). the samples consist of quarterly data from 2005 to 2016, and contain a total of 1,553 entries of records from 33 public banks. in addition to the tej, the data of mainland china also come from the csmar solution journal. these samples are also from 2005 to 2016, and contain a total of 3,840 entries of the observed values of 96 banks, excluding missing values. the boone indictor is used to assess market competition, as well as the correlation between market share and marginal cost. the market share in this paper is the result of the sum of the quarterly loans of each bank for each year divided by the sum of the quarterly loans of an individual bank each year. marginal cost refers to the average cost and income ratio, as proposed by van leuvensteijn et al. (2011). the data of bank interest rates are the monthly average loan interest rates of the banks of taiwan and mainland china. in order to obtain sensitivity, market interest rates are the monthly average weighted overnight lending interest rates. a total of 264 entries of quarterly data from 2005 and 2016 in the markets of mainland china and taiwan are observed. 3.2 model development and research hypotheses 3.2.1 boone indicator model ln(ms�,�)＝α + β� ln(mc�,�)+∑ γ����,…..,(���) d�+ε�,� (3.1a) ln(ms�,�)＝α + β� ln(mc�,�)+fd+gd ln(mc�,�) +ε�,� (3.1b) i: country. t: time. ms: market share of each bank. mc: marginal cost of each bank. ε: residuals. β: boone indicator. (α, β, γ, f, g): parameters. d: virtual variable. (-βt) means the negative parameter of βt, indicating efficient market share. hence, eq. (3.1a) is rewritten as eq. (3.1b) to describe the boone indicator of βt. theoretically, the taiwanese market is more competitive than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. the first hypothesis is proposed: h��: β�(𝑇) − β�(c) < 0; h��: β�(t) − β�(c) ≥ 0. βt (t) and βt (c) represent the boone indicators of taiwan and mainland china. 3.2.2 interest rate spread model if there is a co-integration relationship between bank interest rates and their corresponding market prices, ecm will be the most suitable model. this paper considers the interest rates of two products of relevant banks, and develops the following two model equations. 4 how policy-related interest rates are passed through to the loan interest rates of banks. this paper intends to prove the following three hypotheses: 1) more competitive markets have a smaller boone indicator; 2) in a highly competitive market, spread is low; and 3) in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. 3. data and model 3.1 data sources and description the variable data used in the model of this paper are from the unbalance panel data of the taiwan economic journal (tej). the samples consist of quarterly data from 2005 to 2016, and contain a total of 1,553 entries of records from 33 public banks. in addition to the tej, the data of mainland china also come from the csmar solution journal. these samples are also from 2005 to 2016, and contain a total of 3,840 entries of the observed values of 96 banks, excluding missing values. the boone indictor is used to assess market competition, as well as the correlation between market share and marginal cost. the market share in this paper is the result of the sum of the quarterly loans of each bank for each year divided by the sum of the quarterly loans of an individual bank each year. marginal cost refers to the average cost and income ratio, as proposed by van leuvensteijn et al. (2011). the data of bank interest rates are the monthly average loan interest rates of the banks of taiwan and mainland china. in order to obtain sensitivity, market interest rates are the monthly average weighted overnight lending interest rates. a total of 264 entries of quarterly data from 2005 and 2016 in the markets of mainland china and taiwan are observed. 3.2 model development and research hypotheses 3.2.1 boone indicator model ln(ms�,�)＝α + β� ln(mc�,�)+∑ γ����,…..,(���) d�+ε�,� (3.1a) ln(ms�,�)＝α + β� ln(mc�,�)+fd+gd ln(mc�,�) +ε�,� (3.1b) i: country. t: time. ms: market share of each bank. mc: marginal cost of each bank. ε: residuals. β: boone indicator. (α, β, γ, f, g): parameters. d: virtual variable. (-βt) means the negative parameter of βt, indicating efficient market share. hence, eq. (3.1a) is rewritten as eq. (3.1b) to describe the boone indicator of βt. theoretically, the taiwanese market is more competitive than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. the first hypothesis is proposed: h��: β�(𝑇) − β�(c) < 0; h��: β�(t) − β�(c) ≥ 0. βt (t) and βt (c) represent the boone indicators of taiwan and mainland china. 3.2.2 interest rate spread model if there is a co-integration relationship between bank interest rates and their corresponding market prices, ecm will be the most suitable model. this paper considers the interest rates of two products of relevant banks, and develops the following two model equations. 78 the international journal of banking and finance, vol. 15, no 1, 2020 : 73-88 were lower than those of mainland china. the first hypothesis is proposed: represent the boone indicators of taiwan and mainland china. 3.2.2 interest rate spread model if there is a co-integration relationship between bank interest rates and their corresponding market prices, ecm will be the most suitable model. this paper considers the interest rates of two products of relevant banks, and develops the following two model equations. (3.2a) (3.2b) (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii, t stands for country (i) and time (t) of the boone indicator. the market price of each country = long-term changes in interest rates, and = short-term changes in market interest rates). market interest rates multiply the boone indicator (bi). = boone long-term changes in market interest rates and = short-term changes in boone indicator). country virtual variable (di). the shortterm model of model (3.2b) includes error correction item in models (3.2a) and (3.2b) and various competitive effects [parameters of (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume to obtain interest rate spread model eq. (3.3). (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: 4 how policy-related interest rates are passed through to the loan interest rates of banks. this paper intends to prove the following three hypotheses: 1) more competitive markets have a smaller boone indicator; 2) in a highly competitive market, spread is low; and 3) in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. 3. data and model 3.1 data sources and description the variable data used in the model of this paper are from the unbalance panel data of the taiwan economic journal (tej). the samples consist of quarterly data from 2005 to 2016, and contain a total of 1,553 entries of records from 33 public banks. in addition to the tej, the data of mainland china also come from the csmar solution journal. these samples are also from 2005 to 2016, and contain a total of 3,840 entries of the observed values of 96 banks, excluding missing values. the boone indictor is used to assess market competition, as well as the correlation between market share and marginal cost. the market share in this paper is the result of the sum of the quarterly loans of each bank for each year divided by the sum of the quarterly loans of an individual bank each year. marginal cost refers to the average cost and income ratio, as proposed by van leuvensteijn et al. (2011). the data of bank interest rates are the monthly average loan interest rates of the banks of taiwan and mainland china. in order to obtain sensitivity, market interest rates are the monthly average weighted overnight lending interest rates. a total of 264 entries of quarterly data from 2005 and 2016 in the markets of mainland china and taiwan are observed. 3.2 model development and research hypotheses 3.2.1 boone indicator model ln(ms�,�)＝α + β� ln(mc�,�)+∑ γ����,…..,(���) d�+ε�,� (3.1a) ln(ms�,�)＝α + β� ln(mc�,�)+fd+gd ln(mc�,�) +ε�,� (3.1b) i: country. t: time. ms: market share of each bank. mc: marginal cost of each bank. ε: residuals. β: boone indicator. (α, β, γ, f, g): parameters. d: virtual variable. (-βt) means the negative parameter of βt, indicating efficient market share. hence, eq. (3.1a) is rewritten as eq. (3.1b) to describe the boone indicator of βt. theoretically, the taiwanese market is more competitive than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. the first hypothesis is proposed: h��: β�(𝑇) − β�(c) < 0; h��: β�(t) − β�(c) ≥ 0. βt (t) and βt (c) represent the boone indicators of taiwan and mainland china. 3.2.2 interest rate spread model if there is a co-integration relationship between bank interest rates and their corresponding market prices, ecm will be the most suitable model. this paper considers the interest rates of two products of relevant banks, and develops the following two model equations. 4 how policy-related interest rates are passed through to the loan interest rates of banks. this paper intends to prove the following three hypotheses: 1) more competitive markets have a smaller boone indicator; 2) in a highly competitive market, spread is low; and 3) in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. 3. data and model 3.1 data sources and description the variable data used in the model of this paper are from the unbalance panel data of the taiwan economic journal (tej). the samples consist of quarterly data from 2005 to 2016, and contain a total of 1,553 entries of records from 33 public banks. in addition to the tej, the data of mainland china also come from the csmar solution journal. these samples are also from 2005 to 2016, and contain a total of 3,840 entries of the observed values of 96 banks, excluding missing values. the boone indictor is used to assess market competition, as well as the correlation between market share and marginal cost. the market share in this paper is the result of the sum of the quarterly loans of each bank for each year divided by the sum of the quarterly loans of an individual bank each year. marginal cost refers to the average cost and income ratio, as proposed by van leuvensteijn et al. (2011). the data of bank interest rates are the monthly average loan interest rates of the banks of taiwan and mainland china. in order to obtain sensitivity, market interest rates are the monthly average weighted overnight lending interest rates. a total of 264 entries of quarterly data from 2005 and 2016 in the markets of mainland china and taiwan are observed. 3.2 model development and research hypotheses 3.2.1 boone indicator model ln(ms�,�)＝α + β� ln(mc�,�)+∑ γ����,…..,(���) d�+ε�,� (3.1a) ln(ms�,�)＝α + β� ln(mc�,�)+fd+gd ln(mc�,�) +ε�,� (3.1b) i: country. t: time. ms: market share of each bank. mc: marginal cost of each bank. ε: residuals. β: boone indicator. (α, β, γ, f, g): parameters. d: virtual variable. (-βt) means the negative parameter of βt, indicating efficient market share. hence, eq. (3.1a) is rewritten as eq. (3.1b) to describe the boone indicator of βt. theoretically, the taiwanese market is more competitive than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. the first hypothesis is proposed: h��: β�(𝑇) − β�(c) < 0; h��: β�(t) − β�(c) ≥ 0. βt (t) and βt (c) represent the boone indicators of taiwan and mainland china. 3.2.2 interest rate spread model if there is a co-integration relationship between bank interest rates and their corresponding market prices, ecm will be the most suitable model. this paper considers the interest rates of two products of relevant banks, and develops the following two model equations. 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank comparison of the pass-through speed models of different markets: an empirical 79 study of the markets of mainland china and taiwan: 73-88 means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows the descriptive statistics of all samples of taiwan, including bank interest rates (br), market interest rates (mr), boone indicator (bi), bi*mr, br-mr, market share (ms), and marginal cost (mc). table 2 displays the descriptive statistics of all samples of mainland china, including mean, standard deviation, median, maximum, and minimum. based on the observations of tables 1 and 2, the boone coefficient of taiwan was higher, indicating that it was less competitive, while that of mainland china was lower, indicating that it was more competitive. table 1. descriptive statistics of sample from taiwan taiwan market bank interest rates market interest rates boone indicator bi*mr spread market share marginal cost mean 3.2686 0.8138 0.1112 2.8986 2.4547 0.0436 0.4342 standard deviation 0.6316 0.7040 4.8837 5.7725 0.1974 0.0339 0.1270 median 2.8830 0.3880 -2.1324 -0.5852 2.4565 0.0269 0.4237 maximum 4.4430 2.4960 9.2208 18.4704 3.6770 0.1377 0.7819 minimum 2.5630 0.0970 -6.5300 -2.2572 1.5600 0.0041 0.1950 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 5 br�,� = ζbi�,� + η�mr�,� + θbi�,�mr�,� + k�d� + μ�,� (3.2a) ∆br�,� = λ�μ�,��� + μ�∆mr�,� + ξbi�,�∆mr�,� + v�,� (3.2b) br�,� = c + hbi�,�mr�,� + jd� + mdbi�,�mr�,� + μ�,� (3.2c) i means country (i =1,...,n). t means month (t = 1,...,t). model (3.2a) reflects the long-term equilibrium pass-through. model (3.2b) indicates the long-term equilibrium and short-term adjustment via bank interest rates. this study first discusses the long-term influence of model (3.2a), as the short-term influence of model (3.2b) is determined by the error of μ�,� of model (3.2a). br means bank interest rates (loan interest rates). ∆br refers to the monthly changes in bank interest rates. bii,t stands for country (i) and time (t) of boone indicator. the market price of each country (η�mr�,� = long-term changes in interest rates, and μ�∆mr�,� = short-term changes in market interest rates). market interest rates (mri,t) multiply the boone indicator (bi). (θbi�,�mr�,� = boone long-term changes in market interest rates and ξb�,�∆mr�,� = short-term changes in boone indicator). country virtual variable (di). the short-term model of model (3.2b) includes error correction item (λ�μ�,���). in models (3.2a) and (3.2b) and various competitive effects [parameters of ((ζ,ξ,θ) (panel)], eq. 3.3 can be deduced by eq. 3.2a. the pass-through effect of interest rate spread model eq. (3.2a) may be direct and complete. we assume η� = 1; θ = 0 to obtain interest rate spread model eq. (3.3). (br�,� − mr�,�) = c + δbi�,� + k�d� + ζd� ∗ bi�,� + μ�,� (3.3) the spread model is adopted. hypothesis 2: in a highly competitive market, the spread (bank interest rates market interest rates) is low. theoretically, the competitive spread of taiwan is smaller than that of mainland china, as the loan interest rates of taiwan during the research period were lower than those of mainland china. see hypothesis 2, as follows: h��: ζ(𝑇) − ζ(𝐶) > 0; h��: ζ(𝑇) − ζ(𝐶) ≤ 0. ζ(t) and ζ(c) means the spreads of taiwan and mainland china. 3.2.3 pass-through speed model in order to verify the speed equation, this paper rewrites eq. (3.2b) into eq. (3.2c); and mdbi_(i,t) mr_(i,t), virtual variable (d), boone indicator(bi), and market interest rates (mr) are multiplied together to obtain coefficient m, in order to determine when the competition of taiwan is greater than mainland china; if the pass-through speed is faster, while the p value of coefficient m is significant. hypothesis 3: in a highly competitive market, short-term bank interest rates are more sensitive than market interest rates. see hypothesis 3, as follows: h��: ξ(𝑇) − ξ(c) > 0; h��: ξ(𝑇) − ξ(c) ≤ 0. ξ(t) and ξ(c) represent the pass-through speeds of taiwan and mainland china, respectively. 4. analysis of empirical results 4.1 descriptive analysis table 1 shows descriptive statistics of all the samples of taiwan, including bank 80 the international journal of banking and finance, vol. 15, no 1, 2020 : 73-88 table 2. descriptive statistics of sample from mainland china mainland china market bank interest rates market interest rates boone indicator bi*mr spread market share marginal cost mean 5.8418 2.7550 -2.4491 -11.2121 3.0867 0.0272 0.4928 standard deviation 0.4865 1.2102 4.2567 15.8631 1.1326 0.0430 0.1547 median 6.0000 2.7700 -3.9605 -10.4027 3.0072 0.0108 0.4581 maximum 7.2900 5.6036 3.7828 6.7369 4.8172 0.1872 0.9710 minimum 4.3500 1.0600 -9.7 -54.3552 0.3963 0.0001 0.2163 4.2 heterogeneous variation robustness test in order to avoid model distortion when descriptive analysis, correlation coefficient, covariance, and variables are extremely similar, this study added the newey-west test, which could exclude self-correlation and heterogeneous variation to reach the best unbiased estimator. the first step included heterogeneous variation test bp (breusch) and the white test. table 3 shows that all three equations reject the null hypothesis (h0) of homogeneity and variation, implying that all three models have heterogeneous variations. table 3. test of heterogeneous variations of three models boone model speed model interest rate spread model bp test bp test bp test f statistics 22.6074 p value 0 1362.569 p value 0 283.075 p value 0 lm statistics 67.0279 p value x (3) 0 241.1677 p value x (2) 0 181.355 p value x (2) 0 white test white test white test f statistics 24.1204 p value 0 7.275 p value 0 20.201 p value 0 lm statistics 118.0894 p value x (5) 0 32.6216 p value x (5) 0 74.276 p value x (5) 0 comparison of the pass-through speed models of different markets: an empirical 81 study of the markets of mainland china and taiwan: 73-88 4.3 results of regression analysis 4.3.1 boone model of taiwan and mainland china in a competitive market, the boone indicator, meaning the marginal cost coefficient, is negative, as shown by g in eq. (3.1b). this paper supposes that, if taiwan entered a developing country earlier than mainland china, the loan interest rates of taiwan during the research period would be lower than those of mainland china. hence, the taiwanese market should be more competitive. however, the test results (table 4) of this paper rejects h0. the results show that mainland china is more competitive than taiwan in terms of the loan market. therefore, h10 is rejected, implying that lower intermediary profits lead to stronger price competitiveness, which is similar to klein (1971). table 4. regression analysis results of boone model dependent variable: market share (general regression) independent variable correlation coefficient standard deviation t statistics p value intercept 0.004083 0.000789 5.171947 0.0001 marginal cost 0.021126 0.002092 10.0994 0.0005 dummy variable 0.001285 0.002998 0.428676 0.6682 dummy variable multiplies marginal cost 0.039935 0.006841 5.837611 0.006 r-squared 0.1218 in order to prevent model distortion, this paper also tested the neweywest model, as shown in table 5. the results also rejected h0, that taiwan is more competitive than mainland china. as mainland china is more competitive than taiwan, thus, h10 is also rejected. 82 the international journal of banking and finance, vol. 15, no 1, 2020 : 73-88 table 5. results of newey-west regression of boone model dependent variable: market share (newey-west) independent variable correlation coefficient standard deviation t statistics p value intercept 0.004083 0.001538 2.654738 0.008 marginal cost 0.021126 0.004543 4.649663 0.0005 dummy variable 0.001285 0.004219 0.304538 0.7607 dummy variable multiplies marginal cost 0.039935 0.010863 3.676347 0.0002 r-squared 0.1218 4.3.2 interest rate spread models of taiwan and mainland china table 6 demonstrates the general regression of the speed model. the boone indicator points out that, in a more competitive market, the pass-through between the loan interest rates of banks and market interest rates is faster. the virtual variable, the boone indicator, and market interest rates are multiplied together. coefficient m in eq. (3.2c) is used. this paper supposes that, if taiwan entered a developing country earlier than mainland china, the loan interest rates of taiwan during the research period would be lower than those of mainland china. hence, taiwan should be more competitive than mainland china; and the pass-through speed of the former should be faster than the latter. the results show that, a more competitive loan market has more complete pass-through speed. the pass-through speed of mainland china is faster than that of taiwan. thus, h20 is rejected. table 6. regression analysis results of the speed model dependent variable: bank interest rates (general regression) independent variable correlation coefficient standard deviation t statistics p value intercept 5.812833 0.058696 99.03251 0.0007 (continued) comparison of the pass-through speed models of different markets: an empirical 83 study of the markets of mainland china and taiwan: 73-88 dependent variable: bank interest rates (general regression) independent variable correlation coefficient standard deviation t statistics p value boone indicator multiplies market interest rates -0.017685 0.004437 -3.986113 0.0002 dummy variable -2.799504 0.079158 -35.3659 0.0005 dummy variable, boone indicator, and market interest rates multiply together 0.105759 0.009365 11.29316 0.002 r-squared 0.8692 table 7 shows the speed model newey-west adjustment, and indicates that a more competitive loan market has more complete pass-through speed. the passthrough speed of mainland china is faster than that of taiwan, thus, h20 is also rejected. table 7. newey-west regression analysis results of the speed model dependent variable: bank interest rates (newey-west) independent variable correlation coefficient standard deviation t statistics p value intercept item 5.812833 0.140954 41.2393 0.0007 boone indicator multiplies market interest rates -0.017685 0.007965 -2.220332 0.0273 virtual variable -2.799504 0.156414 -17.89808 0.0005 virtual variable, boone indicator, and market interest rates multiply together 0.105759 0.010366 10.20277 0.002 r-squared 0.8692 84 the international journal of banking and finance, vol. 15, no 1, 2020 : 73-88 4.3.3 interest rate indicator pass-through models of taiwan and mainland china the boone indicator refers to the spread of loan interest rates of banks minus the market interest rates in a competitive market. the boone indicator multiplies virtual variables to reach a positive coefficient, such as the ζ value in eq. (3.3). this paper supposes that if taiwan entered a developing country earlier than mainland china, the loan interest rates of taiwan during the research period would be lower than those of mainland china. hence, the spread of taiwan should be smaller than that of mainland china; however, table 8 shows that the spread (loan interest rates of banks minus market interest rates in the loan market) of mainland china is smaller than that of taiwan, thus, h30 is rejected. table 8. regression analysis results of the spread model dependent variable: loan interest rates of banks minus market interest rates (general regression) independent variable correlation coefficient standard deviation t statistics p value intercept 3.599982 0.061925 58.13471 0.0001 boone indicator 0.195465 0.017029 11.47845 0.000 dummy variable -1.143672 0.080582 -14.19264 0.0001 dummy variable multiplies boone indicator -0.209357 0.020056 -10.43853 0.003 r-squared 0.4796 in order to prevent model distortion, newey-west regression is tested. table 9 shows the test of the spread model. the results reject the null hypothesis that taiwan is more competitive than mainland china, and that the former's spread is smaller than that of the latter. the market in mainland china is more competitive than taiwan, thus, the former's spread is smaller than that of the latter, and therefore h30 is rejected. comparison of the pass-through speed models of different markets: an empirical 85 study of the markets of mainland china and taiwan: 73-88 table 9. newey-west regression analysis results of the spread model dependent variable: loan interest rates of banks minus market interest rates (newey-west) independent variable correlation coefficient standard deviation t statistics p value intercept 3.599982 0.127965 28.13257 0.0007 boone indicator 0.195465 0.03238 6.036665 0.0002 dummy variable -1.143672 0.131338 -8.707852 0.0005 dummy variable multiplies boone indicator -0.209357 0.032782 -6.386281 0.002 r-squared 0.4796 4.4 empirical results based on the robustness analysis, this paper reached the following conclusions. eq. (3.1b) in the boone model of hypothesis 1 assumes that taiwan is more competitive than mainland china. however, the results showed the opposite. eq. (3.3) in the spread model of hypothesis 2 assumes that the spread of the loan market of taiwan is smaller than that of mainland china. however, the result indicated the opposite. eq. (3.2c) in the speed model of hypothesis 3 assumes that the pass-through speed of taiwan is faster than that of mainland china. however, the result revealed the opposite. 5. conclusion and suggestion china’s loan market is more competitive which implies a lower spread; in other words, competition may lower interest rates. however, bank competition may lead to more dangerous behavior, resulting in financial turmoil. there is no recent literature that explored in depth the correlation between bank competition and interest rates in emerging markets. some literature adopted the lerner indicator, a traditional indicator and panzar-rosse h statistics or hhi concentration, which are limited by the regulations of the interest rate system, and thus, fail to effectively assess bank competition. van leuvensteijn et al. (2013) assessed the competition among banks in china, and deemed that the boone indicator was the best way to assess such competition. however, the boone indicator is a relatively 86 the international journal of banking and finance, vol. 15, no 1, 2020 : 73-88 immature tool, and is not supported by literature. therefore, this paper adopted the boone indicator to compare bank competition in the markets of mainland china and taiwan, and reached three important empirical conclusions, which can be compared with leuvensteijn, bikker, adrian, rixtel and sorensen (2013), cottarelli and kourelis (1994), and borio and fritz (1995). during the research period, the loan interest rates of taiwan were lower than those of mainland china; however, the deregulation of taiwan was earlier than that of mainland china. theoretically, the loan market of taiwan should be more competitive than that of mainland china. however, the empirical results indicated that, the competition among banks in the loan market of mainland china is more intense than that of taiwan. in addition, china has smaller spread and faster pass-through speed than taiwan, which is consistent with klein (1971). the results of this paper could serve as complementary reference for literature on the boone indicator. overall, after comparing the empirical analysis results with the mundellflemming model, we can see that china adopts a fixed exchange rate system featuring high interest rate sensitivity, fast regulation and small implicit lm slope. when the people’s bank of china adopts an expansionary monetary policy, global imbalances would be adjusted through foreign exchange reserves, without impacting on long-term balanced national income. in contrast, the taiwan market features low interest rate sensitivity and slow regulation. as the taiwan market adopts a floating exchange rate, global imbalances as a result of expansionary monetary policy would be improved through devaluation, leading to significant increase in long-term balanced gdp. even so, based on policy dynamics of the mundell-flemming model, national income growth is possible for both mainland china and the taiwan market, if fiscal policy is followed. references bikker, j. a., & van leuvensteijn, m. (2008). competition and efficiency in the dutch life insurance industry. applied economics, 40, 2063–84. d o i . org/10.1080/00036840600949298 boone, j. (2008). a new way to measure competition. the economic journal, 118, 1245–1261. doi.org/10.1111/j.1468-0297.2008.02168.x| borio, c., & fritz, w. (1995). the response of short-term bank lending rates to policy rates: a cross-country perspective. bis working paper no. 27. bredin, d., & o’reilly, g. (2004). an analysis of the transmission mechanism of monetary policy in ireland. applied economics, 36, 49–58. d o i abs/10.1080/0003684042000177198 cottarelli, c., ferri, g., & generale, a. (1995). bank lending rates and financial structure in italy: a case study. imf working paper no. 38. cottarelli, c., & kourelis, a. (1994). financial structure, bank lending rates and the transmission of monetary policy. imf staff paper, 42, 670–700. comparison of the pass-through speed models of different markets: an empirical 87 study of the markets of mainland china and taiwan: 73-88 de bondt, g. j. (2005). interest rate pass-through: empirical results for the euro area. german economic review, 6, 37–78. doi.org/10.1111/j.14656485.2005.00121.x| de graeve, f., de jonghe, o., & vander vennet, r. (2007). competition, transmission and bank pricing policies: evidence from belgian loan and deposit markets. journal of banking and finance, 31, 259–78. gual, j. (1999). deregulation, integration and market structure in european banking. journal of the japanese and international economies, 13, 372–96. doi.org/10.1006/jjie.1999.0428 gropp, r., kok sørensen, c., & lichtenberger, j. (2007). the dynamics of bank spreads and financial structure. european central bank working paper series no. 714. heinemann, f., & schu¨ ler, m. (2002). integration benefits on eu retail credit markets – evidence from interest rate pass-through, mimeo zew. heffernan, s. a. (1997). modelling british interest rate adjustment: an errorcorrection approach. economica, 64, 211–31. kuan-min, w., & thanh-binh nguyen, t. (2010). asymmetric passthrough and risk of interest rates: an empirical exploration of taiwan and hong kong. applied economics, 42, 659–70. doi: 10.1080/00036840701704444 kok sørensen, c., & werner, t. (2006). bank interest rate pass-through in the euro area: a cross-country comparison. ecb working paper series no. 580. klein, m. (1971). a theory of the banking firm. journal of money, credit, and banking, 3, 205–18. lago-gonza´ lez, r., & salas-fuma´ s, v. (2005). market power and bank interest rate adjustments. banco de espan˜ a working paper no. 0539. maudos, j., & fernández de guevara, j. (2004). factors explaining the interest margin in the banking sectors of the european union. journal of banking & finance, 28(9), 2259–2281.doi.org/10.1016/j.jbankfin.2003.09.004 mojon, b. (2001), financial structure and the interest rate channel of ecb monetary policy. economie et provision, 147, 89–115. monti, m. (1972). deposit, credit, and interest rate determination under alternative bank objectives, in g. p. sezgo & k. shell (eds.), mathematical methods in investment and finance. amesterdam: north holland. scholnick, b. (1996). asymmetric adjustment of commercial bank interest rates: evidence from malaysia and singapore. journal of international money and finance, 15, 485–96. sander, h., & kleimeier, s. (2002). asymmetric adjustment of commercial bank interest rates in the euro area: an empirical investigation into interest rate pass through. kredit und kapital, 35, 161–92. 88 the international journal of banking and finance, vol. 15, no 1, 2020 : 73-88 van leuvensteijn, m., j. a. bikker, adrian a. r. j. m. van rixtel., & c. k. sorensen (2011). a new approach to measuring competition in the loan markets of the euro area. applied economics, 43, 3155-3167. doi. org/10.1080/00036840903493234 van leuvensteijn, m., c. k. sorensen, j. a. bikker, adrian a. r. j. m., & van rixtel (2013), impact of bank competition on the interest rate passthrough in the euro area. applied economics, 45, 1359-1380. doi.org/10. 1080/00036846.2011.617697 van leuvensteijn, m., bing, x., & adrian van rixtel (2013). measuring bank competition in china: a comparison of new versus conventional approaches applied to loan markets. monetary and economic department, bis working papers no 422. weth, m. a. (2002). the pass-through from market interest rates to bank lending rates in germany. deutsche bundesbank discussion paper no. 11/02. dx.doi.org/10.2139/ssrn.320112. 27 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: zahra, m., kee, d. m. h., teh, s. s., & paul, g. d. (2022). psychological capital impact on extra role behaviour via work engagement: evidence from the pakistani banking sector. international journal of banking and finance, 17(1), 27-52. https:// doi.org/10.32890/ ijbf2022.17.1.2 psychological capital impact on extra role behaviour via work engagement: evidence from the pakistani banking sector 1munazza zahra, 2daisy kee mui hung, 3teh shan shan & 4gadi dung paul school of management universiti sains malaysia, penang, malaysia 1corresponding author: munazza.zahra89@gmail.com received: 30/3/2021 revised: 17/4/2021 accepted: 4/5/2021 published: 2/12/2021 abstract the banking sector in pakistan has always been one of the most significant economic sectors of the country. the purpose of this paper has been to identify the factors urging bank employees to volunteer themselves for activities outside of their formal job obligations. a total of 315 responses were gathered from the employees of the four private commercial banks in pakistan. the researchers tested the relationship between the four facets of psychological capital (i.e., efficacy, resilience, optimism, and hope) and the two dimensions of extra role behaviour (individual and organisational). the researchers used the spss version 23 to analyse the demographic profiles of the respondents, and the smartpls version 3 to test the hypotheses formed 28 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 using the structural equation modelling technique. the researchers also investigated the mechanism through which psychological impacts extra role behaviour. under the job demands-resources (jdr) model, work engagement is employed as a mediating variable. the results of the study revealed a significant positive relationship between efficacy, optimism, and resilience with extra role behaviours. however, hope remained insignificant to extra role behaviours. the study outcomes also supported the major hypotheses on the mediating role of work engagement in the relationship between the psychological capital and extra role behaviour. recommendations for future studies and the limitations of the present study were aslo discussed. keywords: psychological capital, extra role behaviour, work engagement, bank employees, jd-r model. jel classification: m0. introduction the world is currently looking for high-performance organisations to improve the global economy. banks are service-oriented organisations, and they are the prime source of financial support for all other economic sectors (i.e., manufacturing and agriculture) of the country (waseem et al., 2020). the banking sector of pakistan has always been one of the most important sectors of the economy (waseem et al., 2020). its significance as the lifeblood of economic growth, especially in the collection of deposits and the provision of loans to states and individuals, households and industries has been inarguable (waseem et al., 2020). in the era of globalization and ultimate competition, banks have persisted and prospered on the optimal exploitation of their employees (siddiqui, 2019). consequently, it is essential for the banks to recognize the factors causing employees to volunteer themselves for activities outside of their formal job requirements. this type of optional behaviour of employees is referred to as organisational citizenship behaviour (ocb), or extra role behaviour. such behaviours are of vital importance, especially in the era which has introduced technological robustness and a competitive environment in the banking sector of 29 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 pakistan (siddiqui, 2019). thus the ocb is one of the important factors in efforts to increase the overall effectiveness of banks. the term organisation citizenship behaviour was first introduced by organ (1988) as “an individual behaviour which is not rewarded by a formal reward system, but that, when combined with the same behaviour in a group, results in effectiveness.” organ’s idea of the ocb was based upon the chester barnard’s concept (barnard, 1938) a concept which stated it was one’s will to cooperate, but in addition to the influence of social behaviour (chaitanya & tripathi, 2001). organ’s perspective was that such behaviours could improve the overall performance of the employees. organ explained that the ocb was not an obligatory requirement of an employee, instead it was a matter of a person’s preference. williams and anderson (1991) provided a more detailed definition, pointing out the ocb was the set of employee behaviours that directly or indirectly contributed to organisational goal accomplishment. williams and anderson (1991) conceptualized the ocb as having two dimensions; individual-directed organisational citizenship behaviour (ocbi) as the behaviours that immediately benefitted specific individuals and, in this way, indirectly contributed to the organisational success, and organisation-directed organisational citizenship behaviour (ocbo) were behaviours that benefitted the organisation in general, for example, the employee would give advance notice when unable to come to work, or adhered to informal rules devised to maintain order. previous studies have shown a positive relationship between positive personal resources and the ocbs (gupta et al., 2017). among such resources was the recently evolved idea of psychological capital (psycap), which has become a catchphrase in the field of positive organizational behaviour. luthans et al. (2015, p. 2) described the psycap as “an individual’s positive state of development that is defined by; (1) having faith (efficacy) in taking on and making the required effort to excel in difficult tasks; (2) making a positive allocation (optimism) of progress, both now and in the future; (3) determined towards goals and, where required, redirecting the path to success; (4) when beset by problems and adversity, sustaining, and bouncing back and even beyond (resilience) to attain success. when employees were hopeful, confident, optimistic, and resilient, they would achieve more 30 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 (ribeiro et al., 2021). these positive personal resources would help in developing the organizational citizenship behaviour for organizations, as well as individuals (chamisa et al., 2020). organisational psychologists believed that the ocb was shown more by engaged employees as compared to unengaged employees (thakre & mathew, 2020). this was because the engaged employees concurrently spent their intellectual, physical, emotional, and psychological energies in their job (bakker & albrecht, 2018). likewise, the psychological capital (psycap) has been regarded as a combination of important personal resources (such as hope, optimism, resilience, and efficacy) that would enable employees to accomplish their desired objectives. this accomplishment would create a feeling of engagement and stimulate positivity in employees, which in turn, would persuade them to become involved in extra-role behaviours. in the context of pakistan, several studies have examined the relationship between the psychological capital and task performance or in role behaviour (latif, 2017), counterproductive behaviour (imran & shahnawaz, 2020) and project success (sarwar et al., 2017). however, scant attention has been paid to the need for more studies investigating the mediating role of work engagement between the psycap and extra role behaviours. additionally, there is also the need to study the effect of work engagement on the two facets of the ocb, which comprised the individual-directed organisation citizenship behaviour (ocbi) and the organisation-directed organisation citizenship behaviour (ocbo). these studies are especially important for developing nations such as pakistan, where the banking sector is becoming more competitive, and in light of this development, bank management expects its employees to be involved in extra role behaviours to improve the banking sector. furthermore, the present study has used a back translation process to translate the psychological capital questionnaire (pcq-12) into urdu to help the respondents have a better understanding of the questionnaire items. literature review the psycap is the cutting-edge need for handling multiple behavioural concerns of employees (alola & atsa’am, 2020). it is based on the ‘positive organisational behaviour’ (pob) theory, which has 31 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 focused on building up the individual personal resources to reach their maximum capacity in order to perform well and prosper in organisational settings (luthans & youssef-morgan, 2017). the psycap is accepted as a blend of four distinct positive personal resources outlined in the hero model. according to luthans et al. (2007), these personal resources were hope, efficacy, resilience and optimism (collectively called hero). these four positive personal resources have beeb viewed as ‘state-like’, which implied that they could be improved instead of being attributes that were viewed as fixed (luthans et al., 2007). the psycap is a mixture of having an elevated level of each resource and that together the resources are seen as having a more significant impact on performance. besides, newman et al. (2014) have conducted extensive research on the psycap, centred on dependent variables, and the results have shown that the psycap was one of the variables that could increase employee efficiency. moreover, it is significant to distinguish the elements driving employees to volunteer themselves for tasks outside their formal job obligations. such behaviours, though not included in the job descriptions, have become vital in the context of enhancing employee performance and efficiency (podsakoff et al., 2018). thus, the present study has aimed at assessing the role of the four facets of the psycap in extra role behaviours, and investigating the mediating role of work engagement in psycap-ocb relationships. efficacy and the ocbs efficacy referred to an individual’s belief in their capacity to execute behaviours necessary to produce specific performance attainments (bandura, 1997). in the studies of the relationship between efficacy and the ocb by pradhan et al. (2020) and beauregard (2012), found a consistent and strong relationship between efficacy and performance. according to cohen and abedallah (2015), efficacy meant that individuals wanted to do more than what was expected, as they were confident of their abilities. this efficacy would lead them to extra role behaviour. based on the findings of these studies, the present study has proposed the following hypotheses: h1: efficacy has a positive relationship with the ocbi. h2: efficacy has a positive relationship with the ocbo. 32 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 hope and the ocbs according to snyder (1995), hope should be seen as a positive cognitive state based on a sense of successful goal-directed determination and planning to meet these goals. hope has proven to be effective in explaining the ocbi and the ocbo in several studies (jung & yoon, 2015; ozyilmaz, 2020). akhras and alam (2020) conducted a case study on business work units in the middle east and the north african area. they concluded that hope was the enhancer of the ocbs in employees. likewise, the hope theory by snyder (2002) postulated that employees with high hopes were determined to engage physically, cognitively, and emotionally in their work, thus enhancing their ocbs at their work place. jung and yoon (2015) suggested that one of the positive constructs of the psychological capital, i.e., hope, positively affected the ocbs. thus, in light of the findings in these prvious studies, the present study has come up with the following hypotheses: h3: hope has a positive relationship with the ocbi. h4: hope has a positive relationship with the ocbo. optimism and the ocbs the term optimism “refers to a positive frame of mind and means that a person takes the view of expecting the best outcome from any given situation. an optimistic person is one who sees the glass as ‘half-full’ instead of ‘half-empty’” (seligman, 2006, p. 05). optimism has been frequently related to the ocbs in various researches (ugwu & igbende 2017; jabbar et al., 2019; hough et al., 2020). ugwu and igbende (2017) documented that management should assess optimism during the selection process and hire only highly optimistic workers, by screening out the pessimistic ones. smithikrai (2020) acknowledged optimism as the only possible option that could deliver the best result in the employee’s citizenship behaviour. smithikrai (2020) further suggested to examine the relationship in other work settings such as in banks. thus, the proposed hypotheses for the present study were as follows: h5: optimism has a positive relationship with the ocbi. h6: optimism has a positive relationship with the ocbo. 33 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 resilience and the ocbs the american psychological association (2014, para. 4) has defined resilience as “the process of adapting well in the face of adversity, trauma, tragedy, threats or even significant sources of stress.” kašpárkováet al. (2018) also found that resilience and perceived performance were positively linked. in their study,data was collected from 360 workers in the czech republic. paul et al. (2019) examined in an integrative model, the role of resilience on organizational citizenship behaviours and found that there was a significant impact of resilience on the ocbs. in another study conducted by paul et al. (2016), using a sample of 345employees working in the manufacturing industries of uttarakhand and himachal pradesh in india, the researchers found empirical evidence on the positive relationship between resilience and the ocbs. in light of the findings in these previous studies, the present study has proposed the following hypotheses: h7: resilience has a positive relationship with the ocbi. h8: resilience has a positive relationship with the ocbo. the point of highlighting some of the main findings in this brief background of the study variables is to show the research gap in the existing studies. these studies lacked insights into the mechanism found in the positive constructs and how it is related to extra role behaviour. furthermore, the majority of the studies related to the ocbs were mostly conducted in the western context. furthermore, the studies conducted in the context of pakistan had used the english version of the psycap tool. however, it has been recommended by researchers to use the translated version of the tool when conducting a cross-sectional research (tyupa, 2011; son, 2018). to fill these gaps, the present study has brought together a potential interceding variable of work engagement in its investigation into the relationship between the four facets of the psycap and the two dimensions of extra role behaviour in context of the banking sector in pakistan. additionally, a back translation process has been implemented to translate the tool into the local language, i.e., urdu, in order to facilitate respondent comprehension of the items in the study questionnaire. 34 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 work engagement as mediator work engagement has been defined as a positive, fulfilling, affectivemotivational state of work-related well-being (bakker et al., 2007). this study has utilized the job demands resources (jd-r) model (bakker et al., 2007) to create and test a procedural model that has linked the psycap to work engagement and performance. work engagement has become a central idea in work on organisational behaviour (wirtz et al., 2017). studies showed that engaged workers were more energetic and satisfied in their work, and they considered themselves to be able to deal with in-role and extra-role behaviours (iqbal et., 2017). work engagement was seen as a positive behaviour or outlook at work that would prompt positive work-related results (bakker & albrecht, 2018). work engagement has been empirically tested with a focus on an assortment of its impacts on employees. at the individual level, it has been identified with the psychological and physical well-being of employees (joo et al., 2017). previous studies showed that employees with more elevated work engagement levels had less level of dissatisfaction and higher levels of performance (leijten et al., 2015; shimazu et al., 2016). work engagement has also been related to job satisfaction and turnover intention (ofei-dodoo et al., 2020). there are many reasons to choose work engagement as a mediator. the first and the foremost is that it is one of the important predictors of individual and team outcomes. as engaged employees are focused and dedicated in their work often show better ocbs. secondly, engaged workers are open to gaining experience, they like to support their colleagues and enjoy participating in organisational meetings as opportunities to provide their valuable inputs. finally, work engagement becomes even more critical when employees arefaced with increasingly greater job demands (competition in the banking sector), but have adequate personal resources (psycap) that are available to deal with these job demands (jdr model). moreover, the research on engagement has often been built on the jdr model (bakker & albrecht, 2018). these features have underscored the significance of work engagement at the workplace. in this regard, the present research was aimed at investigating work engagement as a mediating variable that could lead to the ocbs. thus, the following hypothesis has been proposed for this study: 35 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 h9: work engagement mediates the relationship between the psycap and the ocbs. application of the jdr model the jd-r (job demands-resources) model proposed that every organizational setting could be classified into two general aspects; job demands and job resources (bakker & demerouti, 2007). thus, establishing a model that may be applied to many work-related settings regardless of particular demands and resources involved (bakker et al., 2014). job resources were seen as referring to the physical, social, organizational or personal aspects of work that could help to achieve objectives/goals, minimize job demands, and also promoted personal growth and development. where jobs demands referred to the physical, social, or organizational aspects of work that involved continuous physical or mental effort (bakker et al., 2014). the jdr model has suggested an essential role for the personal resources that employees could use to maximize their ability to cope with the demands of the job. when demands caused anxiety and burnout, (personal) resources would lead to engagement and extra role behaviour (schaufeli & taris, 2014). personal resources were therefore, considered essential antecedents of engagement and extra role behaviour. in the context of the banking sector of pakistan, the ongoing pace and demand of work had consistently added responsibilities and a heavy workload, however, personal resources (such as hope, efficacy, resilience, and optimism) could mitigate the job demands (stress, workload etc.), because personal resources were positively linked to employee well-being and mental health (schaufeli & taris, 2014). thus, based on the jdr model, the present study has focused on the personal resources of the employees, such that employees who possessed high personal resources of hope, efficacy, resilience, and optimism, might display higher work engagement, which in turn, would lead to better ocb for the employees in the banking sector of pakistan. 36 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 methodology the present study has used a quantitative research methodology. it distributed the research questionnaire to collect the relevant data. the research used a purposive sampling technique and distributed the questionnaire to the following respondents: officer grade (og) 1, 2 and 3 of the banking sector in pakistan; og1, 2 and 3 were the assistant manager, operation manager, cash management officer, cashier, cheque processing officer, credit officer, compliance officer, management trainee officer, card operation officer etc. (obaid, 2013; ali, & chaudhry, 2017). these employees were considered as image builders and representatives of the banks, as they were in direct contact with customers and customers were in direct contact with them (ali & chaudhry, 2017). a total of 315 usable responses were collected from the employees working in the commercial banks of pakistan. the descriptive statistics showed that 74.3 percent were male and 25.7percent were female, and among them 40.3 percent were single, and 59.7 percent were married. in terms of education, about 60.6 percent had a masters, 36.5 percent had a bachelor’s degree and 2.9 percent were diploma holders in banking. in terms of positions held, 56.5 percent were the og3, 23.3 percent were the og2 and 20.3 percent were the og1. table 1 demographic profiles of respondents variable category frequency percentage gender male female 234 81 74.3 25.7 marital status married single 188 127 59.7 40.3 education masters bachelors diploma 191 115 9 60.6 36.5 2.9 officer grade og1 og2 og3 64 73 178 20.3 23.2 56.5 37 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 measurement tool the psycap was measured with an urdu version of the 12-item psycap questionnaire “pcq-12” and it was adapted from the original english language version developed by luthans et al. (2007). a back translation process was used to translate the tool into local language i.e., urdu. three language experts from the department of urdu, forman christian college, lahore, who had a good knowledge of both the languages, i.e., english and urdu.were selected they translated the english version pcq-12 into the urdu language. then, two experts from the urdu department at gomal university, dera ismail khan, pakistan, translated the urdu version of the questionnaire into the english version. in the final step, the translated urdu version and the original english version were compared, to ensure the quality and accuracy of the back translation. in sum, the pcq-12-urdu was scored using a 6-point likert scale ranging from 1 “strongly disagree” to 6 “strongly agree”. work engagement was measured with a short 9-item utrecht work engagement scale by schaufeli et al. (2006).it howver, used a 7-point likert scale ranging from 1 “never” to 7 “always”. the ocbs were measured using the 11 items suggested by williams and anderson (1991). a seven point likert scale ranging from 1 “strongly disagree” to 7 “strongly agree” was used to measure the individualdirected citizenship behaviour (ocbi) and the organisation-directed citizenship behaviour (ocbo). analysis of results following the strong recommendations by abdullahi et al. (2015), the present study has also utilized the structural equation modelling (sem) technique to research the effect of the psycap on the ocbs. the sem technique has been used to examine the relationships among variables. it was a multivariate data analysis technique which would enable the researcher to examine the relationships among multiple independent and dependent variables simultaneously, including a mechanism that could eliminate measurement errors in the observed variables through the application of bootstrapping, the mediating effect of work engagement was also tested as part of the comprehensive model (hair et al., 2014). 38 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 the pls-sem has gained popularity as a key multivariate analysis method in all disciplines including management, marketing, human resource management, organizational behavior and development, global strategy and other research fields (hair et al., 2014). the main aim of the pls has been the prediction of variance explained in the dependent variable(s), was comparable to theory development and testing in the form of structural relationship (i.e. parameter estimation) in the sem, hence putting emphasis on prediction (hair et al., 2014; hair et al., 2011). the present study has utilized the pls-sem to test its proposed hypotheses. the pls-sem was carried out utilizing the smartpls version 3.2.7 (ringle et al., 2015), which contained the measurement model (outer) and structural model (inner) analyses. evaluation of the measurement model table 2 shows the constructs’ reliabilities and convergent validities of the present study. according to hair jr et al. (2016), the construct values which were less than 0.60 had to be deleted from the model to get the average variance extracted (ave) to be more than 0.5. thus, the ocbo3 was deleted because the outer loading was less than 0.60 and was unable to attain the desired level of the ave i.e., 0.5. table 2 construct validity and reliability cronbach’s alpha rho_a composite reliability ave efficacy 0.811 0.818 0.888 0.726 hope 0.74 0.747 0.839 0.568 ocbi 0.822 0.833 0.87 0.529 ocbo 0.778 0.788 0.857 0.601 optimism 0.545 0.719 0.798 0.669 resilience 0.669 0.68 0.819 0.602 we 0.888 0.897 0.91 0.53 39 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 this study has also measured the heterotrait-monotrait ratio (htmt) criterion, as is shown in table 3. the heterotrait-monotrait (htmt) ratio of correlations was used to assess discriminant validity in the variance-based sem. the values of the htmt’s confidence intervals of the correlations between constructs were less than the 0.85 and 0.90 criteria (hair et al., 2016), as such the htmt results further strengthened the suitability of the discriminant validity. table 3 htmt criteria efficacy hope ocbi ocbo optimism resilience we efficacy hope 0.676 ocbi 0.342 0.411 ocbo 0.521 0.569 0.784 optimism 0.504 0.799 0.444 0.567 resilience 0.53 0.733 0.646 0.706 0.755 we 0.439 0.425 0.623 0.608 0.53 0.633 evaluation of the structural model as pointed out by hair et al. (2014), the structural model has been used to verify the link between the dependent and independent variables. structural equation modelling (sem) has been considered a technique which could enable the researcher to examine the relationships among multiple independent and dependent variables simultaneously, as it included a mechanism that would eliminate measurement errors in the observed variables (hair et al., 2014). the researchers used the partial least square (pls) approach, as its main aim was the prediction of variance in the form of structural relationships between the dependent and independent variables. hair et al. (2016) has suggested the following measurements to examine the structural model such as collinearity assessment, effect size, analysis of path coefficient and variance explained. these measurements will now be elaborated in the sub-sections below. 40 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 collinearity assessment as the present study used a self-reporting data collection technique, it was necessary to check for collinearity of the study items. according to hair et al. (2016), the vif values should be less than 5. as shown in table 4, the present study’s vif values ranged from 1 to 2.365, indicating that there were no collinearity problems. table 4 collinearity assessment (vif values) psycap vif ocbs vif we vif eff1 1.993 ocbi1 1.65 we1 2.389 eff2 1.908 ocbi2 1.51 we2 2.5 eff3 1.582 ocbi3 1.461 we3 1.72 hope1 1.182 ocbi4 1.769 we4 2.065 hope2 1.894 ocbi5 1.768 we5 2.314 hope3 1.45 ocbi6 1.752 we6 1.987 hope4 1.84 ocbo1 1.495 we7 2.364 opt1 1.163 ocbo2 1.736 we8 2.365 opt2 1.163 ocbo4 1.656 we9 1.541 res1 1.217 ocbo5 1.349 res2 1.377 res3 1.372 note. eff denotes efficacy, opt denotes optimism, res denotes resilience and we denotes work engagement. effect size the size of the effect is often known as the practical significance and helps to inform practitioners about whether an effect is practically significant. for the f2 values, the rule of thumb for small, medium, and large effect sizes is 0.02. 0.15, and 0.35, respectively, (cohen, 1992). table 5 shows that resilience is the most effective among all other positive personal resources. also, work engagement has shown a medium effect on the ocbi and a small effect on and 0.35 ocbo, thus, supporting the practical significance of the variables. 41 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 table 5 effect size (f2) ocbi f2 ocbo f2 we f2 efficacy 0 no 0.02 small 0.043 small hope 0.001 no 0.008 no 0.001 small optimism 0.001 no 0.006 no 0.029 small resilience 0.083 small 0.052 small 0.124 small we 0.152 medium 0.099 small analysis of path coefficient the use of path analysis to examine causal structures among continuous variables was pioneered by wright (1960). path analysis has allowed researchers to examine the direct impact of a predictor on the dependent variable, and also see the indirect relationships via a mediator. table 6 indicates the direct relations between the independent and dependent variables. the study examined four elements of psycap and two dimensions of ocbs. table 6 analysis of the path coefficient (direct relations) hypotheses beta value mean std dev t-values p-values decision efficacy → ocbi 0.078 0.08 0.026 3.03 0.002 supported efficacy → ocbo 0.061 0.062 0.021 2.899 0.004 supported hope → ocbi -0.014 -0.012 0.029 0.495 0.62 unsupported hope → ocbo -0.011 -0.009 0.023 0.49 0.624 unsupported optimism → ocbi 0.068 0.066 0.029 2.32 0.02 supported optimism → ocbo 0.053 0.051 0.023 2.295 0.022 supported resilience → ocbi 0.134 0.137 0.038 3.555 0 supported resilience → ocbo 0.105 0.107 0.033 3.212 0.001 supported note. std dev denotes standard deviation. “p<0.01” and “p<0.05” the results of direct relationships are as shown in table 6, which showed that there was a statistically significant positive between efficacy and the ocbi (β=0.026, p-value=0.002), ocbo (β=0.061, 42 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 p-value=0.004). the relationship was found be significant between optimism and the ocbi (β=0.068, p-value=0.02), and optimism and the ocbo (β=0.053, p-value=0.022). the results also showed a statistically significant positive impact of resilience and the ocbi (β=0.134, p-value=0.000), and resilience and the ocbo (β=0.105, p-value=0.001). however, the results showed an insignificant relationship between hope and the ocbi (β=-0.014, p-value=0.62), and hope and the ocbo (β=-0.011, p-value=0.624). thus, the three elements of the psycap were found to be significant with two dimensions of the performance. however, hope lost its significance. figure 1 path model table 7 analysis of the path coefficient (indirect relation) beta values sample mean standard deviation tstatistics p-values psycap → we → ocb 0.329 0.340 0.033 10.097 0.000 16 the results of direct relationships are as shown in table 6, which showed that there was a statistically significant positive between efficacy and the ocbi (β=0.026, p-value=0.002), ocbo (β=0.061, p-value=0.004). the relationship was found be significant between optimism and the ocbi (β=0.068, p-value=0.02), and optimism and the ocbo (β=0.053, pvalue=0.022). the results also showed a statistically significant positive impact of resilience and the ocbi (β=0.134, p-value=0.000), and resilience and the ocbo (β=0.105, p-value=0.001). however, the results showed an insignificant relationship between hope and the ocbi (β=0.014, p-value=0.62), and hope and the ocbo (β=-0.011, pvalue=0.624). thus, the three elements of the psycap were found to be significant with two dimensions of the performance. however, hope lost its significance. figure 1 path model table 7 43 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 to test the indirect relations, the study obtained a 95 percent percentile bootstrapping (see figure 2) confidence intervals for the hypothesized indirect effect, using 5000 bootstrap samples. the results showed that the mediating effect of work engagement (β= 0.329, p-value = 0.000) has a significant relationship with the psycap and the ocb. figure 2 bootstrapping variance explained the r2 value defines the variance on the dependent variable because of the independent variables (hair et al., 2014). the r2 effect for each dependent variable is revealed in table 8. according to cohen (1992) the suggested value for r2 was considered low at 0.12, moderate at 0.25, and substantial above 0.25.for the present study, the r2 was 0.318 for the we, 0.40 for the ocbo and 0.354 for the ocbi. what this meant was that the four facets of the psychological capital showed substantial variance in the ocbi and the ocbo. table 8 variance explained r square r square adjusted effect ocbi 0.354 0.344 substantial ocbo 0.40 0.391 substantial we 0.318 0.309 substantial conclusion in the past few years, the emphasis in the field has been on researching, explaining, and understanding the link between employees’ positive personal resources and their impacts on the ocbs. this study has 15.620 15.783 psycap we ocbs 44 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 contributed to a better understanding of the issues by researching the four positive resources of hope, optimism, resilience, and efficacy with the ocbs, alongside the mechanism through which these positive personal resources of the psycap lead to the ocbs. the findings of the present study did not support the view that hope has a relationship with the ocbi and the ocbo. the studies examining the relationship between hope and the ocbs have been scarce in the context of pakistan. however, the findings of the present study were supported in the research by youssef and luthans (2007), who argued that hope was a future-oriented approach. this view implied that the employees might not be happy with their current job position. the findings of the present study also revealed that most of the respondents possessed a master’s degree (66.6%), and about half of them were serving as the og3.these respondents might not be happy serving in their current position, and thus might not engage at work and did not reveal the ocbi and the ocbo. the findings of the study also revealed that efficacy was significantly related to the ocbi and the ocbo. the results also showed that efficacy was strongly linked to the ocbi compared to the ocbo. the results of the study have shown support for previous studies related to efficacy and the ocbs (beauregard 2012; carter et al., 2018). beauregard (2012) pointed out that employees high in efficacy demonstrated supportive behaviour towards their fellow coworkers (ocbi) and the organization (ocbo). the present study has suggested that efficacy could influence the feelings of competence and faith in one’s perceived ability to prove one’s self as an asset to the organisation. in the context of banking, the results of the present study seemed to imply that employees with high efficacy would develop a deeper interest in the tasks they were involved in. they put much effort into their tasks and supported their colleagues in achieving objectives, as they were aware of their ability to produce the desired results. thus, efficacy is one of the important personal resources to attain desired objectives (the application of the jdr) for the employees in the pakistani banking sector. optimism, the third element of the psycap, was also found be significant with the ocbi and the ocbo. the results of the present study also revealed that optimism was strongly linked to the ocbi as compared to the ocbo. the results seemed to imply that employees who were high in optimism did more than what they have been asked 45 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 to do. in the banking culture of pakistan, maybe it was the optimist’s behaviour to be positive ,so that colleagues (the ocbi) would feel comfortable when they decided to approach their colleagues and get support. thus, it implied that optimists went out of the way to help colleagues and expected praise in exchange (busseri et al., 2009). thus, the findings of the study supported the argument that employees who possessed high optimism would go out of their way to help their colleagues and organisation . in sum, optimism was considered significant for the ocbi and the ocbo. the fourth element of the psycap was also found be significant to the ocbi and the ocbo. the results of the present study also indicated that resilience impacted the ocbi the most, as compared to the ocbo. these results were supported by previous research which suggested that resilience had a significant relationship with the ocbi and the ocbo (gupta et al., 2017; kašpárková et al., 2018). luthans et al. (2007) pointed out in their study that the impact of resilience was significant as compared to the other dimensions of the psyacp. it has been necessary to consider resilience since employees faced several stressors in the banking environment. the private commercial banks in pakistan are facing so many challenges, thus in the current banking environment resilience of the employees was seen as critical as with it employees have the ability to bounce back from a crisis. the results of the study also supported the view that work engagement mediated the relationship between the psycap and the ocbs. it means that work engagement is significantly influential in its effect on the ocbi and the ocbo. it implies that the workforce is the most critical asset in developing the banking industry, having a knowledge of the factors that can positively influence engagement with the bank will be very beneficial for the overall efficiency of the banks. thus, it has become imperative that banks consider the factors that can significantly contribute to the engagement of the ir employees . in this regard, it can be suggested to the hr of banks to give attention to the four facets of the psycap of their employees, so that by increasing the engagement of their employees, it will ultimately help to improve the ocbs. recommendations for future research this study was aimed at examining the impact of the four facets of the psycap on two dimensions of the ocbs. the present research 46 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 has been able to make both theoretical and practical contributions to the existing body of knowledge in the field. in particular, current research about the psycap is still new, but this study has been able to fill the existing gap by investigating the impact of the psycap on the ocb, especially in the context of the private commercial banks sector in pakistan. in light of the findings in the present study on the positive direct and indirect effects of the psycap on the ocb, it is recommended that the psycap be seen as having an important role to play in the current banking environment in the south asian setting, i.e., in pakistan. the findings reported in this study seemed to suggest that senior officials and the hr of the banks should make needful changes in policies, particularly when thinking about the psychological aspects of candidates in the recruitment and selection process. for example, aside from the usual focus on academic capabilities, the policymakers can also consider the psycap of potential recruits as a fundamental selection criteria. hr could develop policies to nurture and foster the psycap of the company’s employees, as it has been regarded as a state-like ability, according to luthans et al. (2007). in this regard, the banking sector should structure its professional training and professional development projects to cultivate employees’ psycap, alongside other individual abilities. thus, the banking industry now has the option to have better qualified employees who will be equipped with the necessary psychological strengths to achieve a competitive advantage. the study has utilised a cross sectional study approach, collecting and analysing data simultaneously. it is thus, recommended that a longitudinal research study be carried out to determine the impact of the four facets of the psycap at different points of time. furthemore, as the sample of the present study only consisted of the og1, the og2 and the og3, as they were representative of the field offices, it is further recommended that future studies test the relationships and validate the results with a larger sample size and at the managerial level. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for profit sectors. 47 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 references abdullahi, m. s., abubakar, a., aliyu, r. l., & umar, m. b. (2015). empirical review on the determinants influencing firm performance in developing countries. international journal of scientific and research publications, 5(6), 1-10. https://doi. org/10.1.1.736.3539 akhras, c., & alam, a. s. a. (2020). a hope impact in mena: business leadership and job performance. international journal of psychosocial rehabilitation, 24(1), 898-906. https://doi. org/ 10.37200/ijpr/v24i1/pr200196 ali, h., & sharif chaudhry, i. (2017). effect of human capital on organization performance: an analysis from service sector of punjab, pakistan. european online journal of natural and social sciences, 6(3), 475-481. https://european-science.com/ eojnss/ article/view/5127/pdf alola, u. v., & atsa’am, d. d. (2020). measuring employees’ psychological capital using data mining approach. journal of public affairs, 20(2), e2050. american psychological association (2014). the road to resilience. http://www.apa.org/helpcentre/road-resilience.aspx bandura, a. (1997). self-efficacy: the exercise of control. new york. barnard, c. i. (1938). the functions of the executive. harvard university. bakker, a. b., & demerouti, e. (2007). the job demands-resources model: state of the art. journal of managerial psychology. 22 (3), 309-328. https://doi.org/10.1108/02683940710733115 bakker, a. b., & albrecht, s. (2018). work engagement: current trends. career development international, 23(1), 4-11. https:// doi.org/ 10.1108/cdi-11-2017-0207 bakker, a. b., & demerouti, e. (2014). job demands-resources theory. wellbeing. 3, 1-28. https://doi.org/10.1002/9781118539415. wbwell019 beauregard, t. a. (2012). perfectionism, self-efficacy and ocb: the moderating role of gender. personnel review, 41(5), 590-608. https://doi.org/10.1108/00483481211249120 busseri, m. a., choma, b. l., & sadava, s. w. (2009). functional or fantasy? examining the implications of subjective temporal perspective “trajectories” for life satisfaction. personality and social psychology bulletin, 35(3), 295-308. 48 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 chaitanya, s. k., & tripathi, n. (2001). dimensions of organisational citizenship behaviour. indian journal of industrial relations, 37 (2), 217-230. https://www.jstor.org/stable/27767778 chamisa, s. f., mjoli, t. q., & mhlanga, t. s. (2020). psychological capital and organisational citizenship behaviour in selected public hospitals in the eastern cape province of south africa. sa journal of human resource management, 18(0), a1247, 1-12 https://doi.org/10.4102/sajhrm.v18i0.1247 carter, w. r., nesbit, p. l., badham, r. j., parker, s. k., & sung, l. k. (2018). the effects of employee engagement and selfefficacy on job performance: a longitudinal field study. the international journal of human resource management, 29(17), 2483-2502. cohen, j. (1992). statistical power analysis. current directions in psychological science, 1(3), 98-101. cohen, a., & abedallah, m. (2015) ‘the mediating role of burnout on the relationship of emotional intelligence and self-efficacy with ocb and performance’, management research review, 38(1), 2–28 https://doi.org/10.1108/mrr-10-2013-0238 gupta, m., shaheen, m., & reddy, p. k. (2017). impact of psychological capital on organizational citizenship behavior: mediation by work engagement. journal of management development. 36(7), 973-983. https://doi.org/10.1108/jmd06-2016-0084 hair, j. f., ringle, c. m., & sarstedt, m. (2011). pls-sem: indeed a silver bullet. journal of marketing theory and practice, 19(2), 139-152. https://doi.org/10.2753/mtp1069-6679190202 hair jr, j. f., hult, g. t. m., ringle, c., & sarstedt, m. (2016). a primer on partial least squares structural equation modeling (pls-sem). sage publications. hair jr, j. f., sarstedt, m., hopkins, l., & kuppelwieser, v. g. (2014). partial least squares structural equation modeling (pls-sem): an emerging tool in business research. european business review, 26(2), 106-121. https://doi.org/10.1108/ebr-10-20130128 hough, c., sumlin, c., & green, k. w. (2020). impact of ethics, trust, and optimism on performance. management research review. 43 (9), 1135-1155. imran, m., & shahnawaz, m. g. (2020). psycap and performance: wellbeing at work as a mediator. asia-pacific journal of management research and innovation, 16(2), 93-102. https:// doi.org/10.1177/2319510x20915999 49 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 iqbal, j., shabbir, m. s., zameer, h., khan, i. a., & sandhu, m. a. (2017). antecedents and consequences of employee engagement: evidence from corporate sector of pakistan. paradigms, 11(1), 78. joo, b. k., zigarmi, d., nimon, k., & shuck, b. (2017). work cognition and psychological well-being: the role of cognitive engagement as a partial mediator. the journal of applied behavioral science, 53(4), 446-469. jung, h. s. & yoon, h. h. (2015). the impact of employees’ positive psychological capital on job satisfaction and organisational citizenship behaviours in the hotel. international journal of contemporary hospitality management, 27(6), 1135–1156 https://doi.org/10.1108/ijchm-01-2014-0019 kašpárková, l., vaculík, m., procházka, j., & schaufeli, w. b. (2018). why resilient workers perform better: the roles of job satisfaction and work engagement. journal of workplace behavioural health, 33(1), 43–62. https://doi.org/10.1080/ 15555240.2018.1441719 latif, h. (2017). psychological capital and work engagement in relation to job performance among pakistani teachers. university of nottingham. leijten, f. r., van den heuvel, s. g., van der beek, a. j., ybema, j. f., robroek, s. j., & burdorf, a. (2015). associations of workrelated factors and work engagement with mental and physical health: a 1-year follow-up study among older workers. journal of occupational rehabilitation, 25(1), 86-95. luthans, f., youssef, c. m., & avolio, b. j. (2015). psychological capital and beyond. oxford university press, usa. luthans, f., & youssef-morgan, c. m. (2017). psychological capital: an evidence-based positive approach. annual review of organizational psychology and organizational behavior, 4, 339-366. https://doi.org/10.1146/annurevorgpsych-032516-113324 luthans, f., avolio, b. j., avey, j. b., & norman, s. m. (2007). positive psychological capital: measurement and relationship with performance and satisfaction. personnel psychology, 60(3), 541-572. jabbar, m. n., nawaz, m., rehman, f. u., bhatti, g. a., & choudhary, a. (2019). does optimism and work engagement matter to improve job performance? an empirical study. international journal of information, business and management, 11(4), 170176. 50 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 jung, h. s., & yoon, h. h. (2015). the impact of employees’ positive psychological capital on job satisfaction and organizational citizenship behaviors in the hotel. international journal of contemporary hospitality management, 27(6), 1135-1157. joo, b. k., zigarmi, d., nimon, k., & shuck, b. (2017). work cognition and psychological well-being: the role of cognitive engagement as a partial mediator. the journal of applied behavioral science, 53(4), 446-469. https://doi. org/10.1177/0021886316688780 ofei-dodoo, s., long, m. c., bretches, m., kruse, b. j., haynes, c., & bachman, c. (2020). work engagement, job satisfaction, and turnover intentions among family medicine residency program managers. international journal of medical education, 11, 4753. 10.5116/ijme.5e3e.7f16 obaid, a. (2013). changes in the pay structures and systems in the banking sector of pakistan: implications for a differentiated workforce and the employee psychological contract. the university of manchester (united kingdom). organ, d. w. (1988). organizational citizenship behavior: the good soldier syndrome. lexington books/dc heath and com. ozyilmaz, a. (2020). hope and human capital enhance job engagement to improve workplace outcomes. journal of occupational and organisational psychology, 93(1), 187–214. https://doi. org/10.1111/joop.12289 pradhan, r. k., jena, l. k., & panigrahy, n. p. (2020). do sustainability practices buffer the impact of self-efficacy on organisational citizenship behaviour? journal of indian business research. 12(4), 509-528. https://doi.org/10.1108/jibr-05-2019-0170 paul, h., bamel, u., ashta, a., & stokes, p. (2019). examining an integrative model of resilience, subjective well-being and commitment as predictors of organizational citizenship behaviours. international journal of organizational analysis, 27(5), 1274-1297. https://doi.org/10.1108/ijoa-08-2018-1514 paul, h., bamel, u. k., & garg, p. (2016). employee resilience and ocb: mediating effects of organisational commitment. the journal for decision makers, 41(4), 308–324 https://doi. org/10.1177/0256090916672765 podsakoff, p. m., mackenzie, s. b., & podsakoff, n. p. (eds.). (2018). the oxford handbook of organisational citizenship behaviour. oxford university press. 51 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 ribeiro, n., gupta, m., gomes, d., & alexandre, n. (2021). impact of psychological capital (psycap) on affective commitment: mediating role of affective well-being. international journal of organizational analysis. https://doi.org/10.1108/ijoa-042020-2122 ringle, c., da silva, d., & bido, d. (2015). structural equation modeling with the smartpls. bido, d., da silva, d., & ringle, c. (2014). structural equation modeling with the smartpls. brazilian journal of marketing, 13(2), 56-73. https://ssrn.com/ abstract=2676422 siddiqui, d. a. (2019). impact of motivation on employees performance in pakistan. rida, n., & siddiqui, da (2019). impact of motivation on employees performance in pakistan. business management and strategy, 10(1), 1-22. https://ssrn. com/abstract=3381441 sarwar, h., nadeem, k., & aftab, j. (2017). the impact of psychological capital on project success mediating role of emotional intelligence in construction organisations of pakistan. journal of global entrepreneurship research, 7(22), 1-13. https://doi.org/10.1186/s40497-017-0080-4 schaufeli, w. b., bakker, a. b., & salanova, m. (2006). the measurement of work engagement with a short questionnaire: a cross-national study. educational and psychological measurement, 66(4), 701-716. https://doi.org/10.1177/ 0013164405282471 schaufeli, w. b., & taris, t. w. (2014). a critical review of the job demands-resources model: implications for improving work and health. bridging occupational, organizational and public health, 43-68. https://doi.org/10.1007/978-94-007-5640-3_4 seligman, m. e. (2006). learned optimism: how to change your mind and your life. vintage. shimazu, a., matsudaira, k., de jonge, j., tosaka, n., watanabe, k., & takahashi, m. (2016). psychological detachment from work during nonwork time: linear or curvilinear relations with mental health and work engagement? industrial health, 54(3) 282-292. https://doi.org/10.2486/indhealth.2015-0097 smithikrai, c. (2020). how optimism leads employees to show good citizenship: a serial mediation model. panyapiwat journal, 12(2), 160-173. http://orcid.org/0000-0003-4538-5891 snyder, c. r. (1995). conceptualizing, measuring, and nurturing hope. journal of counseling & development, 73(3), 355-360. https://doi.org/10.1002/j.1556-6676.1995.tb01764.x 52 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 27–52 snyder, c. r. (2002). hope theory: rainbows in the mind. psychological inquiry, 13(4), 249-275. son, j. (2018). back translation as a documentation tool. the translation & interpreting, 10(2), 89–100. https://search. informit.org/ doi/ 10.3316/informit.864953916346703 thakre, n., & mathew, p. (2020). psychological empowerment, work engagement, and organizational citizenship behavior among indian service-sector employees. global business and organizational excellence, 39(4), 45-52. https://doi. org/10.1002/ joe.22003 tyupa, s. (2011). a theoretical framework for back-translation as a quality assessment tool. new voices in translation studies, 7(1), 35-46. ugwu, f. o., & igbende, d. a. (2017). going beyond borders: work centrality, emotional intelligence and employee optimism as predictors of organizational citizenship behavior. cogent psychology, 4(1), 1362805. https://doi.org/10.1080/ 23311908. 2017.1362805 waseem, m., ali, s., bilal, h., & khan, a. a. (2020). high involvement management practices and employees’ ocbs in banking sector of pakistan. journal of accounting and finance in emerging economies, 6(2), 529-538. https://doi.org/10.26710/jafee. v6i2.1229 williams, l. j., & anderson, s. e. (1991) ‘job satisfaction and organisational commitment as predictors of organisational citizenship and in-role behaviours’, journal of management, 17(3), 601–617. https://doi.org/10.1177/014920639101700305 wirtz, n., rigotti, t., otto, k., & loeb, c. (2017). what about the leader? crossover of emotional exhaustion and work engagement from followers to leaders. journal of occupational health psychology, 22(1), 86. wright, s. (1960). the treatment of reciprocal interaction, with or without lag, in path analysis. biometrics, 16(3), 423-445. youssef, c. m., & luthans, f. (2007). positive organizational behavior in the workplace: the impact of hope, optimism, and resilience. journal of management, 33(5), 774-800. 53 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 how to cite this article: hamid, f. s., & harizan, s. h. m. (2023). behavioral biases and credit card repayments among malaysians. international journal of banking and finance, 18(2), 53-78. https://doi.org/10.32890/ ijbf2023.18.2.3 behavioral biases and credit card repayments among malaysians 1fazelina sahul hamid & 2siti haslina md harizan 1bristol business school university of the west of england, united kingdom 2school of distance education, universiti sains malaysia 1corresponding author: fazelina.sahulhamid@uwe.ac.uk received: 10/2/2022 revised: 20/4/2022 accepted: 3/6/2022 published: 25/6/2023 abstract analyses done using the survey response of 451 credit cardholders in the northern region of west malaysia have shown that credit cardholder repayment pattern differed based on their demographic profile. specifically, the study found that credit cardholders who were younger and earned a lower income tended to make poorer repayment decisions. in contrast, those with higher academic qualifications tended to make better repayment decisions. further analyses also revealed that behavioral biases were better at predicting those who made full payment of the outstanding balance and those who paid more than the minimum amount, but failed to predict those who only paid the minimum amount or less. in addition, it was found that overspending emerged as the strongest predictor which discriminated among the three repayment groups, followed by risk aversion, myopia and impulsiveness. https://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 54 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 keywords: credit card repayment, behavioral biases, consumer finance, personal debt. jel classification: g4, g5. introduction household debt-to-gross domestic product (gdp) in malaysia remains the highest among asean countries. the high level of household debt has become a major concern in malaysia because the growth of disposable income is slowing down and the economic condition is less favorable (world bank, 2019). economic analysts and policymakers are concerned about this situation, given the potential risk it may pose to the stability of the financial system and economic growth. one of the issues that have become a growing concern in malaysia is the rising credit card loans over the years. a credit card provides consumers with easy access to credit for consumption. credit cardholders can shop without having enough cash with them. this may lure consumers to spend more than what is necessary and live beyond their means (soman, 2001). the convenience of using a credit card can lead to the problem of massive debt accumulation. borrowing more makes consumers vulnerable to bankruptcies and credit market exclusion (gathergood, 2012). increasingly, the younger generation and low-income individuals in malaysia are turning to credit card loans for discretionary consumptions that are aimed at supporting extravagant lifestyle choices rather than on necessities (world bank, 2019). unlike other loans that require collaterals and fixed settlements, the credit card loan is a revolving loan that is offered based on an individual’s income category. consumers are only required to make minimum payments of five percent of the outstanding amount due every month. given that credit card loans are often very costly, consumers may rapidly go into debt and face financial hardship if they are not able to settle their loans. data released by the credit counselling and debt management agency (akpk) shows that the credit card debt accounted for 55 percent of the total debt portfolio mediated by the agency (the malaysian reserve, 2019). bankruptcies related to credit card loans have also increased sharply over the years, especially among the younger generation (world bank, 2019). this raises the question of 55 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 why malaysians are incurring debts they could not afford and given the known dangers of excessive borrowing. traditional theories in finance like the capital asset pricing model (capm) and the miller-modigliani theorem assume that individuals make portfolio allocation based on a trade-off between expected return and risk. these theories assume that individuals are rational decision-makers who make well-informed decisions that maximize their utility. however, empirical evidence shows that consumers do not always make rational decisions. behavioral finance posits that decision making by individuals is bounded because of inadequate resources and the lack of ability (sharma & kumar, 2019). these lead to judgmental errors and suboptimal decision making that can be detrimental for individuals and society (ceschi et al., 2019). as far as borrowing is concerned, most of the microeconomic theories posit that individuals borrow to smooth their consumption over the life cycle. even though debt allows consumers to improve their standard of living by leveraging on their future income, over-indebtedness becomes a concern because individuals must use much of their current income to settle their debt (gathergood, 2012). as a result, they may not have enough funds for expenditure and savings. this impairs their financial well-being and makes it difficult for them to cope with unforeseen financial shocks due to illness, emergencies or job loss. existing studies on credit card debt in malaysia have mainly focused on macroeconomic factors (theong et al., 2018); attitude (ramayah et al., 2002; chong, 2017); cardholders attributes (wei et al., 2018); financial knowledge (jusoh & lin, 2012), financial literacy (hamid & loke, 2021a) and demographic profile (loke et al., 2011). despite the increasing volume of research on credit card loans in malaysia, not many of these previous studies have investigated the link between behavioral biases and credit card repayment. existing studies on behavioral biases in malaysia have focused on trading and investment in the stock market (hamid et al., 2013; brahmana et al., 2015; jaiyeoba et al., 2020) and asset management industry (ahmad et al., 2018). this study aims to fill this research gap and add to the literature by identifying the differences in credit card repayments based on the demographic profiles of the users. additionally, this study establishes the link between behavioral biases and credit card repayment. the 56 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 findings of this research can help identify the characteristics of vulnerable credit card users and the behavioral biases that they are prone to. in doing so, the present study will enable policymakers and credit card providers to plan for a better card credit risk management framework that can assist vulnerable borrowers to improve their financial decision making. this is in line with the bank negara malaysia (bnm)’s aim of strengthening household resilience and financial stability. literature review behavioral economics postulates that consumers are not always rational in their financial decision making. behavioral biases that have been studied include risk aversion, mental accounting, regret avoidance, overconfidence, naive expectations, myopia, lack of selfcontrol and procrastination (barberis & thaler, 2003; ceschi et al., 2019). the existing literature has used behavioral biases in explaining various financial decision making, such as stock market anomalies (barberis & thaler, 2003), excessive risk-taking (hamid et al., 2013), over-indebtedness (gathergood, 2012; barboza, 2018) and delayed repayments (barboza et al., 2017; barboza, 2018). the latter is true for credit card loans that give consumers the flexibility to decide how much to repay. shefrin and thaler (1981) includes self control to the model of intertemporal choice in the theory of rational behavior. they illustrate that individuals who are myopic (doers) care only about instant gratification. meanwhile, the planners care about the consumers’ present and future. they show that a consumer’s myopic perspective usually results in less constant outcome compared to a farsighted perspective. in the context of a competitive credit market, heidhues and koszegi (2010) demonstrate that self-control bias influences consumer to borrow more and delay repayments, which results in large welfare losses. meanwhile, (baumeister, 2002) defines self-control bias as the individuals’ lack of ability to regulate their cognition, emotion, behavior and responses. individuals with a self-control bias are more impatient in the short-run compared to long-run. as a result, they prefer instant gratification and are more concerned about their present than the future. the existing literature has linked self-control bias to dimensions related to impulsiveness, overspending, myopia and risk-taking. 57 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 existing studies show that credit card repayments vary according to the demographic profiles of the cardholders. barboza (2018) finds that older individuals and those with higher educational attainment make better credit card repayment decisions. stavins (2020) observes that the credit card holders tendency to be a revolve payment is influenced by their demographic characteristics such as age, income, gender, race, education and employment status. wang et al. (2011) and wang et al. (2011) shows that males are more likely to be revolve payment hamid and loke (2021b) finds that those with higher income make better credit card repayments. in line with this, the following hypothesis was derived: hypothesis 1: there is a significant relationship between the demographic profile of credit cardholders and credit card repayments. impulsiveness is identified as a personality trait that is associated with the immediate gratification of desires and needs that lacks consideration for long-term consequences. within the behavioral finance research literature, impulsive buying behavior is associated with a lack of self-control. baumeister (2002) defines it as the unplanned, spontaneous and irresistible buying behavior that happens without proper consideration about whether it is consistent with the individual’s long-term goals. impulsive individuals take more risk. in line with this, impulsive buying behavior has been linked to credit card usage (baumeister, 2002). pirog and roberts (2007) confirm that impulsiveness is linked to credit card misuse. gerardi et al. (2010) and brown and graf (2013) have provided empirical evidence that impulsiveness is associated with a greater tendency to obtain consumer loans. limerick and peltier (2014) find that credit cardholders who are impulsive have a higher debt. peltier et al. (2013) find that impulsive buying behavior leads to a greater credit card debt among young individuals. similarly, the present study expects impulsiveness to be linked to a worse credit card repayment pattern. the ‘pain of payment” is felt more when payment for goods and services are made using cash. purchases made using credit cards are often settled later. as a result, individuals have a greater tendency to spend more than what they can pay in a period because of the availability of credit cards. soman (2001) associates credit card usage to overspending behavior. overspending using a credit card is attributed 58 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 to psychological and social factors (sotiropoulos & d’astous, 2012). empirical evidence shows that there is a link between overspending behavior and credit card usage among consumers in the united kingdom and taiwan (lo & harvey, 2011). a similar association is also noticed in the unites states (huebner et al., 2018). meanwhile, barboza et al. (2017) find that overspending also leads to poor credit card repayment behavior among college students in the united states. the present study also expects overspending to be related to a worse credit card repayment pattern. myopia is another form of self-control bias whereby individuals are more focused on their short-term rather than long-term benefits (sunstein, 2006). as a result, myopic individuals make decisions that are detrimental to their wellbeing over time. among the behaviors that are linked to myopia are excessive borrowing, lack of saving, overeating, drinking alcohol and smoking. brown and graf (2013) confirm that myopic individuals have a higher tendency not to invest in the financial market. also, they consume more now and delay retirement saving (chybalski & marcinkiewicz, 2018). meier and sprenger (2010) find that myopic individuals usually have a credit card and carry a higher credit card debt. similarly, kuchler and pagel (2018) observe that myopic individuals are more likely to delay their credit card debt repayments. in line with this, the present study expects myopia to be linked to a worse credit card repayment pattern. risk-taking relates to the extent an individual is willing to be exposed to an uncertain outcome that may result in financial gains or losses. it varies among individuals. barlow (1991) has identified risk-taking as one of the dimensions of self-control bias. studies show that risktaking behavior influences financial decision making. brown and graf (2013) find that risk-averse individuals have a higher tendency to not invest in the financial market, but have retirement savings. hamid et al. (2013) and pak and mahmood (2015) confirm that individuals who have a higher risk-taking tendency are more likely to invest in risky assets. risk-taking behavior has also been associated with intention to use a credit card and the possibility of overusing it (lin et al., 2015). palan et al. (2011) have pointed out that a higher level of risk-taking is associated with the incidence of greater credit card misuse. in line with this, the present study expects risk aversion to be associated with better credit card repayment. in line with the above arguments, the following hypothesis was derived: 59 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 hypothesis 2: there is a significant relationship between behavioural biases and credit card repayments. there is a dearth of literature examining the role of behavioral bias in influencing credit card debt in malaysia. theong et al. (2018) studied the effect of macroeconomic uncertainties on credit card default in malaysia. chong (2017) scrutinized the role of attitude towards credit, dependability, image consciousness and financial planning in influencing credit card repayment among malaysians. whereas, chin et al. (2017) analysed the impact of personal attitude and spending pattern on credit card repayment pattern among academicians in malaysia. jusoh and lin (2012) investigated the role demographic profile, financial knowledge and financial attitude in influencing credit card practices. teoh et al. (2013) looked at how credit cardholders spending habits vary based on cardholders’ demographic profiles, attitudes toward money, benefits given by banks, and accommodative bank policies. meanwhile, loke et al. (2011) studied the role of demographic factors related to age, race, education, income, number of loan commitments, household size and current account ownership in influencing credit card ownership. hamid and loke (2021a) investigated the role of financial literacy and money management skills in influencing credit card repayments in malaysia. this study aims to fill the gap in the literature by looking at the role of behavioral biases in influencing credit card repayments. methodology data for this study were collected using a self-administered survey. a total of 500 questionnaires were distributed, and 451 completed questionnaires were used for the analysis. data were collected in the northern region of west malaysia. only credit card users who had made payment decisions took part in the study. the questionnaire contained questions on demographic profile, and behavioral biases related to credit card usage and repayment practices. the latter was the dependent variable of this study. it was measured using a categorical scale. this variable was used to assess credit card repayment behavior, with 1 = pay in full, 2 = pay more than minimum and 3 = pay only minimum or less than minimum. in line with the literature, the four dimensions of self-control bias relevant in understanding credit card repayments have been identified 60 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 as impulsiveness, overspending, myopia and risk aversion. in line with brown and graf (2013), impulsiveness is measured by identifying the individual’s response to the following statement: “i am impulsive and tend to buy things that i cannot afford”. this item was measured using a five-point likert scale ranging from 1= “agree” to 5 = “don’t agree”. following barboza et al. (2017), overspending is measured by identifying the individual’s tendency to purchase items using a credit card, knowing that they will not have enough money to pay their bill in full when the balance is due. a five-point likert scale ranging from 1= “never” to 5 = “always” is used. meanwhile, myopia refers to individuals’ tendency to be more concerned about their short-term gain rather than long-term ones. following brown and graf (2013), this variable is assessed using the following statement: “i live for the present and don’t think about my financial future. this item was measured using a five-point likert scale ranging from 1= “agree” to 5 = “don’t agree”. risk aversion refers to the self-assessment of risk that individuals are willing to take in their financial investment. in line with grund and sliwka (2010) and brown and graf (2013), this item was measured using a five-point likert scale ranging from 1= “no risk” to 5 = “high risk”. this study also considers the demographics variables to account for differences in respondents’ gender, age, income, academic qualification, race and marital status. t-test and analysis of variance (anova) were used in testing hypothesis 1. whereas, discriminant analysis was used to find support for hypothesis 2. discriminant analysis has been performed by combrink and lew (2020) in examining the relationships among underdog bias, overconfidence and risk propensity in investor decisionmaking behavior. hamid and loke (2021a) which investigated the relationship between socio-economic factors, financial literacy, money management skill, overspending and impulsiveness on credit card repayment decisions also used discriminant analysis. the advantages of using such a model is that it enables one to predict membership in two or more mutually exclusively groups. in the context of the present study, it is the relationship between those who made full payment of the outstanding balance, paying more than the minimum amount. and paying the minimum amount or less. discriminant analysis will also enable the creation of a regression equation that makes use of a 61 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 dependent (criterion) variable that is discrete rather than continuous. based on the preexisting data in which group membership is already known, a regression equation can be computed that maximally discriminates between two or more groups (george & mallery, 2003). in terms of reliability, the cronbach alpha coefficient value of all variables were above 0.7, indicating an acceptable and good internal consistency reliability of validated measurement scales (sekaran & bougie, 2014). in terms of validity, the kmo measure of sampling adequacy was 0.662, indicating sufficient inter-correlations and the bartlett’s test of sphericity was significant (chi-square=247.41, p<0.01). eigenvalues greater than 1.0 and the total variance explained was 39.14 percent of the total variance. all items loaded into the five components as expected. among the components, myopia was associated with 20.12 percent of the variance in the original data, followed by credit card repayment (20.11%), risk aversion (20.04%), overspending (19.86%) and impulsiveness (19.8%). all independent variables were significantly positive correlated with credit card repayment, as in shown in table 1. table 1 inter-correlations of major variables 1 2 3 4 5 overspending risk aaversion 0.165** myopia 0.248** 0.075* impulsiveness 0.451** 0.142** 0.382** credit card rrepayment 0.319** 0.078* 0.080* 0.192** note. ***p<0.01, **p<0.05, *p<0.1 results descriptive analysis as shown in table 2, the sample comprises 220 males (48.78%) and 231 females (51.22%). a higher percentage of the respondents were aged below 36 (55.43%). the remaining 41.69 percent of them were aged between 36 to 55, while 2.88% were aged 55 and above. 62 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 a smaller percentage of the respondents only had secondary level education (14.2%). respondents with a diploma or degree totalled 262 people (58.22%) and those with masters and above numbered 124 respondents (27.56%). almost half of the respondents earned a lower income (rm2,000 to rm4,000). those earning middle income were 127 respondents (28.22%) and those earning higher income were 109 respondents (24.22%). the majority of the respondents were bumiputra (81.34%) and married (68.07%). table 2 demographic profile of respondents variable number of observations (%) gender male 220 48.78 female 231 51.22 age age up to 35 250 55.43 36 to 55 188 41.69 55 and above 13 2.88 education secondary 64 14.22 diploma and degree 262 58.22 masters and above 124 27.56 income rm2,000 to rm4,000 214 47.56 rm4,001 to rm6,000 127 28.22 rm6,000 and above 109 24.22 ethnicity bumiputra 345 81.34 chinese 42 9.33 indian and others 29 6.44 marital status single 127 28.16 married 307 68.07 divorced / widowed 17 3.77 63 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 differences in credit card repayments based on the demographic profile the t-test results in the last row of table 3 show that there is no significant mean difference in credit card repayment between males and females (p = 0.443). similar results were obtained by aydin (2022) in the case of credit cardholders in turkey and jusoh et al. (2012) in the case of malaysia. the results in table 3 also show, as determined by a one-way anova (f (2,448) = 0.75, p = 0.473), that there is no statistically significant difference in credit card repayments between groups with different age categories. this is in contrast to the findings in wang et al. (2011) and hamid and loke (2021a). meanwhile, the one-way anova analysis done to identify if people with a different academic qualification differ in their credit card repayments showed that there was a statistically significant difference between groups in their credit card repayment (f(2,447) = 6.69, p = 0.001). a scheffe post-hoc test revealed that those with secondary education, and masters and above qualifications differed significantly in their credit card repayment (-0.320 ± -0.57, p = 0.008). also, there was a significant difference in the credit card repayment between those with a diploma and degree qualifications, and masters and above qualifications (-0.23 ± -0.41, p = 0.007). however, no statistically significant difference was observed in the credit card repayment between those with secondary education, and diploma and degree qualifications. barboza (2018); hamid and loke (2021) and salisbury and zhao (2020) also find that individuals with a higher educational attainment made better credit card repayment decisions. additionally, results in table 3 also show that there was a statistically significant difference between income groups in their credit card repayment (f (2,447) = 3.16, p = 0.043). a scheffe post-hoc test revealed that those earning income between rm2,000 to rm4,000, and above rm6,000 differed significantly in terms of their credit card repayment (-0.20 ± -0.39, p = 0.044). however, no statistically significant differences were observed between other groups. salisbury (2014) also finds that credit card repayments vary based on income. results in table 2 show that credit card repayments differed between races (f (2,447) = 4.2, p = 0.016). a scheffe post-hoc test revealed that there was a statistically significant difference in the credit card repayment behavior between bumiputra and chinese respondents 64 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 (-0.30 ± -0.57, p = 0.024). however, no statistically significant differences were observed between other groups. stavins (2020) also finds a significant difference in credit card debt based on the race of the cardholders. the findings also show that credit card repayment varied based on the participants marital status (f (2,448) = 2.90, p =0.056). a scheffe post-hoc test revealed that there was a statistically significant difference in the credit card repayment between single and married individuals (0.17 ± -0.01, p = 0.058). a similar observation was obtained by hamid and loke (2021b). table 3 t-test and anova test for the differences in demographic profile and credit card repayment variables mean f-value p-value gender female 1.66 -0.8 0.422 male 1.71 age less than 35 1.69 0.75 0.47336-55 1.7 55 and above 1.46 academic qualification secondary 1.83 6.69 0.001***diploma and degree 1.74 master and above 1.51 income rm2,000rm4,000 1.76 3.16 0.043**rm4,001rm6,000 1.68 above rm6,000 1.56 race bumiputra 1.73 4.2 0.016**chinese 1.43 indian and others 1.6 marital status single 1.57 2.9 0.056*married 1.74 widowed / divorced 1.65 note. ***p<0.01, **p<0.05, *p<0.1 the above findings haved confirmed that credit card repayments vary based on the demographic profile of the cardholders, thus providing support for hypothesis 1 of this study. 65 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 relationship between credit card holders’ behavioral biases and credit card payment a discriminant analysis can be leveraged to determine which independent variables are related to the dependent variable, as well as to predict the value of the dependent variable based on the values of the independent variables. by performing a discriminant analysis, the present study can address classification problems in which two or more groups, clusters, or populations are known up front, and one or more new observations are placed into one of the known classifications based on measured characteristics. in this study, a discriminant analysis was used to determine the relationship between behavioral biases and credit card payment. the analysis sought to answer the question of whether a combination of overspending, risk aversion, myopia, and impulsiveness could predict whether an individual would make full payment of the outstanding balance, or pay more than the minimum amount, or pay the minimum amount or less. several assumptions must be met for the discriminant analysis to produce a valid result. the multivariate test has a goodness of fit statistics measured by wilks’ lambda as shown in table 4, whereby p-values were less than 0.05 (p < 0.05), meaning that the model was a good fit for the data. a small lambda (< 1.00) indicates that the group means appeared to differ. the associate significance values indicate whether the difference was significant. for those who made full payment of the outstanding balance, the mean for overspending was 1.74, risk aversion was 2.73, myopia was 1.74 and impulsiveness was 1.75. for those who were paying more than the minimum amount, the mean for overspending was 2.55, risk aversion was 2.94, myopia was 1.94 and impulsiveness was 2.20. for those who were paying the minimum or less, the mean for overspending was 2.57, risk aversion was 2.82, myopia was 2.12 and impulsiveness was 2.35. behavioral factors related to overspending, myopia and impulsiveness have shown significant initial differences in predicting whether an individual will make full payment of the outstanding balance, or paying more than the minimum amount, or paying the minimum amount or less. a significant relationship between overspending and credit card debt was also observed by barboza et al. (2017) and wong and lynn (2020). the findings of this study are also in line with those 66 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 in gathergood (2012) and ottaviani and vandone (2018) which had observed a significant relationship between impulsiveness and debt holdings. additionally, the significant link observed between myopia and credit card debt repayment is similar to the findings of kuchler and pagel (2018). table 4 group means, wilks’ lambda (u-statistic) and univariate f-ratio variable group 1 mean group 2 mean group 3 mean total mean wilks’ lambda f signif. overspending 1.74 2.55 2.57 2.2 0.849 39.09 0.000*** risk aversion 2.73 2.94 2.82 2.84 0.993 1.492 0.226 myopia 1.74 1.94 2.12 1.87 0.986 3.029 0.049** impulsive 1.75 2.2 2.35 2.02 0.948 12.049 0.000*** note. group 1 refers to those who are making full payment of the outstanding balance, group 2 refers to those who are paying more than the minimum amount, and group 3 refers those who are paying the minimum or less; ***p<0.01, **p<0.05, *p<0.1 log determinants and box’s m tables box’s m tests the null hypothesis that the covariance matrices do not differ between groups formed by the dependent. box’s m tests the equality of group covariance matrices, and measures the multivariate normality of the data, whereby approximate f is a transformation that tests whether the determinants from each level of the dependent variable differ significantly from each other. it is desirable that this test not be significant so that the null hypothesis that the groups do not differ can be retained. for this assumption to hold, the log determinants should be equal. table 5 log determinants table log determinants group rank log determinant full payment of the outstanding balance 5 0.091 paying more than the minimum amount 5 -0.242 paying the minimum amount or less 5 0.439 note. the ranks and natural logarithms of determinants printed are those of the group covariance matrices. 67 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 table 6 box’s m test result table test results box’s m 35.629 f approx. 1.158 df1 42 df2 75247.506 sig. 0.253 note. tests null hypothesis of equal population covariance matrices. in box’s m test, a non-significant m is required to show similarity and a lack of significant differences. in this case, the log determinants appeared similar and box’s m was 35.629, with f = 1.158. the p-value of 0.253 indicates that the data did not differ significantly from the multivariate normal. eigenvalues eigenvalues provide information on each of the discriminate functions (equations) produced. the maximum number of discriminant functions produced is the number of groups minus 1. a large eigenvalue is associated with a strong function (1.5 to 2.5 is acceptable). in this study, it was found that the eigenvalue for function 1 was 0.200, which was stronger than function 2 (0.005). canonical correlation the canonical correlation is the multiple correlation between the predictors and the discriminant function. there were three groups tested in the study, namely ‘full payment of the outstanding balance’, ‘paying more than the minimum amount’ and ‘paying the minimum amount or less’. there are two functions displayed in a canonical correlation. the correlation indicates that function 1 moderately discriminates between groups, while function 2 weakly discriminates between groups (1.00 is perfect). in this study, a canonical correlation of .408 suggested that the model in function 1 explained 16.65 percent of the variation in the grouping variable, i.e. whether a respondent made a full payment of the outstanding balance or was paying more than the minimum amount. for function 2, a canonical correlation of 68 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 0.069 suggested that the model explained 0.48 percent of the variation in the grouping variable, i.e. whether a respondent was paying more than the minimum amount or paying the minimum amount or less. wilks’ lambda wilks’ lambda indicates the significance of the discriminant function. in this study, wilks’ lambda indicated a highly significant function (p < 0.000) and provided the proportion of total variability not explained, i.e. it was the converse of the squared canonical correlation. for function 1, 83.35 percent of the variation were unexplained while for function 2, 99.5 percent of the variation in the grouping variable were unexplained. a small lambda in function 1 indicated that the three group means appeared to differ between those who made full payment of the outstanding balance, paid more than the minimum amount, and paid the minimum amount or less. a high chi-square value indicated that function 1 discriminated well (p ≤ 0.001) while function 2 did not (p = 0.719). table 7 eigenvalues, canonical correlation, and wilks’ lambda table test of function eigen value % of variance cum % canonical correlation test of function wilks’ lambda x2 df sig. 1 0.200 97.7 97.7 0.408 1 through 2 0.829 82.049 10 0.000 2 0.005 2.3 100.0 0.069 2 0.995 2.089 4 0.719 the canonical discriminant function coefficient unstandardized coefficients (b) were used to create the discriminant function (equation). they operated just like a regression equation. the discriminant function coefficients (b) or standardized form beta both indicate the partial contribution of each variable to the discriminate function, controlling for all other variables in the equation. they can be used to assess each iv’s unique contribution to the discriminate function and thus, provide information on the relative importance of each variable. if there are any dummy variables, as in regression, individual beta weights cannot be used and dummy variables must be assessed as a group through hierarchical discriminant analysis; running the analysis, first without the dummy variables then with 69 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 them. the difference in squared canonical correlation indicates the explanatory effect of the set of dummy variables. table 8 unstandardised canonical discriminant function coefficients function 1 2 overspending 0.79 -0.572 risk aversion 0.011 -0.432 myopia -0.011 0.467 impulsive 0.127 0.506 (constant) -2.895 -0.22 based on the list of coefficients (and the constant) of the discriminant equation in table 8, the discriminant function/equations are as follows: function 1 d1 = -2.895 + 0.79(overspending) + 0.011(risk aversion) – 0.011(myopia) +0.127 (impulsiveness) function 2 d2 = -0.220 – 0.572(overspending) -0.432(risk aversion) + 0.467(myopia) + 0.505 (impulsiveness). for function 1, the overspending score was the strongest predictor in importance as a predictor. this variable with a large coefficient stood out as the one that strongly predicted allocation between making full payment of the outstanding balance or paying more than the minimum amount group. impulsiveness, risk aversion, and myopia scores were less successful as predictors. for function 2, all the variables seemed not to be different from each other and may be less able to be predictors which allocate between paying more than the minimum amount or paying the minimum amount or less group. classification table finally, there is the classification stage. the classification table, is a table in which the rows are the observed categories of the dependent 70 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 and the columns are the predicted categories. when prediction is perfect, all cases will lie on the diagonal. the percentage of cases on the diagonal is the percentage of correct classifications. the cross validated set of data is a more authentic presentation of the power of the discriminant function than that provided by the original classifications, and often produces a poorer outcome. the crossvalidation is normally termed a ‘jack-knife’ classification. it develops a discriminant function by successively classifying all cases according to categories. this is repeated until no case is left out in the process. table 9 classification table actual group predicted group membership total full payment of the outstanding balance paying more than the minimum amount paying the minimum amount or less full payment of the outstanding balance 133 (68.6%) 61 (31.4%) 0 (0%) 194 (100%) paying more than the minimum amount 57 (28.8%) 141 (71.2%) 0 (0%) 198 (100%) paying the minimum amount or less 17 (33.3%) 34 (66.7%) 0 (0%) 51 (100%) total 207 236 0 443 (100%) priors 0.438 0.447 0.115 1.00 note. 61.9% of original grouped cases correctly classified. classification gives information about actual group membership vs. predicted group membership. from table 6 it can be seen that 61.9 percent of original grouped cases were correctly classified by all the independent variables (overspending, risk aversion, myopia and impulsiveness). the combination of these variables was better at predicting those who were paying more than the minimum amount (71.2%) than those who made full payment of the outstanding balance (68.6%) and those who were paying the minimum amount or less. this seemed to suggest that other factors could be influencing the payment decision of the last group. ferreira et al. (2021) found that over-indebtedness was caused by the lack of financial literacy and 71 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 social-economics factors. meanwhile, leandro and botelho (2022) found that overindebtedness could be caused by materialism, while (french & mckillop, 2016) found that it could be caused by the lack of money management skills. the findings of this study have reaffirmed the relationship between behavioral biases and credit card repayments, providing support for hypothesis 2. conclusion rising consumer debt has become a major concern in malaysia. bankruptcies related to credit card loan has increased over the years especially among the younger generations. the existing literature in consumer finance highlights the role of behavioral biases in influencing over indebtedness. this study contributes to the literature by analyzing the link between behavioral biases and credit card repayment among malaysians. it has been able to provide support for the proposal that credit card repayments vary according to users’ demographic profile. additionally, the study has shown that behavioral biases can be used to predict credit card repayments. as far as credit card repayment decision is concerned, the study found that it differed according to the academic qualification, income, race and marital status of the respondents. the findings showed that those with a higher academic qualification and income made better repayments decisions. additionally, it was revealed that chinese respondents made significantly better repayments than the malays. similarly, married respondents make better repayments than the singles. moreover, the discriminant analysis showed that 61.9 percent of original grouped cases were correctly classified by the behavioral factors related to overspending, risk aversion, myopia and impulsiveness. however, the findings seemed to suggest that these variables were better at predicting those who were paying more than the minimum amount and those who made full payment of the outstanding balance, but not those who only paid the minimum amount or less. as far as the predictor variables were concerned, it was found that overspending remained as the strongest predictor which discriminated between these groups, followed by risk aversion, myopia and impulsiveness. this study has contributed to the body of knowledge on consumer finance by analysing the link between behavioral biases and credit 72 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 card repayment among malaysians. for examplw, it was shown that behavioral biases have deterred consumers from making an optimal decision. hence, it is very crucial to identify how these biases influence credit card repayments, so as to be able to ensure that appropriate measures can be taken to alleviate them. the existing literature has highlighted that individuals who were less vulnerable to behavioral biases made better financial decisions. consistent with this general consensus, the present study has provided further evidence that credit card holders who earned higher income and had better academic qualifications usually demonstrated better repayment decisions. it is thus, clear that measures taken by the regulators to improve credit card repayments need to focus more on certain group of individuals who are more vulnerable than others. additionally, regulators also need to take measures to educate consumers about behavioral biases related to overspending, impulsiveness, myopia and risk taking. this can be done through awareness campaigns in the regular and social media. overall, the present study has concluded that the behavioral biases studied were able to better predict those who would make better repayment decisions, but has failed to identify those who didn’t. based on its findings, the practical implication that can be derived is that loan providers can benefit from the perceptual map, by mapping loan applicants based on their behavioural biases and consumer traits. one is in a better position to predict whether the loan applicant will make better repayment decisions. this will certainly assist the relevant personnel in charge of loans make better decisions pertaining to loan approval. however, this study has several limitations. firstly, the sample used did not represent the general population in malaysia as it was only focused on the northern region of west malaysia. secondly, the failure of behavioral biases in predicting those who only made the minimum or lower payment seems to suggest that a more comprehensive model that incorporates other factors that influence financial decision making need to be considered in future research. future studies should explore the role of other types of behavioral biases,for example, attitudes and personality traits. lastly, future studies also can look into different types of behavioral biases related to credit card repayment, such as mental accounting, availability bias, limited attention and overconfidence bias. 73 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 acknowledgment this work was supported by a universiti sains malaysia (usm) research university grant no. 1001/pjjauh/8016028. references ahmad, z., ibrahim, h., & tuyon, j. (2018). governance of behavioural biases in asset management industry: insights from fund managers in malaysia. asian academy of management journal of accounting and finance, 14(2), 65–102. https://doi. org/10.21315/aamjaf2018.14.2.4. aydin, a. e. (2022). psychological and demographic factors influencing responsible credit card debt payment. journal of financial services marketing. palgrave macmillan uk. 27(1), 17–26. https://doi.org/10.1057/s41264-021-00094-0. barberis, n., & thaler, r. (2003). a survey of behavioral finance. handbook of the economics of finance, 1, 1053–1128. http:// www.nber.org/papers/w9222. barboza, g. (2018). i will pay tomorrow, or maybe the day after. credit card repayment, present biased and procrastination. economic notes, 47(2–3), 455–494. https://doi.org/10.1111/ecno.12106. barboza, g. a., smith, c., & boubacar, i. (2017). a contribution to the empirics of consumers’ anxiety behavior on and in credit card repayment. credit card management and financial literacy among college student. journal of financial management, markets and institutions, (1), 35–66. https://doi. org/10.12831/87059. barlow, h. d. (1991). explaining crimes and analogous acts, or the unrestrained will grab at pleasure whenever they can. the journal of criminal law and criminology, 82(1), 229-297. https://scholarlycommons.law.northwestern.edu/jclc/vol82/ iss1/11 baumeister, r. f. (2002). yielding to temptation: self-control failure, impulsive purchasing, and consumer behavior. the journal of consumer research, 28(4), 670–676. brahmana, r., hooy, c. w., & ahmad, z. (2015). monday irrationality of investors in bursa malaysia: the role of psychological biases. malaysian journal of economic studies, 52(2), 227–243. 74 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 brown, m., & graf, r. (2013). financial literacy, household investment and household debt: evidence from switzerland. working papers on finance. 13/1. http://www1.vwa.unisg.ch/ repec/usg/sfwpfi/wpf-1301.pdf. ceschi, a., costantini, a., sartori, r., weller, j., & di fabio, a. (2019). dimensions of decision-making: an evidencebased classification of heuristics and biases. personality and individual differences, 146(november 2017), 188–200. https:// doi.org/10.1016/j.paid.2018.07.033. chin, k. y., yi, w. c., kwan, c. l., & si, c. m. (2017). an empirical investigation on credit card repayment pattern of academicians in malaysia. the international journal of banking and finance, 13(2), 21–42. https://doi.org/10.4018/978-1-4666-6268-1.ch065. chong, f. n. (2017). factors affecting credit card debt payment in malaysia. international journal of business and economics, 16(1), pp. 21–30. chybalski, f., & marcinkiewicz, e. (2018). myopic behaviour and state involvement in a pension system: a cross-section study for oecd countries. ekonomista, 1, 68–88. combrink, s., & lew, c. (2020). potential underdog bias, overconfidence and risk propensity in investor decision-making behavior. journal of behavioral finance. 21(4), 337–351. https://doi.org/10.1080/15427560.2019.1692843. ferreira, m. b., pinto, d. c., herter, m. m., soro, j., vanneschi, l., castelli, m., & peres, f. (2021). using artificial intelligence to overcome over-indebtedness and fight poverty. journal of business research, 131, 411–425. french, d., & mckillop, d. (2016). financial literacy and overindebtedness in low-income households. international review of financial analysis. elsevier b.v., 48, 1–11. https://doi. org/10.1016/j.irfa.2016.08.004. gathergood, j. (2012). self-control, financial literacy and consumer over-indebtedness’. journal of economic psychology. elsevier b.v., 33(3), 590–602. https://doi.org/10.1016/j. joep.2011.11.006. george, d., & mallery, p. (2003). spss for windows step by step: a simple guide and reference (4th ed.). allyn & bacon. gerardi, k., goette, l., & meier, s. (2010). financial literacy and subprime mortgage, federal reserve bank of atlanta working paper. no. 2010-10. 75 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 grund, c., & sliwka, d. (2010). evidence on performance pay and risk aversion. economics letters. elsevier b.v., 106(1), 8–11. https://doi.org/10.1016/j.econlet.2009.09.005. hamid, f. s., rangel, g. j., taib, f. m., & thurasamy, r.. (2013). the relationship between risk propensity, risk perception and risk-taking behaviour in an emerging market. the international journal of banking and finance, 10(1), 134–146. hamid, f. s., & loke, yiing jia. (2021a). financial literacy, money management skill and credit card repayments. international journal of consumer studies, 45(2), 235–247. https://doi. org/10.1111/ijcs.12614. hamid, f. s., & loke, yiing jia. (2021b). information disclosure and credit card repayments: evidence from malaysia. international journal of bank marketing, 39(7), 1479–1497. https://doi. org/10.1108/ijbm-03-2021-0082. heidhues, p., & koszegi, b. (2010). exploiting naivete about selfcontrol in the credit market. american economic review, 100(5 december), 2279-2303. https://www.jstor.org/stable/41038763 huebner, j., fuchs, k., & ilic, a. (2018). preventing over-spending: increasing salience of credit card payments through smartphone interventions. mkwi 2018 multikonferenz wirtschaftsinformatik, 1075–1082. jaiyeoba, h. b., abdullah, m. a., & ibrahim, k. (2020). institutional investors vs retail investors: are psychological biases equally applicable to investor divides in malaysia? international journal of bank marketing, 38(3), 671–691. https://doi. org/10.1108/ijbm-07-2019-0242. jusoh, z. m., & lin, l. y. (2012). personal financial knowledge and attitude towards credit card practices among working adults in malaysia. international journal of business and social science, 3(7), 176–185. kuchler, t., & pagel, m. (2018). sticking to your plan: the role of present bias for credit card paydown. national bureau of economic research. no. w24881. leandro, j. c., & botelho, d. (2022). consumer over-indebtedness: a review and future research agenda. journal of business research, 145(march), 535–551. https://doi.org/10.1016/j. jbusres.2022.03.023. limerick, l., & peltier, j. w. (2014). the effects of self-control failures on risky credit card usage. marketing management journal, 24(2), 149–161. 76 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 lin, w.-r., wang, y.-h., yang, f. -j., & sekarini, n. s. (2015). the effect of credit card usage on compulsive buying behavior intention. journal of accounting, finance & management strategy, 8(2), 45–74. https://doi.org/10.5824/1309-1581.2015.3.002.x. lo, h. y., & harvey, n. (2011). shopping without pain: compulsive buying and the effects of credit card availability in europe and the far east. journal of economic psychology. 32(1), 79–92. https://doi.org/10.1016/j.joep.2010.12.002. loke, y. j., yen, s. t., & tan, a. k. g. (2011). credit card averters and amassers: evidence from malaysia. asian economic journal, 25(4), 397–412. https://doi.org/10.1111/j.14678381.2011.02068.x. meier, s., & sprenger, c. (2010). present-biased preferences and credit card borrowing. american economic journal: applied economics, 2(1), 193–210. https://doi.org/10.1257/ app.2.1.193. ottaviani, c., & vandone, d. (2018). financial literacy, debt burden and impulsivity: a mediation analysis. economic notes, 47(2– 3), 439–454. https://doi.org/10.1111/ecno.12115. pak, o., & mahmood, m. (2015). impact of personality on risk tolerance and investment decisions: a study on potential investors of kazakhstan. international journal of commerce and management, 25(4), 370–384. https://doi.org/10.1108/ ijcoma-01-2013-0002. palan, k. m., morrow, p. c., trapp, a., & blacburn, v. (2011). compulsive buying behavior in college students: the mediating role of credit card misuse. journal of marketing theory and practice, 19(1), 81–96. https://doi.org/10.2753/mtp10696679190105. peltier, j. w., pomirleanu, n., endres, m., & markos, e. (2013). psycho-social factors impacting credit acquisition and use by college students. journal of financial services marketing, 18(4), 271–284. https://doi.org/10.1057/fsm.2013.20. pirog, s. f., & roberts, j. a. (2007). personality and credit card misuse among college students: the mediating role of impulsiveness. journal of marketing theory and practice, 15(1), 65–77. https://doi.org/10.2753/mtp1069-6679150105. ramayah, t., noor, n., & lim, h. c. (2002). cardholders’ attitude and bank credit card usage in malaysia: an exploratory study. asian academy or management journal, 7(1), 75–102. 77 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 salisbury, l. c. (2014). minimum payment warnings and information decisions. journal of public policy & marketing, 33(1), 49–64. salisbury, l. c., & zhao, m. (2020). active choice format and minimum payment warnings in credit card repayment decisions. journal of public policy and marketing, 39(3), 284–304. https://doi. org/10.1177/0743915619868691. sekaran, u., & bougie, r. (2014). research methods for business: a skill-building approach (6 th ed.). john wiley & sons. sharma, a., & kumar, a. (2019). a review paper on behavioral finance: study of emerging trends. qualitative research in financial markets. https://doi.org/10.1108/qrfm-06-20170050. shefrin, h. m., & thaler, r. (1981). an economic theory of selfcontrol. journal of political economy, 89(2), 392–406. soman, d. (2001). effects of payment mechanism on spending behavior: the role of rehearsal and immediacy of payments. journal of consumer research, 27(4), 460–474. https://doi. org/10.1086/319621. sotiropoulos, v., & d’astous, a. (2012). social networks and credit card overspending among young adult consumers. journal of consumer affairs, 46(3), 457–484. https://doi.org/10.1111/ j.1745-6606.2012.01239.x. stavins, j. (2020). credit card debt and consumer payment choice: what can we learn from credit bureau data? journal of financial services research, 58(1), 59–90. https://doi.org/10.1007/ s10693-019-00330-8. sunstein, c. r. (2006). boundedly rational borrowing. university of chicago law review, 73(1), 249–270. teoh, w. m., chong, s., & yong, s. m. (2013). exploring the factors influencing credit card spending behavior among malaysians. international journal of bank marketing, 31(6), 481–500. https://doi.org/10.1108/ijbm-04-2013-0037. the malaysian reserve. (2019). credit cards, personal loans landing malaysians in debt trap. 8 august. https://themalaysianreserve. c o m / 2 0 1 9 / 0 8 / 0 8 / c r e d i t c a r d s p e r s o n a l l o a n s l a n d i n g malaysians-in-debt-trap/ theong, m. j., osman, a. f., & yap, s. f. (2018). household indebtedness: how global and domestic macro-economic factors influence credit card debt default in malaysia. institutions and economies, 10(3), 37–56. 78 the international journal of banking and finance, vol. 18, number 2 (july) 2023, pp: 53–78 wang, l. l., lu, w., & malhotra, n. k. (2011). demographics, attitude, personality and credit card features correlate with credit card debt: a view from china. journal of economic psychology. 32(1), 179–193. https://doi.org/10.1016/j.joep.2010.11.006. wang, l., lv, w., & jiang, l. (2011). the impact of attitude variables on the credit debt behavior. nankai business review international, 2(2), 120–139. https://doi.org/10.1108/20408741111139909. wei, c.-y., chin, l. -k., kuah, y. -c., & chia, m. -s. (2018). attributes of academicians’ credit card usage behaviors in malaysia. journal of contemporary issues and thought, 8, 73–85. https:// doi.org/10.37134/jcit.vol8.8.2018. wong, k. y., & lynn, m. (2020). credit card cue effect: how mere exposure to credit card cues promotes consumers’ perceived financial well-being and spending. international journal of bank marketing, 38(2), 368–383. https://doi.org/10.1108/ ijbm-01-2019-0010. world bank. (2019). malaysian economic monitor: making ends meet. https://doi.org/10.1787/g2355e556-en. 25 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 how to cite this article: ariff, m., & zarei, a. (2022). are domestic firms exposed to similar currency risk as international trading firms? international journal of banking and finance, 17(2), 25-56. https://doi. org/10.32890/ ijbf2022.17.2.2 are domestic firms exposed to similar currency risk as international trading firms? 1ariff mohamed & 2alireza zarei 1 department of economics and finance sunway university, malaysia 2coventry university, united kingdom corresponding author: ariff13@gmail.com received: 15/4/2021 revised: 19/5/2021 accepted: 20/5/2021 published: 27/6/2022 abstract this paper reports key findings about currency risk using two samples of listed firms: one sample with zero foreign currency revenues, hence having zero-currency risk; and the other sample with positive revenues in foreign currencies from foreign transactions. the latter is therefore, exposed to currency risk. asset pricing theories predict that stocks of currency-risk-exposed firms should suffer significant currency risk, while those firms with zero-currency-risk should not have any effect from currency risk since currency transactions across borders is nil. the latter hypothesis has yet to be tested explicitly, so there is a gap in the literature. we report stock returns are significantly affected not just for firms with foreign-currency revenues but also for firms with zero foreign-currency transactions. these findings are useful to top http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 26 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 management of all businesses to undertake currency-hedge plans for both domestic and international trading firms. keywords: exchange rates, direct vs indirect exposure, panel regression, australian dollar, pooled vs fixed vs random effects. jel classification: f23, f31, g12. currency exposure risk is a top management concern this paper reports new findings on how the australian dollar movements over 37 years affect both australia’s exchange-listed multinational firms, as well as domestic firms with zero foreignsource cash flows. being an asia pacific country, a research using australian data in the year after a major regime change to free-float currency could yield lessons for businesses in the region’s developed countries: japan, malaysia, south korea, taiwan and singapore, as well as emerging economies. currency-risk-effect on stock prices in the asia pacific region has not been studied using theories developed in jorion (1990) or solnik (1974), two main asset pricing theories about currency as price-relevant factor for stocks. in contrast, there have been lots of studies using the parity theorems, for example, as in ariff and zarei (2019). top management of businesses in all countries continues to grapple with the significant currency risk to their firms ever since 1973, after the demise of the breton woods fixed-exchange-gold-backed monetary system. a free-floating regime started resulting in upticks in volatility of exchange rates across all currencies. “do currency movements affect stock returns because a business dependent on foreign-origin cash flows necessarily will reduce/appreciate the value of such flows depending on how the a$ moves up/down when cash flows received or paid are converted to other currencies? should it affect stocks of international trading australian firms earning multicurrency cash flows? a neglected research question is the converse: are stocks of domestic firms with zero exposure to foreign cash flows into their accounts affected by currency movements? this latter question has yet to be addressed satisfactorily, hence businesses generally do not take action to offset currency-originated losses if a 27 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 typical firm has zero transactions in the books. a number of offsetting hedging plans are executed by the management of firms transacting in foreign currencies. australia adopted a clean free-floating regime in 1984.1 this led to increased volatility of the australian dollar (a$). this also happened to several other currencies switching to free floating. volatility on the one hand and the periodic depreciation of the currency value when down-cycles in economic outputs introduce greater business risk from currency movements to both types of firms. thus, the top management of investing firms faced with increased currency risk incurs greater costs, which they have to monitor and, if possible to put in place expensive currency hedge management to reduce the impact from the currency risk. if top management ever had evidence that domestic firms are prone to suffer currency risk a topic on which there has been a lack of research evidence businesses could have taken remedial actions to engage in hedging the risk. thus, the domestic firm’s zeroforeign cash flow exposure to currency is an area to clarify and also show evidence to the practitioners of business management. the motivation for this research has been, thus from the need to test the three strands of currency theories, study the currency risk of zeroforeign-source firms, since theories tested to-date assume that such firms are unaffected by currency risk, and adopt a fama-mcbeth type portfolio aggregation method in order to control the idiosyncratic errors in estimated parameters. failure to control idiosyncratic errors when individual firm observations were used has been shown in the literature as a possible source of error in almost all past studies to-date. the value of a listed firm with transactions in foreign currencies depends on how much there is direct currency exposure. it is reasonable to argue that firms with zero transaction in foreign currencies in the book may have the currency risk coming through other sources, such as when a domestic firm buys items from wholesalers importing foreign items, as it will usually charge a higher price if the currency has depreciated. what management knows from received theories and management literature is that each time a currency value changes, the overall value of a firm’s cross-border earnings/funds change. this impacts stock returns and has been studied using samples of firms with direct exposures to currency risk in the united states (us) listed firms (agrawal & 28 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 harper, 2010). such is not the case for other asia pacific nations of which the australian economy represents a developed economy larger than most, except china, india and japan. this paper has been motivated to contribute to the management literature on this important management issue in the asia pacific region, especially its focus on the stock market impact of currency movements of two types of firms.2 adler and dumas (1984) was an early study showing domestic us stocks have a negative influence from currency risk, although ther``e were two other macro-factor that have yet to be tested for the us firms in that same study. our study hopes to get new findings on whether currency risk is significant in both the domestic zero-exposed and the directly exposed firms listed on the australian stock exchange over the 2000-2019 period, this time frame was selected because it covered a turbulent exchange rate period. it was during this period, that the value of the usd had depreciated from its pre-2014 rate of us$1.31=a$1.00 to us$ 0.77=a$1.00 in 2019. this paper reports a more refined finding on the degree of currency impacts on different degrees of revenue-exposures by selecting the five portfolios of zero to high-risk-exposure firms. the present study found that stock price reaction was greater if a firm had greater cross-border transactions: importantly zero-exposure firms had as much exposure as directly exposed firms to currency risk. this last aspect, namely against the positive exposure of international transacting firms has yet been known for any country using, as in this study, two key asset pricing theories at macro-level instead of at firm level, as relevant for currency studies. the stock prices of firms exposed directly to the exchange rate risk are evidently affected by currency exchange rate changes, as has been predicted by the adler-dumas theorem or solnik (1974). there has been no prior evidence of a significant currency effect on the zeroexposed domestic firms nor has there been a study testing the degree of exposure using portfolio aggregation method. khoo (1994) is a study of currency exposure of selected australian industries, but not domestic or international trading firms. the econometric method applied in this study also leads to quite robust results. the present study has applied panel regression, using pooled, random and fixed effects procedures, as well as lm tests (hausman, 1978; breusch & pagan, 1980). 29 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 the rest of the paper is organized into five sections. the next section is a brief summary of australian dollar studies using usand ukcurrencies. the third section contains a brief discussion on the underlying theories of asset pricing with exchange rate as a macro factor, while in the ensuing methodology section, details of the test models and research hypotheses, as well as data collection and methodology are discussed. the results are presented and discussed in the section before the conclusion. the interesting new finding is that the presumed zeroexposed domestic firms are also significantly affected by exchange rate changes, as are the directly-exposed international firms. in addition, the impact on stock price depends on the degree of currency exposure ranging from heavily-exposed to zero-exposed portfolios of listed firms, unless hedge is in place. this new finding ought to urge top management of zero-currency exposed firms to take cover currency hedging, as is already occurring in the case of directly exposed firms. australian dollar (a$) exposure the aud exchange rate has been subjected to large fluctuations ever since the commodity boom of the 1963-1980 period ended, as well as the abandonment of managed exchange rate policy of the reserve bank of australia in 1984. this policy change was the first attempt to establish an independent control on monetary policy, with a view to eliminating speculative attacks on the then much stronger aud prior to 1984.3 the aud was overvalued then and was a favorite reserve currency of central bankers before its significant reform. ever since 1984, mainly as a result of the reform, firms began to be substantially exposed to exchange rate fluctuations, although no study to-date has fully revealed exchange rate effects on indirectly-exposed domestic firms, indeed even the directly exposed firms (except for one study? citation at industry level). the total value of the stock market was aud 1.63 trillion with 2,186 listed firms as at 2019. this made the market one of the biggest in the world. figure 1 is a plot of aud against the usd and the british pound over the long period from 1971 to 2019. 30 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 figure 1 a$1.00 equivalent to us$ market closing rate at end of each year, 1983 and 2019 high volatility was quite evident in the aftermath of 1983, especially in favor of the usd and great britain sterling pound until recent few years. on the onset of the establishment of the free-floating exchange rate regime in 1984 and up till the late 1980s, the aud faced relatively higher inflation compared to most other oecd countries. high inflation at that time was associated with a long-run nominal exchange rate depreciation against major currencies such as the yen, deutschemark (a free-floating former imf currencies) and the trade weighted basket of currencies. table 1 descriptive statistics of exchange rates mean median std. dev. skewness kurtosis a$gbp 2.072 2.080 0.380 0.124 2.071 a$us$ 1.207 1.250 0.305 0.322 2.572 over the period shown in the figure, the uk pound is worth on average about 2.1 usd, and the usd is worth 1.2 aud. starting with the year 2001 onwards, the aud appreciated significantly on the back of a commodity boom worldwide, as was also the case prior to 1980. thus, a long period of depreciation was followed by an appreciation of 4 the aud exchange rate has been subjected to large fluctuations ever since the commodity boom of the 1963-1980 period ended, as well as the abandonment of managed exchange rate policy of the reserve bank of australia in 1984. this policy change was the first attempt to establish an independent control on monetary policy, with a view to eliminating speculative attacks on the then much stronger aud prior to 1984.3 the aud was overvalued then and was a favorite reserve currency of central bankers before its significant reform. ever since 1984, mainly as a result of the reform, firms began to be substantially exposed to exchange rate fluctuations, although no study to-date has fully revealed exchange rate effects on indirectly-exposed domestic firms, indeed even the directly exposed firms (except for one study? citation at industry level). the total value of the stock market was aud 1.63 trillion with 2,186 listed firms as at 2019. this made the market one of the biggest in the world. figure 1 is a plot of aud against the usd and the british pound over the long period from 1971 to 2019. figure 1 a$1.00 equivalent to us$ market closing rate at end of each year, 1983 and 2019 high volatility was quite evident in the aftermath of 1983, especially in favor of the usd and great britain sterling pound until recent few years. on the onset of the establishment of the free-floating exchange rate regime in 1984 and up till the late 1980s, the aud faced relatively higher inflation compared to most other oecd countries. high inflation at that time was associated with a long-run nominal exchange rate depreciation against major currencies such as the yen, deutschemark (a freefloating former imf currencies) and the trade weighted basket of currencies. over the period shown in the figure, the uk pound is worth on average about 2.1 usd, and the usd is worth 1.2 aud. starting with the year 2001 onwards, the aud appreciated significantly on the back of a commodity boom worldwide, as was also the case prior to 1980. thus, a long period of depreciation was followed by an appreciation of the currency in part in the recent 12-year period ending in 2017. both upand down-movements would severely limit the value of firms, since the perverse effect of currency moves is in both directions. the period chosen is a very recent period, when the aud appreciated only to depreciate later. table 1 is 31 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 the currency in part in the recent 12-year period ending in 2017. both upand down-movements would severely limit the value of firms, since the perverse effect of currency moves is in both directions. the period chosen is a very recent period, when the aud appreciated only to depreciate later. table 1 is summary of descriptive statistics of the australian exchange rate against the us dollar and the great britain pound. the figures are in currency values and not in log change. two world-renowned examples are the bhp-billiton and nab (national australia bank) with more than half their trading revenues coming from non-domestic operations. any sort of fluctuations in the behavior of aud will therefore influence the profitability of such companies, thus also affecting the asset value of such firms. on the other hand, domestic australian firms whose cash flows were fully dominated in the local currency were assumed to have no effect from currency movements. a 2010 article by aggarwal & harper op cit., was about the us firms with zero foreign incomes. other than that paper on us firms, no studies have pointed out the behavior of indirectly exposed domestic firms. asset pricing theories and evidence asset pricing theories the present study has examined how stock returns computed from stock prices were influenced by currency exchange rate movements, as well as by other theories that have been based on macro level factors, often ignored by finance researchers. there are two streams of literature on this topic: one that uses macro factors to explain movements in a currency and the other that uses firm-specific factors (di iorio & faff, 2002; aggarwal & harper, 2010) as driving stock returns. the exchange rate is a macro variable, not a firm-specific variable. there is legitimacy in using macro factors alone – unlike the urge by scholars to choose the beaten track of firm-specific factors (aggrawal-harper op cit.) using asset pricing theories. in this study, the jorion and/or solnik model was applied to study, within the macro factor framework, how the known macro factors together will influence stock returns as suggested by the currency-relevant theories. adler and dumas (1984) has provided a start in the attempt to develop 32 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 a model of macro factors affecting stock prices. in their model, the currency’s impact on the firm’s stock returns is as expressed in equation (1): (1) where, measures the total exposure of firm to the exchange rate and denotes the returns of the firm over time period to the country currency changes. this is similar to the approach in the earlier market model (sharpe, 1964), which has aimed at showing how another macro factor systematic risk on the stock market as a whole, could influence individual share values of firms. the adler-dumas model has considered the exchange rate risk as an international macro factor in equation (1), while in equation (2) from sharpe (1974), it was considered a domestic macro factor. the domestic risk variable has been modelled as in equation (2) in sharpe’s (1964) market model for an individual firm, as the marginal sensitivity of individual stock returns to a macro stock market factor: (2) where, is the domestic systematic risk and are the residuals of the regression. the residual captures the excluded firm-specific factor effects, whereas the systematic macro factor effect on stock returns is the risk, measured in the stock return sensitivity to market returns as the beta, . solnik (1974) has specified a model independently – in a sense amalgamating the two equations into one – in which both the macro factors of stock market and exchange rate, a second macro factor, have been incorporated. this is the two-factor model, with a macro stock market as the first factor and the international macro factor as sourced from currency risk represented as equation (3): (3) where, denotes foreign currency risk premium (that is, the rate of change in foreign currency minus the country’s risk-free rate) as an additional factor to the market factor risk premium , with i 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 33 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 i indicating individual stocks similar to that of sharpe op cit. thus, in solnik’s model, an international systematic risk (isr) factor is added as the currency risk while the macro market risk is the this model derives a risk pricing relation for individual stocks as coming from exchange rate risk as in adler and dumas (1984). this has been modeled and tested for a large number of countries (ariff & marisetty, 2012). the measure of the isr, is computed by running a regression using stock returns of a country’s market index against the returns from an international stock market index. that is, by rerunning the equation (2) with returns from an international market index, one may get a measure of the isr, . in other words, the isr is estimated first using equation (4): (4) where, is the international index return; and is the market index return for a country j. (jorion, 1991) has specified a direct method of testing the two macro factor effects on the individual stocks as in equation (5): (5) equation (5) incorporates the exchange rate as an additional macro risk factor in an extended international market model. the exchange rate is the additional risk factor and (recall adler & dumas op cit.) is the marginal sensitivity of a firm’s stock returns to the changes, theorized as having an impact on stock returns. the result is a two-factor market model with two macro factors affecting stock returns. the actual return is a proxy using the market index, and is the exchange rate. the impact of the currency exposure is measured by the value of as in equation (5). are there other macro factors that could be added to this two-factor macro model? to answer this, reference is made to the intuition in ross (1976). at this point, there is the need to review theories using macro factors to build a model in order to link the market factor and the exchange rate factors in an extended jorion’s model. chen et al. (1986) operationalised a much more general multi-factor test model using ross’s (1976) theory. chen identified three more macro factors as 1 are domestic firms exposed to similar currency risk as internationally-trading firms? 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝑣𝑣𝑖𝑖,𝑡𝑡 (1) , γi 𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 and 𝑅𝑅𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑡𝑡 𝑗𝑗’s 𝛽𝛽𝑖𝑖 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (2) 𝛽𝛽𝑖𝑖 𝜀𝜀𝑖𝑖,𝑡𝑡 𝐸𝐸(𝑅𝑅𝑖𝑖 ) = 𝑅𝑅𝑓𝑓 + 𝛽𝛽𝑖𝑖,1(𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝛽𝛽𝑖𝑖,2(𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹)𝑖𝑖 (3) 𝐹𝐹𝐹𝐹𝑅𝑅𝐹𝐹𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝛽𝛽𝑖𝑖,1. 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 34 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 having influences on stock returns by identifying the risk premium (mp), industrial production as growth in gdp, and term structure of interest rate (uts) and unanticipated inflation (ui), as well as changes in expected inflation (dei) as significant contributory factors for stock returns, notably all these are macro-level variables as in equation (6): (6) the term structure of interest rates is is the lending and savings rate; inflation; and growth in gdp. term structure is measured in that article as the difference between (a) yield on a long-dated treasury, and (ii) yield on a short-dated treasury, the rate of change in the industrial production index (ipi), on which there are data series readily available as a proxy for economy-wide earnings, can be used as a substitute for gdp growth in the chen et al. (1986) study. gdp growth rate and the ipi are highly correlated, so they are multicollinear, thus one factor has to be dropped. this line of reasoning provides more macro factors to the already specified two factors in jorion’s (1991) and solnik’s (1974) models. thus, there are four macro factors all chosen as non-firm-specific factors, unlike in aggarwal-harper (op. cit.), that could be used in building a test model to examine their joint impacts on stock prices, with the exchange rate impact being the main focus of this study. the present study intends to develop a test model in the next section using this line of reasoning, which is consistent with that in adler-dumas, jorion and chen-roll-ross. firm-specific factors in the model it is instructive at this juncture to also point to another approach taken by researchers to investigate how the exchange rate exposure on stock returns may be measured using firm-specific factors mixed with the exchange rate as a macro factor. the fama-french model is used by inserting the exchange rate, into the fama-french model, along with , while the smb and hml are firm-specific factors. another is di iorio and faff (2002). (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 2 𝛽𝛽𝑖𝑖,2, 𝑅𝑅𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (4) 𝑅𝑅𝐼𝐼𝐼𝐼,𝑡𝑡 𝑅𝑅𝑗𝑗,𝑡𝑡 𝑅𝑅𝑖𝑖,𝑡𝑡 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽𝑖𝑖 𝑅𝑅𝑚𝑚,𝑡𝑡 + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5) 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑖𝑖,𝑡𝑡 𝑅𝑅𝑚𝑚,𝑡𝑡 𝑋𝑋𝑅𝑅 𝛾𝛾𝑖𝑖 𝑅𝑅 = 𝑎𝑎 + 𝑏𝑏𝐼𝐼𝑀𝑀 𝑀𝑀𝑀𝑀 + 𝑏𝑏𝐷𝐷𝐷𝐷𝐼𝐼 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝑏𝑏𝑈𝑈𝐼𝐼 𝑈𝑈𝐷𝐷 + 𝑏𝑏𝑈𝑈𝑀𝑀𝑈𝑈 𝑈𝑈𝑀𝑀𝑅𝑅 + 𝑏𝑏𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 + 𝑒𝑒 (6) (𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑈𝑈 ) 𝑖𝑖𝐿𝐿 𝑖𝑖𝑈𝑈. 𝑋𝑋𝑅𝑅, 𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 𝑅𝑅𝑖𝑖 − 𝑅𝑅𝑓𝑓 = 𝑎𝑎 + 𝑏𝑏𝑖𝑖 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) + 𝑠𝑠𝑖𝑖 𝐷𝐷(𝑈𝑈𝑀𝑀𝑆𝑆) + ℎ𝑖𝑖 𝐷𝐷(𝐻𝐻𝑀𝑀𝐻𝐻) + 𝛾𝛾𝑖𝑖 𝑋𝑋𝑅𝑅𝑗𝑗,𝑡𝑡 + 𝜀𝜀𝑖𝑖 (7) 35 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 where, the is the market risk premium factor; is the firm-specific variable of market capitalization of small minus big firms; hml is also a firm-specific variable of high minus low bookto-market value firms, and xr is the macro factor of exchange rate changes.4 hence there are two different approaches in theory-building on how the exchange rate may affect stock returns. the model to be developed in the next section uses only the macro factors from macro theories and excludes all firm-specific factors just described. in that way, the findings in the present study are based on a multi-factor approach consistent with ross (1976) and as operationalised in chen et al. (1986) within the framework of solnik’s and jorion’s contributions to the exchange rate behavior. as can be deduced from this brief review on asset pricing theories, the concern of the present study is more with the macro factors that affect stock returns in the australian context, and perhaps this is the first such attempt ever made. empirical evidence on exchange rate and asset prices among the various studies on foreign exchange rate exposure, most identified only very low correlation between change in the stock prices and the change in the exchange rates; (e.g. bartov & bodnar, 1994; mok, 1993; griffin & stulz, 2001; and ariff & zarei, 2019) on low causal link. furthermore, despite the fact that several studies had focused on the foreign exchange exposure of multinational corporations, little is known of the domestic firm with zero-foreign currency exposure (exception is the us study using individual stocks). there appears to be a false belief, because it is presumed that non-exposure to foreign currency cash flows makes domestic firms protected from currency risk, thus it gives a wrong lesson to the top management of domestic firms. obviously this is not sound reasoning since costs of input items in a modern economy are affected by exchange rate changes, even if a firm does not have cross-country transactions in foreign currencies. mok (1993) was an example of building a causality model linking the exchange rate, interest rate and stock price to see their interdependence. the results of that study seemed to suggest that there was an only a weak bidirectional causality between the exchange rate and stock returns. similar results have been observed in past studies of the exchange rate impact on stock returns. 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 36 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 adler and dumas (1984) stated some 36 years ago that firms with no foreign operations, foreign currency assets, and foreign currency liabilities might also be exposed to the exchange rate risk, a claim which has not been followed up. their intuition has been largely ignored by later researchers. accordingly, aggarwal and harper (2010) refocused on domestic firms and showed that us domestic firms have been equally exposed to the risk of currency risk. therefore, in reviewing the empirical literature, only those studies that had looked at multinational and domestic firms to measure the currency impact on stock returns have been included. following the collapse, in 1973, of the breton woods rules on exchange rate management and subsequent introduction of newer exchange rate systems, empirical studies began to focus on the role of the exchange rate in asset pricing valuation in the international context. jorion (1990) found that the exchange rate risk was priced. dumas and solnik (1995) applied the generalized method of moments or gmm to make a conditional specification of the single factor model developed by adler and dumas (1984), which had assumed a nonstochastic inflation factor. they found that the model held statistically, and the currency effect could not be rejected. likewise, de santis and gerard (1998) and de santis et al. (2003) applied a multivariate garch-in-mean specification for a conditional estimation of the single factor model. their findings also verified the usefulness of the model. further support for the single exchange rate factor model has been reported in the studies of capiello and fearnley (2000) and dahlquist and saellstrom (2002). the two factor model of jorion (1991) has also been applied in a number of studies, such as in bodnar and gentry (1993), bartov and bodnar (1994), bartov et al. (1996), and griffin and stulz (2001), which covered non-asia pacific economies. these studies reported a high-to-modest explanatory power of the model for other-than-market-factors on stock return; except one which reported no seasonal effect. while the results were not strongly significant in terms of the exchange rate exposure, the authors explained that the lack of evidence was due to operational hedges by firms in their samples, which included multinational firms. that had a confounding effect in the results. evidence on the exchange rate and asset pricing relationship is scant in the australian market. one primary research on foreign exchange 37 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 exposure (loudon, 1993) has provided evidence of a significant sensitivity of around one in three australian industries to exchange rate movements. that study applied a multifactor asset pricing approach, using the procedure of adler and dumas (1984) and jorion (1990,1991); the study was carried out before the aud appreciated. the presence of any sort of premium for currency risk in equity returns was found to be negligible. the finding was considered a bit strange, but this was perhaps due to the simpler statistical methodology. a second study by khoo (1994) focused on the foreign exchange exposure of australian mining firms and reported only a small degree of stock return sensitivity to exchange rate movements. these are very dated studies and recent data with much stronger volatility have been ignored, leaving the possibility of observing a richer effect of the enhanced volatility of the aud in the recent two decades. likewise, other studies like the one by di iorio and faff ( 2002) found the stability of exchange rate exposure of industries by implementing a basic augmented market model, using two series of exchange rates, namely the usd and the japanese ¥. a cross-sectional test model was used and the foreign exchange exposure was also taken into consideration. while the results seemed to suggest a strong relationship between australian industry returns and usd-aud fluctuations, it was not directly evident for the japanese ¥. furthermore, the use of bilateral exchange rates was found to be more insightful than the aggregate exchange rate indices.5 a recent paper by kang et al. (2016) showed findings which corroborated the claim that there was a significant exposure of firms using two types of risk factors in their estimation procedure. data, variables, hypotheses and methodology in this section, there will be a description of the research design to study the relationship between currency exchange rate changes and stock returns of australian firms, with a focus on identifying the exchange rate as a significant factor for both zero-exposed and directly exposed firms transacting in foreign currencies. we have a large the latter firms, which have been ranked according to their extent of currency transactions into four directly exposed portfolios of stocks from high-to-low exposure. to this the research also added the sample of domestic firms with zero-currency exposure as the next 38 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 portfolio. there were thus five portfolios of companies used: the first four comprised about 30 stocks in each portfolio, sorted on the percentage of foreign-sourced revenues, and the fifth portfolio was a group of 50 domestic companies with no foreign-sourced revenues or assets. data have been sourced from the e-ikon thompson-reuters database. a few major hypotheses have been developed to address the research questions on whether australian firms are exposed to currency risk. if exposed, do domestic firms experience currency risk exposure? in selecting the firms for greatest exposure to currency risk, the study had to choose very large firms over others to set up a sample of highly-exposed and less exposed firms. similarly, it had to choose firms with zero-exposure by targeting those firms that had no foreign transactions. this had limited the study samples which were finally chosen for the study. the following hypotheses have been proposed: h1 : there is no significant association between the aud exchange rate changes against the usd, and stock returns of directly exposed multinational corporations represented by the four portfolios. h2 : there is no significant association between the aud exchange rate changes against the usd, and stock prices of indirectly exposed domestic corporations. h3 : the greater is the exposure of balance sheet and profit-and-loss items to the currency changes, the greater is the effect of currency risk on the stock prices. therefore, the highly exposed firms have greater impact from currency movements compared to the less exposed firms. a . the first two hypotheses follow the approach using macro factors as criterion variables in a model consistent with that found in adlerdumas, jorion and chen-roll-ross. the exchange rate factor is included alongside a combination of two other macro control variables, which potentially play crucial roles in determining the stock returns. equation (8) is the main focus of this study: (8) where denotes the value of assets (stocks), represents the nominal exchange rate as in equation (5), represents market return, is the 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 39 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 long-term domestic interest rate measured as the government bond rate, as the domestic short-term rate of treasury bills, used to measure the term structure of interest. is a proxy for corporate earnings since the industrial production index over time , while is the subscript for the companies. see table 2 for the theory-predicted signs of the variables. the first two hypotheses were tested by examining the significance of the coefficients on the four exposed firm-portfolios and the fifth zero-exposed domestic portfolio. hypothesis h3 could be tested by observing monotonically increasing coefficients in the four ranked portfolios representing exposed firms to currency risk, with the domestic firms having least exposure (assuming it has a zero coefficient). an increase in term structure and/or depreciation in currency will have a negative impact on share returns; share returns react positively to increased growth of incomes (ipi), unless the test period is one where the ipi has a significant declining trend, when the sign may then turn negative. table 2 variable specification, definitions and expected signs no. variables definition expected sign 1. ln. log difference of stock prices over time periods dependent variable 2. tsir long-term minus short –term interest rate – 3. 4. lnipi lnner log difference of industrial production over time periods log difference of aud/usd nominal exchange rate over time periods + _ the last two factors were used as control variables. the model was tested first with the first two factors as in jorion (1991), and then separately using the full model. data sources, variable identification and method selection data employed in this study were the stock prices of companies, nominal exchange rate, long-term and short-term risk-free interest 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 3 (𝑅𝑅𝑚𝑚 − 𝑅𝑅𝑓𝑓 ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑙𝑙𝑙𝑙 ( 𝑉𝑉𝑡𝑡 𝑉𝑉𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 = 𝛼𝛼𝑖𝑖 + 𝛾𝛾2 𝑙𝑙𝑙𝑙 ( 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝛾𝛾3𝑆𝑆𝑅𝑅𝑗𝑗𝑗𝑗 + 𝛾𝛾4(𝑖𝑖𝐿𝐿 − 𝑖𝑖𝑆𝑆 )𝑗𝑗𝑗𝑗 + 𝛾𝛾5𝑙𝑙𝑙𝑙 ( 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡 𝐼𝐼𝐼𝐼𝐼𝐼𝑡𝑡−1 ) 𝑗𝑗𝑗𝑗 + 𝜀𝜀𝑗𝑗𝑗𝑗 (8) 𝑉𝑉 𝑁𝑁𝑁𝑁𝑅𝑅 𝑆𝑆𝑅𝑅 𝑖𝑖𝐿𝐿 𝑖𝑖𝑆𝑆 𝐼𝐼𝐼𝐼𝐼𝐼 𝑡𝑡 𝑗𝑗 40 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 rates, industrial production index. the major source of data was from the international financial statistics (ifs) cd-rom. furthermore, the s&p capital iq source was used as data on stock prices. the datastream financial statements and price series were also used. data was collected from the monthly and yearly series. the monthly data were the month-end values of the factors over the period 2000 to 2016, and the yearly data were the year-end observations for the same period. the present study has applied both intervals. given recent findings such as in the study by ariff and zarei (2015), it has become clear that the use of low frequency data such as the yearly data series would enable one to observe a high coefficient of variation. this was because such intervaling would remove the temporary components of the time series. this is seen as desirable as the low frequency data series tend to have less noise, so the underlying relationship is more evident. it is also consistent with the length of time for currencies to reach equilibrium values as has been reported in the literature (manzur & ariff, 1995). in the present study the decision to use the usd is based on the premise that it is the international currency most used in australia’s trade, while it is also true that the usd (and to some extent the japanese ¥) is the currency of hedging in the international futures and forward markets by australian firms. though trade statistics suggest that australian trade with the us is the dominant one, however, invoices for sales and payments are normally done by australian firms (as do across the world) in usd. hence, using the usd exchange rate is valid. the present study has used a panel or longitudinal data set up, with an estimation method based on the choice from pooled, random and fixed effects. this method was selected because it could produce robust parameter estimates compared to cross-sectional or time series regressions. the panel data set has been used in more and more studies in recent years, and is becoming a mainstay of stock market research because of its superior accuracy. it was therefore, only to be expected that this method has been the prime choice over other methods in most exchange rate studies. in addition, this approach allows for the inclusion of data for cross-sections as well as time-series. the lagrange multiplier (lm) test is used to determine the optimum model for estimation based on the choice between pooled and random 41 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 effects. in a linear model, these tests are conducted based on the information on pooled ols residuals, while on the alternative model (random effect model) the estimation would involve generalized least square procedure based on the two-step or maximum likelihood or mle procedure. next, in choosing between fixed and random effect, the hausman (1978) test value was used: the test is based on confirming if the explanatory variables are correlated with individualspecific effects. findings on direct and indirect exposure of stock returns the central research question is: does the aud exchange rate movements have a significant impact on the stock returns of listed firms grouped as (i) four portfolios of currency-exposed firms and (ii) a portfolio of zero-exposed domestic firms? the findings are discussed in this section, starting first with the summary descriptions of variables and data preparation tests. the results from test models follow in subsequent sub-sections. descriptive statistics on variables the mean values of most of the variables were very close to zero, due to the data transformation and with the final data used being log change ratios. the exchange rate variable (ner) was calculated using one period log change of domestic over foreign (usd) exchange rate. similarly, the rate of stock return (v) and that of industrial production (ipi) were calculated using one period log change of ipi index values, respectively. table 3 is a summary of the descriptive statistics on the variables used in the study. this table reports summary of descriptive statistics for the stock return of five quintiles on all the variables. the first four quintiles are constructed based on revenue and total asset as percentage denominated in foreign sources. this table reports summary of descriptive statistics for the stock return of five quintiles on all the variables. the first four quintiles are constructed based on revenue and total asset as percentage denominated in foreign sources. 42 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 table 3 descriptive statistics of annual and monthly series on the quartile basis panel a (annual series) mean median std. dev. skew ξ kurt ♠ obs. ♦ return (g1ϯ) 0.063 0.065 0.393 0.467 9.512 336 g2 0.043 0.095 0.456 -1.20 6.277 336 g3 0.012 0.054 0.620 -0.287 4.781 336 g4 -0.222 -0.218 0.674 -0.447 5.474 336 gdomestic -0.030 0.041 0.674 -0.522 6.965 600 market 0.033 0.116 0.219 -1.542 4.872 336 fx -0.052 -0.054 0.134 0.117 2.786 336 ipi 0.022 0.021 0.029 -0.046 1.757 336 tsir 0.004 -0.001 0.013 0.832 2.545 336 panel b (monthly series) mean median std. dev. skew ξ kurt ♠ obs. ♦ return (g1) 0.005 0.009 0.101 -1.296 24.585 4592 g2 0.002 0.006 0.110 -0.637 8.654 4592 g3 0.003 0.014 0.161 0.123 12.138 4592 g4 -0.014 -0.022 0.213 0.189 6.656 4592 gdomestic -0.004 0.002 0.180 -1.126 39.807 8200 market 0.003 0.011 0.040 -1.047 5.275 4592 fx -0.002 -0.003 0.039 0.799 5.497 4592 ipi 0.002 0.002 0.004 0.101 3.601 4592 tsir 0.006 0.004 0.014 0.851 3.29 4592 ϯ g1: group one; gdomestic: domestic group; market: market return; fx: foreign exchange changes; ξ skew: skewness; ♠ kurt: kurtosis; ♦ obs: observations. the stock return of the fourth portfolio (fourth quintile) showed a negative and relatively bigger mean value (-22 %) change, perhaps due to an overall lower return associated with the stocks included in this portfolio. it may also be interpreted as showing the greatest exposure of portfolio 4 since this sample of firms might have been small enough to have no hedge management for currency risk. the means of other variables were within the expected range, as for example the yearly. 43 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 table 4 panel unit root tests on the variables used in test models variables deterministic terms annual monthly ll statistics ips statistics ll statistics ips statistics levels return (g1) constant, trend -15.68*** -8.49*** -82.54*** -74.85*** g2 constant, trend -16.10*** -9.21*** -75.35*** -66.63*** g3 constant, trend -13.22*** -5.74*** -79.67*** -66.27*** g4 constant, trend -13.40*** -7.54*** -90.70*** -76.18*** gdomestic constant, trend -20.10*** -10.54*** -104.60*** -87.60*** market constant, trend -18.97*** -9.53*** -56.47*** -56.62*** fx constant, trend -16.73*** -8.02*** -83.80*** -64.34*** ipi constant, trend -27.90*** -17.34*** -4.17*** -21.37*** tsir constant, trend -6.26*** -3.65*** 5.54 -0.37 first differences return (g1) constant -18.09*** -14.19*** 2.68 -60.42*** g2 constant -22.58*** -16.55*** -17.26*** -63.01*** g3 constant -21.70*** -13.90*** -4.38*** -57.75*** g4 constant -26.66*** -18.63*** 13.35*** -55.93*** gdomestic constant -27.26*** -19.50*** -25.80*** -82.23*** market constant -19.33*** -11.47*** -30.24*** -70.02*** fx constant -27.37*** -15.81*** -94.64*** -93.21*** ipi constant -6.32*** -11.88*** 29.33 -30.09*** tsir constant -17.49*** -13.37*** -56.20*** -49.56*** note: the number of lags is determined by the criterion of schwarz with maximum of five. ** indicates the significance level at 5%. *** indicates the significance level at 1%. table 4 is a summary of the panel unit-root test results of levin et al. (2002) (ll) and im et al. (2003) (ips) for monthly and yearly data to deal with stationary of criterion factors. the panel unit-root tests showed robust properties compared to the pure time-series test, as they provided consistent estimates of the true values of parameters when both time and cross-sections tended to infinity. as is evident from the statistics, the variables were stationary at level, i.e., i (0), due to data transformation, except the term structure of the interest rate (tsir) for only monthly series. hence, the tsir had to be excluded from the panel regression using monthly data (see table 4). this table 44 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 reports statistics on unit root and order of integration of variables. test for level variables use constant and trend while applying only constant term for first differenced variables. to provide evidence that the factors were not correlated with one another, the variance inflation factor (vif) was computed, as is shown on table 5. table 5 variance inflation factor and tolerance values for multicollinearity. multicollinearity yearly data monthly data variables vifa tolerance vif tolerance mr 2.12 0.472 1.36 0.733 ipi 1.43 0.697 1.06 0.946 fx 2.42 0.414 1.37 0.732 tsir 1.67 0.600 1.08 0.928 since these values were less than 10 (10 being the critical value), there was no multicollinearity effect on the estimated parameters. hence, the test results on parameters in the following tests were found to be robust estimates. this table reveals the information about multicollinearity of the variables used in the study, where low values of vif are always more accepted, suggesting lower correlation among the variables in the multiple regressions. results from the analysis of the monthly and yearly interval data the statistics on the entire sample are as reported in tables 6 and 7. the present study has taken into account two different estimation models: one based on the approach proposed by jorion (1991) (equation 5) and the other based on an extended version developed in this study (equation 7). table 6 shows the results from the analysis of data with monthly and yearly frequencies from the entire sample of companies. the analysis was carried out using the approach in jorion (1991). the foreign exchange rate and market return variables showed a statistically significant relationship with stock returns for the entire sample. the coefficient for the exchange rate risk was -0.179, with a t-value of -5.62, which is significant at or above the 0.01 probability level. this was also the case in the larger sample using monthly interval data and the smaller sample of yearly interval 45 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 data. the coefficient in the yearly data set was -0.43 with a t-value of -3.50, which is a statistically significant evidence of a currency effect. the adjusted r-squared value of 22 per cent in the yearly data indicated that the two factors together explained the 22 percent of the variation in stock returns: the corresponding value in the monthly data set was 8 percent. the model fit is suggesting that there is a large currency effect on stock returns while the market factor effect is as expected, around 1. the beta around 1.00 indicates that the sample is a representative market sample. table 6 reports results based on three specifications of fixed effects models specified in the methodology section of this paper. table 6 results of entire sample of australian firms based on jorion’s twofactor model intercept mr fx adjusted r-squared yearly (fixed effect) -0.082 (-5.82)*** 1.027 (13.59)*** -0.430 (-3.50)*** 0.22 monthly (fixed effect) -0.0046 (-4.29)*** 0.978 (31.46)*** -0.179 (-5.62)*** 0.08 note: the model fit is significant at 0.01 or better using the chi-squared value of 184.657 (yearly) and 1607.734 (monthly). table 7 is a summary of statistics of the entire sample of firms using the model developed in the present study, using the four theorysuggested (relying on ross, 1976) macro factors as indicated in the description of the factors in row 1. the additional factors were for earnings across the economy over the test period (recall this period includes several years of negative earnings following the 9/11 event and the global financial crisis years). as in the previous jorion-model results, the first two coefficients on the market and exchange rate factors were showing signs that were theory-consistent and the coefficients were statistically significant at 0.01 or better p-values. the ipi was not significant in both samples, so its negative signs have been ignored. the earnings of the firms 46 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 in the study sample declined substantially in the test period, such that this factor which would be significant in other studies turned out to be insignificant. the term structure factor on interest rate as a control factor had a theory-consistent sign, and was significant at 0.05 acceptance level. the coefficient of variation in this repeated test with more variables remained the same as in jorion’s two-factor model results. table 7 reports results based on three specifications of pooled, random and fixed effects models specified in the methodology section of this paper. the unexpected negative sign on ipi is due to the firms reporting losses during most of the post financial crisis period. table 7 results of entire sample of australian firms, 2000-2016 with four factors intercept mr fx ipi tsir r-sq. yearly (fixed effect) -0.071 (-3.98)*** 1.023 (12.10)*** -0.508 (-3.55)*** -0.445 (0.85) -1.280 (-1.77)* 0.22 monthly (fixed effect) -0.004 (-3.39)*** 0.978 (31.46)*** -0.184 (-5.73)*** -0.335 (-1.22) 0.08 note: “-“ indicates that the regression test run excluded this factor due to its non-stationarity property. the model fit is significant at 0.01 or better using the chi-squared value of 187.903 (yearly) and 1609.277 (monthly). the marginal effect of forex risk ranged from as high as -0.508 (yearly data) in the entire sample to -0.184 in the monthly data set. these were considered significant, given the large t-values of -3.55 and -5.73. the coefficient of determination was 22 per cent in the yearly data sample for the entire sample; it was 8 percent in the monthly data set. portfolio-based results for degrees of exposure the results from the five portfolios of companies ranked by the degree of exposure to foreign cash flows are as presented in the ensuing tables 8 and 9. the first four portfolios were directly exposed samples of firms while the fifth was the zero-exposed domestic sample firms. 47 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 table 8 results of five quartiles of australian firms based on jorion’s two-factor mod intercept mr fx r-sq. group 1(de ♠) (pooled ols) 0.0017 (1.28) 0.953 (24.44)*** -0.123 (-3.07)*** 0.16 group 2(de) (pooled ols) -0.0012 (-0.83) 0.916 (20.97)*** -0.080 (-1.79)* 0.12 group 3(de) (pooled ols) -0.0006 (-0.27) 0.873 (13.16)*** -0.184 (-2.70)*** 0.06 group 4(de) (pooled ols) -0.018 (-6.07)*** 1.170 (13.42)*** -0.330 (-3.96)*** 0.06 group 5 (ie ♣ ) (pooled ols) -0.008 (-4.64)*** 1.085 (19.77)*** -0.289 (-5.12)*** 0.07 group 1(de) (random effect) -0.392 (1.60) 0.998 (9.69)*** 0.171 (1.02) 0.27 group 2(de) (pooled ols) -0.011 (-0.53) 1.009 (8.51)*** -0.412 (-2.13)** 0.32 group 3(de) (pooled ols) -0.063 (-1.96)* 0.888 (5.11)*** -0.90 (-3.17)*** 0.21 group 4(de) (pooled ols) -0.292 (-8.33)*** 1.211 (6.46)*** -0.582 (-1.90)* 0.22 group 5 (ie) (pooled ols) -.121 (-4.75)*** 1.086 (7.95)*** -1.072 (-4.81)*** 0.26 note: ♠ de: directly exposed firms; ♣ ie: indirectly exposed firms, adjusted r-squared values. the model fit is significant at 0.01 or better using the chi-squared values computed for each of the samples. the computed chi-squared values are greater than the critical values. table 8, panel a, contains results from data set used at monthly intervals (hence containing more white noise) while the results shown under panel b are from yearly interval data. the foreign exchange exposure coefficients of all five portfolios were found to be statistically significant as predicted by theory with the correct signs. 48 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 this finding is seen as remarkably different from the weak support in the literature for currency risk in prior studies. this is perhaps due to the use of panel regression combining the portfolio grouping of fama-mcbeth. the only exception to the support for theory is in portfolio 1, which represents the largest listed firms with an ability to hedge currency risk management successfully. it is evident from the statistics that the coefficients of the five ranked portfolios were significant with the domestic firms having the most impact, as judged by its relatively large marginal effect of around -0.289. the exception was portfolio 1, which seemed to have less impact than the other large firms on average, despite being significant, because these firms had active repo hedging procedure in place. table 8 reports results from the two factor jorion’s model on 5 groups. the results are indicative of significant market return and foreign exchange exposure of all groups, an exception being group 1 when using yearly series. the size of the coefficients ranged between -0.08 and -0.33. it appears that the greatest impact of currency effect was on the fourth portfolio, which represented firms with the lower direct exposure to crossborder transactions. note also that the zero-exposed fifth portfolio of firms had the highest coefficient. this might be due to the fact that the fourth portfolio of firms and the fifth portfolio were smaller firms relative to the first three, with lack of full-hedging for currency exposure risk. the highly exposed portfolio 1 and portfolio 2 had smaller coefficients, perhaps for the opposite reason that these larger firms had successful hedging that reduced the size of the currency risk via hedges. table 9 reports results from the two factor jorion model on 5 groups. the results affirm evidence of significant foreign exchange exposure when using monthly data. as for the yearly data, there is no significant foreign exchange exposure for the group 1 only, which can be due to operational hedging of those companies. aggarwal & harris op cit. also reported this peculiarity of smaller coefficients for larger firms, which they attributed to the ability of larger firms having effective hedges in place against currency risk that reduced or nullified the size of the negative effect of currency risk for such firms. 49 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 table 9 results of five quartiles of australian firms based on extended jorion’s model intercept mr fx ipi tsir r-sq. group 1(de♠) (pooled ols) 0.003 (1.99)** 0.954 (24.45)*** -0.133 (-3.30)*** -0.680 (-1.98)** n/a 0.16 group 2(de) (pooled ols) -0.001 (-0.58) 0.916 (20.97)*** -0.082 (-1.83)* -0.162 (-0.42) n/a 0.12 group 3(de) (pooled ols) -0.0002 (-0.08) 0.873 (13.16)*** -0.187 (-2.72)*** -0.224 (-0.70) n/a 0.06 group 4(de) (pooled ols) -0.018 (-5.37)*** 1.171 (13.42)*** -0.335 (-3.70)*** -0.273 (-0.36) n/a 0.06 group 5 (ie♣) (pooled ols) -0.009 (-4.50)*** 1.085 (19.76)*** -0.284 (-4.99)*** 0.324 (0.67) n/a 0.07 group 1(de) (fixed effect) 0.086 (3.57)*** 1.001 (8.55)*** -0.111 (-0.55) -2.111 (-3.00)*** -4.111 (-2.40)** 0.29 group 2(de) (pooled ols) 0.017 (0.61) 1.089 (7.98)*** -0.432 (-1.82)* -1.438 (-1.72)* -0.322 (-0.16) 0.32 group 3(de) (pooled ols) -0.086 (-2.09)** 0.895 (4.45)*** -0.746 (-2.13)** 1.037 (0.84) 2.242 (0.76) 0.21 group 4(de) (pooled ols) -0.310 (-6.94)*** 1.125 (5.19)*** -0.639 (-1.69)* 0.965 (0.73) -0.800 (-0.25) 0.22 group 5 (ie) (pooled ols) -0.150 (-4.64)*** 1.008 (6.74)*** -1.046 (-4.26)*** 1.385 (1.49) 0.573 (0.28) 0.26 note: ♠ de: directly exposed firms; ♣ ie: indirectly exposed firms, adjusted r-squared values. the model fit is significant at 0.01 or better using the chi-squared values computed for each of the samples. the computed chi-squared values are greater than the critical values. the summary statistics from the extended model used in this study were from the four-factor model. the results were seen as slightly improved in the case of foreign exchange exposure with significant impact from the ipi and the tsir for the portfolio 1 companies. note that the ipi earnings impact was significant for portfolio 1 and portfolio 2 in the yearly data results and only for the case of portfolio 1 using monthly data. the size of the impact was also substantial in the monthly data sets. it is reasonable to explain the yearly-data results as the consequence of stock returns already incorporating annual reports data information on hedge gains and losses reported at the end of the 50 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 financial year, but not so in the study results obtained from monthly data. the results also showed that the domestic firms (portfolio 5) had a significant negative impact of -1.046, which seemed to suggest that the share returns losses for portfolio 5 was about equal to the impact of the currency. this would mean that the top management of these firms did have to have hedge in place for currency risk, an important result for application. investment firms managing portfolios of domestic firms ought to have currency-hedges in place to ensure that investors are protected against possible losses, if such firms have currency exposures. in the monthly data set too, it was noted that a significant effect for this and other groups as having negative influences from us dollar movements against the australian currency. the additional two factors have been used as control factors to refocus the test on the jorion’s two-factor results (refer to the coefficients on the ipi and the tsir). the corresponding r-squared values for the domestic firms in row 5 in panels a and b were 7 and 26 per cent respectively. the annual data sample showed a modestly higher value for adjusted r-squares, meaning that the addition of additional factors from parity theorems helped to improve the explanatory power of the test. the zero-exposed domestic firms (portfolio 5) showed about slightly more than half the marginal effect of currency risk compared to the exposed firms. note that the coefficient of portfolio 5 in the yearly data was -0.282, compared to the slightly higher average for the four portfolios of multinational firms (portfolios 1-4). the statistics from the monthly data set for the groups 1 to 4 had a higher exposure in this model, as had also been observed for the entire sample tests in the previous sub-section. for example, with regard to portfolio 3 in the context of its yearly data set, the middle group of directly exposed firms yielded a marginal value of -0.746, while portfolio 4 had a smaller coefficient. the smallest value of portfolio 1 for the largest exposed portfolio had a coefficient equal to -0.111, which was much smaller than the size observed for the less exposed firms in other three directly exposed portfolios. this again could have been due to the superior skills of the top management of larger firms to hedge most of the risk 51 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 through hedging programs in place. this was obvious from the yearly data tests (see tables 4 and 5), in which the investors would have access to evaluate hedging effectiveness reported in the notes to the accounts in the annual reports, so investors would have factored in that information in changing the stock prices. thus the directly exposed firms all had had significant exposure in during the test period of sixteen turbulent currency years. these results would seem to suggest that the zero-exposed australian listed firms were also significantly affected by currency risk. the smaller size coefficient for portfolio 1 in this study seemed to suggest that part of the reason for this reduced size could be the same risk-offsetting abilities of the largest firms through active hedging. it could also mean that the australian firms could not fully offset, hence the effect was still negative, though smaller than the size theorized for most directly exposed firms in portfolio 1. the industrial production index, as a control variable in the case of the annual data sets, was only statistically significant for the companies classified as portfolios 1 and 2. the r-squared value followed a declining trend from portfolios 1 to 5. this as expected, was a natural occurrence. as in the previous tables, the model fit in the panel regression has been robust, as indicated by the chi-squared values being greater than the critical values. overall, the results have provided empirical support for the two-factor model of jorion (1991). this was even the case when the control variables introduced in this study produced theory-consistent results for australian listed firms. this means that the currency risk has been significantly priced in the stock returns in the australian market for both the directly and the zero-exposed firms. these results are new findings which will make substantial contributions to the asia pacific literature. conclusion this study started with the aim of studying the currency risk of australian companies exposed to the volatility of the exchange rate by using the monthly and also annual data for the last 16 years. more specifically, the aim was to study currency exposure using theory52 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 origin macro factors only, and by using the three well-received asset pricing theories away from the firm-specific-factor-based experiments. the test period was also one that had significant up-trends for the currency compared with the pre-2000 period, when the currency depreciated severely starting from 1983 onward to 1997. a significant reason for this study is to measure the exposure of domestic firms with no foreign currency in their books, comprising a group of firms which have been ignored in almost all studies. another question is to measure the size effects of differing risk exposure via portfolios. the study categorized four portfolios of firms exposed to direct currency risk as shown by their revenue size and asset size. the fifth portfolio was made up of firms with no foreign revenue/capital, as such this portfolio had zero currency risk, and hence the aim was to test currency risk effect on stock returns of this fifth group. the data set was split further into yearly and monthly sets to check if the yearly-interval data series improved the coefficients of variation. the proper econometric procedures required tests on several assumptions (stationarity; multicollinearity; etc.) so that results from a refined econometric model should lead to valid data transformation to comply with panel data regression assumptions in actual regression runs. panel regressions were conducted with individual firms and then in five portfolios forming the panels. the three main findings of this study have made significant contributions to the foreign exchange literature. first, the exchange rate movements do affect significantly the stock returns of directly exposed firms in an orderly manner; the higher the exposure, the higher is the marginal effect. the size of the largest firm’s coefficient is smaller, and this may be due to the very large firms taking hedge protection that reduces the size of the currency effect. the second result appears to be a new finding as the previously held belief that domestic firms ought to have a neutral-effect from currency movements is shown to be rejected. zero-exposed domestic firms have slightly lower risk (when tested as an entire sample), but then in comparison with four portfolios, domestic firms were found to have the highest currency risk effect. it had a coefficient of -0.284 and a significant t-value of 4.99. these findings are important for the business management of currency by the top management of both types of firms operating in a freefloating environment (in australia, indonesia, japan, south korea, 53 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 new zealand, and the philippines), and possibly in a managed float as well (thai bhat and malaysian ringgit). the role of risk management is all the more critical in domestic firms since the tests in this study has revealed that all firms have been significantly affected by the exchange rate movements. the findings presented here on international finance theory seems to suggest that foreign exchange is a pricing factor for stock valuation by investors, as in two rarely tested exchange rate theories. first, this evidence should spur the top management of businesses to examine risk-reducing measures to offset the international systematic currency risk factor (isr). next, it is useful for a policy debate because these findings are expected as natural consequences of the freefloating monetary management of a currency. the operational models developed in this study is worth testing out in more asia-pacific freefloating, as well as managed currency regimes to document more evidence to add to the existing literature and thus, provide further guidance to top management in more countries. these results with their high explanatory power and strong significant effects on both types of firms should be recommended as the way to study and document evidence to guide corporate finance policy on exposure risk management. acknowledgment the authors would like to express their gratitude to the malaysian government which funded this research on currency risk under its 2016-9 fundamental research grant scheme (ref: frgs/1/2016/ ss01/upm/01/1). in addition, we are grateful to the discussants/ participants of the 22nd malaysian finance association conference (17-19 november 2020) for their comments. this paper won the best paper award at the mfa-2020 conference. endnotes 1 the a$ was managed for several decades away from the freefloating exchange rate, oblivious to the fact that this mostly commodity-trading nation’s currency was way above its value after the commodity boom of the post-war era was over by the end of 1970s. the a$ was way above the trade-weighted 54 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 value, so something needed to be done to make the adjustment to an overvalued dollar. hence the free-floating law was put in place some 10 years after the breakdown of the fixed-exchange monetary system in 1973. 2 this research finding is relevant to the top management of firms in general – both domestic and currency-exposed firms – listed on the australian stock exchange. for the first time, there is evidence of an exchange rate movement having devastating effect not just on stock prices of firms exposed to foreigncurrency revenues, but also on stock prices of domestic firms with no cash flow exposure in the books. the australian dollar is a free-floating currency among the 12 so-called clean floating ones, and its behavior changes prior to the 2020 covid-19 has been investigated. this paper is a first country study using the data cited above. 3 it is strange that monetary authorities wait for speculative attacks instead of taking pre-emptive actions before speculative attacks. for instance, while aud was under speculative attack several others, such as the argentine peso was also under attack in the 1980s. much later in 1996, speculative attack on the malaysia’s ringgit also led to a disastrous depreciation in value from myr2.54 to usd1.0, and usd1.00 for myr4.10 in 1999, it has plateau ever since. 4 drew, m.e., naughton, t. & veeraraghavan, m., (2003). firm size, book-to-market equity and security returns: evidence from the shanghai stock exchange. australian journal of management, 28, 119-139, computed the returns of firms listed on the shanghai stock exchange using the multifactor asset pricing approach, to control the impact of size and book to market ratio on the return of investors. again this is a test using firm-specific factors. 5 for this important reason, as supported by the research, the tests in the present study are restricted to bilateral currencies. the test results using trade weighted currency index are available on request, as they are excluded from this report. references adler, m., & dumas, b. (1984). exposure to currency risk: definition and measurement. financial management, 13, 41-50. 55 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 aggarwal, r., & harper, j. t. (2010). foreign exchange exposure of “domestic” corporations. journal of international money and finance, 29, 1619-1636. ariff, m., & marisetty, v. b. (2012). panel data approach to identify factors correlated with equity market risk premiums in developed and emerging markets. quantitative finance, 12, 107-118. ariff, m., & zarei, a. (2015). exchange rate behavior of canada, japan, the united kingdom and the united states. open economies review, 27, 341-357. ariff, m., & zarei, a. (2019). the impact of exchange rates on stock market prices: new evidence from seven free-floating currencies. the european journal of finance, 25(14), 1-29. bartov, e., & bodnar, g. m. (1994). firm valuation, earnings expectations, and the exchange-rate exposure effect. the journal of finance, 49, 1755-1785. bartov, e., bodnar, g. m., & kaul, a. (1996). exchange rate variability and the riskiness of us multinational firms: evidence from the breakdown of the bretton woods system. journal of financial economics, 42, 105-132. bodnar, g. m., & gentry, w. m. (1993). exchange rate exposure and industry characteristics: evidence from canada, japan, and the usa. journal of international money and finance, 12, 29-45. breusch, t. s., & pagan, a. r. (1980). the lagrange multiplier test and its applications to model specification in econometrics. the review of economic studies, 47, 239-253. capiello, l., & fearnley, t. a. (2000). international capm with regime switching garch parameters. international center for financial asset management and engineering chen, n.-f., roll, r., & ross, s. a. (1986). economic forces and the stock market. the journal of business, 59, 383-403. dahlquist, m., & saellstrom, t. (2002). an evaluation of international asset pricing models. cepr discussion papers 3145. https:// ideas.repec.org/p/cpr/ceprdp/3145.html de santis, g., & gerard, b. (1998). how big is the premium for currency risk? journal of financial economics, 49, 375-412. de santis, g., gerard, b., & hillion, p. (2003). the relevance of currency risk in the emu. journal of economics and business, 55, 427-462. di iorio, a., & faff, r. (2002). a test of the stability of exchange rate risk: evidence from the australian equities market. global finance journal, 12, 179-203. 56 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 25–56 drew, m. e., naughton, t., & veeraraghavan, m. (2003). firm size, book-to-market equity and security returns: evidence from the shanghai stock exchange. australian journal of management, 28, 119-139. dumas, b., & solnik, b. (1995). the world price of foreign exchange risk. the journal of finance, 50, 445-479. griffin, j. m., & stulz, r. m. (2001). international competition and exchange rate shocks: a cross-country industry analysis of stock returns. review of financial studies, 14, 215-241. hausman, j. a. (1978). specification tests in econometrics. econometrica, 46, 1251-1271. im, k. s., pesaran, m. h., & shin, y. (2003). testing for unit roots in heterogeneous panels. journal of econometrics, 115, 53-74. jorion, p. (1990). the exchange-rate exposure of u.s. multinationals. the journal of business, 63, 331-345. jorion, p. (1991). the pricing of exchange rate risk in the stock market. the journal of financial and quantitative analysis, 26, 363-376. kang, s., kim, s., & lee, j. w. (2016). reexamining the exchange rate exposure puzzle by classifying exchange rate risks into two types. global economic review, 45, 116-133. khoo, a. (1994). estimation of foreign exchange exposure: an application to mining companies in australia. journal of international money and finance, 13, 342-363. levin, a., lin, c.-f. & james chu, c.-s. (2002). unit root tests in panel data: asymptotic and finite-sample properties. journal of econometrics, 108, 1-24. loudon, g. f. (1993). foreign exchange exposure and the pricing of currency risk in equity returns: some australian evidence. pacific-basin finance journal, 1, 335-354. manzur, m., & ariff, m. (1995). purchasing power parity: new methods and extensions. applied financial economics, 5, 19-26. mok, h. m. (1993). causality of interest rate, exchange rate and stock prices at stock market open and close in hong kong. asia pacific journal of management, 10, 123-143. ross, s. a. (1976). the arbitrage theory of capital asset pricing. journal of economic theory, 13, 341-360. sharpe, w. f. (1964). capital asset prices: a theory of market equilibrium under conditions of risk. the journal of finance, 19, 425-442. solnik, b. h. (1974). an equilibrium model of the international capital market. journal of economic theory, 8, 500-524. 53 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance how to cite this article: ozili, p.k. (2022). does economic policy uncertainty reduce financial inclusion? international journal of banking and finance, 17(1), 53 80. https://doi.org/10.32890/ ijbf2022.17.1.3 does economic policy uncertainty reduce financial inclusion? peterson k. ozili central bank of nigeria, abuja, nigeria corresponding author: petersonkitakogelu@yahoo.com received: 19/2/2021 revised: 5/5/2021 accepted: 7/5/2021 published: 2/12/2021 abstract this study investigates whether the level of economic policy uncertainty (epu) would reduce the level of financial inclusion. it was predicted that a high level of epu could have a negative effect on the level of financial inclusion. it was argued that a high level of epu would discourage financial institutions from providing basic financial services to low end customers and unbanked adults, and this would lead to a decrease in the level of financial inclusion. using a sample of 22 countries, the study found that the level of epu did not have a significant impact on financial inclusion. none of the nine indicators of financial inclusion were found to have a significant direct relationship with epu. however, there was some evidence that the combined effect of a high level of epu and high nonperforming loans could reduce financial inclusion, particularly through bank branch contraction and a reduction in the use of electronic payments. furthermore, the use of formal accounts and credit cards would increase in times of high credit supply and when there was a high level of epu. 54 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 keywords: financial inclusion, policy uncertainty, economic policy uncertainty, non-performing loan, unbanked adults. jel classification: d14, d18, g21, g28. introduction in recent years, economic policy uncertainty has become the focus of economic policy debates. such debates were mostly focused on how policy uncertainty had affected economic agents in the real and financial sectors. economic policy uncertainty (epu) has been seen as uncertainty about changes in fiscal, monetary and regulatory policies of the government (baker et al., 2016). epu might arise from whether there would be unexpected changes in existing government policies (ashraf & shen, 2019; ng et al., 2020). the recent epu literature has shown that epu could affect corporate decisions, financial institutions and the real economy (e.g., caglayan & xu, 2019; lee et al., 2017; gulen & ion, 2016). however, the literature has not examined how epu might affect access to finance or the level of financial inclusion. the present study will contribute to the literature by examining whether epu could reduce or improve the level of financial inclusion. in the financial inclusion literature, financial inclusion has been broadly defined as the provision of affordable formal financial services to households, individuals and small businesses (ozili, 2018; zins & weill, 2016; ozili, 2020b). the goal of financial inclusion is to reduce the number of unbanked adults, and this has been mostly achieved by expanding financial services to unbanked adults in remote areas (collard, 2007; demirgüç-kunt & klapper, 2013; neuberger, 2015; ozili, 2020a). the international development community has considered financial inclusion to be the most significant way to expand financial services in developed and emerging economies (neuberger, 2015; ozili, 2020a). as such, the present study has focused on financial inclusion in developed and emerging economies for two reasons. firstly, individuals or households in developed and emerging economies have not been immune to financial exclusion. the rising cost of 55 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 basic financial services, low income, personal bankruptcy and the desire for financial privacy have led to increased financial exclusion. secondly, financial inclusion has been a major component of the social inclusion programs in most developed and emerging economies. for instance, many social inclusion programs, such as having access to social welfare, community participation, infrastructure, housing, employment or education, are dependent on owning a formal account which individuals could use to receive welfare benefits or to make payment for services. understanding how financial inclusion can be affected by epu is important because individuals and households rely on financial institutions for the supply of basic financial services, and these financial institutions may be severely affected by a high level of epu. such a situation may affect their willingness to reach the unbanked adults, and to serve existing low end customers in times of high levels of epu. in other words, uncertain economic policies can affect the level of financial inclusion through its effect on the financial sector. recent studies have documented that a high level of epu negatively affected the financial sector. such studies showed that financial institutions would increase interest rates, re-price loans, reduce credit supply, and have liquidity shortages in times of high levels of epu (bordo et al., 2016; yung et al., 2019; garcía-kuhnert et al., 2015). high levels of epu could affect a financial institution’s incentive to supply basic financial services to low-end customers and households. financial institutions can increase the interest rate on loans and credit cards, and charge high fees for basic services such as the atm card maintenance fees and other fees, in response to high levels of epu in the business environment. basic financial services will become costly and will severely affect low-income individuals and households, which can make them exit the formal financial sector, thereby reducing financial inclusion. more importantly, high levels of epu in the business environment can lead to difficult business conditions for financial institutions, creating problems such as fewer demand for loans, higher nonperforming loans and liquidity shortage. due to these difficulties, financial institutions will be drawn into providing better financial services to high-end customers who can pay a premium for financial services and reduce the provision of financial services to low-end customers. they will begin to ignore their poorer customers who cannot afford to pay a premium for basic financial 56 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 services induced by high levels of epu in the business environment. therefore, financial inclusion is likely to be lower during periods of high levels of epu. it can therefore, be predicted that, if financial institutions perceive high levels of epu in the business environment and take into account its expected depressive effects in the provision of basic financial services, financial institutions will reduce the supply of basic financial services. in other words, the level of financial inclusion can be seen as being negatively associated with the level of epu. on the other hand, if financial institutions do not take into account the expected depressive effects of epu in the provision of basic financial services, then the level of financial inclusion can be seen as positively associated with the level of epu. using data for 22 countries from 2011 to 2017, the findings revealed that epu has an insignificant effect on financial inclusion. none of the nine indicators of financial inclusion showed a significant direct relationship with epu. in addition, the combined effect of high levels of epu and high non-performing loans lead to bank branch contraction and a reduction in the use of electronic payments. meanwhile, the use of formal accounts and credit cards increased in times of high credit supply and high epu. the present study contributes to the epu and financial inclusion literatures. in three ways. firstly, the study contributes to the epu literature that explore the effects of epu (gulen & ion, 2016; karadima & louri, 2020; ozili, 2021a). this study has extended the scope of the epu literature by focusing on how epu affects the level of financial inclusion. the findings of the present study have shown that high levels of epu had some depressive effects on financial inclusion. secondly, this study contributes to the financial inclusion literature (see, mindra et al., 2017; ozili, 2020a; zins & weill, 2016; ozili, 2020b). the study showed that epu is a determinant of the level of financial inclusion. the study is the first in the literature to identify the level of epu to be a determinant of the level of financial inclusion. finally, this study has also contributed to the literature on the effects of financial inclusion on financial institutions (demetriades & hook law, 2006; king & levine, 1993; rioja & valev, 2004; ozili, 2021c). the findings of the present study have shown that the level of epu, through its effect on banks, would have implications for financial inclusion. 57 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 the rest of the paper is structured in the following way. section 2 reviews the literature and develops the study hypothesis. section 3 presents the data and methodology. section 4 presents the results, and section 5 the conclusions. literature review determinants and consequences of financial inclusion the literature has documented some determinants and consequences of financial inclusion. for example, lópez and winkler (2019) examined whether financial inclusion could mitigate credit downturns and upturns. they found that higher levels of financial inclusion led to a decrease in credit growth. chen and jin (2017) analyzed data from the 2011 china household financial survey, and observed that over half of the sample (53.21%) reported using credit, and only 19.77 percent of the sample used formal credit. they also observed that the use of formal credit was associated with being employed, educated, having a high income, and having a high net-worth. evans and alenoghena (2017) tested whether the gdp per capita translated into higher financial inclusion. they examined 15 african countries from 2005 to 2014. they found that gdp per capita had a positive relationship with financial inclusion, but the relationship was not significant. omar and inaba (2020) investigated the impact of financial inclusion on poverty reduction. they used the gdp per capita to measure poverty. they found that the per capita real gdp had a positive influence on the level of financial inclusion in developing countries. ozili (2020b) investigated financial inclusion through the business cycle. the study used the gdp growth rate to measure the state of the business cycle. the study documented evidence of increased formal savings and active formal account ownership in periods of economic prosperity, and a decrease in formal savings and active formal account ownership in recessionary periods. vo et al. (2019) investigated the linkages between financial inclusion and macroeconomic stability in 22 emerging and frontier economies from 2008 to 2015. they found that financial inclusion, measured as the growth rate in the number of bank branches over 100,000 account holders, improved financial stability only to some extent. 58 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 similarly, machdar (2020) analyzed the effect of financial inclusion on the financial stability of banks in indonesia, and found a negative relationship between financial inclusion and the level of nonperforming loans (npls). morgan and pontines (2018) examined the relationship between financial stability and financial inclusion. they found that increased lending to small and medium-sized enterprises (smes) reduced the size of npls and lower the probability of default by financial institutions. ozili (2021b) showed that greater levels of financial inclusion would improve the cost efficiency of the financial sector in developing countries. markose et al. (2020) examined the economic viability of financial inclusion programs in india, and showed that higher financial inclusion programs, under the pradhan mantri jan-dhan yojana (pmjdy) scheme, led to cost inefficiency among public sector banks. epu and financial institutions a substantial body of literature has examined the effects of economic policy uncertainty (epu) on financial firms, and the firms’ response to policy uncertainty. nguyen et al. (2020) examined the impact of epu on aggregate bank credit growth at domestic and global levels. using different measures of epu, they studied this issue in 22 countries from 2001 to 2015, and documented evidence that a high level epu led to low credit growth, and the negative impact was stronger in emerging economies than in advanced economies. ashraf and shen (2019) examined the effect of government economic policy uncertainty on the pricing on bank loans in 17 countries from 1998 to 2012. they found that banks repriced loans by charging higher interest rate in times of high levels of epu. the implication of their findings was that epu is an important risk factor that banks would take into account when making loan pricing decisions. bordo et al. (2016) examined the impact of epu on bank credit growth for a 50-year period from 1961 q4 to 2014 q3. they found that policy uncertainty, through its effect on loan supply, had a significant negative effect on bank credit growth. hu and gong (2019) empirically tested the association between bank lending and epu. they found that high levels of epu would reduce credit growth, and the negative effect was greater for largersized banks and riskier banks. luo and zhang (2020) examined the impact of epu on firm-specific crash risk among chinese listed firms. they found that high levels of epu would increase the likelihood of firms experiencing stock price crash. karadima and louri (2020) investigated the effect of epu on nonperforming loans. they found 59 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 that high levels of epu would lead to an increase in nonperforming loans. caglayan and xu (2019) examined the effect of epu on loan loss provisions in 18 countries. they found that high levels of epu was associated with the increase in loan loss provisions. berger et al. (2020) found that high levels of epu led to liquidity hoarding by banks. hypothesis development in developed and emerging economies, financial institutions have been the main agents of financial inclusion (chakrabarty, 2011; ghosh, 2013; brown et al., 2016). uncertain economic conditions and uncertainty about economic policies could present difficult business conditions for financial institutions, and this could dampen their incentive to supply basic financial services to unbanked adults and low end customers. when faced with high levels of epu, financial institutions may become unwilling to serve poor individuals and households in order to reduce operating cost and manage risks. rising operating costs, high nonperforming loans, inefficiencies in the distribution of financial services and diseconomies of scale, can hurt financial institutions and create a disincentive to supply financial services to low end customers and unbanked adults in remote communities, thereby reducing the level of financial inclusion. therefore, the present study has predicted that high levels of epu would reduce the level of financial inclusion. as such the following hypothesis was proposed. h1: economic policy uncertainty reduces the level of financial inclusion. methodology sample financial inclusion data was extracted from the global financial development indicators. financial inclusion information in the database was available only for the year 2011, 2014 and 2017. this was because the world bank’s financial inclusion survey was conducted triennially (i.e., every three years). information on the epu index was extracted from the epu database at: https://www.policyuncertainty.com. the epu database develops indices of economic policy uncertainty for 60 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 major developed and emerging economies of the world. the epu index was constructed based on the methodology described in baker, bloom and davis (2016). macroeconomic data was extracted from the world bank database, and the data collected covered the period from 2011 to 2017. the sample consisted of 22 countries. the countries included australia, brazil, canada, chile, china, colombia, france, germany, greece, india, ireland, italy, japan, korea, mexico, netherland, russia, singapore, spain, sweden, uk and the us. table 1 shows the description of the variables. method model the model has conceptualized financial inclusion as a function of a financial institution’s performance, macro-financial linkage, and the macroeconomic variable, and has been expressed in equation (1): (1) where i = country, t = year. fi is a vector of dependent variables. fi includes: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. specifically, acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. atm = atms per 100,000 adults. bbpa = bank branches per 100,000 adults. mpb = adults using a mobile phone to pay bills. elp = adults who use electronic payments to make payments. for the explanatory variables, epud = year-end value of the epu index. npl = ratio of nonperforming loans to gross loans. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. variable justification the dependent variables were the financial inclusion variables, namely: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. these variables are commonly used in the literature to measure financial inclusion (imaeva et al., 2014; chakrabarty, 2011; ozili, 𝐹𝐹𝐹𝐹𝐹𝐹, 𝑡𝑡 = 𝛽𝛽𝛽𝛽 + 𝛽𝛽1𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽2𝑁𝑁𝐸𝐸𝑁𝑁𝐹𝐹, 𝑡𝑡 + 𝛽𝛽3𝛽𝛽𝑂𝑂𝑂𝑂𝑂𝑂𝐹𝐹, 𝑡𝑡 + 𝛽𝛽4𝐸𝐸𝑂𝑂𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽5𝐺𝐺𝐸𝐸𝐸𝐸𝑃𝑃𝐹𝐹, 𝑡𝑡 + 𝛽𝛽6𝑁𝑁𝐸𝐸𝑁𝑁 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽7𝛽𝛽𝑂𝑂𝑂𝑂𝑂𝑂 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽8𝐸𝐸𝑂𝑂𝐸𝐸 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽9𝐺𝐺𝐸𝐸𝐸𝐸𝑃𝑃 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝑒𝑒𝐹𝐹, 𝑡𝑡 (1) where i = country, t = year. fi is a vector of dependent variables. fi includes: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. specifically, acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. atm = atms per 100,000 adults. bbpa = bank branches per 100,000 adults. mpb = adults using a mobile phone to pay bills. elp = adults who use electronic payments to make payments. for the explanatory variables, epud = year-end value of the epu index. npl = ratio of nonperforming loans to gross loans. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. variable justification the dependent variables were the financial inclusion variables, namely: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. these variables are commonly used in the literature to measure financial inclusion (imaeva et al., 2014; chakrabarty, 2011; ozili, 2018; célerier & matray, 2019). the explanatory variables were: epud, npl, dcp, octa and gdpr. the epu variables (i.e., epud and epua) and the associated interaction variables (in equation 1) were the explanatory variables of interest in the model. the epu variables were the epud and epua variables. the epud variable has been measured as the year-end value of the monthly epu index, i.e., the december value of the monthly epu index. the epua variable is the average of the monthly epu index values. a negative relationship between the epud and the epua variables is expected because high levels of economic policy uncertainty would negatively affect the performance of financial institutions. as a result, these financial institutions would be compelled to adjust their business decisions to reduce costs (caglayan & xu, 2019; lee et al., 2017), and this would in turn, affect the supply of basic financial services to individuals and households possibly through the closure of bank branches, discontinuation of certain financial services, higher fees for services, high interest rates, etc. thus, a negative sign on the epud or the epua coefficient would indicate that high levels of epu in the business environment would lead to lower levels of financial inclusion. the npl variable was introduced as a control variable. the npl variable measured the asset quality of the banking sector. a negative relationship between the npl and financial inclusion is expected because large npls would negatively affect bank profitability (ghosh, 2015; ozili, 2019). banks with large npls would expect low profits levels, and can proactively take steps to reduce costs, possibly by reducing the supply of costly financial services, such as, reducing the cost of maintaining bank branches and closing some branches in some rural and urban areas. this would lead to lower financial inclusion. the third explanatory variable is the octa variable, measured as the ratio of bank overhead cost to total asset ratio. the octa was introduced into the model to capture whether the propensity to supply financial services by banks was driven by overhead cost considerations. a negative relationship between octa and 𝐹𝐹𝐹𝐹𝐹𝐹, 𝑡𝑡 = 𝛽𝛽𝛽𝛽 + 𝛽𝛽1𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽2𝑁𝑁𝐸𝐸𝑁𝑁𝐹𝐹, 𝑡𝑡 + 𝛽𝛽3𝛽𝛽𝑂𝑂𝑂𝑂𝑂𝑂𝐹𝐹, 𝑡𝑡 + 𝛽𝛽4𝐸𝐸𝑂𝑂𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽5𝐺𝐺𝐸𝐸𝐸𝐸𝑃𝑃𝐹𝐹, 𝑡𝑡 + 𝛽𝛽6𝑁𝑁𝐸𝐸𝑁𝑁 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽7𝛽𝛽𝑂𝑂𝑂𝑂𝑂𝑂 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽8𝐸𝐸𝑂𝑂𝐸𝐸 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽9𝐺𝐺𝐸𝐸𝐸𝐸𝑃𝑃 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝑒𝑒𝐹𝐹, 𝑡𝑡 (1) where i = country, t = year. fi is a vector of dependent variables. fi includes: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. specifically, acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. atm = atms per 100,000 adults. bbpa = bank branches per 100,000 adults. mpb = adults using a mobile phone to pay bills. elp = adults who use electronic payments to make payments. for the explanatory variables, epud = year-end value of the epu index. npl = ratio of nonperforming loans to gross loans. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. variable justification the dependent variables were the financial inclusion variables, namely: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. these variables are commonly used in the literature to measure financial inclusion (imaeva et al., 2014; chakrabarty, 2011; ozili, 2018; célerier & matray, 2019). the explanatory variables were: epud, npl, dcp, octa and gdpr. the epu variables (i.e., epud and epua) and the associated interaction variables (in equation 1) were the explanatory variables of interest in the model. the epu variables were the epud and epua variables. the epud variable has been measured as the year-end value of the monthly epu index, i.e., the december value of the monthly epu index. the epua variable is the average of the monthly epu index values. a negative relationship between the epud and the epua variables is expected because high levels of economic policy uncertainty would negatively affect the performance of financial institutions. as a result, these financial institutions would be compelled to adjust their business decisions to reduce costs (caglayan & xu, 2019; lee et al., 2017), and this would in turn, affect the supply of basic financial services to individuals and households possibly through the closure of bank branches, discontinuation of certain financial services, higher fees for services, high interest rates, etc. thus, a negative sign on the epud or the epua coefficient would indicate that high levels of epu in the business environment would lead to lower levels of financial inclusion. the npl variable was introduced as a control variable. the npl variable measured the asset quality of the banking sector. a negative relationship between the npl and financial inclusion is expected because large npls would negatively affect bank profitability (ghosh, 2015; ozili, 2019). banks with large npls would expect low profits levels, and can proactively take steps to reduce costs, possibly by reducing the supply of costly financial services, such as, reducing the cost of maintaining bank branches and closing some branches in some rural and urban areas. this would lead to lower financial inclusion. the third explanatory variable is the octa variable, measured as the ratio of bank overhead cost to total asset ratio. the octa was introduced into the model to capture whether the propensity to supply financial services by banks was driven by overhead cost considerations. a negative relationship between octa and 𝐹𝐹𝐹𝐹𝐹𝐹, 𝑡𝑡 = 𝛽𝛽𝛽𝛽 + 𝛽𝛽1𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽2𝑁𝑁𝐸𝐸𝑁𝑁𝐹𝐹, 𝑡𝑡 + 𝛽𝛽3𝛽𝛽𝑂𝑂𝑂𝑂𝑂𝑂𝐹𝐹, 𝑡𝑡 + 𝛽𝛽4𝐸𝐸𝑂𝑂𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽5𝐺𝐺𝐸𝐸𝐸𝐸𝑃𝑃𝐹𝐹, 𝑡𝑡 + 𝛽𝛽6𝑁𝑁𝐸𝐸𝑁𝑁 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽7𝛽𝛽𝑂𝑂𝑂𝑂𝑂𝑂 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽8𝐸𝐸𝑂𝑂𝐸𝐸 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽9𝐺𝐺𝐸𝐸𝐸𝐸𝑃𝑃 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝑒𝑒𝐹𝐹, 𝑡𝑡 (1) where i = country, t = year. fi is a vector of dependent variables. fi includes: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. specifically, acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. atm = atms per 100,000 adults. bbpa = bank branches per 100,000 adults. mpb = adults using a mobile phone to pay bills. elp = adults who use electronic payments to make payments. for the explanatory variables, epud = year-end value of the epu index. npl = ratio of nonperforming loans to gross loans. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. variable justification the dependent variables were the financial inclusion variables, namely: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. these variables are commonly used in the literature to measure financial inclusion (imaeva et al., 2014; chakrabarty, 2011; ozili, 2018; célerier & matray, 2019). the explanatory variables were: epud, npl, dcp, octa and gdpr. the epu variables (i.e., epud and epua) and the associated interaction variables (in equation 1) were the explanatory variables of interest in the model. the epu variables were the epud and epua variables. the epud variable has been measured as the year-end value of the monthly epu index, i.e., the december value of the monthly epu index. the epua variable is the average of the monthly epu index values. a negative relationship between the epud and the epua variables is expected because high levels of economic policy uncertainty would negatively affect the performance of financial institutions. as a result, these financial institutions would be compelled to adjust their business decisions to reduce costs (caglayan & xu, 2019; lee et al., 2017), and this would in turn, affect the supply of basic financial services to individuals and households possibly through the closure of bank branches, discontinuation of certain financial services, higher fees for services, high interest rates, etc. thus, a negative sign on the epud or the epua coefficient would indicate that high levels of epu in the business environment would lead to lower levels of financial inclusion. the npl variable was introduced as a control variable. the npl variable measured the asset quality of the banking sector. a negative relationship between the npl and financial inclusion is expected because large npls would negatively affect bank profitability (ghosh, 2015; ozili, 2019). banks with large npls would expect low profits levels, and can proactively take steps to reduce costs, possibly by reducing the supply of costly financial services, such as, reducing the cost of maintaining bank branches and closing some branches in some rural and urban areas. this would lead to lower financial inclusion. the third explanatory variable is the octa variable, measured as the ratio of bank overhead cost to total asset ratio. the octa was introduced into the model to capture whether the propensity to supply financial services by banks was driven by overhead cost considerations. a negative relationship between octa and 𝐹𝐹𝐹𝐹𝐹𝐹, 𝑡𝑡 = 𝛽𝛽𝛽𝛽 + 𝛽𝛽1𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽2𝑁𝑁𝐸𝐸𝑁𝑁𝐹𝐹, 𝑡𝑡 + 𝛽𝛽3𝛽𝛽𝑂𝑂𝑂𝑂𝑂𝑂𝐹𝐹, 𝑡𝑡 + 𝛽𝛽4𝐸𝐸𝑂𝑂𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽5𝐺𝐺𝐸𝐸𝐸𝐸𝑃𝑃𝐹𝐹, 𝑡𝑡 + 𝛽𝛽6𝑁𝑁𝐸𝐸𝑁𝑁 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽7𝛽𝛽𝑂𝑂𝑂𝑂𝑂𝑂 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽8𝐸𝐸𝑂𝑂𝐸𝐸 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽9𝐺𝐺𝐸𝐸𝐸𝐸𝑃𝑃 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝑒𝑒𝐹𝐹, 𝑡𝑡 (1) where i = country, t = year. fi is a vector of dependent variables. fi includes: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. specifically, acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. atm = atms per 100,000 adults. bbpa = bank branches per 100,000 adults. mpb = adults using a mobile phone to pay bills. elp = adults who use electronic payments to make payments. for the explanatory variables, epud = year-end value of the epu index. npl = ratio of nonperforming loans to gross loans. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. variable justification the dependent variables were the financial inclusion variables, namely: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. these variables are commonly used in the literature to measure financial inclusion (imaeva et al., 2014; chakrabarty, 2011; ozili, 2018; célerier & matray, 2019). the explanatory variables were: epud, npl, dcp, octa and gdpr. the epu variables (i.e., epud and epua) and the associated interaction variables (in equation 1) were the explanatory variables of interest in the model. the epu variables were the epud and epua variables. the epud variable has been measured as the year-end value of the monthly epu index, i.e., the december value of the monthly epu index. the epua variable is the average of the monthly epu index values. a negative relationship between the epud and the epua variables is expected because high levels of economic policy uncertainty would negatively affect the performance of financial institutions. as a result, these financial institutions would be compelled to adjust their business decisions to reduce costs (caglayan & xu, 2019; lee et al., 2017), and this would in turn, affect the supply of basic financial services to individuals and households possibly through the closure of bank branches, discontinuation of certain financial services, higher fees for services, high interest rates, etc. thus, a negative sign on the epud or the epua coefficient would indicate that high levels of epu in the business environment would lead to lower levels of financial inclusion. the npl variable was introduced as a control variable. the npl variable measured the asset quality of the banking sector. a negative relationship between the npl and financial inclusion is expected because large npls would negatively affect bank profitability (ghosh, 2015; ozili, 2019). banks with large npls would expect low profits levels, and can proactively take steps to reduce costs, possibly by reducing the supply of costly financial services, such as, reducing the cost of maintaining bank branches and closing some branches in some rural and urban areas. this would lead to lower financial inclusion. the third explanatory variable is the octa variable, measured as the ratio of bank overhead cost to total asset ratio. the octa was introduced into the model to capture whether the propensity to supply financial services by banks was driven by overhead cost considerations. a negative relationship between octa and 𝐹𝐹𝐹𝐹𝐹𝐹, 𝑡𝑡 = 𝛽𝛽𝛽𝛽 + 𝛽𝛽1𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽2𝑁𝑁𝐸𝐸𝑁𝑁𝐹𝐹, 𝑡𝑡 + 𝛽𝛽3𝛽𝛽𝑂𝑂𝑂𝑂𝑂𝑂𝐹𝐹, 𝑡𝑡 + 𝛽𝛽4𝐸𝐸𝑂𝑂𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽5𝐺𝐺𝐸𝐸𝐸𝐸𝑃𝑃𝐹𝐹, 𝑡𝑡 + 𝛽𝛽6𝑁𝑁𝐸𝐸𝑁𝑁 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽7𝛽𝛽𝑂𝑂𝑂𝑂𝑂𝑂 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽8𝐸𝐸𝑂𝑂𝐸𝐸 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝛽𝛽9𝐺𝐺𝐸𝐸𝐸𝐸𝑃𝑃 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐹𝐹, 𝑡𝑡 + 𝑒𝑒𝐹𝐹, 𝑡𝑡 (1) where i = country, t = year. fi is a vector of dependent variables. fi includes: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. specifically, acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. atm = atms per 100,000 adults. bbpa = bank branches per 100,000 adults. mpb = adults using a mobile phone to pay bills. elp = adults who use electronic payments to make payments. for the explanatory variables, epud = year-end value of the epu index. npl = ratio of nonperforming loans to gross loans. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. variable justification the dependent variables were the financial inclusion variables, namely: atm, bbpa, mpb, elp, acc, dc, cc, sav and bor. these variables are commonly used in the literature to measure financial inclusion (imaeva et al., 2014; chakrabarty, 2011; ozili, 2018; célerier & matray, 2019). the explanatory variables were: epud, npl, dcp, octa and gdpr. the epu variables (i.e., epud and epua) and the associated interaction variables (in equation 1) were the explanatory variables of interest in the model. the epu variables were the epud and epua variables. the epud variable has been measured as the year-end value of the monthly epu index, i.e., the december value of the monthly epu index. the epua variable is the average of the monthly epu index values. a negative relationship between the epud and the epua variables is expected because high levels of economic policy uncertainty would negatively affect the performance of financial institutions. as a result, these financial institutions would be compelled to adjust their business decisions to reduce costs (caglayan & xu, 2019; lee et al., 2017), and this would in turn, affect the supply of basic financial services to individuals and households possibly through the closure of bank branches, discontinuation of certain financial services, higher fees for services, high interest rates, etc. thus, a negative sign on the epud or the epua coefficient would indicate that high levels of epu in the business environment would lead to lower levels of financial inclusion. the npl variable was introduced as a control variable. the npl variable measured the asset quality of the banking sector. a negative relationship between the npl and financial inclusion is expected because large npls would negatively affect bank profitability (ghosh, 2015; ozili, 2019). banks with large npls would expect low profits levels, and can proactively take steps to reduce costs, possibly by reducing the supply of costly financial services, such as, reducing the cost of maintaining bank branches and closing some branches in some rural and urban areas. this would lead to lower financial inclusion. the third explanatory variable is the octa variable, measured as the ratio of bank overhead cost to total asset ratio. the octa was introduced into the model to capture whether the propensity to supply financial services by banks was driven by overhead cost considerations. a negative relationship between octa and 61 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 2018; célerier & matray, 2019). the explanatory variables were: epud, npl, dcp, octa and gdpr. the epu variables (i.e., epud and epua) and the associated interaction variables (in equation 1) were the explanatory variables of interest in the model. the epu variables were the epud and epua variables. the epud variable has been measured as the year-end value of the monthly epu index, i.e., the december value of the monthly epu index. the epua variable is the average of the monthly epu index values. a negative relationship between the epud and the epua variables is expected because high levels of economic policy uncertainty would negatively affect the performance of financial institutions. as a result, these financial institutions would be compelled to adjust their business decisions to reduce costs (caglayan & xu, 2019; lee et al., 2017), and this would in turn, affect the supply of basic financial services to individuals and households possibly through the closure of bank branches, discontinuation of certain financial services, higher fees for services, high interest rates, etc. thus, a negative sign on the epud or the epua coefficient would indicate that high levels of epu in the business environment would lead to lower levels of financial inclusion. the npl variable was introduced as a control variable. the npl variable measured the asset quality of the banking sector. a negative relationship between the npl and financial inclusion is expected because large npls would negatively affect bank profitability (ghosh, 2015; ozili, 2019). banks with large npls would expect low profits levels, and can proactively take steps to reduce costs, possibly by reducing the supply of costly financial services, such as, reducing the cost of maintaining bank branches and closing some branches in some rural and urban areas. this would lead to lower financial inclusion. the third explanatory variable is the octa variable, measured as the ratio of bank overhead cost to total asset ratio. the octa was introduced into the model to capture whether the propensity to supply financial services by banks was driven by overhead cost considerations. a negative relationship between octa and the financial inclusion variables was expected because high overhead costs would negatively affect bank profitability (camanho & dyson, 2005; perera et al., 2007), and banks that had high overhead costs would take proactive steps to reduce overhead costs possibly by closing bank branches, thereby, leading to lower financial inclusion. 62 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 the fourth explanatory variable is the dcp variable which measured credit supply by banks to the private sector. a positive relationship between the dcp and the financial inclusion variables is expected because the high supply of bank credit to the private sector would stimulate the growth of credit-related financial services that are beneficial to households, individuals and small businesses, such as payday loans, instant loans, overdraft, etc. the fifth variable is the gdpr variable which measures the real gdp growth rate. it captures fluctuations in the business cycle. ozili (2020b) found evidence for a positive effect of the gdpr variable on the level of financial inclusion. finally, all models were estimated using the fixed effect regression. all the regression estimations included country and year fixed effects. a number of studies on financial inclusion have used the fixed effect regression approach to investigate the determinants and/or consequences of financial inclusion such as studies by markose et al. (2020), oz-yalaman (2019), anson et al. (2013) and le et al. (2020). accordingly, the present study has also made use of the fixed effect approach. results descriptive results npl was found to average 5.5 percent of the gross loans. the npl ratio was at a double-digit higher in greece, ireland and italy, but was much lower in canada, korea and sweden. the dcp ratio was 105 percent, but exhibited substantial differences across countries in the sample. for instance, the dcps were much lower in mexico, russia and colombia, and higher in the us, uk and japan. on average, the gdpr was 2.5 percent and was higher for banks in china, ireland and india, but lower in greece and italy. the octa ratio was 2.3 percent on average, and was higher in russia and colombia, but lower in japan, singapore and australia. the epud and the epua were higher in the uk, brazil and france compared to the readings in mexico and italy. overall, the mean of the explanatory variables was higher than the median values except for the dcp. finally, all the financial inclusion vector variables were higher in australia, canada and japan compared to those in the other countries in the sample. 63 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 ta bl e 1 d es cr ip tiv e st at is tic s (m ea n va lu es o f t he v ar ia bl es ) c ou nt ry n pl d c p g d pr e pu d e pu a o c ta e l p m pb a c c a t m b b pa c c d c sa v b o r a us tr al ia 1. 3 13 0 2. 7 13 5 12 0 0. 9 93 .3 19 .4 99 16 4 29 61 84 61 20 b ra zi l 2. 9 63 0. 3 23 6 22 0 3. 8 46 .6 1. 5 68 12 7 21 36 58 17 11 c an ad a 0. 8 10 1 2. 5 18 7 18 8 1. 8 95 .4 14 .7 90 19 7 22 68 82 53 22 c hi le 1. 8 11 0 3. 7 18 7 13 1 2. 1 49 .3 3. 4 58 54 16 24 44 17 11 c hi na 1. 5 13 1 7. 2 17 5 17 2 1. 8 43 .8 5. 3 67 59 9 13 45 33 9 c ol om bi a 3. 2 56 3. 1 15 2 13 1 4. 5 25 .0 0. 7 45 49 19 16 32 16 14 fr an ce 3. 8 95 1. 5 25 2 25 1 1. 0 89 .8 3. 4 96 10 8 34 40 79 51 16 g er m an y 4. 8 85 0. 1 15 3 16 2 1. 4 94 .1 4. 0 95 11 3 17 39 82 51 15 g re ec e 30 .0 10 3 -0 .5 12 7 11 8 1. 5 31 .1 0. 9 77 54 26 11 44 14 7 in di a 7. 5 60 6. 1 12 5 11 1 1. 6 16 .7 0. 4 57 27 15 10 26 19 8 ir el an d 17 .5 77 6. 8 17 2 15 1 1. 5 86 .2 11 .5 91 91 27 47 74 44 15 it al y 14 .3 98 -0 .1 11 4 12 4 1. 6 71 .2 2. 1 84 99 48 39 51 32 10 ja pa n 1. 5 15 9 1. 6 13 0 11 8 0. 7 82 .5 0. 8 96 14 9 31 64 62 56 8 k or ea 0. 9 12 3 2. 9 15 7 13 1 1. 9 85 .1 6. 9 84 24 6 16 51 59 44 16 (c on tin ue d) 64 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 c ou nt ry n pl d c p g d pr e pu d e pu a o c ta e l p m pb a c c a t m b b pa c c d c sa v b o r m ex ic o 2. 5 42 2. 5 62 56 3. 2 24 .6 1. 7 43 51 15 18 35 17 9 n et he rl an d 3. 3 10 3 1. 6 11 5 10 7 3. 7 96 .6 10 .5 91 59 18 33 89 51 11 r us si a 6. 8 62 1. 7 24 5 18 5 7. 8 48 .6 2. 9 67 15 0 30 19 41 20 10 si ng ap or e 1. 7 12 8 3. 1 14 3 14 0 0. 9 85 .1 3. 9 96 72 20 38 68 51 12 sp ai n 6. 0 12 7 1. 4 13 2 12 8 1. 3 89 .7 4. 6 96 11 0 62 50 78 46 16 sw ed en 1. 4 13 6 1. 9 13 1 12 0 1. 1 97 .6 14 .2 99 50 20 48 95 67 23 u k 2. 3 14 7 1. 9 28 4 29 8 1. 7 95 .2 12 .4 96 12 8 27 59 89 51 17 u s 2. 2 18 9 2. 1 14 3 13 5 2. 7 88 .4 17 .6 91 . 33 61 74 53 22 m ea n 5. 5 10 5 2. 5 16 2 15 0 2. 3 69 6. 5 81 10 3 25 39 64 40 14 m ed ia n 2. 71 11 1 2. 2 14 2 13 7 1. 6 84 3. 7 94 96 21 38 69 48 13 s. d . 7. 8 42 3. 3 84 72 2. 5 27 6. 9 22 63 15 20 27 21 6 o bs er va tio ns 14 5 14 7 15 4 15 4 15 4 15 4 88 88 15 4 14 7 14 9 15 4 15 4 15 4 15 4 n ot e. a c c = a du lts w ho o w n a fo rm al a cc ou nt . d c = d eb it ca rd o w ne rs hi p. c c = c re di t c ar d ow ne rs hi p. s a v = a du lts w ho h av e fo rm al s av in gs . b o r = a du lts w ho h av e fo rm al b or ro w in gs . a t m = a t m s pe r 10 0, 00 0 ad ul ts . b b pa = b an k br an ch es p er 1 00 ,0 00 a du lts . m pb = a du lts u si ng a m ob ile p ho ne to p ay b ill s. e l p = ad ul ts w ho u se e le ct ro ni c pa ym en ts to m ak e pa ym en ts . e pu d = y ea ren d va lu e of th e m on th ly e pu in de x. e pu a = av er ag e va lu e of th e m on th ly e pu in de x. n pl = n on pe rf or m in g lo an s ra tio . o c ta = th e ra tio o f ba nk o ve rh ea d co st to to ta l a ss et r at io . d c p = cr ed it su pp ly to th e pr iv at e se ct or b y ba nk s to g d p ra tio . g d pr = re al g d p gr ow th ra te 65 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 correlation analysis table 2 reports the pearson correlation results. the npl was negatively correlated with the epud and the epua variables. this indicated that a high npl was associated with a low epu. the gdpr was negatively correlated with the epud and the epua variables. the gdpr was significantly correlated with the epua, which seemed to suggest that a high epu was associated with economic downturns. the dcp was positively correlated with the epud and the epua, but the correlation coefficient was insignificant. similarly, the octa was positively correlated with the epud and the epua variables, but the correlation coefficient was insignificant. overall, the correlations were low, and indicated that multi-collinearity was not a problem in the analysis. table 2 correlation of epu and the explanatory variables (pearson correlation) variables epud epua npl gdpr dcp octa epud 1.000 ----epua 0.839*** 1.000 (0.00) ----npl -0.079 -0.065 1.000 (0.35) (0.44) ----gdpr -0.121 -0.156* -0.205** 1.000 (0.15) (0.06) (0.02) ----dcp 0.046 0.093 -0.129 -0.188*** 1.000 (0.59) (0.27) (0.13) (0.03) ----octa 0.071 0.023 -0.027 0.020 -0.463*** 1.000 (0.41) (0.78) (0.74) (0.81) (0.00) ----note. p-value is reported in parenthesis. ***, **, * represent statistical significance at the 1%, 5% and 10% levels. epud = year-end value of the monthly epu index. epua = average value of the monthly epu index. npl = nonperforming loans ratio. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. 66 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 (c on tin ue d) ta bl e 3 c or re la tio n of e p u a nd th e d ep en de nt v ar ia bl es (p ea rs on c or re la tio n) v ar ia bl e e pu d e pu a a t m b b pa m pb e l p a c c c c d c sa v b o r e pu d 1 .0 00 --e pu a 0. 79 ** * 1. 00 0 (0 .0 0) --a t m 0. 30 ** 0. 31 6* ** 1. 00 0 (0 .0 1) (0 .0 0) --b b pa 0. 03 0. 03 2 0. 22 5* * 1 .0 00 (0 .7 9) (0 .7 7) (0 .0 4) --m pb -0 .1 2 0. 02 2 0. 26 9* * -0 .0 79 1. 00 0 (0 .2 9) (0 .8 5) (0 .0 2) (0 .4 8) --e l p -0 .0 2 0. 12 0. 42 ** * 0. 24 ** 0. 57 2* ** 1. 00 0 (0 .8 5) (0 .2 8) (0 .0 0) (0 .0 3) (0 .0 0) -- 67 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 v ar ia bl e e pu d e pu a a t m b b pa m pb e l p a c c c c d c sa v b o r a c c -0 .0 1 0. 13 0. 38 ** * 0. 27 ** 0. 46 3* ** 0. 87 0* ** 1. 00 0 (0 .9 1) (0 .2 5) (0 .0 0) (0 .0 1) (0 .0 0) (0 .0 0) --c c 0. 01 0. 14 0. 66 ** * 0. 32 ** * 0. 50 4* ** 0. 84 8* ** 0. 72 5* ** 1. 00 0 (0 .9 4) (0 .2 3) (0 .0 0) (0 .0 0) (0 .0 0) (0 .0 0) (0 .0 0) --d c -0 .0 5 0. 13 0. 29 ** 0. 19 * 0. 55 1* ** 0. 95 1* ** 0. 91 9* ** 0. 79 1* ** 1. 00 0 (0 .6 6) (0 .2 7) (0 .0 1) (0 .0 8) (0 .0 0) (0 .0 0) (0 .0 0) (0 .0 0) --sa v -0 .1 5 -0 .0 2 0. 33 ** * 0. 10 3 0. 57 8* ** 0. 90 3* ** 0. 83 7* ** 0. 81 0* ** 0. 87 8* ** 1. 00 0 (0 .1 6) (0 .8 9) (0 .0 0) (0 .3 5) (0 .0 0) (0 .0 0) (0 .0 0) (0 .0 0) (0 .0 0) --b o r 0. 03 0. 10 0. 38 ** * 0. 10 1 0. 58 0* ** 0. 64 6* ** 0. 47 1* ** 0. 65 4* ** 0. 57 9* ** 0. 62 6* ** 1. 00 0 (0 .7 5) (0 .3 6) (0 .0 0) (0 .3 7) (0 .0 0) (0 .0 0) (0 .0 0) (0 .0 0) (0 .0 0) (0 .0 0) --n ot e. p -v al ue s ar e re po rt ed in p ar en th es is . * ** , * *, * re pr es en t s ig ni fic an ce a t t he 1 % , 5 % a nd 1 0% le ve l. a c c = a du lts w ho o w n a fo rm al a cc ou nt . d c = d eb it ca rd o w ne rs hi p. c c = c re di t ca rd o w ne rs hi p. s a v = a du lts w ho h av e fo rm al s av in gs . b o r = a du lts w ho h av e fo rm al b or ro w in gs . a t m = a t m s pe r 10 0, 00 0 ad ul ts . b b pa = b an k br an ch es p er 1 00 ,0 00 a du lts . m pb = a du lts u si ng a m ob ile p ho ne to p ay b ill s. e l p = ad ul ts w ho us e el ec tr on ic p ay m en ts to m ak e pa ym en ts . e pu d = y ea ren d va lu e of th e m on th ly e pu in de x. e pu a = a ve ra ge v al ue o f th e m on th ly e pu in de x 68 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 table 3 shows the pearson correlation result for each of the financial inclusion variables. the two epu variables (i.e., the epud and the epua) were significant and negatively correlated with atm, which seemed to suggest that a higher economic policy uncertainty was associated with a lower supply of atms per 100,000 adults. on the other hand, the epud and the epua were positively correlated with some bor, cc and bbpa, but the correlation was insignificant. similarly, the epud and the epua were negatively correlated with sav and the correlation was insignificant. the remaining dependent variables (mpb, elp, acc and dc) showed conflicting signs when correlated with the epud and the epua. overall, the correlations were low, and indicated that multi-collinearity was not a problem in the analysis. regression results effect of epu on financial inclusion the results are reported in table 4 and table 5. only the significant results were interpreted. the epud coefficient was insignificant in columns 1 to 9 in table 4 and table 5. this indicated that economic policy uncertainty was not significantly related to the nine financial inclusion indicators. regarding the control variables, the npl coefficient was negative, as was expected in six of the nine models. this has confirmed the prediction that the npl would have a negative relationship with financial inclusion. more specifically, a high npl would lead to a decrease in atm supply and bank branch contraction. in contrast, a high npl was associated with an increase in formal accounts and the use of electronic payments. the gdpr coefficient was significant and negatively related to cc in column 7. this seemed to suggest that the use of credit cards was higher during periods of economic prosperity. the dcp coefficient was significant and positively related to the acc and the bbpa. this result supported the a priori expectation, and indicated that a higher supply of credit to the private sector would lead to a significant increase in atm supply and bank branch expansion, thereby, increasing financial inclusion. in contrast, the dcp coefficient was significant and negatively related to the elp and the cc, which indicated that a higher supply of credit to the private sector would lead to a significant decrease in the number 69 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 of adults using electronic payments and a decrease in credit card usage, thereby, decreasing financial inclusion. the octa coefficient was significant and negatively related to atm, acc and cc. this result supported the a priori expectation, and indicated that high overhead costs in banks would lead to a significant decrease in atm supply, formal account ownership and credit card usage, thereby, decreasing financial inclusion. in contrast, the octa coefficient was significant and positively related to the bbpa, which indicated that high overhead costs in banks would lead to a significant increase in the number of bank branches. table 4 effect of epu on financial inclusion (1) (2) (3) (4) variable atm bbpa mpb elp c 70.068***(7.57) 11.577*** (3.76) -3.323 (-0.53) 70.109*** (8.56) epud 0.018 (1.31) -0.0005 (-0.10) 0.001 (0.13) 0.011 (1.29) npl -0.628** (-2.29) -0.343** (-4.12) -0.184 (-0.84) 1.673*** (5.79) gdpr 0.260 (0.74) -0.074 (-0.68) 0.040 (0.25) 0.135 (0.65) dcp 0.266*** (3.21) 0.151*** (5.73) 0.101 (1.66) -0.149* (-1.86) octa -1.400*** (-2.89) 0.438*** (2.93) -0.185 (-0.68) -0.107 (-0.29) adjusted r2 98.05 97.73 85.06 98.54 f-statistic 218.90 183.42 16.05 179.68 observation 131 134 75 75 note. regression in table 4 includes country and year fixed effect. t-statistic is reported in parenthesis. *, **, *** represent significance level at 10%, 5% and 1% level. atm = atms per 100,000 adults. bbpa = bank branches per 100,000 adults. mpb = adults using a mobile phone to pay bills. elp = adults who use electronic payments to make payments. epud = year-end value of the monthly epu index. npl = nonperforming loans ratio. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. 70 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 table 5 effect of epu on financial inclusion (5) (6) (7) (8) (9) variable acc dc cc sav bor β0 74.817*** (13.18) 57.946*** (3.28) 45.921*** (10.43) 43.506*** (8.79) 16.309*** (6.54) epud 0.012 (1.51) -0.006 (-0.31) 0.005 (0.71) -0.005 (-0.70) 0.003 (0.73) npl 0.486*** (3.00) 0.445 (1.18) -0.016 (-0.13) -0.139 (-0.99) -0.142 (-1.99) gdpr -0.223 (-1.06) -0.225 (-0.46) -0.501*** (-3.09) -0.143 (-0.79) -0.091 (-0.99) dcp 0.024 (0.49) 0.012 (0.11) -0.074* (-1.97) -0.032 (-0.76) -0.019 (-0.92) octa -0.685** (-2.36) 0.416 (0.62) -0.573** (-2.55) -0.095 (-0.37) -0.089 (-0.70) adjusted r2 95.99 84.95 96.57 96.33 87.86 f-statistic 107.01 25.94 125.77 117.01 32.97 observation 138 138 138 138 138 note. regression in table 5 includes country and year fixed effect. t-statistics is reported in parenthesis. *, **, *** represent significance level at 10%, 5% and 1% level. acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. epud = year-end value of the monthly epu index. npl = nonperforming loans ratio. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. interaction analysis – effect of epu on financial inclusion the interaction results are shown in table 6 and table 7. only the significant results were interpreted. the npl*epud coefficient was significant and negatively related to the bbpa and the elp in columns 2 and 4. this seemed to suggest that the combined effect of high levels of economic policy uncertainty and high non-performing loans lead to bank branch contraction and a reduction in the use of electronic payments, thereby reducing financial inclusion. the gdpr*epud coefficient was significant and positively related with the elp, and negatively related to the acc in columns 4 and 5, respectively. 71 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 table 6 interaction analysis – effect of epu on financial inclusion (1) (2) (3) (4) variable atm bbpa mpb elp β0 93.949*** (6.92) 11.381** (2.59) 3.789 (0.36) 68.598*** (6.18) epud -0.099** (-2.14) 0.004 (0.27) -0.021 (-0.67) 0.012 (1.13) npl -0.395 (-1.39) -0.254** (-2.58) -0.233 (-0.93) 1.820*** (5.86) gdpr -0.411 (-0.94) 0.022 (0.14) 0.014 (0.06) 0.237 (1.07) dcp 0.157 (1.36) 0.152*** (4.21) 0.070 (0.79) -0.077 (-1.21) octa -6.008*** (-6.19) 0.266 (0.81) -1.894 (-1.24) 0.077 (0.91) npl*epud -0.002 (-1.13) -0.001* (-1.85) 0.0001 (0.11) -0.003* (-1.83) gdpr*epud 0.005 (1.52) -0.001 (-0.99) -0.0004 (-0.21) 0.005** (2.17) dcp*epud 0.0005 (1.42) -0.0001 (-0.001) 0.0001 (0.44) -0.001** (-2.09) octa*epud 0.023*** (5.18) 0.001 (0.49) 0.006 (1.15) 0.004 (0.62) adjusted r2 98.46 97.73 84.49 98.68 f-statistic 246.03 164.75 13.59 174.32 observation 131 134 75 75 note. regression in table 4 includes country and year fixed effect. t-statistic is reported in parenthesis. *, **, *** represent significance level at 10%, 5% and 1% level. atm = atms per 100,000 adults. bbpa = bank branches per 100,000 adults. mpb = adults using a mobile phone to pay bills. elp = adults who use electronic payments to make payments. epud = year-end value of the monthly epu index. npl = nonperforming loans ratio. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. this seemed to suggest that high levels of the epu in times of economic boom led to higher electronic payments and a decrease in formal account ownership. 72 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 the dcp*epud coefficient was significant and negatively related to the elp, and positively related to the acc and the cc in columns 4, 5 and 7, respectively. this seemed to suggest that a high credit supply in times of high levels of the epu led to higher formal account ownership, higher credit card usage and a decrease in the use of electronic payments. the octa*epud coefficient was significant and positively related to the atm, the acc and the cc in columns 1, 5 and 7, respectively. this seemed to suggest that high overhead costs in banks in times of high levels of the epu led to higher formal account ownership, greater atm supply and higher credit card usage. table 7 interaction analysis – effect of epu on financial inclusion (5) (6) (7) (8) (9) variable acc dc cc sav bor β0 90.816*** (10.89) 65.705*** (3.28) 55.823*** (8.49) 46.677*** (6.17) 16.465*** (4.32) epud -0.049* (-1.77) -0.028 (-0.41) -0.039 (-1.75) -0.019 (-0.78) 0.001 (0.09) npl 0.345* (1.85) 0.452 (1.01) -0.018 (-0.12) -0.108 (-0.64) -0.017 (-1.99) gdpr -0.036 (-0.12) -0.072 (-0.10) -0.604** (-2.65) -0.260 (-0.99) -0.099 (-0.75) dcp -0.096 (-1.41) -0.077 (-0.47) -0.142*** (-2.63) -0.057 (-0.92) -0.019 (-0.62) octa -1.718*** (-2.76) 0.994 (0.67) -1.475*** (-3.00) -0.268 (-0.48) -0.103 (-0.36) npl*epud 0.001 (1.10) -0.001 (-0.25) -0.0001 (-0.15) -0.0004 (-0.44) 0.0003 (0.68) gdpr*epud -0.003* (-1.64) -0.003 (-0.68) -0.0001 (-0.08) 0.0005 (0.29) 0.0002 (0.18) dcp*epud 0.0004** (2.15) 0.0004 (0.77) 0.0003* (1.83) 0.0001 (0.71) -0.00001 (-0.07) octa*epud 0.005* (1.96) -0.003 (-0.45) 0.004** (2.03) 0.001 (0.29) 0.0001 (0.10) adjusted r2 96.19 84.58 96.63 96.23 87.45 f-statistic 99.89 22.48 113.31 100.82 28.26 observation 138 138 138 138 138 73 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 note. regression in table 5 includes country and year fixed effect. t-statistics is reported in parenthesis. *, **, *** represent significance level at 10%, 5% and 1% level. acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. epud = year-end value of the monthly epu index. npl = nonperforming loans ratio. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. additional analysis an additional analysis was also performed in the present study. the epua variable – the average of the 12-month epu index – was introduced into the model as a time-sensitive alternative proxy of economic policy uncertainty. this was expressed as equation (2): (2) the results are reported in table 8. the npl*epua coefficient was significant and negatively related to the bbpa in column 3 of table 8, and was consistent with the earlier result for the npl*epud reported in column 2 of table 6. the octa*epua coefficient was also significant and positively related to the atm in column 2 of table 6, and was consistent with the earlier result for the octa*epud reported in column 1 of table 4. the gdpr*epua coefficient was significant and positively related to the elp in column 9 of table 8, and was consistent with the earlier result for the gdpr*epud reported in column 4 of table 6. also, the gdpr*epua coefficient was significant and negatively related to the acc in column 5 of table 8, and was consistent with the earlier result for the gdpr*epud reported in column 5 of table 7. on the other hand, the remaining results for the interaction analyses showed conflicting signs when the epud and the epua variables were used as alternative proxies for economic policy uncertainty (epu). interaction analysis – effect of epu on financial inclusion (5) (6) (7) (8) (9) variable acc dc cc sav bor β0 90.816*** (10.89) 65.705*** (3.28) 55.823*** (8.49) 46.677*** (6.17) 16.465*** (4.32) epud -0.049* (-1.77) -0.028 (-0.41) -0.039 (-1.75) -0.019 (-0.78) 0.001 (0.09) npl 0.345* (1.85) 0.452 (1.01) -0.018 (-0.12) -0.108 (-0.64) -0.017 (-1.99) gdpr -0.036 (-0.12) -0.072 (-0.10) -0.604** (-2.65) -0.260 (-0.99) -0.099 (-0.75) dcp -0.096 (-1.41) -0.077 (-0.47) -0.142*** (-2.63) -0.057 (-0.92) -0.019 (-0.62) octa -1.718*** (-2.76) 0.994 (0.67) -1.475*** (-3.00) -0.268 (-0.48) -0.103 (-0.36) npl*epud 0.001 (1.10) -0.001 (-0.25) -0.0001 (-0.15) -0.0004 (-0.44) 0.0003 (0.68) gdpr*epud -0.003* (-1.64) -0.003 (-0.68) -0.0001 (-0.08) 0.0005 (0.29) 0.0002 (0.18) dcp*epud 0.0004** (2.15) 0.0004 (0.77) 0.0003* (1.83) 0.0001 (0.71) -0.00001 (-0.07) octa*epud 0.005* (1.96) -0.003 (-0.45) 0.004** (2.03) 0.001 (0.29) 0.0001 (0.10) adjusted r2 96.19 84.58 96.63 96.23 87.45 f-statistic 99.89 22.48 113.31 100.82 28.26 observation 138 138 138 138 138 note. regression in table 5 includes country and year fixed effect. t-statistics is reported in parenthesis. *, **, *** represent significance level at 10%, 5% and 1% level. acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. epud = year-end value of the monthly epu index. npl = nonperforming loans ratio. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. additional analysis an additional analysis was also performed in the present study. the epua variable – the average of the 12-month epu index – was introduced into the model as a time-sensitive alternative proxy of economic policy uncertainty. this was expressed as equation (2): 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑁𝑁𝐸𝐸𝑁𝑁𝐹𝐹𝐹𝐹 + 𝑂𝑂𝑂𝑂𝑂𝑂𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐷𝐷𝑂𝑂𝐸𝐸𝐹𝐹𝐹𝐹 + 𝐺𝐺𝐷𝐷𝐸𝐸𝐺𝐺𝐹𝐹𝐹𝐹 + 𝑁𝑁𝐸𝐸𝑁𝑁 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑂𝑂𝑂𝑂𝑂𝑂𝐴𝐴 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐷𝐷𝑂𝑂𝐸𝐸 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐺𝐺𝐷𝐷𝐸𝐸𝐺𝐺 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑒𝑒𝐹𝐹𝐹𝐹 (2) the results are reported in table 8. the npl*epua coefficient was significant and negatively related to the bbpa in column 3 of table 8, and was consistent with the earlier result for the npl*epud reported interaction analysis – effect of epu on financial inclusion (5) (6) (7) (8) (9) variable acc dc cc sav bor β0 90.816*** (10.89) 65.705*** (3.28) 55.823*** (8.49) 46.677*** (6.17) 16.465*** (4.32) epud -0.049* (-1.77) -0.028 (-0.41) -0.039 (-1.75) -0.019 (-0.78) 0.001 (0.09) npl 0.345* (1.85) 0.452 (1.01) -0.018 (-0.12) -0.108 (-0.64) -0.017 (-1.99) gdpr -0.036 (-0.12) -0.072 (-0.10) -0.604** (-2.65) -0.260 (-0.99) -0.099 (-0.75) dcp -0.096 (-1.41) -0.077 (-0.47) -0.142*** (-2.63) -0.057 (-0.92) -0.019 (-0.62) octa -1.718*** (-2.76) 0.994 (0.67) -1.475*** (-3.00) -0.268 (-0.48) -0.103 (-0.36) npl*epud 0.001 (1.10) -0.001 (-0.25) -0.0001 (-0.15) -0.0004 (-0.44) 0.0003 (0.68) gdpr*epud -0.003* (-1.64) -0.003 (-0.68) -0.0001 (-0.08) 0.0005 (0.29) 0.0002 (0.18) dcp*epud 0.0004** (2.15) 0.0004 (0.77) 0.0003* (1.83) 0.0001 (0.71) -0.00001 (-0.07) octa*epud 0.005* (1.96) -0.003 (-0.45) 0.004** (2.03) 0.001 (0.29) 0.0001 (0.10) adjusted r2 96.19 84.58 96.63 96.23 87.45 f-statistic 99.89 22.48 113.31 100.82 28.26 observation 138 138 138 138 138 note. regression in table 5 includes country and year fixed effect. t-statistics is reported in parenthesis. *, **, *** represent significance level at 10%, 5% and 1% level. acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. epud = year-end value of the monthly epu index. npl = nonperforming loans ratio. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. additional analysis an additional analysis was also performed in the present study. the epua variable – the average of the 12-month epu index – was introduced into the model as a time-sensitive alternative proxy of economic policy uncertainty. this was expressed as equation (2): 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑁𝑁𝐸𝐸𝑁𝑁𝐹𝐹𝐹𝐹 + 𝑂𝑂𝑂𝑂𝑂𝑂𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐷𝐷𝑂𝑂𝐸𝐸𝐹𝐹𝐹𝐹 + 𝐺𝐺𝐷𝐷𝐸𝐸𝐺𝐺𝐹𝐹𝐹𝐹 + 𝑁𝑁𝐸𝐸𝑁𝑁 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑂𝑂𝑂𝑂𝑂𝑂𝐴𝐴 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐷𝐷𝑂𝑂𝐸𝐸 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐺𝐺𝐷𝐷𝐸𝐸𝐺𝐺 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑒𝑒𝐹𝐹𝐹𝐹 (2) the results are reported in table 8. the npl*epua coefficient was significant and negatively related to the bbpa in column 3 of table 8, and was consistent with the earlier result for the npl*epud reported interaction analysis – effect of epu on financial inclusion (5) (6) (7) (8) (9) variable acc dc cc sav bor β0 90.816*** (10.89) 65.705*** (3.28) 55.823*** (8.49) 46.677*** (6.17) 16.465*** (4.32) epud -0.049* (-1.77) -0.028 (-0.41) -0.039 (-1.75) -0.019 (-0.78) 0.001 (0.09) npl 0.345* (1.85) 0.452 (1.01) -0.018 (-0.12) -0.108 (-0.64) -0.017 (-1.99) gdpr -0.036 (-0.12) -0.072 (-0.10) -0.604** (-2.65) -0.260 (-0.99) -0.099 (-0.75) dcp -0.096 (-1.41) -0.077 (-0.47) -0.142*** (-2.63) -0.057 (-0.92) -0.019 (-0.62) octa -1.718*** (-2.76) 0.994 (0.67) -1.475*** (-3.00) -0.268 (-0.48) -0.103 (-0.36) npl*epud 0.001 (1.10) -0.001 (-0.25) -0.0001 (-0.15) -0.0004 (-0.44) 0.0003 (0.68) gdpr*epud -0.003* (-1.64) -0.003 (-0.68) -0.0001 (-0.08) 0.0005 (0.29) 0.0002 (0.18) dcp*epud 0.0004** (2.15) 0.0004 (0.77) 0.0003* (1.83) 0.0001 (0.71) -0.00001 (-0.07) octa*epud 0.005* (1.96) -0.003 (-0.45) 0.004** (2.03) 0.001 (0.29) 0.0001 (0.10) adjusted r2 96.19 84.58 96.63 96.23 87.45 f-statistic 99.89 22.48 113.31 100.82 28.26 observation 138 138 138 138 138 note. regression in table 5 includes country and year fixed effect. t-statistics is reported in parenthesis. *, **, *** represent significance level at 10%, 5% and 1% level. acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. epud = year-end value of the monthly epu index. npl = nonperforming loans ratio. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. additional analysis an additional analysis was also performed in the present study. the epua variable – the average of the 12-month epu index – was introduced into the model as a time-sensitive alternative proxy of economic policy uncertainty. this was expressed as equation (2): 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑁𝑁𝐸𝐸𝑁𝑁𝐹𝐹𝐹𝐹 + 𝑂𝑂𝑂𝑂𝑂𝑂𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐷𝐷𝑂𝑂𝐸𝐸𝐹𝐹𝐹𝐹 + 𝐺𝐺𝐷𝐷𝐸𝐸𝐺𝐺𝐹𝐹𝐹𝐹 + 𝑁𝑁𝐸𝐸𝑁𝑁 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑂𝑂𝑂𝑂𝑂𝑂𝐴𝐴 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐷𝐷𝑂𝑂𝐸𝐸 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐺𝐺𝐷𝐷𝐸𝐸𝐺𝐺 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑒𝑒𝐹𝐹𝐹𝐹 (2) the results are reported in table 8. the npl*epua coefficient was significant and negatively related to the bbpa in column 3 of table 8, and was consistent with the earlier result for the npl*epud reported interaction analysis – effect of epu on financial inclusion (5) (6) (7) (8) (9) variable acc dc cc sav bor β0 90.816*** (10.89) 65.705*** (3.28) 55.823*** (8.49) 46.677*** (6.17) 16.465*** (4.32) epud -0.049* (-1.77) -0.028 (-0.41) -0.039 (-1.75) -0.019 (-0.78) 0.001 (0.09) npl 0.345* (1.85) 0.452 (1.01) -0.018 (-0.12) -0.108 (-0.64) -0.017 (-1.99) gdpr -0.036 (-0.12) -0.072 (-0.10) -0.604** (-2.65) -0.260 (-0.99) -0.099 (-0.75) dcp -0.096 (-1.41) -0.077 (-0.47) -0.142*** (-2.63) -0.057 (-0.92) -0.019 (-0.62) octa -1.718*** (-2.76) 0.994 (0.67) -1.475*** (-3.00) -0.268 (-0.48) -0.103 (-0.36) npl*epud 0.001 (1.10) -0.001 (-0.25) -0.0001 (-0.15) -0.0004 (-0.44) 0.0003 (0.68) gdpr*epud -0.003* (-1.64) -0.003 (-0.68) -0.0001 (-0.08) 0.0005 (0.29) 0.0002 (0.18) dcp*epud 0.0004** (2.15) 0.0004 (0.77) 0.0003* (1.83) 0.0001 (0.71) -0.00001 (-0.07) octa*epud 0.005* (1.96) -0.003 (-0.45) 0.004** (2.03) 0.001 (0.29) 0.0001 (0.10) adjusted r2 96.19 84.58 96.63 96.23 87.45 f-statistic 99.89 22.48 113.31 100.82 28.26 observation 138 138 138 138 138 note. regression in table 5 includes country and year fixed effect. t-statistics is reported in parenthesis. *, **, *** represent significance level at 10%, 5% and 1% level. acc = adults who own a formal account. dc = debit card ownership. cc = credit card ownership. sav = adults who have formal savings. bor = adults who have formal borrowings. epud = year-end value of the monthly epu index. npl = nonperforming loans ratio. octa = the ratio of bank overhead cost to total asset ratio. dcp = credit supply to the private sector by banks to gdp ratio. gdpr = real gdp growth rate. additional analysis an additional analysis was also performed in the present study. the epua variable – the average of the 12-month epu index – was introduced into the model as a time-sensitive alternative proxy of economic policy uncertainty. this was expressed as equation (2): 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑁𝑁𝐸𝐸𝑁𝑁𝐹𝐹𝐹𝐹 + 𝑂𝑂𝑂𝑂𝑂𝑂𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐷𝐷𝑂𝑂𝐸𝐸𝐹𝐹𝐹𝐹 + 𝐺𝐺𝐷𝐷𝐸𝐸𝐺𝐺𝐹𝐹𝐹𝐹 + 𝑁𝑁𝐸𝐸𝑁𝑁 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑂𝑂𝑂𝑂𝑂𝑂𝐴𝐴 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐷𝐷𝑂𝑂𝐸𝐸 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝐺𝐺𝐷𝐷𝐸𝐸𝐺𝐺 ∗ 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐹𝐹𝐹𝐹 + 𝑒𝑒𝐹𝐹𝐹𝐹 (2) the results are reported in table 8. the npl*epua coefficient was significant and negatively related to the bbpa in column 3 of table 8, and was consistent with the earlier result for the npl*epud reported 74 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 ta bl e 8 e ffe ct o f e p u o n f in an ci al in cl us io n a dd iti on al a na ly si s (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) (9 ) v ar ia bl e a t m b b pa m pb e l p a c c d c c c sa v b o r β0 73 .3 75 ** * (7 .6 7) 11 .1 47 ** * (3 .3 3) 0. 18 9 (0 .0 3) 77 .7 8* ** (8 .8 8) 82 .0 29 ** * (1 3. 14 ) 60 .8 25 ** * (4 .1 0) 48 .5 64 ** * (9 .8 5) 43 .3 83 ** (7 .7 5) 14 .1 17 ** * (5 .0 9) e pu a -0 .0 14 (0. 59 ) 0. 00 6 (0 .6 8) -0 .0 13 (1. 03 ) -0 .0 13 (0. 84 ) -0 .0 12 (0. 82 ) -0 .0 07 (0. 19 ) -0 .0 07 (0. 63 ) -0 .0 06 (0. 43 ) 0. 01 3 (1 .9 2) n pl -0 .2 78 (0. 97 ) -0 .2 61 ** (2. 72 ) -0 .2 73 (1. 08 ) 1. 53 4* ** (4 .8 2) 0. 42 0* * (2 .2 7) 0. 49 3 (1 .1 2) -0 .0 39 (0. 27 ) -0 .0 78 (0. 47 ) -0 .1 76 (2. 14 ) g d pr -0 .1 44 (0. 33 ) 0. 02 5 (0 .1 7) 0. 07 3 (0 .3 4) -0 .3 63 (1. 35 ) 0. 09 2 (0 .3 3) -0 .0 01 (0. 00 2) -0 .4 99 ** (2. 26 ) -0 .2 12 (0. 84 ) -0 .0 75 (0. 61 ) d c p 0. 31 5* ** (3 .5 7) 0. 15 1* ** (5 .2 1) 0. 09 9 (1 .4 5) -0 .1 25 (1. 45 ) -0 .0 31 (0. 56 ) -0 .0 41 (0. 32 ) -0 .0 89 ** (2. 04 ) -0 .0 32 (0. 64 ) -0 .0 08 (0. 32 ) o c ta -4 .8 51 ** * (5. 99 ) 0. 26 9 (0 .9 6) -1 .1 28 (0. 89 ) -3 .1 99 ** (2. 01 ) -1 .1 78 ** (2. 23 ) 1. 29 4 (1 .0 3) -1 .0 39 ** (2. 49 ) -0 .0 61 (0. 13 ) -0 .0 23 (0. 09 ) n pl *e pu a -0 .0 03 ** (2. 06 ) -0 .0 01 * (1. 78 ) -0 .0 00 5 (0. 46 ) -0 .0 02 (1. 34 ) 0. 00 03 (0 .3 3) -0 .0 01 (0. 46 ) -0 .0 01 (0. 91 ) -0 .0 00 8 (0. 83 ) 0. 00 05 (1 .1 6) (c on tin ue d) 75 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) (9 ) v ar ia bl e a t m b b pa m pb e l p a c c d c c c sa v b o r g d pr *e pu a 0. 00 5 (1 .5 1) -0 .0 01 * (0. 98 ) -0 .0 01 (0. 45 ) 0. 00 5* (1 .9 9) -0 .0 03 (1. 69 ) -0 .0 03 (0. 70 ) -0 .0 00 2 (0. 14 ) 0. 00 05 (0 .2 5) 0. 00 03 (0 .3 1) d c p* ep u a -0 .0 00 1 (0. 62 ) 0. 00 01 (0 .1 2) 0. 00 00 4 (0 .3 3) -0 .0 00 1 (0. 54 ) 0. 00 02 (1 .4 8) 0. 00 02 (0 .8 3) 0. 00 08 (0 .8 4) 0. 00 00 3 (0 .2 7) -0 .0 00 1 (1. 01 ) o c ta *e pu a 0. 01 7* ** (4 .8 9) 0. 00 1 (0 .5 6) 0. 00 3 (0 .8 5) 0. 00 9* * (2 .0 5) 0. 00 3 (1 .2 5) -0 .0 05 (0. 84 ) 0. 00 2 (1 .0 3) -0 .0 00 3 (0. 13 ) -0 .0 00 4 (0. 35 ) a dj us te d r 2 98 .3 9 97 .7 4 84 .7 1 98 .6 2 96 .1 9 84 .5 7 96 .5 4 96 .2 1 87 .8 8 fst at is tic 23 5. 54 16 5. 42 13 .8 1 16 6. 09 99 .8 9 22 .4 5 11 0. 37 10 0. 40 29 .3 8 o bs er va tio n 13 1 13 4 75 75 13 8 13 8 13 8 13 8 13 8 n ot e. r eg re ss io n in t ab le 6 in cl ud es c ou nt ry a nd y ea r fi xe d ef fe ct s. t -s ta tis tic s ar e re po rt ed in p ar en th es is . *, * *, * ** re pr es en t s ig ni fic an ce le ve l a t 10 % , 5 % a nd 1 % . a c c = a du lts w ho o w n a fo rm al a cc ou nt . d c = d eb it ca rd o w ne rs hi p. c c = c re di t c ar d ow ne rs hi p. s a v = a du lts w ho h av e fo rm al sa vi ng s. b o r = a du lts w ho h av e fo rm al b or ro w in gs . a t m = a t m s pe r 10 0, 00 0 ad ul ts . b b pa = b an k br an ch es p er 1 00 ,0 00 a du lts . m pb = a du lts us in g a m ob ile p ho ne to p ay b ill s. e l p = ad ul ts w ho u se e le ct ro ni c pa ym en ts to m ak e pa ym en ts . e pu a = a ve ra ge v al ue o f t he m on th ly e pu in de x. n pl = n on pe rf or m in g lo an s ra tio . o c ta = th e ra tio o f b an k ov er he ad c os t t o to ta l a ss et ra tio . d c p = cr ed it su pp ly to th e pr iv at e se ct or b y ba nk s to g d p ra tio . g d pr = re al g d p gr ow th ra te . 76 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 conclusion the present study analyzed the effect of epu on the level of financial inclusion. there were three main findings. one, epu did not have a significant impact on financial inclusion. none of the nine indicators of financial inclusion were found to have a significant direct relationship with epu. two, the combined effect of high levels of epu and high non-performing loans would lead to bank branch contraction and a reduction in the use of electronic payments. third, the use of formal accounts and credit cards would increase in times of high credit supply and high levels of epu. the implication of these findings is that economic policy uncertainty affects financial inclusion through its effect on financial institutions. as financial institutions intensified their effort to reduce cost in times of high levels of epu, such cost reduction could affect the supply of basic financial services to customers and unbanked adults, thereby reducing financial inclusion. policy makers should design policies that promote high levels of financial inclusion in times of rising levels of epu. policy makers, particularly bank regulators, should formulate policies that prevent banks from closing rural bank branches in times of high epu. however, the effect of such a policy in individual countries may differ due to differences in the national financial inclusion strategy, the current level of financial inclusion, the number of bank branch networks, and level of financial development and regulatory frameworks. the main limitation of the study was the sample period. the sample period is small, and this was due to the few number of reported data in the existing database. future studies should investigate the impact of each epu component on the level of financial inclusion. future studies should also examine whether strong bank supervision in times of high levels of epu will have a positive or negative effect on financial inclusion. finally, the analysis in the present study can be extended by investigating how the level of the epu will affect the propensity of women to use financial services. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for profit sectors. 77 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 references anson, j., berthaud, a., klapper, l., & singer, d. (2013). financial inclusion and the role of the post office. the world bank. ashraf, b. n., & shen, y. (2019). economic policy uncertainty and banks’ loan pricing. journal of financial stability, vol 44, 100695. https://doi.org/10.1016/j.jfs.2019.100695 baker, s. r., & bloom, n. (2013). does uncertainty reduce growth? using disasters as natural experiments (no. w19475). national bureau of economic research. baker, s. r., bloom, n., & davis, s. j. (2016). measuring economic policy uncertainty. the quarterly journal of economics, 131(4), 1593-1636. berger, a. n., guedhami, o., kim, h. h., & li, x. (2020). economic policy uncertainty and bank liquidity hoarding. journal of financial intermediation, 100893. https://doi.org/10.1016/j. jfi.2020.100893 bordo, m. d., duca, j. v., & koch, c. (2016). economic policy uncertainty and the credit channel: aggregate and bank level us evidence over several decades. journal of financial stability, 26, 90-106. brown, m., guin, b., & kirschenmann, k. (2016). microfinance banks and financial inclusion. review of finance, 20(3), 907946. caglayan, m., & xu, b. (2019). economic policy uncertainty effects on credit and stability of financial institutions. bulletin of economic research, 71(3), 342-347. camanho, a. s., & dyson, r. g. (2005). cost efficiency, production and value-added models in the analysis of bank branch performance. journal of the operational research society, 56(5), 483-494. célerier, c., & matray, a. (2019). bank-branch supply, financial inclusion, and wealth accumulation. the review of financial studies, 32(12), 4767-4809. chakrabarty, k. c. (2011). financial inclusion and banks: issues and perspectives. rbi bulletin, november. chen, z., & jin, m. (2017). financial inclusion in china: use of credit. journal of family and economic issues, 38(4), 528-540. collard, s. (2007). toward financial inclusion in the uk: progress and challenges. public money and management, 27(1), 13-20. 78 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 demetriades, p., & hook law, s. (2006). finance, institutions and economic development. international journal of finance & economics, 11(3), 245-260. demirgüç-kunt, a., & klapper, l. (2013). measuring financial inclusion: explaining variation in use of financial services across and within countries. brookings papers on economic activity, 2013(1), 279-340. evans, o., & alenoghena, o. r. (2017). financial inclusion and gdp per capita in africa: a bayesian var model. journal of economics & sustainable development, 8(18), 44-57. garcía-kuhnert, y., marchica, m. t., & mura, r. (2015). shareholder diversification and bank risk-taking. journal of financial intermediation, 24(4), 602-635. ghosh, j. (2013). microfinance and the challenge of financial inclusion for development. cambridge journal of economics, 37(6), 1203-1219. ghosh, a. (2015). banking-industry specific and regional economic determinants of non-performing loans: evidence from us states. journal of financial stability, 20, 93-104. gulen, h., & ion, m. (2016). policy uncertainty and corporate investment. review of financial studies, 29(3), 523-564. hu, s., & gong, d. (2019). economic policy uncertainty, prudential regulation and bank lending. finance research letters, 29, 373-378. imaeva, g., lobanova, i., & tomilova, o. (2014). financial inclusion in russia: the demand-side perspective. consultative group to assist the poor, world bank, moscow. karadima, m., & louri, h. (2020). economic policy uncertainty and non-performing loans: the moderating role of bank concentration. finance research letters, vol. 38, p. 1-5, https:// doi.org/10.1016/j.frl.2020.101458. king, r. g., & levine, r. (1993). finance and growth: schumpeter might be right. the quarterly journal of economics, 108(3), 717-737. le, t. h., le, h. c., & taghizadeh-hesary, f. (2020). does financial inclusion impact co2 emissions? evidence from asia. finance research letters, 34, 1-7. https://doi.org/10.1016/j. frl.2020.101451 lee, c. c., lee, c. c., zeng, j. h., & hsu, y. l. (2017). peer bank behavior, economic policy uncertainty, and leverage decision of financial institutions. journal of financial stability, 30, 79-91. 79 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 lópez, t., & winkler, a. (2019). does financial inclusion mitigate credit boom-bust cycles? journal of financial stability, 43, 116-129. luo, y., & zhang, c. (2020). economic policy uncertainty and stock price crash risk. research in international business and finance, 51, 1-14. https://doi.org/10.1016/j.ribaf.2019.101112 machdar, n. m. (2020). financial inclusion, financial stability and sustainability in the banking sector: the case of indonesia. international journal of economics and businees administration, 8(1), 193-202. markose, s., arun, t., & ozili, p. (2020). financial inclusion, at what cost? quantification of economic viability of a supply side roll out. the european journal of finance, 1-27. https://doi.org/10 .1080/1351847x.2020.1821740 mindra, r., moya, m., zuze, l. t., & kodongo, o. (2017). financial self-efficacy: a determinant of financial inclusion. international journal of bank marketing, 35(3), 338-353. https://doi. org/10.1108/ijbm-05-2016-0065 morgan, p. j., & pontines, v. (2018). financial stability and financial inclusion: the case of sme lending. the singapore economic review, 63(01), 111-124. neuberger, d. (2015). financial inclusion, regulation, and education in germany. asian development bank institute research paper series working paper 530. http://dx.doi.org/10.2139/ ssrn.2627488 ng, j., saffar, w., & zhang, j. j. (2020). policy uncertainty and loan loss provisions in the banking industry. review of accounting studies, 25, 726–777 https://doi.org/10.1007/s11142-01909530-y nguyen, c. p., le, t. h., & su, t. d. (2020). economic policy uncertainty and credit growth: evidence from a global sample. research in international business and finance, 51, 101118. omar, m. a., & inaba, k. (2020). does financial inclusion reduce poverty and income inequality in developing countries? a panel data analysis. journal of economic structures, 9, 1-25. ozili, p. k. (2018). impact of digital finance on financial inclusion and stability. borsa istanbul review, 18(4), 329-340. ozili, p. k. (2019). non-performing loans and financial development: new evidence. the journal of risk finance. 20(1), 59-81. https://doi.org/10.1108/jrf-07-2017-0112. 80 the international journal of banking and finance, vol. 17, number 1 (january) 2022, pp: 53–80 ozili, p. k. (2020a). financial inclusion research around the world: a review. in forum for social economics (pp. 1-23). routledge. https://doi.org/10.1080/07360932.2020.1715238 ozili, p. k. (2020b). financial inclusion and business cycles. journal of financial economic policy, 13(2), 180-199. ozili, p. k. (2021a). economic policy uncertainty: are there regional and country correlations? international review of applied economics, 1-15. https://doi.org/10.1080/02692171.2020.185 3075 ozili, p. k. (2021b). has financial inclusion made the financial sector riskier? journal of financial regulation and compliance. 29(3), 237-255. https://doi.org/10.1108/jfrc-08-2020-0074. ozili (2021c). financial inclusion in nigeria: an overview. international journal of banking and finance. forthcoming. oz-yalaman, g. (2019). financial inclusion and tax revenue. central bank review, 19(3), 107-113. perera, s., skully, m., & wickramanayake, j. (2007). cost efficiency in south asian banking: the impact of bank size, state ownership and stock exchange listings. international review of finance, 7(1-2), 35-60. rioja, f., & valev, n. (2004). finance and the sources of growth at various stages of economic development. economic inquiry, 42(1), 127-140. vo, a. t., van, l. t. h., vo, d. h., & mcaleer, m. (2019). financial inclusion and macroeconomic stability in emerging and frontier markets. annals of financial economics, 14(02), 1950008. https://doi.org/10.1142/s2010495219500088 yung, k., & root, a. (2019). policy uncertainty and earnings management: international evidence. journal of business research, 100, 255-267. zins, a., & weill, l. (2016). the determinants of financial inclusion in africa. review of development finance, 6(1), 46-57. 91 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 how to cite this article: aziz, m. i. a., azhari, a., & mobin, m. a. (2022). detecting asset price bubbles during the covid-19 crisis and its implications: evidence from the stock and oil market. international journal of banking and finance, 17(2), 91-114. https://doi. org/10.32890/ ijbf2022.17.2.4 detecting asset price bubbles during the covid-19 crisis and its implications: evidence from the stock and oil market 1mukhriz izraf azman aziz, 2adilah azhari & 3m ashraful mobin 1&2school of economics, finance and banking, universiti utara malaysia 3managing director, ifintell ltd, malaysia 1corresponding author: mukhriz@uum.edu.my received: 5/7/2021 revised: 1/9/2021 accepted: 4/9/2021 published: 27/6/2022 abstract this study investigates whether the covid-19 pandemic has caused asset price bubbles in the stock and oil markets in the united states and malaysia. more specifically, the study seeks to detect the onset and end of possible speculative bubbles and their causes in these markets. it also examines the existence of a contagion effect between the stock and oil markets during the covid-19 pandemic. to achieve these objectives, the study used the generalized sadf (gsadf) developed by phillips et al. (2015) in order to check for existence of bubbles within the time frame from january 1, 2020, to april 24,2020. this technique allows one to look for the occurrence of multiple bubbles during the sample period with great precision. the findings http://e-journal.uum.edu.my/index.php/ijbf international journal of banking and finance 92 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 showed that five out of the six equities, including the oil price indices had multiple bubbles. evidence was also obtained which linked the explosive activity episodes between the crude oil market and the us stock markets from the start and end point of each bubble event. these findings add not only to the literature on the existence of bubbles in the financial and energy markets during the initial outbreak of covid-19, but also to the significance of the negative impact of pandemics on bubble contagion effects under extreme market conditions. keywords: covid-19, bubble, stock market, oil price, gsadf, financial crisis. jel classification: c33, e44, g15 g21 g32. introduction the world is currently paralyzed by the covid-19 outbreak that began in wuhan, china in late december 2019. the stock markets plummeted across the world as investors fled the markets. as the novel covid-19 virus scare continues to create havoc in the capital markets, it is important to investigate the market crashes due to the pandemic. baker et al. (2020) provides several reasons why such a pandemic has a powerful impact on financial markets. they argue that the gravity of this pandemic, its high contagiousness and the large number of infections and deaths resulting from it, have all contributed to making the stock market shock critical. to underscore the gravity of the pandemic, figure 1 shows the covid-19 cases for china and selected euro nations. countries which have been recording the highest number of daily cases are china, germany, italy, spain, and the uk. except for china, the peak for these countries was detected between 9 march to 14 april 2020. in malaysia, the jump in the number of daily new cases was detected from 13 march 2020 until 14 april 2020 (see figure 2). as for the us, it started to show a jump in the daily new cases on 17 march and continued until late april 2020. the us reported the highest market turbulence since the global financial crisis in december 2008 (baker et al., 2020). this was following the us stock market crash by 20 percent on 11 and 12 march 2020 (giglio et al., 2020). 93 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 figure 1 figure 2 covid-19 cases for usa and covid-19 cases for china malaysia and euro countries while the global economy and financial markets have been in disarray, the recent oil price crisis caused by the russia-saudi arabia price war has worsened the situation. against the backdrop of lower oil demand due to the global economic lockdown1, the russia-saudi price war has resulted in a 65 percent quarterly decline in oil prices2. uncertainty in the real economy which was exacerbated by volatile oil prices has diminished economic growth and adversely impacted the stock markets (degiannakis et al., 2018; kilian & park, 2009). since oil price fluctuations have long been tied to stock market movements (moya-martínez et al., 2014; waheed et al., 2018; arouri et al., 2011; kumar et al., 2012; gomes & chaibi, 2014), falling prices induced by covid-19 (ozili & arun, 2020) has increased market risk aversion to levels not seen since the gfc (yousef et al., (2021). the implied volatility of equities and oil rose to crisis levels as stock markets collapsed3. as a result, speculative bubbles have been common in significant parts of the financial markets, particularly during the early months of the pandemic outbreak. motivated by these tumultuous events, the present study was aimed at examining the explosive behavior episodes in selected us and malaysia stock markets during the first four months of the covid-19 outbreak, which was from 2 january 2020 to 24 april 2020. this period was selected because it covered the non-covid-19 phase (2 january, 2020 – 16 january, 2020) and the early pandemic phase (17 january– 24 april 2020). since the period of study also coincided with the saudi-russia oil price war in early march 2020, the study also investigated the bubble episodes in crude oil market and determined whether there existed a contagion effect between the bubbles in the equity markets and the oil markets. 2 to underscore the gravity of the pandemic, figure 1 shows the covid-19 cases for china and selected euro nations. countries which have been recording the highest number of daily cases are china, germany, italy, spain, and the uk. except for china, the peak for these countries was detected between 9 march to 14 april 2020. in malaysia, the jump in the number of daily new cases was detected from 13 march 2020 until 14 april 2020 (see figure 2). as for the us, it started to show a jump in the daily new cases on 17 march and continued until late april 2020. the us reported the highest market turbulence since the global financial crisis in december 2008 (baker et al., 2020). this was following the us stock market crash by 20 percent on 11 and 12 march 2020 (giglio et al., 2020). figure 1 figure 2 covid-19 cases for usa and malaysia covid-19 cases for china and euro countries while the global economy and financial markets have been in disarray, the recent oil price crisis caused by the russia-saudi arabia price war has worsened the situation. against the backdrop of lower oil demand due to the global economic lockdown1, the russia-saudi price war has resulted in a 65 percent quarterly decline in oil prices2. uncertainty in the real economy which was exacerbated by volatile oil prices has diminished economic growth and adversely impacted the stock markets (degiannakis et al., 2018; kilian & park, 2009). since oil price fluctuations have long been tied to stock market movements (moya-martínez et al., 2014; waheed et al., 2018; arouri et al., 2011; kumar et al., 2012; gomes & chaibi, 2014), falling prices induced by covid-19 (ozili & arun, 2020) has increased market risk aversion to levels not seen since the gfc (yousef et al., (2021). the implied volatility of equities and oil rose to crisis levels as stock markets collapsed3. as a result, speculative bubbles have been common in significant parts of the financial markets, particularly during the early months of the pandemic outbreak. motivated by these tumultuous events, the present study was aimed at examining the explosive behavior episodes in selected us and malaysia stock markets during the first four months of the covid-19 outbreak, which was from 2 january 2020 to 24 april 2020. this period was selected because it covered the non-covid-19 phase (2 january, 2020 – 16 january, 2020) and the early pandemic phase (17 january– 24 april 2020). since the period of study also coincided with the saudi-russia oil price war in early march 2020, the study also investigated the bubble episodes in crude oil market and determined whether there existed a contagion effect between the bubbles in the equity markets and the oil markets. 2 to underscore the gravity of the pandemic, figure 1 shows the covid-19 cases for china and selected euro nations. countries which have been recording the highest number of daily cases are china, germany, italy, spain, and the uk. except for china, the peak for these countries was detected between 9 march to 14 april 2020. in malaysia, the jump in the number of daily new cases was detected from 13 march 2020 until 14 april 2020 (see figure 2). as for the us, it started to show a jump in the daily new cases on 17 march and continued until late april 2020. the us reported the highest market turbulence since the global financial crisis in december 2008 (baker et al., 2020). this was following the us stock market crash by 20 percent on 11 and 12 march 2020 (giglio et al., 2020). figure 1 figure 2 covid-19 cases for usa and malaysia covid-19 cases for china and euro countries while the global economy and financial markets have been in disarray, the recent oil price crisis caused by the russia-saudi arabia price war has worsened the situation. against the backdrop of lower oil demand due to the global economic lockdown1, the russia-saudi price war has resulted in a 65 percent quarterly decline in oil prices2. uncertainty in the real economy which was exacerbated by volatile oil prices has diminished economic growth and adversely impacted the stock markets (degiannakis et al., 2018; kilian & park, 2009). since oil price fluctuations have long been tied to stock market movements (moya-martínez et al., 2014; waheed et al., 2018; arouri et al., 2011; kumar et al., 2012; gomes & chaibi, 2014), falling prices induced by covid-19 (ozili & arun, 2020) has increased market risk aversion to levels not seen since the gfc (yousef et al., (2021). the implied volatility of equities and oil rose to crisis levels as stock markets collapsed3. as a result, speculative bubbles have been common in significant parts of the financial markets, particularly during the early months of the pandemic outbreak. motivated by these tumultuous events, the present study was aimed at examining the explosive behavior episodes in selected us and malaysia stock markets during the first four months of the covid-19 outbreak, which was from 2 january 2020 to 24 april 2020. this period was selected because it covered the non-covid-19 phase (2 january, 2020 – 16 january, 2020) and the early pandemic phase (17 january– 24 april 2020). since the period of study also coincided with the saudi-russia oil price war in early march 2020, the study also investigated the bubble episodes in crude oil market and determined whether there existed a contagion effect between the bubbles in the equity markets and the oil markets. 94 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 as the epidemic wreaked havoc on world markets, determining the exact date when the initial bubbles occurred, how long each bubble lasted and the possible causes for each bubble became important for policy makers and investors alike. this is especially true given that the pandemic is still having an impact on the equities markets, any significant increase in covid-19 cases is likely to spark future episodes of explosive behavior in the stock markets. therefore, to test for bubbles in the stock markets and energy sector, the supremum augmented dickey–fuller (sadf) test introduced by phillips et al. (2011) and the generalized sadf (gsadf) by phillips et al. (2015) were also used in the present study. while the sadf test detects a single bubble episode, the gsadf test overcomes this constraint by assessing the explosive behavior with multiple bubbles, resulting in more robust estimations for the present study. the usa and malaysia stock markets were selected for the following reasons. first, the usa has the largest stock markets in the world and one of the earliest markets to jolt following the coronavirus outbreak in february 20204. second, the usa was also the largest oil producer in 2020, hence was directly affected by the oil price crash. third, malaysia was chosen due to its significant economic relations with the usa, with their bilateral trade volume surpassing rm100 billion in 2020, or 11.1 percent of the total malaysian exports (matrade, 2020). given that malaysia was severely impacted by the gfc (alp et al., 2012), which was precipitated by the collapse of the us subprime market, there was the anticipation that there would be a contagion effect from the asset price bubbles in the usa into the asset markets of malaysia during the covid-19 pandemic. the theorized contagion effect between these markets was seen as highly possible, given how strong the correlation between these markets was at the time of the study (evidence of this correlation is provided in the fourth section of the paper). besides the klse index, the present study also considered the shariah index for the asset price bubbles analysis of malaysia. given the differing rules and regulations that govern these two markets5, this study wanted to examine if there was a major difference (or differences) in how these two indices responded to the turbulent market conditions during the covid-19 epidemic. this was because the islamic stocks were hypothesized to be less vulnerable to shocks due to the lower 95 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 leverage, smaller business size, and under-diversified market (rizvi et al., 2015). it has also been found that the islamic finance portfolio performed better than their conventional counterpart in the early wake of the 2008 gfc (zhang et al., 2020). therefore, findings from the analysis in the present study have important implications for investors who are looking for an alternative form of stock market that will be better in safeguarding investors’ rights from the numerous financial crises that have plagued the world over the past several decades. this study reported in this paper contributes to the literature in several ways. first, it has shed light on the explosive behaviors and contagion effects in the us stock and crude oil markets during the initial phase of the covid-19 outbreak from 2 january 2020 to 24 april 2020. second, it has investigated bubbles episodes in the malaysian stock market and its linkage with the implosion of covid-19 cases from the us stock markets and crude oil markets. third, it has compared the explosive behavior episodes between the conventional stock markets and the islamic stock markets in the markets in malaysia during the early wave of the covid-19 pandemic. with this analysis, it was hoped that researchers can draw new insights about the reactions of investors and traders on these two markets during the initial pandemic outbreak. the analyses have resulted in several significant outcomes. first, there was evidence of multiple bubbles in five out of the six of equity series. second, there was evidence linking explosive behavior episodes between the crude oil market and the us stock markets from the datestamp of the starting and ending points of each bubble incident. in contrast, the stock market in malaysia exhibited explosive behavior more closely to the date-stamp of the covid-19 implosion of local cases around the 13th of march, 2020. there was also evidence linking the bubbles episodes in the us and the crude oil market with the stock market in malaysia. finally, the empirical results showed that multiple bubbles episodes were detected in the shariah stock market, as opposed to a single bubble in the klse index in malaysia. two bubbles episodes that were discovered only in the shariah index corresponded to the initial spread of the covid-19 virus in china, when the global market was still relatively unperturbed by the pandemic. the results of the present study seemed to suggest that the islamic stock market could detect abnormal market behavior earlier, as compared to the conventional stock market. however, these results 96 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 were limited to the malaysian market experience, and more evidence is needed to generalize the findings to other islamic stock markets. the rest of the paper is organized as follows: section 2 presents the literature review; section 3 describes the method and data; section 4 introduces the results and section 5 concludes the paper. literature review the covid-19 pandemic has been regarded as the most devastating worldwide health catastrophe since the spanish flu in 1918. nevertheless, there has been relatively little past research on how epidemics influence financial markets. as a result, the present study offers an important contribution to the literature in the field. according to chen et al. (2009) and loh (2006), the severe acute respiratory syndrome (sars) outbreak had a detrimental impact on industries such as those in aviation, tourism, wholesale, and retail. previous studies have also indicated that the severity of the pandemic might predict the likelihood of an equity market crash. giglio et al. (2020) and wen et al. (2019a, b), for example, demonstrated that short-run investor expectations might correlate with the risk of a stock market crash. however, previous research by giglio et al. (2020, 2021) has also revealed that the likelihood of an equity market crash occurring before a crisis was lower since investors were more bullish about stock market returns. zhang et al. (2020) found that covid-19’s rapid spread had a major impact on financial markets worldwide, resulting in huge increases in the global financial market risk and huge losses to investors for a short period of time. the study by zeren and hizarci (2020) which looked at the impact of covid-19 shock on the stock markets of six countries found a co-integrating relationship between the daily total case and stock markets returns. yilmazkuday (2020) analyzed the covid-19 impact on the s&p 500 index and found a significant relationship between the two. awadhi et al. (2020) also examined the effect of covid-19 on stock market returns and revealed that there were sectoral differences in the market returns. according to mazur et al. (2020), the march 2020 financial market meltdown was caused by government reactions. their analysis also corroborated the findings of mishkin and white (2002), who discovered that a stock market 97 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 crash might result in a loss of 20 percent –25 percent in the united states (u.s.) equities index, compared to past crises (e.g., world war i, world war ii, and so on) owing to a series of panic selling. several studies have looked at asset price bubbles when markets were in distress. a recent study by chang et al. (2021) showed the existence of bubbles in the us stock market during the early months of the covid-19 outbreak by using the gsadf test. focusing on the gold and crude oil markets, gharib et al. (2021) discovered that there existed mild explosive price behaviors in the wti and gold markets from january 4, 2010, to may 4, 2020. the most notable finding was that a bilateral contagion impact of bubbles was found during the recent covid-19 outbreak in the oil and gold markets. zhao et al. (2020) examined the bilateral contagion effect of bubbles between oil price and the chinese stock markets. they detected six bubbles in the oil and stock markets from september 1, 2004, to july 9, 2018. another study by sharma and escobari (2018) have highlighted the existence of bubbles in the energy sector. they found that a strong spike in oil prices had caused the bubble to explode in 2015. li et al. (2020), in their study of natural gas price bubbles in three major economies from 1996 to 2017, found that the 2008 global financial crisis (gfc) contributed to the rapid variations in natural gas prices in all three countries. research method and data theory and method a large set of studies in the existing literature have investigated asset price bubbles using different asset-pricing models (gürkaynak, 2008, de long et al., 1990, tirole, 1985). price bubbles occur when commodity prices deviate from their true values. many researchers have considered the model used by gürkaynak (2008) as the most reliable one from among the others highlighted in the aforementioned studies. three main assumptions formed the basis of the gürkaynak model. first, market information is assumed to be non-asymmetric, such that uninformed traders are unable to influence asset price by manipulating information about prices. second, consumers are assumed to be risk neutral with no premium on risk. this means that fluctuations in prices are not caused by risk premiums that varies 98 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 with time and price variations. third, the discount rate is assumed to remain constant. in short, the general model which allows for the presence of bubbles is given as follows: [1] where is the asset price in time t and signifies the expectation. the coefficients and represent the return and hidden component for time t+1, respectively. for the present study equation [1] has been transformed into the following equation [2]: [2] where is the underlying stock market (crude oil) index (price) and is the stock market (crude oil) return in period equation [2] represents the components of the underlying price without bubbles. the asset-pricing model with the presence of bubbles is depicted by the following equation [3]: [3] equation [3] is represented by two elements. the first element represents the present value of the projected capital return, and the second element captures the asset price bubbles. in equation [3], price bubbles are treated as a factor in the pricing of an asset, rather than a collective error by traders. thus, when assuming an interesting premise can be articulated. evans (1991) opines that the traditional unit root tests to ascertain price bubbles are ineffective when cyclical unsustainable behavior occurs in the period. therefore, phillips and yu (2011) constructed the supremum augmented dickey–fuller (sadf hereafter) to detect explosive behaviors. the sadf test carries out the adf test repetitively on a forward recursive sample sequence that is also connected to the right-sided adf and sup tests. when the bubbles burst, the sadf test is shown to be more effective in detecting structural breaks than other tests (homm & breitung, 2012). the null hypothesis of is rejected when the largest right tailed adf is bigger than the critical value, i.e. there is evidence of an asset price bubble. details of the estimation process are given as follows in equation [4]: [4] 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 99 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 where n is the sample size, is a constant, and assuming that the regression sample begins from and ends at while is the window size of the regression, such that , then equation [4] can be written as equation [5]: [5] where k is the lag order and the window size lies between and 1. as suggested by phillips et al. (2015), the window size for sample size n is determined as follows, using equation [6]: [6] for example, a sample size of requires a window size of 16. following phillips et al. (2015), the lag order is set to zero using the bic information criterion. the critical values are calculated using the monte carlo simulation with 1000 replications. in addition, the parameters and of the data generating process (whereby c is the constant drift factor and η is a coefficient that controls the magnitude of the drift) is set according to phillips et al. (2015) where the asymptotic distribution of the sadf statistic may be summarized as follows in equation [7]: [7] where w is the standard wiener process. despite the superiority of the sadf over traditional unit root tests in distinguishing bubbles, it lacks the ability to detect several bubbles across an extended horizon. as shown by phillips et al. (2015), the sadf is unable to pick up multiple occurrences of bubbles when the period is prolonged. correspondingly, phillips et al. (2015) has introduced the generalized sup adf (gsadf) by adjusting the beginning and ending points, thus allowing for a more flexible window size in the estimation process. the gsadf is useful when analyzing the volatile behavior of an asset price with several parts in a long period which includes several subsamples for this procedure. the gsadf is represented in the equation [8]: [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 1 detecting asset price bubbles during covid-19 crisis and its implications: evidence from stock and oil market 𝑃𝑃𝑡𝑡 = (1 + 𝑟𝑟𝑓𝑓)−1𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+1) [1] 𝑃𝑃𝑡𝑡 t 𝛦𝛦𝑡𝑡 𝜕𝜕𝑡𝑡+1 and ℧𝑡𝑡+1 t+1, 𝑃𝑃𝑡𝑡 𝑓𝑓 = ෎ ൬ 11+𝑟𝑟𝑓𝑓൰ 𝑖𝑖∞ 𝑖𝑖=0 𝛦𝛦𝑡𝑡(𝜕𝜕𝑡𝑡+1 + ℧𝑡𝑡+𝑖𝑖), 𝑖𝑖 = 0,1,2…𝑛𝑛 [2] 𝑃𝑃𝑡𝑡 𝑓𝑓 𝜕𝜕𝑡𝑡+1 𝜕𝜕𝑡𝑡+1. 𝑃𝑃𝑡𝑡 = 𝑃𝑃𝑡𝑡 𝑓𝑓 + 𝐵𝐵𝑡𝑡 [3] 𝐵𝐵𝑡𝑡 ≠ 0, 𝐻𝐻0:𝛣𝛣 = 1 𝑦𝑦𝑡𝑡 = 𝑐𝑐𝑁𝑁−𝜂𝜂 + 𝜃𝜃𝑦𝑦𝑡𝑡−1 + 𝜀𝜀𝑡𝑡 [4] n 𝑐𝑐 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 2 𝜂𝜂>1/2, 𝜀𝜀𝑡𝑡~ nid(0,𝜃𝜃2), 𝜃𝜃=1. 𝑟𝑟2 while 𝑟𝑟𝑤𝑤 𝑟𝑟2 = 𝑟𝑟1 + 𝑟𝑟𝑤𝑤. ∆𝑦𝑦𝑡𝑡 = 𝛽𝛽𝑟𝑟1,𝑟𝑟2 + 𝛽𝛽𝑟𝑟1,𝑟𝑟2𝑆𝑆𝑦𝑦𝑡𝑡−1 σ 𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ∆𝑦𝑦𝑡𝑡−𝑖𝑖 + 𝜇𝜇𝑡𝑡𝑘𝑘𝑖𝑖=1 [5] k 𝜇𝜇𝑡𝑡~nid(0,𝜑𝜑𝑟𝑟1,𝑟𝑟2𝑖𝑖 ) 𝑟𝑟𝑤𝑤 𝑟𝑟0 𝑟𝑟0 𝑟𝑟0 = 0.01 + 1.8.ξ𝑁𝑁 [6] 𝑁𝑁 = 79 𝑐𝑐 and 𝜂𝜂 𝑐𝑐 = 𝜂𝜂 = 1 𝐴𝐴𝐴𝐴𝐴𝐴 → 𝐿𝐿 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑟𝑟∈ሾ𝑟𝑟0,1ሿ ቐ 𝑟𝑟𝑤𝑤ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑊𝑊−12𝑟𝑟𝑤𝑤ቃ−−∫0 𝑟𝑟𝑤𝑤𝑊𝑊𝑊𝑊𝑟𝑟.𝑊𝑊(𝑟𝑟𝑤𝑤) 𝑟𝑟𝑤𝑤 1 2τ ൜𝑟𝑟𝑤𝑤∫0 𝑟𝑟𝑤𝑤𝑊𝑊2𝑊𝑊𝑟𝑟−ቂ∫0 𝑟𝑟𝑤𝑤𝑊𝑊(𝑟𝑟)𝑊𝑊𝑟𝑟ቃ 2 ൠ 1 2τ ቑ [7] 𝐺𝐺𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝑟𝑟𝑤𝑤∈ሾ𝑟𝑟0,1ሿ ൛𝑠𝑠𝑠𝑠𝑠𝑠𝑟𝑟1∈ሾ0,1−𝑟𝑟𝑤𝑤ሿ𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟1 𝑟𝑟𝑤𝑤ൟ [8] 100 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 data the present study has examined the asset price bubbles due to the covid-19 outbreak during the period from january 2, 2020, to april 24, 2020. four daily equities, namely two american stock markets indices (s&p 500 (^gspc) and dow jones (^dji)), two malaysian stock market indices (ftse bursa malaysia (^klse?p=^klse) and emas shariah (fbms.fgi)) were sourced from yahoo finance (https://finance.yahoo.com). for oil prices, two indices (wti spot crude and nymex futures) were sourced from the us energy information administration (https://www.eia.gov/petroleum/data. php). table 1 summary statistics on stock market indices and oil markets dow jones s&p 500 klse shariah nymex spot crude mean 25801.71 2985.72 1461.79 11043.17 40.56 40.36 maximum 29551.42 3386.15 1611.38 12104.30 63.27 63.27 minimum 18591.93 2237.40 1219.72 9120.49 -37.63 -36.98 std. dev. 3344.77 340.71 109.40 865.65 17.66 17.99 skewness -0.42 -0.46 -0.39 -0.51 -1.25 -1.20 kurtosis 1.73 1.83 1.86 2.00 6.01 5.50 jarque-bera 7.61*** 7.33*** 6.41*** 6.87*** 48.41** 37.96** observations 79 79 81 81 76 76 note. dow jones denotes dow jones industrial average, klse denotes ftse bursa malaysia klci, shariah denotes ftse bursa malaysia emas sharia, nymex denotes nymex futures, spot crude denotes wti spot price. note. **, *** indicates statistical significance at 5% and 1% level, respectively. table 1 shows the summary statistics for the four stock market indices and two oil price measures. the range of stock indices in percentage terms was 37 percent for the dow jones and 34 percent for the s&p 500. for the klse and the shariah, the variation was 24 percent each from january,1 to april,24 2020. fluctuations in crude oil prices were more severe, accounting for 158 percent during the sample period, largely attributed to the unprecedented drop to negative value for first time in history on 20th april 2020 at usd37. these large and drastic fluctuations of prices were the result of the severity of the impact of covid-19 on these markets, and which increased the likelihood of causing bubbles. furthermore, all prices exhibited negative skewness, 101 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 with similar skewness value for the four stock market indices and two oil price measures. according to the jacque-bera test of normality, none of the prices were normally distributed. the present study then tested the pairwise correlation between prices of different equity pairs, as is shown in table 2. the observed correlations between the markets in the usa and malaysia were high and significant at the 1 percent level of significance. similar positively significant correlations between the crude oil prices and the stock markets in the usa and malaysia were recorded. the significant correlations between the six equities seemed to suggest the possibility of contagion effects from speculative bubbles in the stock and oil markets during the epidemic. table 2 correlation statistics of stock market indices and oil price correlation dow jones nymex s&p 500 klse shariah spot crude dow jones na nymex 0.78719* na s&p 500 0.99049* 0.76149* na klse 0.87351* 0.90628* 0.84603* na shariah 0.88625* 0.89566* 0.86239* 0.98933* na spot crude 0.79376* 0.99045* 0.76916* 0.91176* 0.90353* na note. dow jones denotes dow jones industrial average, klse denotes ftse bursa malaysia klci, shariah denotes ftse bursa malaysia emas sharia, nymex denotes nymex futures, spot crude denotes wti spot price. note. * denotes significant at 1% level of significance. results table 3 presents the sadf and the gsadf test results for the six asset prices. each test was achieved by performing 1000 replications of the monte carlo simulations. both the statistical values of the gsadf and the sadf for dow jones and the s&p 500 indices were > 99 percent of the threshold level. for example, in the case of the s&p 500, the statistical value of the gsadf test and the sadf test were 2.19 and 1.93, respectively and exceeded the 99 percent threshold level of 1.57 for the gsadf and 1.08 for the sadf. for the klse and shariah markets, the gsadf tests values exceeded 95 percent, while the sadf statistical values were > 99 percent of the threshold. regarding oil prices, the statistical values of the gsadf were 1.92 for the nymex and 1.12 for the spot crude, respectively. the former exceeded the 99 percent threshold, whilst the latter exceeded the 95 102 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 percent threshold. the statistically significant results indicate the presence of bubbles between january 2020 and april 2020, hence supporting the notion of covid-19 induced uncertainty shocks in these markets. table 3 results of the gsadf and sadf tests for stock market indices and oil market market test method statistical value critical level 99% 95% 90% dow jones gsadf 1.71*** gsadf threshold 1.57 1.13 0.93 sadf 1.71*** sadf threshold 1.08 0.45 0.23 s&p 500 gsadf 2.19*** gsadf threshold 1.57 1.13 0.93 sadf 1.93*** sadf threshold 1.08 0.44 0.22 klse gsadf 1.43** gsadf threshold 1.54 1.09 0.86 sadf 1.38*** sadf threshold 0.9 0.39 0.15 shariah gsadf 1.14** gsadf threshold 1.55 1.12 0.89 sadf 1.09*** sadf threshold 0.9 0.4 0.18 nymex gsadf 1.92*** gsadf threshold 1.55 1.1 0.91 sadf 0.56** sadf threshold 0.94 0.45 0.2 spot crude gsadf 1.12** gsadf threshold 1.55 1.1 0.91 sadf -0.04 sadf threshold 0.94 0.45 0.2 note. dow jones denotes dow jones industrial average, klse denotes ftse bursa malaysia klci, shariah denotes ftse bursa malaysia emas sharia, nymex denotes nymex futures, spot crude denotes wti spot price. note. **, *** indicates statistical significance at 5% and 1% level, respectively. since the gsadf test is prone to succeeding bubbles and results from the gsadf tests have shown that each price index has surpassed the threshold values at 95 percent level, the analysis in the present study had proceeded by comparing the gsadf statistical values with the gsadf sequence threshold’s critical value for the four stock markets and two oil price indices at the 95 percent sequence. the graphical assessment in figures 3-8 shows the date-stamp for the beginning and end of each bubble. for the purpose of comparison, the original data series were plotted in the same figures and corresponds to the right vertical axis. figures 3-4 plot the corresponding gsadf statistics (the brown line) against the corresponding 95 percent threshold (the green line) for the dow jones and the s&p500 during the covid-19 pandemic. the date 103 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 of beginning is identified as the first point when the gsadf statistics surpassed the critical value. the end date is equivalent to the point after the date of origin when the gsadf statistics dropped below the threshold value. these periods are represented by the shaded areas as identified by equation (8). figure 3 and figure 4 show evidence of two statistically significant bubbles, with the first bubble detected on the 21st of february and lasted around seven days. during this short span of time, both indices suffered a 12 percent drop in the index. the second successive bubble was detected on the 12th of march and lasted for two days, when both markets registered a 10 percent drop in the index. these results are in line with the findings in chang et al. (2021). these two consecutive short bubbles in these two us stock markets coincided with the declining global oil demand, as reflected by the persistent drop in crude oil prices as much as 22 percent from 20th february to 13th march 2020. this was largely due to the weakened global oil demand, as china shut its economy to curb the spread of the virus and many countries began to limit or suspend air travels to and from china. the explosive behavior episodes of the dow jones and the s&p500 were a manifestation of the adverse economic impact of the coronavirus outbreak and this was made worse after the russia– saudi arabia oil price war which started on the 8th of march 2020. figure 3 bubbles date stamp for dow jones index note. gsadf denotes gsadf sequence; gsadf_cv denotes 95% critical value sequence. 9 sequence. the graphical assessment in figures 3-8 shows the date-stamp for the beginning and end of each bubble. for the purpose of comparison, the original data series were plotted in the same figures and corresponds to the right vertical axis. figures 3-4 plot the corresponding gsadf statistics (the brown line) against the corresponding 95 percent threshold (the green line) for the dow jones and the s&p500 during the covid-19 pandemic. the date of beginning is identified as the first point when the gsadf statistics surpassed the critical value. the end date is equivalent to the point after the date of origin when the gsadf statistics dropped below the threshold value. these periods are represented by the shaded areas as identified by equation (8). figure 3 and figure 4 show evidence of two statistically significant bubbles, with the first bubble detected on the 21st of february and lasted around seven days. during this short span of time, both indices suffered a 12 percent drop in the index. the second successive bubble was detected on the 12th of march and lasted for two days, when both markets registered a 10 percent drop in the index. these results are in line with the findings in chang et al. (2021). these two consecutive short bubbles in these two us stock markets coincided with the declining global oil demand, as reflected by the persistent drop in crude oil prices as much as 22 percent from 20th february to 13th march 2020. this was largely due to the weakened global oil demand, as china shut its economy to curb the spread of the virus and many countries began to limit or suspend air travels to and from china. the explosive behavior episodes of the dow jones and the s&p500 were a manifestation of the adverse economic impact of the coronavirus outbreak and this was made worse after the russia–saudi arabia oil price war which started on the 8th of march 2020. figure 3 bubbles date stamp for dow jones index note. gsadf denotes gsadf sequence; gsadf_cv denotes 95% critical value sequence. 104 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 figure 4 bubbles date stamp for the s&p500 index note. gsadf denotes gsadf sequence; gsadf_cv denotes 95% critical value sequence. figure 5 and figure 6 show the bubble episodes for the stock market in malaysia, in particular, the klse and the shariah. following the earlier mentioned similar approach, the shaded areas corresponded to the starting and ending episodes of exuberance in stock prices, as identified by the gsadf using the 95 percent threshold values derived from the monte carlo simulations. figure 5 bubbles date stamp for the klse index note. gsadf denotes the gsadf sequence; gsadf_cv denotes the 95% critical value sequence. 10 figure 4 bubbles date stamp for the s&p500 index note. gsadf denotes gsadf sequence; gsadf_cv denotes 95% critical value sequence. figure 5 and figure 6 show the bubble episodes for the stock market in malaysia, in particular, the klse and the shariah. following the earlier mentioned similar approach, the shaded areas corresponded to the starting and ending episodes of exuberance in stock prices, as identified by the gsadf using the 95 percent threshold values derived from the monte carlo simulations. for the klse, only one episode of bubble was detected between march,13 and march 21, 2020. the inception date of the bubble for this period corresponded to the sudden spike in covid-19 cases in malaysia, from the daily confirmed cases of 9 on 13th march to 190 confirmed cases on 16th march 2020 (see figure 2). although the shariah index experienced an explosive episode around the same starting date as that of the klse, the gdasf test picked up two prior bubbles for the shariah as is shown in figure 6. there was a double sudden explosion in prices on 31st january and 24th february 2020. the implosion on 31st january was likely a response to the surge in the covid-19 confirmed cases in china by 10,000 percent between 20th january-31st january 2020 (see figure 1), while the explosive behavior on the 24th of february might be rooted in the continuous drop in oil prices after hovering around 53 us dollars for several weeks and has never recovered since. the second bubble on february 24 might also be due to the political upheaval in malaysia when the pakatan harapan coalition government collapsed on 24 february 2020 (ho wah foon), it created a power vacuum when the then prime minister, tun dr. mahathir mohamad resigned. the stock market slid down further and subsequently led to the third bubble in the shariah index following the sudden surge in covid-19 cases and the imposition of a nationwide partial lockdown by the malaysian government. figure 5 bubbles date stamp for the klse index 11 note. gsadf denotes the gsadf sequence; gsadf_cv denotes the 95% critical value sequence. figure 6 bubbles date stamp for the sharia gold index note. gsadf denotes the gsadf sequence; gsadf_cv denotes the 95% critical value sequence. figure 7 and figure 8 display the bubbles episodes for the crude oil market. they show the multiple explosive behavior episodes for the nymex and the spot crude prices between 18th february and 20th april 2020. both the oil price measures exhibited two bubble incidents on the 18th and 26th of february 2020 and these lasted for three days, respectively. the collapse of the oil price bubble in february was initiated by the abrupt decline in the world oil demand. this was a consequence of the widespread shutdown of china’s economy, which was hit hard by covid-19. according to the iea6, the coronavirus outbreak had caused the first quarterly contraction of oil demand by 435 kb/d in 10 105 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 for the klse, only one episode of bubble was detected between march,13 and march 21, 2020. the inception date of the bubble for this period corresponded to the sudden spike in covid-19 cases in malaysia, from the daily confirmed cases of 9 on 13th march to 190 confirmed cases on 16th march 2020 (see figure 2). although the shariah index experienced an explosive episode around the same starting date as that of the klse, the gdasf test picked up two prior bubbles for the shariah as is shown in figure 6. there was a double sudden explosion in prices on 31st january and 24th february 2020. the implosion on 31st january was likely a response to the surge in the covid-19 confirmed cases in china by 10,000 percent between 20th january-31st january 2020 (see figure 1), while the explosive behavior on the 24th of february might be rooted in the continuous drop in oil prices after hovering around 53 us dollars for several weeks and has never recovered since. the second bubble on february 24 might also be due to the political upheaval in malaysia when the pakatan harapan coalition government collapsed on 24 february 2020 (ho wah foon), it created a power vacuum when the then prime minister, tun dr. mahathir mohamad resigned. the stock market slid down further and subsequently led to the third bubble in the shariah index following the sudden surge in covid-19 cases and the imposition of a nationwide partial lockdown by the malaysian government. figure 6 bubbles date stamp for the sharia gold index note. gsadf denotes the gsadf sequence; gsadf_cv denotes the 95% critical value sequence. 11 note. gsadf denotes the gsadf sequence; gsadf_cv denotes the 95% critical value sequence. figure 6 bubbles date stamp for the sharia gold index note. gsadf denotes the gsadf sequence; gsadf_cv denotes the 95% critical value sequence. figure 7 and figure 8 display the bubbles episodes for the crude oil market. they show the multiple explosive behavior episodes for the nymex and the spot crude prices between 18th february and 20th april 2020. both the oil price measures exhibited two bubble incidents on the 18th and 26th of february 2020 and these lasted for three days, respectively. the collapse of the oil price bubble in february was initiated by the abrupt decline in the world oil demand. this was a consequence of the widespread shutdown of china’s economy, which was hit hard by covid-19. according to the iea6, the coronavirus outbreak had caused the first quarterly contraction of oil demand by 435 kb/d in 10 106 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 figure 7 and figure 8 display the bubbles episodes for the crude oil market. they show the multiple explosive behavior episodes for the nymex and the spot crude prices between 18th february and 20th april 2020. both the oil price measures exhibited two bubble incidents on the 18th and 26th of february 2020 and these lasted for three days, respectively. the collapse of the oil price bubble in february was initiated by the abrupt decline in the world oil demand. this was a consequence of the widespread shutdown of china’s economy, which was hit hard by covid-19. figure 7 bubbles date stamp for the nymex price note. gsadf denotes gsadf sequence; gsadf_cv denotes 95% critical value sequence. according to the iea6, the coronavirus outbreak had caused the first quarterly contraction of oil demand by 435 kb/d in 10 years. the oil market took another major blow in march when saudi arabia engaged in a price war with russia on march 8, 2020. a movement of panic ensued in the energy market, triggering yet another implosion of the oil price bubble, highlighted in the green-shaded area in figure 7 and figure 8. the gsadf test picked up another short-lived bubble for the nymex index on march 18, 2020 (purple-shaded area in figure 7), as crude oil price tumbled by another 23 percent during an overnight trade. as oil traders began to accommodate the free fall of the crude oil price over the next four weeks, in the midst of the ongoing price 12 years. the oil market took another major blow in march when saudi arabia engaged in a price war with russia on march 8, 2020. a movement of panic ensued in the energy market, triggering yet another implosion of the oil price bubble, highlighted in the green-shaded area in figure 7 and figure 8. the gsadf test picked up another short-lived bubble for the nymex index on march 18, 2020 (purple-shaded area in figure 7), as crude oil price tumbled by another 23 percent during an overnight trade. as oil traders began to accommodate the free fall of the crude oil price over the next four weeks, in the midst of the ongoing price war between russia and saudi arabia reached a stalemate, there was no episode of explosive behavior in the energy market until the 20th of april when oil price turned negative for the first time in history. both the nymex and the spot crude prices were traded at 37 us dollars a day before the may futures contracts expired on april 20th. the gsadf statistics for the nymex and the spot crude were above its critical values for one day on the 20th of april 2020, as is shown in the red-shaded region in figure 6 and figure 7, consistent with the monthlong crisis in the oil markets. overall, findings for the crude oil markets are consistent with the results in the study by gharib et al. (2021), as it also identified several explosive episodes in the west texas light (wti) oil price during the corresponding period. figure 7 bubbles date stamp for the nymex price note. gsadf denotes gsadf sequence; gsadf_cv denotes 95% critical value sequence. figure 8 bubbles date stamp for the spot crude oil price 107 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 war between russia and saudi arabia reached a stalemate, there was no episode of explosive behavior in the energy market until the 20th of april when oil price turned negative for the first time in history. both the nymex and the spot crude prices were traded at 37 us dollars a day before the may futures contracts expired on april 20th. the gsadf statistics for the nymex and the spot crude were above its critical values for one day on the 20th of april 2020, as is shown in the red-shaded region in figure 6 and figure 7, consistent with the monthlong crisis in the oil markets. overall, findings for the crude oil markets are consistent with the results in the study by gharib et al. (2021), as it also identified several explosive episodes in the west texas light (wti) oil price during the corresponding period. figure 8 bubbles date stamp for the spot crude oil price note. gsadf denotes gsadf sequence; gsadf_cv denotes 95% critical value sequence. conclusion the present study has focused on how the covid-19 pandemic has caused the asset bubbles in the stocks markets of the usa and malaysia, as well as in the global oil markets. it found that the increase 13 note. gsadf denotes gsadf sequence; gsadf_cv denotes 95% critical value sequence. conclusion the present study has focused on how the covid-19 pandemic has caused the asset bubbles in the stocks markets of the usa and malaysia, as well as in the global oil markets. it found that the increase in covid-19 cases has exacerbated the volatility in the financial markets, as well as in the oil market. the gsadf has provided a powerful explanatory story by documenting a series of bubble in the period between january 1, 2020, to april 24, 2020, for six equity series. the findings showed that the major bubble episodes in the us stock markets were precipitated by the slump in the oil market, and to a lesser extent was due to covid-19. this finding is consistent with the position of the usa as a major oil producer striding past saudi arabia and russia since 2017. the analysis carried out in this study revealed that the volatility of the us stock markets was more influenced by oil price movements rather than the shock created by covid-19. this is in line with the findings of gao et al. (2021). investors can utilize this information as a prediction and diagnostic tool for future stock prices, as well as an early warning system for economic instability, notably in the us stock market. in contrast, the stock market in malaysia exhibited explosive behaviour episodes that correlated strongly with covid-19 cases. this is not surprising because of the high degree of market integration between malaysia and its major trading partners, particularly china and the usa. there was also evidence of bilateral contagion effects of bubbles between the oil stock market and the shariah stock market in malaysia. multiple bubble episodes began simultaneously in the crude oil markets (nymex and spot crude) following the oil price slump in february 2020 and then spread to the shariah market in malaysia in march 2020. the explosive behaviour of the stock markets in malaysia is also explained by the political instability as a result of the collapse of the pakatan harapan government in march 2020. these findings have important implications for government leaders, the 108 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 in covid-19 cases has exacerbated the volatility in the financial markets, as well as in the oil market. the gsadf has provided a powerful explanatory story by documenting a series of bubble in the period between january 1, 2020, to april 24, 2020, for six equity series. the findings showed that the major bubble episodes in the us stock markets were precipitated by the slump in the oil market, and to a lesser extent was due to covid-19. this finding is consistent with the position of the usa as a major oil producer striding past saudi arabia and russia since 2017. the analysis carried out in this study revealed that the volatility of the us stock markets was more influenced by oil price movements rather than the shock created by covid-19. this is in line with the findings of gao et al. (2021). investors can utilize this information as a prediction and diagnostic tool for future stock prices, as well as an early warning system for economic instability, notably in the us stock market. in contrast, the stock market in malaysia exhibited explosive behaviour episodes that correlated strongly with covid-19 cases. this is not surprising because of the high degree of market integration between malaysia and its major trading partners, particularly china and the usa. there was also evidence of bilateral contagion effects of bubbles between the oil stock market and the shariah stock market in malaysia. multiple bubble episodes began simultaneously in the crude oil markets (nymex and spot crude) following the oil price slump in february 2020 and then spread to the shariah market in malaysia in march 2020. the explosive behaviour of the stock markets in malaysia is also explained by the political instability as a result of the collapse of the pakatan harapan government in march 2020. these findings have important implications for government leaders, the single most important was the need to create a stable climate for the financial system so that all other sectors of the economy could thrive, especially during the pandemic. in comparing the conventional (klse) market and the islamic (shariah) market, it was found that there was a single bubble episode in the klse index, but multiple bubbles in the shariah index. in particular, the gsadf test was able to pick up two bubble episodes earlier in the shariah index, both of which coincided with the initial implosion of the covid-19 outbreak in china, and the impending collapse of the crude oil price. the noteworthy differences of bubble episodes in malaysia has revealed the sensitivity of the shariah 109 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 index to market shocks when the asset prices were heated. the early detection of a bubble episode due to the covid-19 pandemic from the shariah index could help protect investors and fund managers in both markets (conventional and islamic) from impending market crashes. this may be attributed to the rigorous screening of shariah compliant companies included in the islamic index, which has excluded gambling and riba (unproportionate interest). furthermore, the strict benchmarking criteria, namely the nature of asset (jakhura & mangera, 2010) and the low debt-to-equity ratio (zandi et al., 2014) for shariah-compliant companies would ensure that investors’ rights were protected. consequently, this will make investment in islamic financial markets a viable feature for diversification. however, since the present findings have been limited to the market in malaysia, it is suggested that further investigation be carried out on the shariah index of other countries to confirm the current results. acknowledgment this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. endnotes 1 oil market report. (april 2020). by the international energy agency.https://www.iea.org/reports/oil-market-reportapril-2020. 2 jacobs, trent. opec+ moves to end price war with 10 million b/d cut. pubs.spe.org. journal of petroleum technology. archived from the original on 10 april 2020. 3 retrieved from https://www.oecd.org/coronavirus/policyresponses/global-financial-markets-policy-responses-to-covid19-2d98c7e0/ 4 (24 february 2020). the dow jones industrial average and ftse 100 dropped more than 3%. retrieved from “global stock markets plunge on coronavirus fears”. bbc news. 24 february 2020. 5 the klse index is derived based on shariah and non-shariah (conventional) compliant stocks; the shariah index is calculated 110 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 solely on stocks which are shariah compliant. information on the listing criteria can be found in https://www.bursamalaysia. com. 6 retrieved from https://www.iea.org/reports/oil-market-reportfebruary-2020 references al-awadhi, a. m., al-saifi, k., al-awadhi, a., & alhamadi, s. (2020). death and contagious infectious diseases: impact of the covid-19 virus on stock market returns. journal of behavioral and experimental finance, 100326–100326. pubmed. https:// doi.org/10.1016/j.jbef.2020.100326 alp, m. h., elekdag, s., & lall, m. s. (2012). an assessment of malaysian monetary policy during the global financial crisis of 2008-09. international monetary fund. arouri, m. e. h., foulquier, p., & fouquau, j. (2011). oil prices and stock markets in europe: a sector perspective. recherches économiques de louvain, 77(1), 5. https://doi.org/10.3917/ rel.771.0005 bahmani-oskooee, m., ghodsi, s. h., & hadzic, m. (2019). asymmetric causality between oil price and stock returns: a sectoral analysis. economic analysis and policy, 63, 165–174. https://doi.org/10.1016/j.eap.2019.06.002 baker, s., bloom, n., davis, s., kost, k., sammon, m., & viratyosin, t. (2020). the unprecedented stock market impact of covid-19. national bureau of economic research. https://doi. org/10.3386/w26945 baldwin, r., & di mauro, b. w. (2020). economics in the time of covid-19. a voxeu. org book, centre for economic policy research, london. accessed, 26. chang, t., hsu, c. m., & wang, m. c. (2021). bubbles during covid-19 period: evidence from the united states using the generalized sub adf test. holistica–journal of business and public administration, 12(1), 49-56. chen, c. d., chen, c. c., tang, w. w., & huang, b. y. (2009). the positive and negative impacts of the sars outbreak: a case of the taiwan industries. the journal of developing areas, 281-293. de long, j. b., shleifer, a., summers, l. h., & waldmann, r. j. (1990). positive feedback investment strategies and destabilizing 111 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 rational speculation. the journal of finance, 45(2), 379–395. https://doi.org/10.1111/j.1540-6261.1990.tb03695.x degiannakis, s., filis, g., & arora, v. (2018). oil prices and stock markets: a review of the theory and empirical evidence. the energy journal, 39(01). https://doi. org/10.5547/01956574.39.5.sdeg gao, x., ren, y., & umar, m. (2021). to what extent does covid-19 drive stock market volatility? a comparison between the us and china. economic research-ekonomska istraživanja, 1-21. gharib, c., mefteh-wali, s., & jabeur, s. b. (2021). the bubble contagion effect of covid-19 outbreak: evidence from crude oil and gold markets. finance research letters, 38, 101703. giglio, s., maggiori, m., stroebel, j., & utkus, s. (2020). inside the mind of a stock market crash. arxiv preprint arxiv:2004.01831. giglio, s., maggiori, m., stroebel, j., & utkus, s. (2021). five facts about beliefs and portfolios. american economic review, 111(5), 1481-1522. gilbert, t. (2011). information aggregation around macroeconomic announcements: revisions matter. journal of financial economics, 101(1), 114–131. https://doi.org/10.1016/j. jfineco.2011.02.013 gomes, m., & chaibi, a. (2014). volatility spillovers between oil prices and stock returns: a focus on frontier markets. journal of applied business research (jabr), 30(2), 509. https://doi. org/10.19030/jabr.v30i2.8421 gormsen, n. j., & koijen, r. s. j. (2020). coronavirus: impact on stock prices and growth expectations. ssrn electronic journal. https://doi.org/10.2139/ssrn.3555917 gürkaynak, r. s. (2008). econometric tests of asset price bubbles: taking stock. journal of economic surveys, 22(1), 166–186. https://doi.org/10.1111/j.1467-6419.2007.00530.x helali, s. m. (2019). detecting and date-stamping rational bubbles in asset price: an empirical investigation in the tunisian stock market. international journal of economics and finance, 11(8), 91. https://doi.org/10.5539/ijef.v11n8p91 henriques, i., & sadorsky, p. (2008). oil prices and the stock prices of alternative energy companies. energy economics, 30(3), 998– 1010. https://doi.org/10.1016/j.eneco.2007.11.001 ho wah foon. (2020). malaysia’s dr mahathir quits as premier. the star. https://www.thestar.com.my/news/regional/2020/02/24/ malaysias-dr-mahathir-quits-as-premier 112 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 homm, u., & breitung, j. (2011). testing for speculative bubbles in stock markets: a comparison of alternative methods. journal of financial econometrics, 10(1), 198–231. https://doi. org/10.1093/jjfinec/nbr009 jakhura, s., & mangera, m. d. (2010). share trading. december 26, (online) http://www.albalagh.net/qa/0143.shtml (accessed on 23 august, 2020). kilian, l., & park, c. (2009). the impact of oil price shocks on the u.s. stock market. international economic review, 50(4), 1267–1287. https://doi.org/10.1111/j.1468-2354.2009.00568.x kumar, s., managi, s., & matsuda, a. (2012). stock prices of clean energy firms, oil and carbon markets: a vector autoregressive analysis. energy economics, 34(1), 215–226. https://doi. org/10.1016/j.eneco.2011.03.002 loh, e. (2006). the impact of sars on the performance and risk profile of airline stocks. the impact of sars on the performance and risk profile of airline stocks, 1000-1022. liu, h., manzoor, a., wang, c., zhang, l., & manzoor, z. (2020). the covid-19 outbreak and affected countries stock markets response. international journal of environmental research and public health, 17(8), 2800. https://doi.org/10.3390/ ijerph17082800 lv, x., lien, d., & yu, c. (2020). who affects who? oil price against the stock return of oil-related companies: evidence from the u.s. and china. international review of economics & finance, 67, 85–100. https://doi.org/10.1016/j.iref.2020.01.002 masson, r., & winter, j. (2020). energy and environmental policy trends: addressing the threat of covid-19 and the oil price war in the petroleum sector. the school of public policy publications, 13. moya-martínez, p., ferrer-lapeña, r., & escribano-sotos, f. (2014). oil price risk in the spanish stock market: an industry perspective. economic modelling, 37, 280–290. https://doi. org/10.1016/j.econmod.2013.11.014 mylenka, t., & novyk, b. (2020). impact of covid-19 on the global energy sector. https://www.pv-magazine.com/2020/04/24/ impact-of-covid-19-on-the-global-energy-sector/ mzoughi, h., urom, c., uddin, g. s., & guesmi, k. (2020). the effects of covid-19 pandemic on oil prices, co2 emissions and the stock market: evidence from a var model. ssrn electronic journal. https://doi.org/10.2139/ssrn.3587906 onali, e. (2020). covid-19 and stock market volatility. ssrn electronic journal. https://doi.org/10.2139/ssrn.3571453 113 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 ozili, p. k., & arun, t. (2020). spillover of covid-19: impact on the global economy. ssrn electronic journal. https://doi. org/10.2139/ssrn.3562570 phan, d. h. b., sharma, s. s., & narayan, p. k. (2015). oil price and stock returns of consumers and producers of crude oil. journal of international financial markets, institutions and money, 34, 245–262. https://doi.org/10.1016/j.intfin.2014.11.010 phillips, p. c. b., & yu, j. (2011). dating the timeline of financial bubbles during the subprime crisis. quantitative economics, 2(3), 455–491. https://doi.org/10.3982/qe82 phillips, p. c. b., shi, s., & yu, j. (2015). testing for multiple bubbles: limit theory of real-time detectors. international economic review, 56(4), 1079–1134. https://doi.org/10.1111/iere.12131 rizvi, s. a. r., arshad, s., & alam, n. (2015). crises and contagion in asia pacific—islamic v/s conventional markets. pacific-basin finance journal, 34, 315-326. https://doi.org/10.1016/j.pacfin.2015.04.002 sadorsky, p. (2008). assessing the impact of oil prices on firms of different sizes: its tough being in the middle. energy p o l i c y , 3 6 ( 1 0 ) , 3 8 5 4 – 3 8 6 1 . h t t p s : / / d o i . o r g / 1 0 . 1 0 1 6 / j . enpol.2008.07.019 sharif, a., aloui, c., & yarovaya, l. (2020). covid-19 pandemic, oil prices, stock market, geopolitical risk and policy uncertainty nexus in the us economy: fresh evidence from the waveletbased approach. international review of financial analysis, 101496. https://doi.org/10.1016/j.irfa.2020.101496 sharma, s., & escobari, d. (2018). identifying price bubble periods in the energy sector. energy economics, 69, 418–429. https://doi. org/10.1016/j.eneco.2017.12.007 surico, p., & galeotti, a. (2020). the economics of a pandemic: the case of covid-19. wheeler institute for business and development, lbs. london: london business school. tagliapietra, s. (2020). covid-19 is causing the collapse of oil markets: when will they recover? https://www.bruegel. org/2020/04/covid-19-is-causing-the-collapse-of-oilmarkets-when-will-they-recover/ tchatoka, f. d., masson, v., & parry, s. (2019). linkages between oil price shocks and stock returns revisited. energy economics, 82, 42–61. https://doi.org/10.1016/j.eneco.2018.02.016 tetlock, p. c. (2010). does public financial news resolve asymmetric information? review of financial studies, 23(9), 3520–3557. https://doi.org/10.1093/rfs/hhq052 114 the international journal of banking and finance, vol. 17, number 2 (july) 2022, pp: 91–114 tirole, j. (1985). asset bubbles and overlapping generations. econometrica, 53(6), 1499. https://doi.org/10.2307/1913232 waheed, r., wei, c., sarwar, s., & lv, y. (2017). impact of oil prices on firm stock return: industry-wise analysis. empirical economics, 55(2), 765–780. https://doi.org/10.1007/s00181017-1296-4 wen, f., xu, l., ouyang, g., & kou, g. (2019). retail investor attention and stock price crash risk: evidence from china. international review of financial analysis, 65, 101376. wen, f., xu, l., chen, b., xia, x., & li, j. (2020). heterogeneous institutional investors, short selling and stock price crash risk: evidence from china. emerging markets finance and trade, 56(12), 2812–2825. https://doi.org/10.1080/154049 6x.2018.1522588 ye, z., & florescu, i. (2020). covid-19 and the equity market. a march 2020 tale. ssrn electronic journal. https://doi.org/10.2139/ ssrn.3566281 youssef, m., mokni, k., & ajmi, a. n. (2021). dynamic connectedness between stock markets in the presence of the covid-19 pandemic: does economic policy uncertainty matter? financial innovation, 7(1), 1-27. yilmazkuday, h. (2020). covid-19 effects on the s&p 500 index. ssrn electronic journal. https://doi.org/10.2139/ssrn.3555433 zandi, g., razak, d. a., & hussin, n. h. (2014). stock market screening: an analogical study on conventional and shariahcompliant stock markets. asian social science, 10(22), 270. zeren, f., & hizarci, a. (2020). the impact of covid-19 coronavirus on stock markets: evidence from selected countries. muhasebe ve finans i̇ncelemeleri dergisi. https://doi.org/10.32951/ mufider.706159 zhang, s. x., huang, h., & wei, f. (2020). geographical distance to the epicenter of covid-19 predicts the burnout of the working population: ripple effect or typhoon eye effect? psychiatry research, 288, 112998. https://doi.org/10.1016/j. psychres.2020.112998 zhao, z., wen, h., & li, k. (2020). identifying bubbles and the contagion effect between oil and stock markets: new evidence from china. economic modelling. https://doi.org/10.1016/j. econmod.2020.02.018